+1 (208) 254-6996 [email protected]
  

Please Review the items in the Quiz 4 folder and Gruber Chapter 10 & 11, incorporating these to answer the four questions below:

“The argument that a voucher system will increase the social efficiency of primary schooling is weaken if there is a high degree of Tiebout sorting on the basis of  income status and race.”  Agree or Disagree with statement above, given that there are significant positive externalities associated with primary education and local property taxes are a major source of primary school funding in the United States.

Don't use plagiarized sources. Get Your Custom Essay on
Discussion Quiz
Just from $13/Page
Order Essay

As described in the text, Fischel (1989) argued that California’s Serrano v. Priest school finance equalization induced voters to limit property taxes in California. Following this argument, would an alternative school finance equalization that produced increased spending for low-wealth communities using state funds be more, less, or equally likely to induce a property tax limitation in California? Explain.

Express your own view, choosing to be in favor, or against a progressive voucher program (see Robert Reich).  What problems would this proposal present from a public choice perspective?  And do you believe that Moving to Opportunity Programs offer a better solution to the problem of  quality primary schooling for low-income? 

Post one thoughtful reply to another student concerning their answer #3 above.  Try to pick a student who has no reply.

1. “The argument that a voucher system will increase the social efficiency of primary schooling is weaken if there is a high degree of Tiebout sorting on the basis of  income status and race.”  Agree or Disagree with statement above, given that there are significant positive externalities associated with primary education and local property taxes are a major source of primary school funding in the United States.

2. As described in the text, Fischel (1989) argued that California’s Serrano v. Priest school finance equalization induced voters to limit property taxes in California. Following this argument, would an alternative school finance equalization that produced increased spending for low-wealth communities using state funds be more, less, or equally likely to induce a property tax limitation in California? Explain.

3. Express your own view, choosing to be in favor, or against a progressive voucher program (see Robert Reich).  What problems would this proposal present from a public choice perspective.  And do you believe that Moving to Opportunity Programs offer a better solution to the problem of  quality primary schooling for low-income?

4. Post one thoughtful reply to another student concerning their answer #3 above.  Try to pick a student who has no reply.

Discussion Prep

For Quiz 4

Cost – Benefit Aspect

Educational Outcomes are correlated with neighborhood locations

Raj Chetty demonstrates greater educational benefits for the same $45 billion spent (costs) on housing location of low-income families

11.4

Better-educated people…

Participate more politically

Perform fewer criminal acts

Have better health and healthier children

Have better educated children

Have more productive coworkers

See Enrico Morretti in Quiz 4 Folder

Effect of Education Levels on Other Outcomes (Positive Externalities/Benefits of

Educational Achievement

3

10

Fiscal Federalism

The United States has a federal system, dividing activity between a national government and state and local governments.

Education, for example, is often provided by state governments.

Optimal fiscal federalism: The question of which activities should take place at which level of government.

5

10.1

The type of spending by state and local governments differ from that of the U.S. federal government.

State and local governments spend the majority of revenue on education, followed by health care and public safety.

The federal government spends the majority of revenue on health care, social security, and national defense.

Spending and Revenue of State and Local Governments

6

10.1

State and local governments rely on multiple sources of revenues.

State governments use sales and income taxes primarily.

Local governments use property taxes heavily. They comprise about half of local government revenue.

Property tax: The tax on land and any buildings on it, such as commercial businesses or residential homes.

Spending and Revenue of State and Local Governments

7

10.2

People not only vote with their feet, they also vote with their pocketbook, in the form of house prices.

House price capitalization: Incorporation into the price of a house the costs (including local property taxes) and benefits (including local public goods) of living in the house.

Areas with relatively generous public goods (given taxes) should have higher house prices.

Evidence on the Tiebout Model: Capitalization of Fiscal Differences into House Prices

8

10.2

Tiebout model predicts that local spending should focus on broad-based programs with few externalities and relatively low economies of scale. Local communities should play a limited role in providing public goods that are redistributive, have large spillovers, and have large economies of scale.

If taxes and benefits are linked, and there are no spillovers or economies of scale, then local public good provision is close to optimal.

Optimal Fiscal Federalism

9

10.3

If residents perceived that property taxes were “too high” in California, why did they wait until 1978 to lower them?

Proposition 13 was actually a response to school finance equalization in California.

Taxes no longer financed local school spending; just taxes, rather than prices. Tax price became infinite.

Voters were happy to limit property taxes once those taxes no longer brought them any benefit.

APPLICATION: School Finance Equalization and Property Tax Limitations in California (Serrano v Priest, Gruber page 299

10.3

School finance equalization: Laws that mandate redistribution of funds across communities in a state to ensure more equal financing of schools.

Generally, studies conclude that spending equalization has led to an equalization in student outcomes.

Finance equalization schemes differ across states:

California redistributes effectively all revenues.

New Jersey redistributes most revenue from towns with revenue above the 85th percentile.

Redistribution in Action: School Finance Equalization

10.3

Different structures result in different tax prices.

Tax price: For school equalization schemes, the amount of revenue a local district would have to raise in order to gain $1 more of spending.

If half of revenue is redistributed, tax price is $2.

If all revenue is redistributed, tax price is infinite.

Evidence suggests that extreme equalization schemes with very high tax prices may lead to an overall reduction in per-pupil spending. In California Proposition 30, approved by voters in 2012, provided critical revenues to California at a time when the state faced daunting fiscal challenges. These revenues increased school funding and allowed for reinvestment in other public services after years of cuts (with sales tax and income tax increase)

Redistribution in Action: School Finance Equalization

See article on California’s Education Financing

In the Quiz 4 Folder: “Financing Education in California”

Robert Reich proposal for a Progressive Public School/Public Charter Schools Voucher Program is proposed as a remedy to the “cherry picking” problem that exists with standard vouchers for financing public schools. Does it run into scale problems? Equity issues? Public Choice issues?

See “Robert Reich on School Choice” in the Quiz 4 Folder

11.2

Educational vouchers: A fixed amount of money given by the government to families with school-age children, who can spend it at any type of school, public or private.

Vouchers put private schools, non-profit and public schools on equal footing.

Vouchers

14

11.2

Voucher proponents make two arguments for them:

Consumer sovereignty

Vouchers allow individuals to more closely match their educational choices with their tastes.

Competition

Vouchers allow the education market to benefit from the competitive pressures that make private markets function efficiently.

Vouchers

15

11.2

Critics make several arguments against vouchers.

Vouchers may lead to excessive school specialization.

By focusing on particular market segments, schools give less focus to the key elements of education.

Vouchers will lead to segregation.

Critics of voucher systems argue that vouchers have the potential to reintroduce segregation along many dimensions, such as race, income, or child ability.

Problems with Educational Vouchers

16

11.2

Vouchers are an inefficient and inequitable use of public resources.

With vouchers, total public-sector costs would rise, as the government would pay part of the private school costs that families currently pay.

The education market may not be competitive.

The education market is described more closely by a model of natural monopoly, with efficiency gains to having only one monopoly provider of the good.

Problems with Educational Vouchers

17

11.3

Evidence is mixed.

Generally suggests that vouchers improve educational outcomes.

They come at the cost of potentially increasing inequality in educational achievement.

Some sort of guarantee of access must be provided to ensure all students have an education.

Bottom Line on Vouchers and School Choice

18

Quiz #4

ECON 3314

Cost-Benefit Revolution

See Cass Sunstein and Gruber Chapter 8

This is a Hicks-Kaldor definition of efficiency, Not Pareto Efficiency

Benefit Cost formula

3 Steps to Benefit Cost Analysis

Identify the impacts

Classify the impacts as Benefits, Costs or Transfer

Evaluate the impacts

Gruber Chapter 10 & 11 Reading Assigned

Chapter 10 pp. 281-291 and pp. 296-300.

Chapter 11 pp. 305-310 and pp. 314-323

Tiebout Sorting/Vouchers

Question: Tiebout, see Gruber Ch. 10 Voucher Debate in Gruber Ch. 11

“The argument that a voucher system will increase the social efficiency of primary schooling is weaken if there is a high degree of Tiebout sorting on the basis of race and income status.”

Agree or Disagree with statement above, given that there are significant positive externalities associated with primary education and local property taxes are the main source of primary school funding in the United States.

California’s Serrano v. Priest

See Application Box on page 299 of Gruber pdf for Chapters 10 & 11

Copyright © 2010 Worth Publishers

‹#› of 41

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

1

10

Heather Luea and Dan Sacks

P R E P A R E D B Y

10.1 Fiscal Federalism in the United States and Abroad

10.2 Optimal Fiscal Federalism

10.3 Redistribution Across Communities

10.4 Conclusion

State and Local Government Expenditures

‹#› of 41

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

2

10

Fiscal Federalism: No Child Left Behind Act

In 2002, President Bush signed into law the No Child Left Behind (NCLB) Act.

The goal of NCLB was to address the poor educational opportunities for low income and minority students by requiring standardized testing.

NCLB required schools to publish scores, and penalties were to be imposed on schools that did not show student progress.

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

3

10

Fiscal Federalism: No Child Left Behind Act

Intense controversy existed during the first few years between states and the federal government.

Proponents of the Act felt states had ignored educational quality for low income and minority students while opponents felt the states knew best how to address any educational deficiencies in its student population.

In 2015, the House passed a rewrite which reduces the federal role in supervising education. The bill prohibits federal government from mandating or encouraging specific curricula among a host of other revisions.

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

4

10

Fiscal Federalism

The United States has a federal system, dividing activity between a national government and state and local governments.

Education, for example, is often provided by state governments.

Optimal fiscal federalism: The question of which activities should take place at which level of government.

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

5

10.1

The distribution of government spending has changed dramatically over time in the United States.

1902: Federal government accounted for about 34% of total government spending.

2012: Federal government accounted for about 65% of total government spending.

Local and state spending have declined considerably.

Much state and local spending is now supported by intergovernmental grants.

Intergovernmental grants: Payments from one level of government to another.

Fiscal Federalism in the United States and Abroad

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

6

10.1

State and Local Spending in the United States, 1902−2012

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

7

10.1

Three primary factors are behind the change in the composition of government spending.

Sixteenth amendment which allowed the federal government to levy income taxes on citizens

New Deal programs of the 1930s in response to the Great Depression

Introduction of social insurance and welfare programs

Fiscal Federalism in the United States and Abroad

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

8

10.1

The type of spending by state and local governments differ from that of the U.S. federal government.

State and local governments spend the majority of revenue on education, followed by health care and public safety.

The federal government spends the majority of revenue on health care, social security, and national defense.

Spending and Revenue of State and Local Governments

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

9

10.1

State and local governments rely on multiple sources of revenues.

State governments use sales and income taxes primarily.

Local governments use property taxes heavily. They comprise about half of local government revenue.

Property tax: The tax on land and any buildings on it, such as commercial businesses or residential homes.

Spending and Revenue of State and Local Governments

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

10

10.1

Spending and Revenue of State and Local Governments

SpendingRevenue
State$/PCState$/PC
Education spendingAK4,672Income taxesNY2,431
MA2,848MT854
TN1,993Many0
Health care spendingDC10,349Sales taxesDC1,904
LA6,759Iowa698
UT5,031Many0

Expenditures and revenues vary greatly across states.

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

11

10.1

Fiscal Federalism Abroad

Spending (% of all)Revenue (% of all)
Greece0.00.8
Portugal13.75.5
France20.312.1
Norway33.511.9
United States47.036.8
Denmark63.324.7
OECD Average30.827.2

Compared to subnational governments of other nations, U.S. state and local governments account for a relatively large portion of total government activity.

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

12

10.1

Fiscal Federalism Abroad

Higher levels of centralization exist in many countries because subnational governments have no power to tax citizens.

Many countries engage in fiscal equalization.

Fiscal equalization: Policies by which the national government distributes grants to subnational governments in an effort to equalize differences in wealth.

In many other countries, the central government redistributes a much larger share of revenues to subnational governments.

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

13

10.2

What determines how much and how efficiently local governments provide public goods?

The private market provides the optimal amount of private goods.

Why does the market do so well for private goods but not public goods?

Tiebout’s insight: shopping and competition are missing from the market for pubic goods.

Optimal Fiscal Federalism: The Tiebout Model

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

14

10.2

There is neither shopping nor competition for public goods provided by the federal government.

But, when public goods are provided at the local level, competition arises.

Individuals can “vote with their feet.”

This threat of exit can induce efficiency in local public goods production.

Under certain conditions, public goods provision at the local level will be fully efficient.

The Tiebout Model: Shopping and Competition

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

15

10.2

Competition across towns can lead to the optimal provision of public goods.

Towns determine public good levels and tax rates.

People move freely across towns, picking their preferred locality.

People with similar tastes end up together, paying the same amount in taxes and receiving the same public goods.

There is no free riding because everyone pays the same amount in each town.

Optimal Fiscal Federalism: The Formal Model

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

16

10.2

The Tiebout model requires a number of assumptions that may not hold in reality.

People are perfectly mobile.

People have full information on taxes and benefits.

People must be able to choose among a range of towns that might match their taste for public goods.

The provision of some public goods requires sufficient scale or size.

There must be enough towns so that individuals can sort themselves into groups with similar preferences for public goods.

Problems with Tiebout Competition

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

17

10.2

The Tiebout model requires equal financing of the public good among all residents.

Lump-sum tax: A fixed taxation amount independent of a person’s income, consumption of goods and services, or wealth.

Lump-sum taxes are often infeasible/unfair, so taxes are income or wealth based.

But then the rich pay more than the poor, so the poor chase the rich.

Problems with Tiebout Financing

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

18

10.2

To keep poor people from chasing rich people, towns enact zoning.

Zoning: Restrictions that towns place on the use of real estate.

Zoning regulation establishes, for example, minimum lot sizes.

Zoning regulations protect the tax base of wealthy towns by pricing lower-income people out of the housing market.

Problems with Tiebout Financing

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

19

10.2

The Tiebout model assumes that public goods have effects only in a given town and that the effects do not spill over to neighboring towns.

Many local public goods have similar externality or spillover features: police, public works, education.

If there are spillovers, then low-tax, low-benefit municipalities can free-ride off of high-tax, high-benefit ones.

Problems with the Tiebout Model:

No Externalities/Spillovers

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

20

10.2

Tiebout competition works through sorting.

A testable implication: When people have more choice of local community, the tastes for public goods will be more similar among town residents than when people do not have many choices.

Comparing larger and smaller metropolitan areas (with more and less choice), this seems to be true.

Evidence on the Tiebout Model:

Resident Similarity Across Areas

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

21

10.2

People not only vote with their feet, they also vote with their pocketbook, in the form of house prices.

House price capitalization: Incorporation into the price of a house the costs (including local property taxes) and benefits (including local public goods) of living in the house.

Areas with relatively generous public goods (given taxes) should have higher house prices.

Evidence on the Tiebout Model: Capitalization of Fiscal Differences into House Prices

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

22

Evidence on the Tiebout Model:

California’s Proposition 13

10.2

California’s Proposition 13 became law in 1978.

Set the maximum amount of any tax on property at 1% of the “full cash value.”

Full cash value: Value as of 1976, with annual increases of 2% at most.

Reduced property taxes immensely in some areas, little change in others.

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

10.2

Each $1 of property tax reduction increased house values by about $7, about equal to the PDV of a permanent $1 tax cut.

In principle, the fall in property taxes would result in a future reduction in public goods and services, which would lower home values. This occurred in San Jose where the public school system declared bankruptcy.

The fact that house prices rose by almost the present discounted value of the taxes suggests that Californians did not think that they would lose many valuable public goods and services when taxes fell. This was the case in areas such as San Francisco.

Evidence on the Tiebout Model:

California’s Proposition 13

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

24

10.2

Tiebout model implies that three factors should determine local public good provision:

Tax-benefit linkages: Goods with strong tax-benefit linkages should be provided locally.

Cross-municipality spillovers: If local public goods have large spillover effects on other communities, the goods will be underprovided by any locality.

Economies of scale: Public goods with large economies of scale are not efficiently provided by many competing local jurisdictions.

Optimal Fiscal Federalism

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

25

10.2

Tiebout model predicts that local spending should focus on broad-based programs with few externalities and relatively low economies of scale. Local communities should play a limited role in providing public goods that are redistributive, have large spillovers, and have large economies of scale.

If taxes and benefits are linked, and there are no spillovers or economies of scale, then local public good provision is close to optimal.

Optimal Fiscal Federalism

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

26

10.3

Enormous inequality in revenue across municipalities:

Carlisle, MA raises $22,472/student while Lakeville, MA raises $13,932.

Should we care about the inequality?

If Tiebout is right, then this reflects optimal sorting and financing. If a town has low revenues or low spending, it is because residents chose to provide low level of public goods. This is efficient given their tastes and redistribution should not occur.

Redistribution Across Communities

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

27

10.3

Should we care about the inequality?

If Tiebout does not perfectly reflect reality, redistribution from high-revenue, high-spending communities to low-revenue, low-spending communities is supported for two reasons:

People may not be able to “vote with their feet.”

Externalities may be present.

Redistribution Across Communities

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

28

10.3

The main tool of redistribution is intergovernmental grants—cash transfers from one level of government to another.

Grants are a large and growing share of federal spending and come in multiple forms, with different implications.

Matching grant: A grant, the amount of which is tied to the amount of spending by the local community.

Block grant: A grant of some fixed amount with no mandate on how it is to be spent.

Conditional block grant: A grant of some fixed amount with a mandate that the money be spent in a particular way.

Tools of Redistribution: Grants

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

29

Private goods spending (thousands)

Education spending (thousands)

0

Tools of Redistribution: Grants

10.3

$1,000

500

500

$1,000

A

B

X

IC1

Consider a community’s budget constraint AB and spending at point X.

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

$2,000

Private goods spending (thousands)

Education spending (thousands)

0

$1,000

500

500

1,000

A

B

X

IC1

IC2

Y

750

625

C

Matching Grants

10.3

A matching grant reduces the cost of education by half. The budget constraint pivots from AB to AC and increases spending to point Y. Both income and substitution effects occur.

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

31

E

1,375

D

$2,000

Private goods spending (thousands)

Education spending (thousands)

0

1,000

500

500

1,000

A

B

X

IC1

IC3

Y

750

625

C

Z

575

800

$1,375

Income effect

Substitution effect

Block Grant

10.3

A block grant shifts budget constraint from AB to DE and increases spending to point Z. Only income effect occurs.

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

$2,000

Private goods spending (thousands)

Education spending (thousands)

0

1,000

500

500

1,000

A

B

X

IC1

IC3

Y

750

625

C

Z

E

1,375

575

800

$1,375

D

Income effect

Substitution effect

10.3

F

375

Conditional Block Grant

A conditional block grant shifts budget constraint from AB to AFE and increases spending to point Z. Effect of conditional grant depends on the amount the town would have spent without the condition.

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

10.3

Implications of Different Grant Types

Different grant types affect incentives in different ways.

Matching grants rotate out the budget constraint, acting like a subsidy.

Help with externalities, since they are targeted.

Block grants shift out the entire budget constraint, raising spending on all goods.

Good for redistribution.

Conditional block grants only differ from block grants if the amount of the grant is greater than the initial educational spending.

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

10.3

School finance equalization: Laws that mandate redistribution of funds across communities in a state to ensure more equal financing of schools.

Generally, studies conclude that spending equalization has led to an equalization in student outcomes.

Finance equalization schemes differ across states:

California redistributes effectively all revenues.

New Jersey redistributes most revenue from towns with revenue above the 85th percentile.

Redistribution in Action: School Finance Equalization

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

10.3

Different structures result in different tax prices.

Tax price: For school equalization schemes, the amount of revenue a local district would have to raise in order to gain $1 more of spending.

If half of revenue is redistributed, tax price is $2.

If all revenue is redistributed, tax price is infinite.

Evidence suggests that extreme equalization schemes with very high tax prices may lead to an overall reduction in per-pupil spending.

Redistribution in Action: School Finance Equalization

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

10.3

Theory implies that conditional grants crowd-out local spending one-for-one. Do they?

When examining how states spend grant money, the flypaper effect seems to matter: “The money sticks where it hits.”

However, the positive correlation between grants and spending may be because states that get grants are the ones that like spending the most.

Redistribution in Action: The Flypaper Effect

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

Redistribution in Action: The Flypaper Effect

10.3

Knight attempted to measure the importance of the flypaper effect.

Looked at how spending changes as states’ congressional delegations gain or lose power.

Each additional $1 of federal grant money increase due to rising congressional power leads to a $0.90 reduction in the state’s own spending.

Additional studies also find evidence inconsistent with the flypaper effect.

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

10.3

If residents perceived that property taxes were “too high” in California, why did they wait until 1978 to lower them?

Proposition 13 was actually a response to school finance equalization in California.

Taxes no longer financed local school spending; just taxes, rather than prices. Tax price became infinite.

Voters were happy to limit property taxes once those taxes no longer brought them any benefit.

APPLICATION: School Finance Equalization and Property Tax Limitations in California

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

10.4

Central governments collect only part of total tax revenues and spend only part of total public spending.

The United States places a large share of governmental responsibilities on its subnational governments relative to other developed countries.

The Tiebout model suggests that the spending should be done locally when:

Spending is on goods for which local preferences are relatively similar.

Most residents can benefit from those goods.

Conclusion

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

10.4

Conclusion

Higher levels of government may not believe the conclusions of the idealized Tiebout model.

They will want to redistribute across lower levels of government.

If the higher-level government decides that it wants to redistribute across lower levels, it can do so through several different types of grants.

Appropriate choice of grants depends on goal of government financing.

‹#› of 41

C H A P T E R 1 0 ■ S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

Copyright © 2010 Worth Publishers

‹#› of 34

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

1

11

Heather Luea and Dan Sacks

P R E P A R E D B Y

11.1 Why Should the Government Be Involved in Education?

11.2 How Is the Government Involved in Education?

11.3 Evidence on Competition in Education Markets

11.4 Measuring the Returns to Education

11.5 The Role of the Government in Higher Education

11.6 Conclusion

Education

‹#› of 34

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

2

11

Education is a hotly debated topic in public policy.

Education is the single largest expenditure item for state and local governments.

The United States spends more on education than most countries…

…but lags behind leading countries, and even much less wealthy ones, on standardized test scores.

Education

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

3

11

Education

The United States spends more money per pupil than nearly every country on Earth, but its educational outcomes are only average.

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

4

11.1

There are a number of public benefits (positive externalities) to education that might justify a government role in its provision.

Productivity

Society can benefit from the higher standard of living that comes with increased productivity.

Citizenship

Education may make citizens more informed and active voters, improving the quality of the democratic process.

Why Should the Government Be Involved in Education?

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

5

11.1

Educational credit market failure: The failure of the credit market to make loans that would raise total social surplus by financing productive education.

Without public education, many families would borrow money for their children’s education.

This market likely would not function well.

Why Should the Government Be Involved in Education?

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

6

11.1

Failure to maximize family utility

Parents may not choose an appropriate level of education for their children.

Redistribution

As long as education is a normal good, higher-income families would provide more education.

Income mobility has long been a stated goal for most democratic societies, and public education supports this goal.

Why Should the Government Be Involved in Education?

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

7

11.2

Most public education is provided through free public schools.

This system may crowd out private education provision.

Absent free public schools, some parents would send their children to expensive, high-quality schools.

With free public schools, parents can reduce quality by a small amount but save a lot.

Free Public Education and Crowding Out

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

8

11.2

One solution to the crowd-out problem is educational vouchers.

Educational vouchers: A fixed amount of money given by the government to families with school-age children, who can spend it at any type of school, public or private.

Vouchers put private schools and public schools on equal footing.

Solving the Crowd-Out Problem: Vouchers

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

9

11.2

Voucher proponents make two arguments for them:

Consumer sovereignty

Vouchers allow individuals to more closely match their educational choices with their tastes.

Competition

Vouchers allow the education market to benefit from the competitive pressures that make private markets function efficiently.

Solving the Crowd-Out Problem: Vouchers

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

10

11.2

Critics make several arguments against vouchers.

Vouchers may lead to excessive school specialization.

By focusing on particular market segments, schools give less focus to the key elements of education.

Vouchers will lead to segregation.

Critics of voucher systems argue that vouchers have the potential to reintroduce segregation along many dimensions, such as race, income, or child ability.

Problems with Educational Vouchers

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

11

11.2

Vouchers are an inefficient and inequitable use of public resources.

With vouchers, total public-sector costs would rise, as the government would pay part of the private school costs that families currently pay.

The education market may not be competitive.

The education market is described more closely by a model of natural monopoly, with efficiency gains to having only one monopoly provider of the good.

Problems with Educational Vouchers

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

12

11.2

The costs of special education

Special education: Programs to educate disabled children.

Each child would be worth a voucher amount that represents the average cost of educating a child in that town in that grade, but all children do not cost the same to educate.

Students with disabilities cost about twice as much to educate as students without.

Problems with Educational Vouchers

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

13

11.3

Direct comparison of voucher users and nonusers misleading.

Rouse (1998): Oversubscribed schools randomly admit students.

Vouchers increased test scores by 1 to 2% points.

Angrist et al. (2002): Columbian vouchers distributed by lottery.

Voucher winners 10% more likely to finish 8th grade, scored higher on standardized tests.

EVIDENCE: Estimating the Effects of Voucher Programs

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

14

11.3

Neilson (2013): Examined voucher program in Chile.

Vouchers reduced the achievement gap between poor and nonpoor students by about one-third.

Muralidharan and Sundararaman (2013): Studied lottery-based allocation of students to private schools in India.

Found modest improvements in test scores at much lower education cost.

EVIDENCE: Estimating the Effects of Voucher Programs

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

15

11.3

Some school districts have not offered vouchers for private schools but have instead allowed students to choose freely among public schools.

Magnet schools: Special public schools set up to attract talented students or students interested in a particular subject or teaching style.

Charter schools: Schools financed with public funds that are not usually under the direct supervision of local school boards or subject to all state regulations for schools.

Experience with Public School Choice

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

16

11.3

The United States has implemented many programs making schools accountable for student performance.

Accountability encourages schools to increase quality.

Many states laws, and No Child Left Behind passed in 2001, create accountability measures.

These laws appear to improve math scores, although the effect on reading is unclear.

Experience with Public School Incentives

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

17

11.3

Accountability programs have unintended effects.

They encourage “teaching to the test,” possibly improving test scores without learning.

Schools can manipulate the pool of test takers and the conditions under which they take tests to maximize success.

They may encourage schools or teachers to cheat.

Experience with Public School Incentives

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

18

11.3

Evidence is mixed.

Generally suggests that vouchers improve educational outcomes.

They come at the cost of potentially increasing inequality in educational achievement.

Some sort of guarantee of access must be provided to ensure all students have an education.

Bottom Line on Vouchers and School Choice

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

19

11.4

Measuring the returns to education is a key, but difficult, empirical question.

Returns to education: The benefits that accrue to society when students get more schooling or when they get schooling from a higher-quality environment.

More education clearly leads to higher wages.

Interpretation of this correlation is controversial.

Measuring the Returns to Education

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

20

11.4

There are two main interpretations:

Education as human capital accumulation

Human capital: A person’s stock of skills, which may be increased by further education.

Education as a screening device

Screening: A model that suggests that education provides only a means of separating high-ability individuals from low-ability individuals and does not actually improve skills.

Effects of Education Levels on Productivity

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

21

11.4

Policy implications

Human capital: Government would want to support education to raise their productivity.

Screening: Education does not raise productivity, so there is no reason to support it.

Differentiating the theories

Most of the returns to education reflect accumulation of human capital.

Some screening value to obtaining a high school or higher education degree.

Effects of Education Levels on Productivity

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

22

11.4

Comparing wages of people with more and less education suffers from ability bias.

More education may reflect intelligence or motivation, biasing comparison.

Duflo (2004): Look at school construction, comparing cohorts and regions with more schools.

Use laws that force people to stay in school longer.

Most approaches suggest that each extra year of schooling increases wages 7 to 10%.

EVIDENCE: Estimating the Return to Education

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

23

11.4

Better-educated people…

Participate more politically

Perform fewer criminal acts

Have better health and healthier children

Have better educated children

Have more productive coworkers

Effect of Education Levels on Other Outcomes

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

24

11.5

Some randomized experiments assign students to small or large classes.

Project STAR randomly assigned students to small classes, regular classes, or regular classes with teacher’s aides.

Small class size produced test score gains that persisted well into the future.

Overall rate of return to small classes: 5.5%.

EVIDENCE: Estimating the Effects of School Quality

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

25

11.5

An alternative approach is a quasi-experimental analysis of changes in school resources:

Mid-1990s California had the largest class sizes in the nation, 29 students per class on average.

The state government in 1996 paid schools to reduce their class size to 20 students per class.

Schools that implemented this quickly produced little gain, relative to schools that took longer.

EVIDENCE: Estimating the Effects of School Quality

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

26

11.5

The Role of the Government in Higher Education

Much of the discussion is on primary and secondary education.

In fact, about 43% of spending is for higher education.

Higher education in the United States is viewed as an enormous success.

Receives three sources of government financing:

$242 billion from state and local funding

$32 billion in Pell grants

$19.5 billion in tax breaks

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

27

11.5

State provision

The primary form of government financing of higher education is direct provision of higher education through locally and state-supported colleges and universities.

Pell Grants

The Pell Grant program is a subsidy to higher education administered by the federal government that provides grants to low-income families to pay for their educational expenditures.

Current Government Role

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

28

11.5

The federal government also provides support with loans.

Direct student loans: Loans taken directly from the U.S. Department of Education.

Guaranteed student loans: Loans taken from private banks for which the banks are guaranteed repayment by the government.

Means-tested subsidies:

Low interest rate guarantee

Defer repayment of the loan until graduation

Loans

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

29

11.5

The final way in which the government finances higher education is through a series of tax breaks for college-goers and their families.

Tax credits for families that send their children to college

Alternatively, deductions for educational expenses

These tax breaks add up to about $19.5 billion per year in forgone government revenue.

Tax Relief

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

30

11.5

The major motivation for government intervention in higher education is not to produce positive externalities but rather to correct the failure in the credit market for student loans.

Given that the major market failure for higher education is in credit markets, shifting state resources away from direct provision and toward loans would likely improve efficiency.

What Is the Market Failure, and How Should It Be Addressed?

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

31

11.6

The provision of education, an impure public good, is one of the most important governmental functions in the United States and around the world.

The optimal amount of government intervention in education markets depends on the extent of market failures in private provision of education and on the public returns to education.

Conclusion

‹#› of 34

C H A P T E R 1 1 ■ E D U C A T I O N

Public Finance and Public Policy Jonathan Gruber Fifth Edition Copyright © 2016 Worth Publishers

32

275

CHAPTER

6 10.1 Fiscal Federalism

in the United States and Abroad

10.2 Optimal Fiscal Federalism

10.3 Redistribution Across Communities

10.4 Conclusion

In 2002, President George W. Bush signed into law what would become one of his most significant contributions to domestic policy: the No Child Left Behind (NCLB) Act. NCLB sought to address the problem of substandard educational opportunities for poor and minority children by requiring standardized testing starting in Grade 3 and continuing through high school. In addition, NCLB mandated that schools publish their scores categorized by race and ethnic group. Harsh penalties, including the pos- sibility of the elimination of principals and teachers and the installation of new management, were to be imposed on schools that failed to show prog- ress. NCLB represented the greatest expansion of federal power over schools in half a century.

The first years of NCLB were marked by intense controversy nationwide and a fierce battle between the states and the federal government. While some concerns arose due to technical shortcomings with the law and lower-than- expected federal funding, the central issue of contention was the intervention on the part of the federal government into public education, a domain that has historically been reserved for the local and state governments. On the one

Questions to keep in mind

■ How does the provision of public goods at the local level affect their opti-

mality?

■ How should different types of spending be shared between different levels

of government?

■ What are the impacts of government grants designed to share spending

across levels of government?

State and Local Government Expenditures

CHAPTER

10 RO

BE RT

D O

DG E/

E+ /G

ET TY

IM AG

ES

276 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

hand, supporters of the law applauded the federal government for intervening when it is clear that many states have either failed or not even tried to close the achievement gap between white and minority students. An April 2005 editorial in the New York Times supporting NCLB stated that, historically, “the federal government has looked the other way when the states have damaged the national interest by failing to educate large swaths of the population. That approach has left us with one of the weakest educational systems in the devel- oped world . . . the Bush administration must stand firm against the districts that simply don’t want to make the effort.”1

On the other side, critics of NCLB countered that the federal govern- ment’s imposition of a standardized criterion across the nation interferes with ongoing local attempts to improve educational systems in a manner most suitable for each state. After a string of challenges to NCLB’s requirements, in 2011 the Obama administration announced that it would issue waivers to qualifying states. These waivers exempt states from the now-100% proficiency standards in math and reading, provided that the states meet certain criteria for improvement and accountability and accept certain education reforms (such as the Common Core program discussed in Chapter 11). The majority of states (42 as of April 2015) now have these waivers.2

On July 8, 2015, the House of Representatives narrowly passed a rewrite of NCLB. This new version would significantly reduce the federal role in super- vising education, instead placing that responsibility in the hands of state and local governments. The bill prohibits the federal government from mandating or encouraging specific curricula like the Common Core; allows parents to opt their children out of the tests without sanctions; and ties Title 1 funding to students, rather than schools.

Critics of this bill worried that allowing parents to decide whether their children take the tests could camouflage educational disparities because the lowest scoring students could stay home rather than take the tests. Rep. Bobby Scott (D-Va.) pointed out that the ability to opt out would skew the results of the tests: “If you’re not measuring the achievement gap, you can’t deal with the achievement gap.” Supporters countered that testing shifts the focus from learning to “teaching to the test.” “Because of this frenzied obsession with high-stakes testing, more and more time is being usurped from actual classroom learning,” says Rep. Matt Salmon (R-Ariz.).3

There was further controversy over the provision for the “portability” of Title 1 funds. Previously, Title 1 federal funding was allocated to schools with the highest concentration of poverty. The rewritten bill instead has Title 1 dollars follow poor students regardless of where they enroll.4 Opponents of the bill argued that the funding for low-income students would instead go to more affluent schools as students transferred out of their high-poverty schools,

1 New York Times (2005b). 2 Klein (2015). 3 Brown and Layton (2015). 4 Brown (2015).

S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S ■ C H A P T E R 1 0 277

stripping these schools of much-needed money. President Obama has threat- ened to veto the House bill.5

At the heart of the debates over NCLB is the question of who should control educational policy.6 This debate raises the important issue of optimal fiscal federalism, the question of which activities should take place at which level of government. Those who opposed NCLB were correct in asserting that local provision of government services allows communities to choose the package of services that best matches the tastes of their residents, potentially improving the efficiency of public goods delivery. The Bush administration was also correct in asserting that in some cases, programs that best serve local interests may not be in the national interest.

In this chapter, we discuss the set of issues surrounding state and local, or “subnational,” government spending, and the division of responsibilities across different levels of government. We begin with a discussion of the current division of responsibilities in the United States and other developed nations. We then turn to a discussion of whether local government provision of public goods solves the problems with government provision of public goods high- lighted in the previous chapter. In particular, by allowing individuals to choose the jurisdiction that best matches their tastes, local government provision of public goods may allow local governments to provide the optimal amount of public goods, surmounting the problems of preference revelation and prefer- ence aggregation that hamper decisions about national public goods provision.

The remainder of the chapter asks whether and how the government should redistribute resources across communities. There are enormous differ- ences across U.S. communities in the ability to finance local public goods, largely due to differences in the value of property on which local taxes are levied. Should the state and federal governments care about these differences? If so, what tools can these higher levels of government use to redistribute resources across communities?

10.1 Fiscal Federalism in the United States and Abroad

The last amendment (Amendment X) in the Bill of Rights of the U.S. Constitution states: “The powers not delegated to the United States by the Constitution, nor prohibited by it to the States, are reserved to the States respectively, or to the people.” Early in the history of the United States, therefore, the federal government played a relatively limited role in many aspects of the nation’s life, including the economy. As Figure 10-1 shows, in 1902 the federal government accounted for only 34% of total government spending, with local governments accounting for 58% and state govern- ments accounting for the remaining 8%. The federal government limited

5 Brown and Layton (2015). 6 Dillon (2005).

optimal fiscal federalism The question of which activities should take place at which level of government.

278 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

itself to spending on national defense, foreign relations, judicial functions, and the postal service. State and local governments were responsible for education, police, roads, sanitation, welfare, health, hospitals, and so on. The various levels of government operated in their own spheres, rarely over- lapping or interfering with each other. Furthermore, the state and local governments funded their spending largely from their own sources. Less than 1% of state and local revenues at the time came from federal govern- ment grants. Intergovernmental grants are payments from one level of government to another.

Over the next 50 years, the situation changed dramatically. By 1952, the fed- eral government accounted for 69% of total government spending, while local and state governments accounted for 20% and 11%, respectively. In addition, 10% of state and local revenue now came from federal grants. This change was largely due to three factors. The first was the Sixteenth Amendment to the Constitution (enacted in 1913), which allowed the federal government to levy income taxes on individual citizens (before this amendment, the Constitution had basically forbidden such taxation), thus providing a centralized source of revenue. The second factor in the growth of the federal government was the New Deal programs of the 1930s, which were the federal government’s response to the Great Depression. These programs initiated a number of federal govern- ment projects that fundamentally changed the relationship between the federal government and state and local governments. Federal grants to lower govern- ments ballooned, and many of the new programs, like the Works Progress (later Work Projects) Administration (WPA) and highway programs, were funded by the federal government but administered locally. The third factor in the growth of federal government was the introduction of large social insurance and welfare programs by the federal government, most notably the Social Security old-age income support program and the system of matching grants to encourage states to provide assistance to the elderly, blind, and disabled.

The share of spending done at the local, state, and federal levels has remained fairly constant over the past 50 years. There has been a growth in

intergovernmental grants Payments from one level of government to another.

■ ■ ■ ■ FIGURE 10-1

In the last hundred years, the federal government has grown significantly relative to state and local governments.

Data from: 1902–1977 data from Wallis and Oates (1998), Table 5.1; Bureau of the Census (2012).

0

20

40

60

80

100%

Federal

Local

State

Share of total government

spending

1902 1927 1952 1977 2012

S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S ■ C H A P T E R 1 0 279

7 Lee et al. (2015), Figure 1. 8 Barnett et al. (2014).

the share of state financing coming from the federal government, largely due to the introduction in the 1960s of jointly federal- and state-financed welfare programs such as cash welfare and public Medicaid insurance for the poor. Federal grants now account for 30% of state and local revenues in 2013.7

Spending and Revenue of State and Local Governments As noted in Chapter 1, the sources of revenue and the types of spending done by state and local governments differ dramatically from those of the U.S. federal government. On the spending side, the largest element of state and local spending is education, followed by health care and transportation; the largest elements of federal spending are health care, Social Security, and national defense. The federal government plays a very small role in financing education. On the revenue side, states receive only 20.7% of their revenues from income taxes, while the federal government obtains nearly half its revenues from income taxation.

A major source of revenue raising at the local level is the property tax, the tax on land and any buildings on it, such as commercial businesses or residential homes. Property taxes raised $433 billion in revenue in 2012 and accounted for almost half of the (nongrant) revenues of local governments.8

We discuss property taxation at length in Chapter 23 as part of the broader discussion of wealth taxation.

There is tremendous variety in spending and revenue raising behavior across U.S. states. Table 10-1 illustrates this variation by showing for a

property tax The tax on land and any buildings on it, such as commercial businesses or residential homes.

■ ■ ■ TABLE 10-1 Comparison of State Spending and Revenue Across the United States

State Dollars Per Capita

Spending

Education Alaska Massachusetts Tennessee

$4,672 (high) 2,848 (median) 1,993 (low)

Health Care District of Columbia Louisiana Utah

10,349 (high) 6,795 (median) 5,031 (low)

Taxes

Income Taxes New York Montana AK/SD/FL/NV/WY/WA/TX

2,431(high) 899 (median)

0 (low)

Sales Taxes Washington Connecticut AL/DE/OR/MT/NH

1,904 (high) 908 (median)

0 (low)

Data from: National Center for Education Statistics (2013); Centers for Medicare and Medicaid Services (2011); Malm (2015); and Tax Foundation (2014).

280 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

number of fiscal measures the state with the highest value, the median value, and the lowest value. For example, the state of Alaska has the highest education spending per capi- ta in the nation, at $4,672 while the median state, Massachusetts, spends $2,848 per capita, and the lowest state, Tennessee, spends $1,993 per capita. Health care spending per capita in the District of Columbia, the highest state, is a little more than twice that in Utah, the lowest state. Income taxes per capita are highest in the state of New York, at $2,431, and are zero in the seven states without an income tax, while sales taxes are highest in Washington, at $1,904, and are zero in the five states without a sales tax.

Fiscal Federalism Abroad Compared with most other developed nations, U.S. subnational (state and local) governments collect a much larger share of total (national plus subnational) government revenues and spend a somewhat larger share of total government spending. A recent survey of OECD nations, summarized in Table 10-2, showed that the average nation’s subnational governments collect only 27.2% of total government revenue, while in the United States subnational governments collect 36.8% of total revenue. The cross-national differences on the spending side are even more significant: the average OECD nation’s subnational government accounts for 30.4% of spending, compared to 46.7% in the United States.

The higher level of centralization in other nations exists because, in many countries, such as Mexico, Austria, and Norway, subnational governments have almost no legal power to tax citizens: this power is reserved for the central government. Moreover, in most countries, central governments redistribute a larger share of their revenues to subnation- al governments. Many countries practice fiscal equalization, whereby the national government distributes grants to subnational governments in an effort to equalize differences in wealth. This can be accomplished by providing larger national grants on a per capita basis to poorer subna- tional areas. In Austria, for example, the federal government offsets more than half the difference across subnational areas in the revenues that they are able to raise through taxation. The federal government in the United States is notable because it does not use grants for equalization; the only such program, initiated by President Richard Nixon in the early 1970s, was eliminated by 1986.9

fiscal equalization Policies by which the national government distributes grants to subnational governments in an effort to equalize differences in wealth.

9 Some implicit equalization still exists through the joint federal and state financing of social insurance and welfare programs because the federal share of those costs rises as state income falls.

■ ■ ■ TABLE 10 -2

Subnational Government Spending/Revenue as Share of Total Government Spending/Revenue in 2011

Spending % Revenue %

Greece 5.6 3.7 Portugal 13.5 6.6 France 20.2 13.1 Norway 33.6 12.1 United States 46.7 36.8 Denmark 61.2 26.7

OECD Average 30.4 27.2

Data from: “OECD Fiscal Decentralisation database“ (2013), Table 1, Table 5.

Compared to the subnational governments of other nation state and local governments in the United States account for a relatively large protion of total government activity.

S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S ■ C H A P T E R 1 0 281

Other nations also have a very different distribution of spending across national and subnational governments. In the United States, for instance, 30 to 40% of state and local spending is devoted to education, while the average in OECD nations is about 20%, highlighting the larger role that the central government plays in education in other countries.10

Recent years have seen a move toward fiscal decentralization around the globe. In the United States, there have been increased efforts to shift control and financing of public programs to the states, as demonstrated by the welfare reform example. In countries as diverse as Hungary, Italy, South Korea, Mexico, and Spain, there have been efforts to shift responsibility for health care, education, and welfare from national to subnational governments. Thus, in most countries, spending by subnational governments has increased over the past couple of decades, often financed through grants from the national gov- ernment. This increased funding and control has typically been accompanied by increasing imposition of national norms and quality standards on locally provided goods (such as increasingly rigid national curricula in education).

10.2 Optimal Fiscal Federalism

The different approaches to fiscal federalism seen in various nations raise a natural question: What is the optimal division of responsibilities across different levels of government? Why should anything be done by local governments? Alternatively, why is there any role for a central government? And which particular types of programs are most appropriately administered at which level of government? A theory of how the efficiency of public goods provision may differ at different levels of government will help answer these questions.

The Tiebout Model Two major problems with government provision of public goods, as discussed in the previous chapter, are the problems of preference revelation and preference aggregation: it is difficult to design democratic institutions that cause individuals to honestly reveal their preferences for public goods, and it is also difficult to aggregate individual preferences into a social decision. As a result, governments are often unable to deliver the optimal level of public goods in practice.

In 1956, economist Charles Tiebout (pronounced TEE-bow) asked: What is it about the private market that guarantees optimal provision of private goods that is missing in the case of public goods?11 His insight was that the factors missing from the market for public goods were shopping and competition. Shopping is the fundamental force that induces efficiency in private goods markets. If a firm is selling an inferior good relative to its competitors, consumers will purchase from the competitors, not from the firm. This competition leads firms to produce efficiently in the perfectly competitive private goods market.

10 Joumard and Kongsrud (2003), Table 4. 11 For the original paper, see Tiebout (1956).jdulgeroffHighlight

282 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

With many public goods, however, there is no shopping. Individuals don’t debate whether to live in the United States or in Canada based on whether the marginal missile is produced by the federal government. Voters can shop across political parties based on their promises to provide public goods, but this is only one of a large number of factors that determine votes for federal office, and the process of changing federal decision making is slow. Because there is little real competition facing the federal government when it makes its decisions to provide public goods, the decisions can result in inefficient public goods provision (as we saw in Chapter 9).

Tiebout pointed out, however, that the situation is different when public goods are provided at the local level by cities and towns (and to a lesser extent, states). In this case, he argued, competition will naturally arise because individuals can vote with their feet: if they don’t like the level of public goods provision in one town, they can move to the next town over, without nearly as much disruption to their lives as moving to another country.

Suppose, for example, that you read that the U.S. Department of Defense was spending $110 on an electronic diode worth $0.04, or $435 on a single claw hammer, or $437 on a measuring tape (as was revealed to be true in the United States in the 1980s).12 What could you do about this? You are unlikely to move to another nation. You could vote out the party in power, but your vote for congressperson or president is based on a large number of factors, of which this is only one. So there is really little you can do to end such inefficiency.

Now suppose instead that you found out that your local public library was spending $37,000 a year to supply the library’s director with a sports car, as happened in New York City in 2014.13 This waste clearly raises the prop- erty taxes that you pay to finance the town government. In this case, you have a realistic option: you can move to the town next door, which may be similar along most dimensions but better in terms of fiscal discipline. With local public goods, we have a new preference revelation device: mobility.

Tiebout argued that this threat of exit can induce efficiency in local public goods production. Indeed, he went one step further and argued that under certain conditions, public goods provision will be fully efficient at the local level. By the same logic that the competitive equilibrium delivers the efficient level of private goods, competition across localities in public goods provision will deliver the efficient level of public goods. Towns that don’t provide efficient levels of public goods will lose citizens to towns that do achieve efficiency— and will eventually go out of business.

The Formal Model In this section, we discuss the formal model that underlies Tiebout’s intuition. This model makes a number of assumptions that are unrealistic, as we discuss in the next section. Yet the main message of the model, that competition across local jurisdictions places competitive pressures on the provision of local public goods, is an important one that is consistent with the evidence that we review later in this chapter.

12 Barron (1983). 13 Gonzalez (2014).

S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S ■ C H A P T E R 1 0 283

The Tiebout model assumes that there are many people who divide them- selves up across towns that provide different levels of public goods. Each town i has Ni residents and finances its public goods spending, Gi, with a uniform tax on all residents of Gi/Ni. Tiebout showed that in this model, individuals will divide themselves up so that each resident in any town has the same taste for public goods, and so demands the same level of public goods spending, Gi.

This model solves the problems of preference revelation and aggregation that cause difficulties with public provision of public goods. There is no problem of revelation because there is no incentive for people to lie with a uniform tax that finances the public goods. To illustrate this, let’s return to the example of Jack and Ava from Chapter 9 (Figure 9.1), but now let’s assume that fireworks cost 75¢ each. Suppose that Jack joins a town of 100 individuals identical to himself. Such a town would vote to have 75 fireworks, with each person paying 56¢ to finance the fireworks. Now suppose that once again Jack lies by saying that he has the same preferences as Ava. In Tiebout’s model, to carry out that lie, he would have to actually move to a town of individuals like Ava (because those in his town want 75 fireworks, so that is the level provided). In Ava’s town, they choose to only purchase 25 fireworks, with each individual paying 19¢ for each firework. By moving to Ava’s town, Jack pays only one-third as much for fireworks—but he only gets one-third as many fireworks as a result. Jack has no incentive to lie because he must act on his lie by moving to a different town that matches his stated preferences. That is, Jack can’t free ride when individuals in each town are identical and equally share the financing of the public good. The problem of preference aggregation is also solved because everyone in town wants the same level of public goods Gi, and the town government can simply divide that amount by the population to get the appropriate financing.

With the preference revelation and aggregation problems solved, Lindahl pricing works in the Tiebout model. Each individual reports her true valua- tion of the public good, the valuations are added, and then each individual is billed for the total cost of the public good divided by population size. This is an equilibrium because every person is happy to pay her share of the tax to get the public good, and the condition for optimal public goods provision is met because the level of public goods provided is determined by the sum of the individual benefits.

Problems with the Tiebout Model Although the Tiebout model is interesting, it is obviously extreme. A number of problems stand in the way of the prediction of the Tiebout model that local public goods provision will be efficient.

Problems with Tiebout Competition The Tiebout model requires a num- ber of assumptions that may not hold in reality. The first assumption is perfect mobility: individuals must not only want to vote with their feet, they must be able to actually carry out that vote. This is difficult in practice. For example, I am now quite settled in my home town in Massachusetts, with many friends and other comforts. It would take a lot more than inappropriate funding of a library director’s sports car to get me to move now.

284 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

Perhaps even more implausible is the assumption that individuals have perfect information on the benefits that they receive from the town and the taxes that they pay. Even if the local public library were buying all its librarians sports cars, I would never find out unless it was somehow exposed by the local media (and I was paying attention).

Moreover, for the Tiebout model to hold, I must be able to choose freely among a range of towns that might match my taste for public goods. This range exists in the suburbs of Boston, where there are many towns that are fairly close to my job at MIT. But it might not be true in other areas, where towns are more spread out and voting with my feet would mean moving considerably farther from my job. Such restrictions on suitable substitutes for one’s town could limit the usefulness of the Tiebout mechanism for smaller or declining metropolitan areas.

Finally, the provision of some public goods requires sufficient scale or size. It is not efficient to run a school with only a few students or to build a park that will be used by only a few residents because of the large fixed costs of constructing the school or the park. These fixed costs lead to efficiencies of scale, whereby the efficiency of a public good is much higher if it is used by many rather than few. A school that is used by 1,000 students can be financed by a much lower property tax per household than a school that is used by 10 students because the large fixed costs of schooling (e.g., the building, the principal) can be spread among the larger set of households.

At the same time, the Tiebout model requires that there be enough towns so that individuals can sort themselves into groups with similar preferences for public goods. This raises a clear tension: Can we divide the population into groups of people who all have similar preferences for public goods, yet also ensure that these groups are large enough to support the economies of scale required by public goods?

Problems with Tiebout Financing A second major problem with the operation of the Tiebout model is that it requires equal financing of the public good among all residents. This kind of financing is called a lump-sum tax, a fixed sum that a person pays in taxation independent of that person’s income, consumption of goods and services, or wealth. As we discuss in the tax chapters, this form of taxation is viewed as highly inequitable by the public because both rich and poor pay the same amount of tax (most forms of taxation place higher tax burdens on the rich than on the poor). As a result, lump-sum taxa- tion is very rarely used to finance government expenditures. Indeed, the most high-profile attempt to impose lump-sum taxes, by the British government of Margaret Thatcher in 1990, resulted in major riots that led to the resignation of the once incredibly popular prime minister.

Towns typically finance their public goods instead through a property tax that is levied in proportion to the value of homes. The problem that this property taxation causes is that the poor chase the rich. Richer people pay a larger share of the public goods bill than do poorer people, so people who value those goods would like to live in a community with people richer than they are. That way, the poorer people can benefit from the higher taxes paid by

lump-sum tax A fixed taxation amount independent of a per- son’s income, consumption of goods and services, or wealth.

S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S ■ C H A P T E R 1 0 285

their richer neighbors. In other words, everyone wants to live in towns with people who are richer than they are so that they can free ride on their neigh- bors’ higher tax payments.14

One way that towns have endeavored to solve this problem is through the use of zoning. Zoning regulations are restrictions that towns place on how real estate property can be used, ostensibly with the goal of preserving the character of the local community. For example, one common zoning regulation requires that houses be built a certain distance back from the street to preserve some yard space and thus the aesthetic character of the neighborhood. Other examples of zoning regulations include prohibitions against using one’s home to run a business in a residential neighborhood, restrictions on the maximum number of occupants a lot or building may house, requirements for minimum lot sizes, constraints on the maximum size of buildings, and bans on multifamily housing.

Zoning regulations protect the tax base of wealthy towns by pricing lower- income people out of the housing market. For example, a town that prohibits multifamily dwellings (such as two-family houses and apartment buildings) lowers the available amount of housing and thus inflates the value of existing housing so that poor people can’t afford to move in and free ride on the tax payments of higher-income neighbors. Indeed, Glaeser and Gyourko (2002) compared areas with different zoning laws and found that the prices of land in zoned areas are higher by a factor of ten than prices in unzoned markets.

No Externalities/Spillovers A third problem with the Tiebout model is that it assumes that public goods have effects only in a given town and that the effects do not spill over to neighboring towns. If such spillovers exist, there is a case for provision of public goods at a higher level of government, or grants that subsidize local purchases.

Imagine that my town is considering building a large new public park. This park will be enjoyed primarily by individuals in my town, but many people from neighboring towns will visit its beautiful grounds as well. Under the Tiebout mechanism, when my town decides whether to build the park, it will consider only the preferences of residents in my town, not the preferences of residents of other towns who might enjoy the park. Thus, we face the standard problem with public goods provision: because people in other towns are free riding on my town’s park, my town will underprovide park services. If the social benefits (to my town and all surrounding towns) exceed the cost of building the park, it should be built, but if the private benefits to my town are smaller than the costs of building the park, then it will not be built, which is socially inefficient.

Many local public goods have similar externality or spillover features: police (if my town’s police department is not large enough, criminal activity in my town might spill over to other towns); public works (if my town’s streets are covered in potholes, the drivers from neighboring towns might suffer as they drive through my town); education (the entire nation benefits from a more

zoning Restrictions that towns place on the use of real estate.

14 See Problem 14 at the end of the chapter for how this might play out in practice.

286 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

educated citizenry); and so on. Thus, there is a fundamental trade-off with the Tiebout approach. There are advantages to locally provided public goods for small towns of similar individuals, but it may be optimal to provide public goods that have external effects or spillovers to other towns at a higher level of government that can internalize the externalities.

Evidence on the Tiebout Model The Tiebout model clearly imposes a very restrictive set of assumptions if taken literally, yet the basic intuition that individuals vote with their feet is still a strong one. Indeed, two types of tests reveal that the provision of local public goods is generally consistent with the Tiebout description.

Resident Similarity Across Areas A clear prediction of the Tiebout model is that people living in a given local community (such as a town) should have similar preferences for local public goods. The more local communities there are from which to choose, by the logic of this model, the more residents can sort themselves into similar groupings. If a city has only one suburb within commuting distance, it will be hard for residents working in the city to vote with their feet if they don’t like the level of public goods provision in that one suburb. Thus, a testable implication of the Tiebout model is that when people have more choice of local community, the tastes for public goods will be more similar among town residents than when people do not have many choices (and so can’t sort themselves into like-minded Tiebout communities).

Supportive evidence on this point comes from Gramlich and Rubinfeld (1982), who surveyed Michigan households on their demand for public goods. They found that in larger metropolitan areas (i.e., in suburbs near cities), where people have greater choice of which community they can live in, preferences for public goods were more similar within towns than in smaller areas with fewer independent towns to choose from. Moreover, in urban/suburban areas, residents were much more satisfied with the level of public goods spending than in nonurban areas where there are fewer ways to vote with one’s feet because there are fewer towns to move to. Bergstrom et al. (1988) used the data from Michigan suburbs to estimate individual demands for public goods and showed that the provision of local public goods appeared to satisfy the efficiency condition that the marginal cost equals the sum of marginal rates of substitution of residents.

Capitalization of Fiscal Differences into House Prices For many indi- viduals, the decision about where to live is not primarily determined by the level of local public goods. Indeed, many residents don’t even demonstrate the basic knowledge of local taxes and spending that is required for the Tiebout mechanism to operate.15 At the same time, the Tiebout mechanism requires not that all residents are willing to vote with their feet, but that enough residents are willing to do so to enforce the optimal provision of public goods.

15 See Dowding et al. (1994) for a review of the evidence on Tiebout and Teske et al. (1993), and see Dowding et al. (1995) for evidence on knowledge of public services and taxes among movers.

S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S ■ C H A P T E R 1 0 287

A town does not have to empty out completely before local officials get the message that the residen s are unhappy with public goods provision; all that is required is that there be sufficient mobility among an informed minority in response to public goods decisions.

In fact, very little actual mobility is required for the Tiebout mechanism to operate because people not only vote with their feet, they also vote with their pocketbook, in the form of house prices. The Tiebout model predicts that any differences in the fiscal attractiveness of a town will be capitalized into house prices. The price of any house reflects the cost (including local property taxes) and benefits (including local public goods) of living in that house. Thus, towns that have a relatively high level of public goods, given taxes paid, will have more expensive housing; conversely, towns that have relatively high property taxes, given the public goods provided, will have less expensive housing. House pricing, therefore, represents voting with your pocketbook: people will pay more for a house in a town that more efficiently delivers local public goods.

There is strong evidence for voting with the pocketbook, as reviewed in the Empirical Evidence box. Thus, even if some residents do not choose their location based on Tiebout factors, enough residents do make choices that way that it drives the pricing of housing across local communities.

Optimal Fiscal Federalism Although the Tiebout model is an imperfect description of reality, changes in local taxation and spending do affect mobility and house prices. Given these positive findings (i.e., they support the predictions of the model about behavior), what are the normative implications of the Tiebout model for the optimal design of fiscal federalism? That is, what does the Tiebout model imply should be the principles that guide the provision of public goods at different levels of government?

The Tiebout model implies that the extent to which public goods should be provided at the local level is determined by three factors. The first is tax-benefit linkages, the extent to which residents view their tax payments as directly tied to goods and services that they receive. Goods with strong tax-benefit linkages, such as local roads, should be provided locally. There is a direct tax-benefit linkage to spending on local roads: higher property taxes fund better-quality roads that benefit most residents of a town. Goods with weaker tax-benefit linkages, such as welfare payments to the lowest income residents of a town, should be provided at the state or federal level. There is a very limited tax-benefit linkage to spending on welfare: the majority of residents in a town do not benefit from redistribution to low-income groups (unless they have altruistic preferences toward the local poor).

If residents can see directly the benefits they are buying with their property tax dollars, they will be willing to pay local taxes. If they cannot see a benefit from their property tax payments, they will vote with their feet by moving to a town that has lower property taxes. If a town instituted a cash welfare program, higher-income residents would have an incentive to leave and move to a town

house price capitalization Incorporation into the price of a house the costs (including local property taxes) and benefits (including local public goods) of living in the house.

tax-benefit linkages The rela- tionship between the taxes that people pay and the government goods and services that they get in return.

288 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

A Property taxes are levied on the assessed value of a house, which can differ substantially from the house’s market value, as discussed in more detail in Chapter 23. B Sears and Citrin (1982).

Evidence for Capitalization from California’s Proposition 13

There is a large literature in state and local public finance that tests for capitalization effects. Typically, this literature pro- ceeds by regressing house prices on school quality or on local property tax rates and assessing whether higher-quality schools lead to higher house prices and higher taxes lead to lower house prices. These simple comparisons are poten- tially biased, however. For example, towns with better public schools may attract higher-income families, so finding that house prices are higher where schools are better does not prove that higher-quality schools are causing higher house prices. This correlation could just reflect that higher-income groups pay more for houses.

More convincing evidence for capitalization effects comes from Rosen’s (1982) study of the effects of California’s Propo- sition 13, a voter initiative that became law in 1978 and has proved to be one of the defining events of state and local public finance of the past half century. Proposition 13 was the first of a series of state laws that limit the ability of localities in a state to levy property taxes. Since its passage, nearly 40 statewide tax-limiting measures have been passed by voters in 18 states through the initiative process.

Proposition 13 mandated that the maximum amount of any tax on property could not exceed 1% of the “full cash

value” of the property. The full cash value was defined as the value as of 1976, with annual increases of 2% at most, unless the property was sold, in which case its full cash value would just be its sale value.A Thus, Proposition 13 restricted local property tax collections in two ways. First, it limited the rate that could be charged: the rate could not exceed 1% of a home’s assessed value. Second, despite the high inflation of the late 1970s, it limited the rate at which the tax base (the house’s value) could be increased to 2% per year. This was a strict limitation: the typical Los Angeles home saw its property tax increase 80% between 1973 and 1977.B

Rosen (1982) studied more than 60 municipalities in the San Francisco metropolitan area, examining tax rates and housing prices six months before and six months after the vote on Proposition 13. He compared towns with high property tax rates before 1978 (the treatment group), which were mandat- ed by Proposition 13 to have large reductions in their property tax rates, to towns with lower property tax rates before 1978 (the control group), which did not see much change in their property tax rates. So long as there was nothing else changing differently between treatment and control towns at this time, the passage of Proposition 13 provides a quasi-experiment for assessing the impact of property taxes on house values.

EMPIRICAL E V I D E N C E

that did not have such a program and had lower local property taxes as a result. The ability of individuals to vote with their feet is a fundamental limitation on a town’s ability to pursue programs that benefit only a minority of residents.

The second factor that determines the optimal level of decentralization is the extent of positive externalities, or spillovers, in public goods provision. If local public goods have large spillover effects on other communities, the goods will be underprovided by any locality. In this case, higher levels of government have a role in promoting the provision of these public goods (e.g., through grants).

The third factor that determines the optimal level of decentralization is the economy of scale in the nature of public goods. Public goods that have large economies of scale, such as national defense, are not efficiently provided by many competing local jurisdictions; public goods without large economies of scale, such as police protection, may be provided more effectively in Tiebout competition.

RO BE

RT D

O DG

E/ E+

/G ET

TY IM

AG ES

S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S ■ C H A P T E R 1 0 289

C Washington Post (1983).

As a result of these factors, the Tiebout model predicts that local spending should focus on broad-based programs with few externalities and relatively low economies of scale, such as road repair, garbage collection, and street clean- ing. Similarly, local communities should play a more limited role in providing public goods that are redistributive (such as cash welfare), have large spillovers (such as education), and have very large economies of scale (such as national defense). The nature of fiscal federalism in the United States is largely consis- tent with this prediction. Public works are financed primarily at the local level, redistributive programs are financed at the state and federal levels, and defense is a national program. Education is roughly one-half financed by localities and one-half financed by higher levels of government (mostly state government), which is consistent with the spillovers associated with education. The only question here is whether the externalities from education are sufficiently large on a nationwide basis that the federal government, which currently provides

Rosen found a strong association between reductions in prop- erty taxes and increases in house values: each $1 of property tax reduction increased house values by about $7. Given that buying a house means committing to a stream of future prop- erty tax payments, full capitalization of lower property taxes into house prices would imply that house prices should rise by the present discounted value of reduced future tax payments (the price should rise today to reflect the entire future benefit from lower property taxes). Mathematically, full capitalization would require that house prices rise by 1/r for each dollar reduction in property taxes, where r is the interest rate (recall from Chapter 4 that the present discounted value of a long future stream of payments is 1/r times the payment). Interest rates at the time of Proposition 13 were about 12%, so a $7 rise in house prices for each $1 reduction in property taxes suggests close to full capi- talization (with full capitalization, house prices would have risen by 1/0.12 5 $8.33 for each dollar reduction in property taxes).

This result implies very large capitalization of this policy change because, in principle, the fall in property taxes would result in a future reduction in public goods and services, which would lower home values. If, for example, each $1 of taxes was going to finance public goods and services worth $1 to resi- dents, then house prices should not have changed because

the gain of lower property taxes would be offset by falling local goods and services. The fact that house prices rose by almost the present discounted value of the taxes suggests that Cali- fornians did not think that they would lose many valuable pub- lic goods and services when taxes fell.

Rosen conjectures that Californians were not worried about falling public goods and services because the state used sup- plementary funds to offset the losses to local communities. Residents apparently perceived that these state offsets would continue or, alternatively, that the cut in taxes was simply reduc- ing “wasteful” local spending. That optimism appears to have been unfounded, however. San Jose, a fairly prosperous area with good public services, found itself having to cut services dramatically in the wake of Proposition 13. The school district laid off art and music teachers in the elementary schools, cut bus transportation, fired school nurses and guidance counsel- ors, and shortened the school day from six to five periods— all to no avail. In 1983, the district became the first American public school system in 40 years to declare bankruptcy. The rest of the town suffered too, as library hours shortened, parks became overgrown, and mental health nurses were fired. A poll of San Jose residents showed that a majority believed that Proposition 13 had worked out “unfavorably” for most people.C

290 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

16 Data from the “Public School District Finance Peer Search” (2015) provided by the Department of Education’s Education Finance Statistics Center, at http://nces.ed.gov/edfin/search/search_intro.asp. 17 Data from “Median Home Prices by Town,” provided by the Boston Globe (2013) at http://www .bostonglobe.com. 18 New York State Education Department (2014).

less than 10% of educational spending, should play a larger role in financing education, as is true in most other industrialized nations. The remainder of this chapter discusses the financing of education in more detail.

10.3 Redistribution Across Communities

The Tiebout model provides a framework for considering one of the most important problems in fiscal federalism: Should there be redistribution of public funds across communities? There is currently enormous inequality in both the ability of local communities to finance public goods (the value of the property tax base) and the extent to which they do so. For example, in the commonwealth of Massachusetts, the town of Lakeville raises only $13,932 in local tax revenue per public school student, while the town of Carlisle raises $22,472.16 Some of this difference comes from decisions about the level of local taxation: the tax per $1,000 of property value is $14.25 in Lakeville and $17.68 in Carlisle. Most of the difference in revenue, however, comes from underlying differences in the values of taxed property: the median single- family home is worth $367,000 in Lakeville and $851,722 in Carlisle.17 In the state of New York, one study found that the property values per public school student varied by a factor of almost six, with the poorest 10% of districts having property values per student of less than $313,891 and the richest 10% of districts having property values per student of more than $1,871,956.18

Should We Care? Should this inequality in revenue bases (as reflected in property values) or revenues raised (the product of property values and property tax rates) across communities concern public policy makers? Should higher levels of govern- ment mandate redistribution across lower levels of government to offset these differences? As noted earlier, such redistribution is an important feature of fiscal federalism in some nations, where the national government distributes grants to poorer communities that largely offset differences in revenues across communities.

The broad answer to the “Should we care?” question is that it depends on the extent to which the Tiebout model describes reality. In a perfect Tiebout world, we would not redistribute across communities: communities would have formed for the efficient provision of public goods, and any redistribution across them would impede efficiency. If a town has low revenues or low spending, it is because the residents of the town have chosen to provide a low level of public goods, and this is the efficient outcome given their tastes. Government redistribution in this case should focus on individuals, not on communities.

S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S ■ C H A P T E R 1 0 291

To the extent that Tiebout does not perfectly describe reality, however, there are two arguments for redistributing from high-revenue, high-spending communities to low-revenue, low-spending communities. The first is failures of the Tiebout mechanism. For example, suppose that there are reasons people cannot effectively vote with their feet, such as restrictive zoning rules that cause houses to be very large and expensive in communities with high public goods (e.g., each house must be on at least a one-acre lot). In this situation, there may be people who desire high levels of public goods but who cannot afford the high quality of house mandated by the zoning rules. These people could remain stuck in a town with low public goods provision, the only place where they can afford a house. In this case, it could be efficient to redistribute to the low public goods towns to help the individuals stuck in a situation where they are forced to underconsume public goods.

The second reason for redistribution is externalities. If a large share of local tax revenue is spent on local public goods with spillovers or externalities for other communities, there is a standard externality argument for higher levels of government to subsidize spending in the communities providing the externalities. For example, suppose that high-quality elementary education in a town leads to lower crime rates in both that town and neighboring towns. In this case, it may be optimal for the state government to tax high-revenue towns and redistribute to low-revenue towns to ensure that low-revenue towns can provide a high-quality elementary education.19

Tools of Redistribution: Grants If higher levels of government decide for one of the two reasons stated to redistribute across lower levels of government, they do so through intergovernmental grants, which are cash transfers from one level of government to another. Grants are a large and growing share of federal spending. From 1960 to 2014, grants to lower levels of government grew from 7.6 to 17% of federal spending.20 State governments, however, have always sent a large portion of the budget to local governments. From 1960 to 2002, state grants to local governments actually dropped slightly, from 34.1 to 28.1% of state spending, the bulk of which funded local education.21 Higher levels of gov- ernment use several different types of grants. In defining these types, we use the example of a state redistributing to local communities (although the same description applies to other forms of higher-to-lower level of government redistribution, such as national to state).

Suppose that the town of Lexington provides only one public good to its residents—education. It finances education through property taxes, and any money families have after taxation is spent on private goods (such as cars or clothing). Figure 10-2 shows the situation in Lexington before any grant is provided. Residents of Lexington have a total budget of $1 million to spend

19 For a sophisticated discussion of the implications of one set of “place-based” local policies, economic development, see Kline and Moretti (2014). 20 Office of Management and Budget (2014), Table 12.1. 21 U.S. House of Representatives Committee on the Budget (2012).

292 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

on education and other private goods, and we model how they choose to divide this budget. At point A, Lexington residents choose to spend nothing on education and spend their entire $1 million budget on private goods. At point B, Lexington spends its entire budget of $1 million on education and nothing on private goods.

The voters of Lexington have some preferences for education and private goods that can be represented as an indifference curve IC1 between these two sets of goods. That is, we can analyze Lexington’s choice between educa- tion and private goods in the same way that we might analyze an individual’s choice between these same items; IC1 represents the aggregation of the indif- ference curves of the voters through a voting mechanism. Before there are any state grants in place, Lexington chooses to spend $500,000 per year on educa- tion and $500,000 per year on private goods. This spending combination is represented by point X, where the town’s indifference curve is tangent to its budget constraint.

Matching Grants One type of grant the state government might use is a matching grant, which ties the amount of funds transferred to the local community to the amount of spending it currently allocates to public goods. For example, a one-for-one matching grant for education would provide $1 of funding from the state for each $1 of education spending by the local community. While we use a one-for-one match as the example here, match rates can vary from 0.01 to more than 1.

matching grant A grant, the amount of which is tied to the amount of spending by the local community.

■ ■ ■ ■ FIGURE 10-2

A Town’s Choice Between Education and Private

With $1 million to spend on some combination of education and private goods, Lexington chooses point X on its budget constraint AB, spending $500,000 on each, at the point where its indifference curve, IC1, is tan- gent to its budget constraint.

Private goods spending

(thousands)

$1,000$500

$1,000

500

Education spending

(thousands)

IC1

A

X

B

0

S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S ■ C H A P T E R 1 0 293

This one-for-one matching grant reduces the price of education by half; each dollar of education spending now costs Lexington only $0.50 because the commonwealth of Massachusetts provides the other $0.50. This change pivots the budget constraint outward from AB to AC in Figure 10-3. This grant unambiguously increases spending on education through both the income and substitution effects. In our example, total education spending increases from $500,000 to $750,000 at point Y. Lexington contributes $375,000 toward education and receives the other $375,000 in matching grants. Of its original $1 million budget, Lexington now has $625,000 to spend on private goods (the original $500,000 it was spending plus the $125,000 it no longer spends on education). As a result of the matching grant, then, total spending on both education and private goods has increased.

Block Grant Another grant option is a block grant, whereby the commonwealth simply gives the local community some grant amount G with no mandate on how it is to be spent. To keep the cost to the government constant, suppose that the government gives Lexington a $375,000 block grant. Because the block grant makes Lexington wealthy enough to afford to spend up to $1.375 million on either education or private goods, it shifts the budget constraint out from AB to DE, as Figure 10-4 illustrates.

block grant A grant of some fixed amount with no mandate on how it is to be spent.

■ ■ ■ ■ FIGURE 10-3

When Lexington is offered a matching grant for educational spending, with $1 of grant for each $1 of local spending, the budget constraint pivots outward from AB to AC. Lexington chooses point Y on AC, as it spends $250,000 more on education (with education spending rising from $500,000 to $750,000) and $125,000 more on private goods.

Private goods spending

(thousands)

$2,000$1,000$750$500

$1,000

500

625

Education spending

(thousands)

IC2

IC1

A

X

B C

Y

0

294 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

While Massachusetts is giving Lexington the same amount of money with the block grant, it has a very different effect on the town’s behavior. Some of this newfound wealth will be used to increase education spending, while some will be used to increase consumption of private goods. In this example, the town moves to point Z, raising education spending by only $75,000 and private goods spending by $300,000 (from $500,000 to $800,000).

The increase in education spending is lower with the block grant ($75,000) than it was with the matching grant ($250,000) because there is now only an income effect on education spending for Lexington, whereas the matching grant had both a substitution and an income effect. The income effect raises spending on education from $500,000 to $575,000, moving the town from point X to point Z. The substitution effect that is added with the matching grant then raises education spending by an additional $175,000 to $750,000, as reflected by the move from point Z to point Y.

On the other hand, Lexington has been made better off with the block grant than with the matching grant. This can be seen graphically by the fact that, with the new budget constraint under the block grant (DE ), the town could have afforded its choice at point Y, with education spending rising to $750,000 and private goods spending rising to $625,000, but it chose a different combination. Because the town chose point Z instead, it must be

When Lexington is offered an unconditional block grant of $375,000, the budget constraint shifts outward from AB to DE. Lexington chooses point Z on DE, as it spends $75,000 more on education (with education spending rising from $500,000 to $575,000) and $300,000 more on private goods.

■ ■ ■ ■ FIGURE 10-4

Private goods spending

(thousands)

$2,000$1,000 $1,375$750$575$5000

1,000

$1,375

625

500

800

Education spending

(thousands)

IC3

IC1

D

A

X

B E C

Z

Y

Income effect Substitution effect

S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S ■ C H A P T E R 1 0 295

on a higher indifference curve. That is, given the freedom to spend its grant money as it likes, without the restriction of a matching condition, the town would rather spend most of the money on private goods and relatively little on education. The matching grant leads to more spending on education than the town would otherwise choose given that amount of money, so it leaves the town on a lower indifference curve.

Thus, the optimal choice of grant mechanisms for higher levels of gov- ernment (such as states) depends on the goal of the grant program. If the goal is to maximize the welfare of the lower level of government, block grants will be most effective. If the goal is to encourage spending on public goods such as education, matching grants will be most effective because they will put both income and substitution effects to work to increase town spending.

Conditional Block Grant Suppose that Massachusetts likes the fact that it has made Lexington better off with a block grant than with a matching grant, but it doesn’t like the fact that education spending hasn’t gone up as much. One way the commonwealth could try to remedy this is through a conditional block grant, a fixed amount of money distributed to the town with a mandate that the money be spent only on education. In this case, the commonwealth could provide Lexington with a $375,000 block grant and mandate that it spend the entire grant on education.

The effect of this conditional block grant is illustrated in Figure 10-5. Lexington can now spend up to $375,000 (the grant amount) on education, while continuing to spend its original $1 million budget on private goods. Thus, the first segment on the budget constraint is now AF. Once Lexington spends beyond $375,000 on education, however, it faces the same trade-off between spending on education and spending on private goods that it did when it got the unconditional grant: the condition imposed on this grant doesn’t matter if the town is already spending more than $375,000 on education. The new budget constraint is, therefore, AFE. Beyond the $375,000 point on the horizontal axis, this new budget constraint is the same as the budget constraint from the unconditional block grant.

As is clear from Figure 10-5, adding this condition has no effect on Lexington’s behavior: the town still chooses to spend the same $575,000 on education that it spent with the unconditional block grant (at point Z ). Because Lexington was already spending more than $375,000 on education, this grant is effectively not conditional for the town—it has the same effect as if the commonwealth had simply given it $375,000 to spend on anything. Thus, the town has undone the mandate to spend the money on education by reallocating existing spending to meet the mandate. This is an example of the type of crowd-out that we discussed in Chapter 7. The government gave the town $375,000 to spend on education, but the town spent only $75,000 net of that money on education; it spent the remaining $300,000 on private goods. Thus, 80% ($300,000/$375,000) of the commonwealth’s spending was crowded out by the town’s reaction. Despite a large state grant, local education spending rose by only a small amount.

conditional block grant A grant of some fixed amount with a mandate that the money be spent in a particular way.

296 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

The effect of a conditional block grant will differ from that of an unconditional block grant only if the town receiving the grant would have spent less than the grant amount without the condition being imposed. That is, adding the condition to the block grant would affect Lexington’s behavior only if it would have chosen to spend less than $375,000 on education with the unconditional block grant. In that case, making the block grant conditional would increase Lexington’s educational spending by more than just $75,000. If towns such as Lexington would spend more than $375,000 on education regardless of this restriction, then there is no effect of imposing the restriction.

Redistribution in Action: School Finance Equalization Perhaps the most dramatic examples of attempts of higher levels of gov- ernment to use grants to influence lower levels of government are school finance equalization laws that mandate redistribution across communities in a state to ensure more equal financing of schools. Local school districts in the United States receive about 44% of their funding from local sources, primarily from local property taxes.22 This dependence on property taxes can

school finance equalization Laws that mandate redistribution of funds across communities in a state to ensure more equal financing of schools.

When the town is offered a conditional block grant for education spending, it can spend up to $375,000 on education while still spending $1 million on private goods. Beyond point F, the conditional block grant operates like the unconditional block grant, so the budget constraint is AFE. For towns that already have high educational spending, like Lexington, the conditional grant has the same effect as the unconditional grant, causing education spending to rise by $75,000.

■ ■ ■ ■ FIGURE 10-5

Private goods spending

(thousands)

$2,000$1,000 $1,375

1,000

$1,375

Education spending

(thousands)

IC3

IC1

D

$3750

A F

X

B E C

Z

Y

$750

625

500

800

Income effect Substitution effect

$575$500

22 New America Foundation (2014).

S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S ■ C H A P T E R 1 0 297

23 Data from the “Public School District Finance Peer Search” provided by the Department of Education’s Education Finance Statistics Center (2015), at http://nces.ed.gov/edfin/search/search_intro.asp. 24 Data from “A Guide to California’s School Finance System,” provided by EdSource Online (2015) at http://www.edsource.org. 25 Data available from Education Law Center (2015) at http://www.edlawcenter.org/index.htm.

lead to vast disparities in the revenue base from which towns fund education because of the wide variation in property values across towns. As a result of the disparity in property values, levels of education funding can differ sub- stantially across localities within a state. In Texas, for example, the Tom Bean district spends $10,116 per student, while the Borden district spends $58,960, or more than five times as much.23

States can try to offset these inequities by using the types of grants just dis- cussed. By collecting tax revenues from all communities, then redistributing the revenues in block or matching grants to particular communities with low property values or low education spending, the state can attempt to equalize education spending across districts. Since 1970, every state has made at least one attempt at school finance equalization, some prompted by state courts, others by the voting public.

The Structure of Equalization Schemes School finance equalization schemes can take very different forms. Some states have systems that attempt to completely or nearly completely equalize spending across school districts. California, for example, provides a base level of education financing for its school districts and prohibits differences between school districts of more than $350 in per-pupil spending. Once a district is spending $350 more than the lowest-spending district, all additional property taxes raised by the town are given to the state for distribution to other districts. Thus, under this scheme, a town receives no benefit from raising its own local property taxes because the extra revenue is divided among districts across the state.24

Less extreme are states that have instituted a statewide property tax that is redistributed in a way that guarantees a certain “foundation level” of per-pupil funding for each town. For example, in the state of New Jersey, towns with property values above the 85th percentile of the property values in the state simply receive a small foundational grant from the state and have to raise other educational revenues locally. Towns with property values below the 85th per- centile of the property values in the state receive a matching grant that is a multiple of their own educational spending, which thus gives towns an incen- tive to raise their spending.25

The Effects of Equalization A number of economics studies have evalu- ated the effects of school finance equalization. These studies generally agree that equalization laws have had the intended effect of equalizing school spending across communities, and spending equalization appears to have led to an equalization in student outcomes as well. Murray, Evans, and Schwab (1998), for example, concluded that court-ordered equalizations reduced in-state spending inequality by 19 to 34%. Card and Payne (2002) found that equalizations narrowed the gap in average SAT scores between children with highly educated and children with poorly educated parents by 8 points, or roughly 5% of the gap.

298 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

There is less agreement about whether this equalization has come about by raising spending among low-spending districts, lowering spending among high-spending districts, or both. A careful study of this question is provided by Hoxby (2001), who computed the tax price of school equalization schemes, the amount of revenue a local district would have to raise in order to gain $1 more of spending. California districts face an infinite tax price: no matter how much revenue they raise through local taxation, they can’t raise their local education spending to more than $350 per pupil above the lowest district. New Jersey’s districts mostly have tax prices of less than 1: a district might raise $0.60 of its own revenue in order to receive $0.40 in

tax price For school equaliza- tion schemes, the amount of revenue a local district would have to raise in order to gain $1 more of spending.

The Flypaper Effect

Block grants are simply income increases to communities if the grants are unconditional or if the grants are conditional but are below the town’s desired level of spending on that public good. As a result, a community should react to a block grant in the same way Lexington did in the example, by sub- stantially reducing its own contribution to the public good (a type of crowding out, as discussed in Chapter 7) so that spending on the public good goes up by only a fraction of the total grant amount.

This theory has been put to the test in the context of fed- eral grants to states. Researchers have compared the spend- ing of states that receive larger and smaller grants from the federal government to assess whether these federal grants largely crowd out the states’ spending. In fact, this does not appear to be the case. Hines and Thaler (1995) reviewed the evidence on this issue and found that the crowd-out of state spending by federal spending is low and often close to zero (so that total spending rises by $1 for each $1 in federal grants). Referring to Figure 10-5, towns such as Lexington appear unlikely to end up at point Z, as the theory implies, and instead seem to spend roughly the same amount on private goods ($500,000 in that example) and to devote the entire block grant to education. Economist Arthur Okun described this as the flypaper effect because “the money sticks where it hits” instead of replacing state spending.

These studies suffer from potential bias, however. As Knight (2002) noted, states that value public goods the most may be the most successful at lobbying for federal grants. If this is true, then there would be a positive correlation between grants and spending—not because of a flypaper effect, but simply because states that get grants are the ones that like spending the most. Thus, states that don’t get grants might not be a good control group for states that do because they might differ in their taste for public spending.

Knight proposed a quasi-experimental approach to solving this problem by noting that highway grants from the federal government to states are determined by the strength of the state’s political representatives. Congress- persons and senators have more power to determine the nature of highway spending if they (1) are on the transpor- tation committees of the House and Senate, (2) are in the majority party, and (3) have long tenures in Congress. In the type of vote-maximizing model discussed in Chapter 9, congresspersons will use this power to bring grants to their states.

Knight compared the level of spending in treatment states that see increases in the power of their congressional delega- tions (e.g., because a senator from that state gets appointed to the Senate transportation committee or because the con- trol of Congress changes to the party of the state’s senator) with the level of spending in control states that see decreases in the power of their congressional delegations (e.g., because a congressperson with long tenure is not reelected). Knight found, as expected, that federal grants rise for states that see increases in the power of their congressional delegations. He also found, as the standard model (but not the flypaper effect) would predict, that this grant money largely crowds out the state’s own spending: each additional $1 of federal grant money increase due to rising congressional power leads to a $0.90 reduction in the state’s own spending (so that total combined state and federal spending rose by only $0.10 per dollar of federal grant).

Knight’s study therefore throws some doubt on the previ- ous literature on the flypaper effect. Additional studies by Gor- don (2004) and Lutz (2010) also find evidence inconsistent with the flypaper effect, suggesting that the traditional conclu- sion of substantial crowd-out from block grants is supported by the evidence.

EMPIRICAL E V I D E N C E

RO BE

RT D

O DG

E/ E+

/G ET

TY IM

AG ES

S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S ■ C H A P T E R 1 0 299

state aid for a total of $1 in increased spending. This district would thus have a tax price of 0.6.

Hoxby found that extreme equalization schemes with very high tax pric- es, such as California’s, lead to an overall reduction in per-pupil spending. Because any taxes that towns raise beyond the minimum level (plus $350) are simply taken by the state and redistributed to other districts, there is an incen- tive to cut taxes and reduce spending. California’s equalization caused a drop in per-pupil spending of 15%; New Mexico’s spending dropped by 13%; and Oklahoma’s, Utah’s, and Arizona’s spending each dropped by 10%. States like California equalized per-pupil spending but only by “leveling down”—that is, lowering the overall education spending across all districts. The result has been a general deterioration in the quality of public schools and a flight to private schools by students who can afford it. Equalization schemes with low tax prices, such as those in New Jersey, New York, and Pennsylvania, actu- ally raised per-pupil spending by 7 to 8%; these states therefore managed to “level up.” Thus, school finance equalization can achieve its intended effects of improving the educational spending of low-wealth districts only if the system is designed in a way that gives those districts incentive to raise their spending without excessively penalizing higher-wealth districts.

Jackson et al. (2014) confirmed Hoxby’s findings about the differential impacts of school finance reforms and then examined the implications later in life for children subject to these reforms. They found that there were little effects on non-poor families, but large effects on the poorest: a 20% rise in spending led to almost a full year more of schooling and 20% higher earnings, enough to eliminate most of the gap between children raised in poor and non-poor families.

School Finance Equalization and Property Tax Limitations in California

William Fischel (1989) asked a very interesting question about the property tax limitations under Proposition 13 in California: If residents perceived that property taxes were “too high” in California, why did they wait until 1978 to lower them? Indeed, earlier referenda in 1968 and 1972 proposing property tax limitations had failed. What had changed by 1978?

Fischel’s answer is that Proposition 13 was actually a response to the court case (Serrano v. Priest) that led to school finance equalization in California in 1976. The key feature of this decision was that it broke the link between local property taxes and spending on schools by imposing the infinite tax price dis- cussed earlier. As a result, Fischel notes, this ruling also broke the Tiebout mecha- nism. Under the Tiebout model, property taxes are essentially prices paid for local services. In this model, individuals shop across communities (much as they shop across goods) to find the package of prices and spending that best matches their tastes. They know that if they choose a community with high taxes, they will be getting a high level of spending as well. Thus, there is a full tax-benefit linkage: their higher taxes buy them better public services (primarily schooling).

APPLICATION

RO BE

RT D

O DG

E/ E+

/G ET

TY IM

AG ES

300 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

The California equalization decision severed the link between taxes paid and benefits received. Taxes were no longer a price: they were just taxes. As a result, it was natural for communities to vote to lower taxes because they no longer perceived any benefit from them. Fischel claimed that wealthy voters would have opposed Proposition 13 in the absence of the school finance equalization because their high taxes were paying for schooling that they desired for their town without subsidizing anyone else’s schooling. School finance equalization changed this so that wealthy property-tax payers now saw that their taxes were paying for benefits accruing to other, poorer citizens in other towns. Thus, these wealthy taxpayers were happy to approve Proposition 13.

10.4 Conclusion

In every country, the central government collects only part of the total national tax revenues and does only part of the national public spending. The remain- der of taxation and spending is done by subnational governments, such as state and local governments in the United States. Relative to other developed countries, the United States places a large share of governmental responsibili- ties on its subnational governments. This chapter presented a theory to explain why spending might be divided between national and subnational govern- ments. When spending is on goods for which local preferences are relatively similar, and where most residents can benefit from those goods, the Tiebout model suggests that the spending should be done locally. When spending is for goods that benefit only a minority of the population, such as income redistri- bution, the Tiebout model suggests that it might be difficult to do this spend- ing locally because the majority of people who do not benefit will “vote with their feet” and move elsewhere. These outcomes are consistent with the divi- sion of responsibility for spending on education and public safety (local) and redistribution (national). In addition, if spending has external effects on other communities, local provision may be inefficient as well, which is consistent with the financing of education in the United States shared between local and state governments, although it raises the question of whether the federal gov- ernment should play a larger role.

Higher levels of government may not believe the conclusions of the ideal- ized Tiebout model, in which case they will want to redistribute across lower levels of government. If the higher-level government decides that it wants to redistribute across lower levels, it can do so through several different types of grants. The appropriate choice depends on the goal (redistributing to offset Tiebout failures or redistributing to offset externalities).

H I G H L I G H T S

■ A large share of public spending and revenue raising is done at the subnational level in the United States, relative to other industrialized countries.

■ The Tiebout model suggests that the provision of local public goods can be efficient if individu- als “vote with their feet” by moving to towns with others who share their tastes for public goods.

S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S ■ C H A P T E R 1 0 301

public services such as education are capitalized into house prices. Why would renters in high-income communities be more likely than owners to support this school choice plan? Why would the reverse be true in low-income communities?

5. Think about two public goods—public schools and food assistance for needy families. Consider the implications of the Tiebout model. Which of the goods is more efficiently provided local- ly? Which is more efficiently provided centrally? Explain.

6. Describe the externalities argument for distrib- uting money from one community to another. Provide an example of this kind of redistribution based on externalities.

7. The state of Minnegan is considering two alter- native methods of funding local road construc- tion, matching grants and block grants. In the case of the matching grant, Minnegan will spend $1 for every $1 spent by localities.

a. What is the price of an additional dollar of local spending in each case?

b. Which of the two methods do you think would lead to higher levels of local spending on roads? Explain your answer.

8. The commonwealth of Massachusetts recently ran an advertising campaign for the lottery which claimed “Even when you lose, you win.” The gist of the advertisement was that lottery revenue was used for particularly good ends, such as educa- tion. Suppose that lottery revenues are indeed

Q U E S T I O N S A N D P R O B L E M S

1. The (identical) citizens of Boomtown have $2 million to spend on either park maintenance or private goods. Each unit of park maintenance costs $10,000.

a. Graph Boomtown’s budget constraint. b. Suppose that Boomtown chooses to purchase

100 units of park maintenance. Draw the town’s indifference curve for this choice.

c. Now suppose that the state government decides to subsidize Boomtown’s purchase of park maintenance by providing the town with one unit of maintenance for every two units the town purchases. Draw the new budget constraint. Will Boomtown purchase more or fewer units of park maintenance? Will Boom- town purchase more or fewer units of the pri- vate good? Illustrate your answer, and explain.

2. Why does the Tiebout model solve the problems with preference revelation that are present with Lindahl pricing?

3. Some have argued that diversity in communities and schools leads to positive externalities. What implications does this view have for the efficiency of a Tiebout equilibrium? What implications does it have for government policy?

4. Brunner, Sonstelie, and Thayer (2001) studied how home ownership and community income influ- enced votes on a proposed initiative in California to allow children to obtain their locally funded educa- tion at any public or private school rather than being districted to their local school. Think about how

■ While the strict version of the Tiebout model is unlikely to hold, there is strong evidence that local spending and taxation respond to local preferences as reflected in mobility (voting with one’s feet) and that the value of local public goods and local tax differences are capitalized into house prices.

■ The Tiebout model suggests that spending with strong tax-benefit linkages (such as public safety) should occur at the local level and that spending with weaker tax-benefit linkages (such as redistribution under cash welfare) should occur at higher levels of government.

■ When higher levels of government want to redistribute to lower levels of government, they use

grants. Matching grants (under which the grant amount matches the amount to be spent by the lower level of government) are the best way to encourage a certain behavior by subnational governments, but unconditional block grants (under which the grant is a fixed dollar amount) maximize the welfare gains to communities from redistribution.

■ A classic example of redistribution across govern- ments is school finance equalization efforts, which have reduced inequality in local school spending but at the cost in some cases of a reduction in over- all educational spending.

302 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

e

e

earmarked for education. How would traditional economic theory evaluate the claim behind the ad campaign? How would an economist who believed in the flypaper effect evaluate it?

9. Why does California’s school finance equalization policy have a high associated marginal tax price? Explain.

A D VA N C E D Q U E S T I O N S

10. Rhode and Strumpf (2003) evaluated a century of historical evidence to investigate the impact of changes in moving costs within the Tiebout model.

a. What does the Tiebout model predict should happen to the similarity of residents within a community as the costs of moving fall?

b. Rhode and Strumpf found that while mobil- ity costs have steadily fallen, the differences in public good provision across communities have fallen as well. Does Tiebout sorting explain this homogenization of public good provision, or must other factors have played a larger role? Explain.

11. The state of Delaland has two types of town. Type A towns are well-to-do, and type B towns are much poorer. Being wealthier, type A towns have more resources to spend on education; their demand curve for education is Q 5 100 2 2P, where P is the price of a unit of education. Type B towns have demand curves for education which are given by Q 5 100 2 4P.

a. The cost of a unit of education is $20 per unit. How many units of education will the two types of town demand?

b. In light of the large discrepancies in educational quality across their two types of town, Delaland decides to redistribute from type A towns to type B towns. In particular, they tax type A towns by $5 for each unit of education they provide, and they give type B towns $5 for each unit of edu- cation they provide. What are the new tax prices of education in the two towns? How many units of education do the towns now purchase?

c. Delaland wants to completely equalize the units of education across towns by taxing type A towns for each unit of education they pro- vide and subsidizing type B towns for each unit of education they provide. It wants to do this in such a way that the taxes on type A towns are

just enough to finance the subsidies on type B towns. If there are three type A towns for every four type B towns, how big a tax should Dela- land levy on type A towns? How big a subsidy should they provide to type B towns?

12. The Individuals with Disabilities Education Act mandates that states and localities pro- vide appropriate education for all students identified as having special needs. States have responded by funding special education using several different mechanisms. Two of these mechanisms are “census” approaches (in which states estimate how many children should have special needs based on student characteristics and allocate money to localities based on these predictions) and “marginal subsidy” approaches (in which states pay localities a percentage of the amount of money that the localities say they spend on special education).

a. It has been found that the marginal subsidy approach leads to more students being classified by their localities as needing special education than does the census approach. Why might this be the case?

b. Suppose that you analyze cross-sectional data on the level of subsidy and the number of students enrolled in special education. You find that, in cross section, states that reimburse localities the most for their special educa- tion students tend to have the highest rates of students enrolled in special education. Think of one possible problem with this analysis.

13. As described in the text, Fischel (1989) argued that California’s Serrano v. Priest school finance equalization induced voters to limit property taxes in California. Following this argument, would an alternative school finance equalization that produced increased spend- ing for low-wealth communities using state funds be more, less, or equally likely to induce a property tax limitation in California? Explain.

S T A T E A N D L O C A L G O V E R N M E N T E X P E N D I T U R E S ■ C H A P T E R 1 0 303

14. There are two types of residents in Brookline and Boston, professors and students. Professors have an income of Y 5 200; students have an income of Y 5 100. Both Brookline and Boston provide road repair services for their citizens. Professors value road repair more than students because they have nicer cars. In fact, the value of road repair to an individual takes the form ((Y 3 R)/10) 2 (R2/2). The per-resident cost of road repair is 5R.

a. What is the marginal value of road repair for each type of individual? What is the marginal cost to each type of individual?

b. How much do professors want to spend on road repair? How much do students want to spend?

c. Assume that residents are distributed as follows:

Brookline Boston

Professors 75 25 Students 25 75

If each town uses majority voting to determine how much road repair to provide, how much will each town provide? Are any residents unsatisfied with the amount of road repair?

d. Now assume that professors and students are able to migrate between Brookline and Boston. Which residents will choose to move?

What will the equilibrium distribution of resi- dents be? Are any residents unsatisfied with the amount of road repair now? Is the provision of road repair efficient? Why or why not?

e. Consider again the premigration equilibrium. The commonwealth of Massachusetts decides to pass a law about road repair. It requires that professors in Massachusetts must contribute 75 units toward road repair in the town where they live; students must contribute 25 units toward road repair in the town where they live. How much road repair will there be in each town under the new regime? Will any residents want to move and, if so, where and why?

f. Now suppose that a few of the professors, in spite of their high incomes, have low demand (10R 2 R2/2), and a few of the students have high demand (20R 2 R2/2). Given these new demands, where would people move if each person in a town pays the same amount toward road repair? If, on the other hand, road repair is funded by a tax that is proportional to income, how much road repair would different groups want, and where would they move?

The e icon indicates a question that requires students to apply the empirical economics principles discussed in Chapter 3 and the Empirical Evidence boxes.

this page left intentionally blank

305

CHAPTER

11 Education

Questions to keep in mind

■ Why and how should the government be involved in the provision

of education?

■ What is the proper role of competition in education markets?

■ What is the current role of the government in the provision of higher

education, and should that be changed?

11.1 Why Should the Government Be Involved in Education?

11.2 How Is the Government Involved in Education?

11.3 Evidence on Competition in Education Markets

11.4 Measuring the Returns to Education

11.5 The Role of the Government in Higher Education

11.6 Conclusion

In the United States, education is the single largest expenditure item for state and local governments: they spend 28% of their budgets to provide their citizens with this service. In fact, the United States spends more money per pupil on education than nearly every other nation on Earth. Yet U.S. students perform only around the international average on tests of reading, math, and science ability. Even worse, U.S. eighth graders are less proficient in math and science than students in much less wealthy countries, such as Hungary, Lithuania, and Slovenia, which have a combined gross domestic product (GDP) that is 1.4% of the U.S. GDP. Figure 11-1 compares the United States with other nations in terms of money spent per pupil and the resulting educational outcomes in mathematics for eighth graders. This comparison reveals that the United States spends much more per student to achieve outcomes that are generally worse than those in these other nations.

While there is widespread agreement on the problematic state of education in the United States today, there is much less agreement about the causes of or solutions to its shortcomings. In 2009, President Obama appointed as his new Secretary of Education Arne Duncan, who, as chief executive officer (CEO) of the Chicago public school system, had earned a strong reputation as an educational reformer, including controversial closings of underperforming schools.1 Duncan brought his reformer spirit to the national stage, calling education reform “the civil rights issue of our generation.” He developed a plan to substantially broaden the

1 Parker (2009).RO BE

RT D

O DG

E/ E+

/G ET

TY IM

AG ES

306 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

federal role in local primary and secondary education through tougher requirements on students and teachers, intensified efforts to assist failing schools, and a growing focus on alternatives to the traditional public school model such as charter schools, which provide free public education but are not bound by many of the regulatory restrictions (and union obligations) of traditional public schools.

The first step in this action plan was taken as part of the stimulus bill passed in early 2009. This legislation sent almost $54 billion to states over a two- year period to prevent layoffs, create jobs, and modernize school buildings and sent another $25 billion to promote the education of disadvantaged students. To receive this aid, however, states had to meet a number of new requirements, including ensuring that the state’s most talented teachers are assigned equitably to both rich and poor students; building sophisticated data systems that link teachers to students and test scores, thus allowing authorities to measure teacher effectiveness; taking vigorous action to assist failing schools; and embracing charter schools as an educational alternative.

In addition, the bill set aside a $4.4 billion Race to the Top fund that Duncan can use to reward states for educational innovations such as collaboration across schools or between schools and nonprofit organizations. States could also use this reward money for teacher pay-for-performance programs. In the first round of this program, 40 states applied and 2 (Delaware and Tennessee) won; in the second round, 47 states applied and 10 states won. Over the course of phases one and two, 35 states and Washington, D.C., have adopted new reading and math standards, and 34 states have “changed laws or policies to improve education.”2

charter schools Schools financed with public funds that are not usually under the direct supervision of local school boards or subject to all state regulations for schools.

2 Department of Education Press Office (2010).

■ ■ ■ ■ FIGURE 11-1

300

350

400

450

500

550

600

650

0

2000

4000

6000

8000

$10,000

U. S.

A.

Sw ed

en

Ja pa

n

UK

So ut

h Ko

re a

Is ra

el

Cz ec

h Re

pu bl

ic

Primary school spending per pupil

(2014 dollars)

8th grade math scores

The United States spends more money per pupil than nearly every country on Earth, but its educational outcomes are only average.

Spending data from: OECD Education at a Glance (2014). Math scores data from: PISA Country Profile (2012).

E D U C A T I O N ■ C H A P T E R 1 1 307

To supporters, these changes and Duncan’s larger plan represent a needed first step toward fundamental educational reform. Representative George Miller, Democrat from California, then chair of the House Education Committee, said, “This is a very serious amount of money . . . both the President and the Secretary do not want to lose a year or two in the efforts to achieve reforms that are necessary to create a modern, effective school system throughout this country.”3 To critics, these aspects of the stimulus bill represented overreaching by the federal government into an area traditionally regulated by state and local governments. Particularly criticized were the aspects of the bill that favored charter schools. One such sticking point was the rule that states that did not embrace charter schools (often derided by the traditional public schools with which they compete) would not be eligible for any Race to the Top funds. Gerald Bracey, an associate at the High/Scope Educational Research Foundation, said, “[Duncan is] blackmailing states, saying you either have to have charters . . . or your stimulus money will be at risk. There’s no evidence out there [about the benefit of charter schools] to justify it.”4

The most recent controversy over education reform pertains to the 2008 introduction of the “Common Core” of educational standards, the federal government’s plan to improve the achievement levels of students across the country. The Common Core is a set of national academic standards that outline what children in grades K–12 ought to know about math and English language arts at the end of each grade.

When the Common Core was initially presented to states, it was fairly uncontroversial. During a 2009 Chicago meeting with governors and school chief officers, many agreed that it was the right thing to do.

The implementation of the Common Core has been rough, however, partly because these national standards (which emphasize depth of knowledge and critical thinking) are more rigorous than standards that were already in place in a number of states. A number of states reacted negatively to their apparent worsening of scores with this new standard. In New York, for example, fewer than one-third of students were found to be up to Common Core English Language Arts standards in the 2012–2013 school year—down from 55% on non-Core standardized tests the previous year.5 Others see the Common Core as the federal government imposing too much power on the education system. Senator Marco Rubio (R-Fla.), a 2016 GOP Presidential hopeful, warned that the White House was attempting to “[turn] the Department of Education into what is effectively a national school board. . . . Empowering parents, local communities and the individual states is the best approach.”6

Is the approach being pursued by Duncan and Obama the right one? Should we go further in moving to a more competitive and accountable education system? Or could educational improvement be better achieved simply by investing more money in our existing system?

3 Glod (2009). 4 Bruce (2009). 5 Resmovits (2014). 6 Williams (2014).

308 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

In this chapter, we review the public finance issues involved in providing education. We begin with the first question of public finance: Why should the government be involved in education at all? We discuss a number of rationales for public involvement and their implications for the second question of public finance: How should the government be involved? We address this question in two steps. First, we consider the structure of government involvement, showing that public provision of a fixed level of education can crowd out private education. This result implies that efficiency may be increased with vouchers that can be used at either public or private schools. We extensively review the debate over school choice and school vouchers, discuss the theo- retical arguments for and against vouchers, and look at the limited empirical evidence available on this debate.

The second step of our analysis is to ask, for a given structure, how much the government should spend on education. A central determinant of how much the government should spend on education is the return provided by this investment. We review the existing evidence on the returns to education and what they imply for government involvement. Finally, we turn to a discussion of higher education, a market that appears to work much better in the United States than elsewhere in the world but that still raises many difficult policy issues.

11.1 Why Should the Government Be Involved in Education?

In the United States, 90% of elementary and secondary students are in public educational institutions instead of privately financed institutions. Should the public sector be so dominant in the provision of education? What failure in the private education market justifies government’s dominant role? Education is not a pure public good because it does not meet the conditions of non-rivalry (that my consumption of the good does not reduce your enjoyment of the good) and non-excludability (I cannot deny you the opportunity to consume or access the good). Education is clearly a rival good: having more children in a classroom may lower the quality of classroom instruction. Education is clearly also to some extent excludable: private schools can decide which students to accept.

At the same time, there are a number of public benefits (positive externalities) to education that might justify a government role in its provision.

Productivity The first potential externality from education is productivity. If a higher level of education makes a person a more productive worker, then society can benefit from education in terms of the higher standard of living that comes with increased productivity. As discussed in Chapter 6, however, this higher standard of living is not an externality if the worker is the only one who reaps the benefits from his or her higher productivity. For example, if more educa- tion raises Stacey’s marginal product of labor, but the increase is fully reflected in her receiving a higher wage from her employer, then there is no positive externality to society from Stacey’s education.

E D U C A T I O N ■ C H A P T E R 1 1 309

Social benefits from higher productivity occur through one of two channels. The first is “spillovers” to other workers: Stacey’s increased productivity could raise the productivity of her coworkers, thus raising their wages and well- being. Because Stacey herself is unlikely to be fully compensated for the rise in her coworkers’ wages, this is a positive externality to her coworkers from her education. The second is through taxes: if Stacey’s higher productivity is reflected in higher pay, then the government collects more tax revenues as a result.

Citizenship Public education may improve the quality of life in the United States in indirect ways as well. Education may make citizens more informed and active voters, which will have positive benefits for other citizens through improving the quality of the democratic process. Education may also reduce the likelihood that people turn to a life of crime, an outcome that has positive benefits for other citizens by improving their safety and reducing the public costs of policing. More generally, education may play a role in enabling immigrants, who are some of the most productive members of U.S. society, to establish themselves in the United States. These arguments are fairly compelling for public intervention in basic education such as elementary school, but they provide less rationale for public financing of secondary and especially higher education.

Credit Market Failures Another market failure that may justify government intervention is the inability of families to borrow to finance education. In a world without government involvement, families would have to provide the money to buy their children’s education from private schools. Suppose, in this private-education-only world, there is a poor family with a talented child, and this child could earn a comfort- able living as an adult if properly educated. It would be socially optimal for this child to be educated, yet the family cannot afford the costs of education.

In principle, the family could borrow against the child’s future labor earnings to finance the education. Yet, in practice, banks and other lenders are unlikely to make such loans because there is no source of collateral (assets owned by a person that the bank can claim if the person doesn’t pay back the loan). If the family takes a loan to finance a home purchase (a mortgage), the collateral is their house; if they don’t repay the loan, the bank can claim their house to offset its losses. Because the bank cannot claim the family’s child if they don’t repay the loan, banks may be unwilling to lend for education; after all, despite the family’s claims, the bank can’t really tell if their child is a good investment or not. This situation is an educational credit market failure: the credit market has failed to make a loan that would raise total social surplus by financing productive education.

The government can address this credit market failure by making loans available to families to finance education. Yet the government in the United States and the governments of most industrialized nations do not play this role except in financing higher education (discussed at the end of this chapter). Instead of providing loans to finance elementary and secondary education, the government directly provides a fixed level of publicly funded education.

educational credit market failure The failure of the credit market to make loans that would raise total social surplus by financing productive education.

310 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

Failure to Maximize Family Utility The reason governments may feel that loans are not a satisfactory solution to credit market failures is that they are concerned that parents would still not choose appropriate levels of education for their children. In a world with well-functioning credit markets (or with government loans available), private education would probably still involve some sacrifice on the part of parents, such as paying the cost of schooling not covered by loans or making interest payments on the loans. Even if total family utility would rise with a more highly educated child, some parents may not be willing to reduce their consumption to finance their children’s education because they care more about their own consumption than their children’s future income. (As noted in Chapter 6, evidence suggests that parents are not maximizing the utility of their entire family.) Children can be harmed by the unwillingness of their parents to finance their education, and making loans available to parents cannot solve that problem. In this case, public provision of education is a better alternative. Otherwise, smart children would be penalized for having selfish parents.

Redistribution A final justification for government involvement is redistribution. In a privately financed education model, as long as education is a normal good (demand for which rises with income), higher-income families would provide more education for their children than would lower-income families. Because more education translates to higher incomes later in life (as we show later in this chapter), this situation would limit income mobility because children of high- income parents would have the best opportunities. Income mobility, whereby low-income people have a chance to raise their incomes, has long been a stated goal for most democratic societies, and public education provides a level playing field that promotes income mobility.

In summary, then, there are various reasons for government involvement in education: potential productivity spillovers, more informed and less criminally inclined citizens, failures in credit markets, failures of family utility maximization, and redistribution. We next turn to the question of how governments are involved in education and what effects their involvement has on educational attainment.

11.2 How Is the Government Involved in Education?

In Chapter 5, we discussed two alternative means for governments to deal with positive externalities: the price mechanism and the quantity mechanism. In the context of education, the price mechanism approach is to offer discounts on private educational costs to students, and the quantity mechanism approach is to mandate that individuals obtain a certain level of education. In practice, the governments of most developed nations pursue neither of these approaches, and instead they provide a fixed level of education for no cost.

E D U C A T I O N ■ C H A P T E R 1 1 311

In this section, we discuss the effects of providing free public education on the level of educational attainment (the amount and quality of education received by individuals) in society.

Free Public Education and Crowding Out We can model public education using the same approach we used to model the provision of a public good (fireworks) in Chapter 7: education is a public good that is provided to some extent by the private sector. As such, an impor- tant problem with the system of public education provision is that it may crowd out private education provision. Indeed, as economist Sam Peltzman argued in 1973, it is possible that providing a fixed amount of public education can actu- ally lower educational attainment in society through inducing choice of lower- quality public schools over higher-quality private schools.7

In Peltzman’s model, individuals are choosing how much to spend on their children’s education. He assumes that the more individuals spend, the higher quality education they can buy for their children (later in the chapter, we review the evidence for the strength of this spending–quality link). The public sector provides some fixed level of expenditure and thus of quality. If parents want higher-quality education than that provided by the public sector, then they must send their children to private school.8 By sending their children to private school, however, parents forgo their entitlement to free public edu- cation for their children. As a result, some parents who might desire higher quality education for their children decide not to use private schools; they reduce their desired education to take advantage of free public education. For this group, free public schools have, therefore, lowered the quality of education they “purchase” for their children.

Figure 11-2 illustrates the choice families face between spending on education and spending on all other goods. Before there is any provision of public educa- tion, families face the budget constraint AB, with a slope that is dictated by the relative prices of private education and other goods. Any money that is spent on a child’s education reduces the family’s budget for purchasing other goods.

The government then provides free public education of a quality that costs EF . For now, we ignore the financing of this educational expenditure (because all the policy alternatives we discuss involve financing as well, we discuss financing separately later in the chapter). The provision of free public education means that individuals can spend their full budget on other goods and still get educational spending of EF (at point C ). To spend more than EF on education, however, the family would have to entirely forgo the free public education; although the public education is free, it can be used only up to amount EF . Thus, the new budget constraint runs from A to C (because education is free to a spending–quality level of EF), then drops down to point

7 See Peltzman (1973). This discussion is couched in terms of school quality, but the same argument could be made about quantity of education received. 8 The model assumes that individuals cannot simply “top off ” public education by supplementing it with private spending, for example, on tutoring.

312 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

D, after which it is the same segment DB as the original budget constraint. What does the provision of free public education do to educational spending (and thus quality) choices?

In Figure 11-2, we compare three families, X, Y, and Z, all of whom have children. Before public schooling is introduced, these families choose different-quality (and thus different-cost) private schools. Family X initially chooses bundle X, the point at which their indifference curve is tangent to the private-market budget constraint AB. This bundle consists of relatively little education spending for their children (a low-quality private school) at E1 and relatively high spending on other goods of G1. After the free public system is introduced, family X moves from point X to point C, a bundle that consists of higher levels of education spending (EF) and higher levels of spending on other goods (G2 at point A). The family is on a higher indifference curve (indicating greater utility) at this kink in the budget constraint because its consumption of both education and other goods has increased.

Family Z initially chooses bundle Z, very high educational spending (a very high-quality private school at E3) and relatively low spending on other goods (G4). When the public system is introduced, there is no change in family Z ’ s spending on either education or other goods; this family wants such a high- quality education for its children that the public school option is irrelevant.

■ ■ ■ ■ FIGURE 11-2

Other goods spending

Education spending

X

C

D

E2E1 EF E3 B

A

G3

G4

G1

G2

EF

Y

Z

When the government introduces free public education in the amount of EF, the budget constraint changes from AB to ACDB. This leads families such as X to increase the amount of education they obtain from E1 to EF and families such as Z to maintain their educational spending of E3. Families such as Y, however, reduce their educational spending from E2 to EF.

E D U C A T I O N ■ C H A P T E R 1 1 313

Family Y initially chooses a medium level of spending at point Y, with initial educational spending of E2 and spending on other goods of G3. After free public education is introduced, however, the family moves to point C, a bundle in which their education spending has fallen a bit from E2 to EF, but their consumption of other goods has increased greatly from G3 to G2. Their utility has increased because point C is on a higher indifference curve than point Y is. Thus, the introduction of free public education has reduced family Y’s spending on education. By spending somewhat less on education, the family can dramatically increase how much they can spend on other goods, and this is a trade-off they are willing to make. It is true that children in family Y would have gotten more education by staying in their original private school (level E2), but this would have required the family to forgo a lot of consumption of other goods; the family is better off by sacrificing a small amount of education to obtain a lot more consumption of other goods. For group Y, public educational spending has crowded out private spending on education as the family reduced their overall education spending levels in response to this free public option.

Thus, free public education increases educational quality for children in families such as X, lowers it for children in families such as Y, and has no effect on families such as Z. In principle, if group Y is big enough relative to group X, total educational spending (and thus educational quality) could actually fall when free public education is introduced.

Solving the Crowd-Out Problem: Vouchers One solution to the crowd-out problem is the use of educational vouchers, whereby parents are given a credit of a certain value (e.g., the average spending on a child of a given age in the public education system) that can be used toward the cost of tuition at any type of school, public or private. Figure 11-3 illustrates how a voucher system could work: families would be given a voucher for an amount EF, which they could either give to their local public schools in return for free education for their children or apply toward private school tuition. The availability of this voucher would lead to a new budget constraint ACE: families get an amount EF to spend on education without lowering other consumption. The voucher has the same effect as a conditional lump-sum grant to local governments: it raises incomes but forces the families to spend a minimum amount on education.

With this system, educational spending (and therefore quality) would increase for all three types of families. Family X would still move to point C, at which both education and other consumption have increased. Once again, family X’s utility has increased, and they will be on a higher indifference curve at this point.

Family Y would no longer move to point C and purchase less education (EF instead of E2) as they did in Figure 11-2 because now they no longer have to forgo the public subsidy to get higher-quality private education. Now, they can move to a point such as Y2 in Figure 11-3, using some of their higher income (from the voucher) to purchase more education (E4 instead of E2) and some to purchase more of other goods (G5 instead of G3). This is a preferred outcome to point C because they get both more education and more consumption of other goods.

educational vouchers A fixed amount of money given by the government to families with school-age children, who can spend it at any type of school, public or private.

314 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

Instead of continuing to purchase the same amount of education as they did in Figure 11-2, family Z would now choose a point such as Z2 and use some of their higher income to purchase more education (E5 instead of E3) and some to purchase more of other goods (G6 instead of G4). Under the voucher program, total education has clearly increased.

This type of analysis motivates support for educational vouchers as a policy option in the United States. A number of analysts have proposed voucher systems whereby individuals are given the choice of either attending free public schools or applying their local public school spending to their private school education. Supporters of vouchers make two arguments in their favor, which mirror the two arguments in favor of free choice in most economic markets.

Consumer Sovereignty The first argument in favor of vouchers is that they allow individuals to more closely match their educational choices with their tastes. By forcing individuals either to choose free public education or to forgo this large public subsidy and choose private education, today’s system does not allow people to maximize their utility by freely choosing the option that makes them best off. This restriction has the unintended consequence of crowd-out that could be solved with vouchers.

■ ■ ■ ■ FIGURE 11-3

When the government provides vouchers in the amount of EF, the budget constraint changes from AB to ACE, leading all families to increase educational spending. Low-spending families like X will spend the full amount (EF) on public schools, families such as Y will switch from public education in an amount EF to private education in an amount E4, and higher-spending families like Z will also increase their educational spending somewhat (from E3 to E5) because the voucher increases their effective incomes.

Other goods spending

Education spending

Y2

Z2

X1

C

E2E1 EF

EF

E4 E5E3 EB

A

G3

G4

G1

G2

G5

G6

EF

Y1

Z1

E D U C A T I O N ■ C H A P T E R 1 1 315

Competition The second argument in favor of vouchers is that they will allow the education market to benefit from the competitive pressures that make private markets function efficiently. Critics contend that the public education sector is rife with inefficiency. They point to the fact that per-pupil spending has more than doubled since 1970, yet the math and reading scores of twelfth graders have risen by only about 2% over that same time period. Furthermore, the number of administrative staff in public schools has grown by 91% since 1970, while the number of enrolled students has grown by only 9%.9

This inefficient bureaucracy has been allowed to grow, critics contend, because there is no competitive pressure to keep it in check. Vouchers would bring that pressure to bear on public schools by making private schools a more affordable option. If students choose schools based on which delivers the best product, not based on the financial advantage of local public schools, then schools that are inefficient will not be chosen because they deliver less education per dollar of spending. If these schools are not chosen, they will be forced out of the education market, just as competition forces inefficient firms out of the market. Thus, competitive pressures will cause schools to serve the needs of students and parents rather than bureaucrats. Vouchers “level the competitive playing field” between private and public schools by removing the financial advantage currently held by public schools.

One response to this claim is to note that there is already competitive pressure on local schools through the Tiebout mechanism (voting with your feet to choose the right mix of property taxes and public goods provision for you). If local schools are inefficient, families will move to other towns where their property-tax dollars are spent more efficiently to produce better education for their children. Indeed, one study found that areas with more school districts from which parents can choose (e.g., many small suburbs) feature both better educational outcomes and lower school spending than do areas with fewer school districts (just several large suburbs).10 This finding is consistent with Tiebout pressures on schools to improve their educational productivity.

It is unlikely, however, that the Tiebout mechanism works perfectly in this case. Individuals choosing towns are choosing a bundle of attributes, not just educational quality. Vouchers allow for Tiebout unbundling. Individuals can live in a town they like for noneducation reasons while sending their children to any public or private school they like.

Problems with Educational Vouchers The consumer sovereignty and competition arguments provide strong support for the use of educational vouchers. However, a number of compelling argu- ments can be made against using education vouchers as a means to improve the quality of education in the United States.

Vouchers May Lead to Excessive School Specialization Many of the arguments in favor of public financing of education relate to the externalities

9 National Center for Education Statistics (2012). 10 Hoxby (2000). This important study has been the subject of some controversy; see Rothstein (2007) and Hoxby (2007).

316 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

that come from having a common educational program, particularly at the elementary school level. The first argument made here for vouchers, that schools will tailor themselves to meet individual tastes, threatens to undercut the benefits of a common program. In principle, a free educational market could produce “football schools,” with little educa- tional provision but excellent football programs, or “art schools,” with little education other than in the arts. By trying to attract particular market segments, schools could give less attention to what are viewed as the central elements of education (such as basic reading, writing, and mathematical skills).11

In principle, this problem could be dealt with through regulations that require all schools to pro-

vide a certain set of common skills. These regulations could also be supported by testing regimes to ensure that students at each school are maintaining an acceptable level of achievement in basic skills.

In practice, however, such regulation could become so onerous and costly to enforce that it would defeat the purpose of school choice. Moreover, as we discuss later, such efforts to hold schools accountable for student performance often have unintended side effects. Ultimately, what determines the optimal level of uniformity across schools is the value to society of educational conformity at each level of schooling. If this value is low, which may be true at the high school level, there will be large gains from free choice. If this value is high, which may be more true at the elementary school level, public provision may be more efficient than private competition with regulation.

Vouchers May Increase School Segregation A major achievement over the past 60 years in the United States, in the eyes of many citizens, is the reduction in segregation in education. A public education system that once provided African Americans and other minorities with separate and unequal educational quality has become more integrated, so in principle, all citizens have the right to high-quality public education. Critics of voucher systems argue that vouchers have the potential to reintroduce segregation along many dimensions, such as race, income, or child ability. These critics envision a world where children of motivated parents move to higher-quality private schools, while children of disinterested or uninformed parents end up in low-quality public schools. If the children of interested and motivated parents differ along the lines of race, income, or child ability from those of uninterested and unmotivated parents, segregation could worsen.

Supporters of vouchers note that, in fact, vouchers may serve to reduce the natural segregation that already exists in our educational system. Currently, students are trapped by the monopoly that their local school system has over

© T

he N

ew Y

or ke

r C ol

le ct

io n

20 02

C ha

rle s

Ba rs

ot ti

fro m

ca

rt oo

nb an

k. co

m . A

ll Ri

gh ts

R es

er ve

d.

“I’m taking my voucher and going to circus school.”

11 A real-world example of this issue arose recently in Texas, where some charter schools were found to be teaching Creationism and to be dismissing the science of evolution. See Koplin (2014) for more information.

E D U C A T I O N ■ C H A P T E R 1 1 317

education production. Vouchers allow motivated students and their parents to choose a better education and end the segregation imposed on them by location. In a famous example, Nobel Prize–winning economist Milton Friedman pointed out that it is unfair that an inner-city child who wants to use his money to buy a new car can do so, but if he wants to use that same money to buy better education, he cannot do so without giving up his public education subsidy.

Both sides of this argument make valid points. It is true that segregation remains a significant problem in the U.S. educational system. Although white students are only just over half of the student population, the typical white student attends a school that is 75% white. Thirty-eight percent of black students and 43% of Latino students attend intensely segregated schools, where 90 to 100% of students are from minorities. The typical black or Latino student today attends school with almost double the share of low-income students in their schools than the typical white or Asian student.12 California is among the states with the most segregated schools, with the typical black student attending a school that is 78% minority.13 Therefore, supporters of vouchers are undoubtedly correct in pointing out that some individuals would benefit from using vouchers to escape to higher-quality education.

At the same time, vouchers might increase segregation by student skill level or motivation. As the motivated and high-skilled students flee poor-quality public schools for higher-quality private schools, the students left behind will be in groups that are of lower motivation and skill. That is, school choice is likely to reduce segregation along some dimensions (e.g., by allowing minority students with greater ability and motivation to mix more with students at higher-quality schools) but increase it along others (e.g., by separating the education system into higher and lower ability/motivation schools).

Vouchers May Be an Inefficient and Inequitable Use of Public Resources One issue that was set aside in the theoretical discussion about the effects of vouchers was the financing of education and of vouchers. Education is financed mostly by local property taxes and state taxes (as discussed in the previous chapter). If the current financing were replaced by vouchers, total public-sector costs would rise because the government would pay a portion of the private school costs that students and their families are currently paying themselves.

For children from families such as X in Figures 11-2 and 11-3, costs would not increase. The children would stay in public school, so costs to the public sector for educating the children in family X would remain at EF. Children from families such as Y would move to private schools, thus spending more on education, but the cost to the public sector would still only be EF because this is the amount of the voucher that is provided; the families are paying the extra costs of the private school. For children such as those from family Z, however, public-sector costs would increase. Previously, family Z was paying the entire

12 Orfield et al. (2012). 13 Orfield and Lee (2006).

318 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

cost of their children’s private education, but now because family Z receives a voucher, the local government is picking up a portion, EF, of this cost. This increased government spending is associated with only a very small rise in their educational attainment, the rise from E3 to E5, which occurs because the families are richer by the amount of the government transfer (EF).

Thus, one cost of substantially increasing the level of education chosen by families such as Y is the cost incurred by providing large new subsidies to families such as Z who don’t much change their educational attainment. That is, crowd-out of private educational spending has been reduced for Y, but it has been introduced for Z. If vouchers are most used by families (like family Z) who were already paying for private school for their children on their own, then this is a fairly inefficient use of public resources. On the other hand, if the vouchers are most used by families (like family Y) who are switching from public education to much higher private levels of education (from EF in Figure 11-2 to E4 in Figure 11-3), this may be an efficient use of public resources. The goal of government policy here is to direct resources to the currently undereducated (such as the children in family Y); if most of the gain from the use of vouchers goes to families such as Z, the goal is not being met.

Equity considerations further strengthen this point. Income and use of private schools are strongly positively correlated; families like Z are much more likely to be high income than families such as Y. Granting much of the voucher expenditure to higher-income families who are already sending their children to private schools is an inefficient and inequitable use of public funds.

Ideally, the government could solve this problem by identifying whether families are in group Y or Z and directing more resources to those in group Y (whose use of more education we want to encourage). Unfortunately, the government cannot perfectly identify which group families are in, so it cannot carry out this type of targeting exactly. One way to approximate this targeting would be to target the voucher’s value to the family’s income. Having vouchers for which the value falls as the family’s income rises would accomplish three goals. First, such a program would target resources to groups who are most likely to use them to increase educational attainment. Second, it would reduce the inequity of a system that mostly benefits higher-income private school attendees. Finally, to the extent that lower-income children are “left behind” in public schools by their higher-ability and more motivated peers, it would provide resources for the remaining public schools to succeed (because income-targeted vouchers would provide higher levels of funding to low-income schools through their larger voucher amounts).

The Education Market May Not Be Competitive The arguments of voucher supporters are based on a perfectly competitive model of the educa- tion market. Yet the education market is described more closely by a model of natural monopoly, in which there are efficiency gains to having only one monopoly provider of the good. Economies of scale in the provision of edu- cation mean that it may not be efficient to have many small schools compet- ing with one another for students; it may be much more (naturally) efficient to have one monopoly provider instead.

E D U C A T I O N ■ C H A P T E R 1 1 319

The fact that education markets may be natural monopolies can lead to failures in the educational market. If a large inner-city school closes due to lack of demand, for example, what happens to its core of unmotivated students who have not taken advantage of choice to enroll elsewhere? There may not be a small school in the city that can meet their needs, and the closing of their school would potentially leave them without educational options. Similarly, how could a rural area without much population density support enough schooling options to effectively introduce competition?

Given these problems, it is unlikely that the government would actually allow certain schools to go out of business and leave local students without educational options. Yet if schools know that they are “too important to fail,” the competitive pressure on the schools would be mitigated: Why should a school work hard to improve its efficiency if it knows it will retain its funding regardless of performance? Thus, there is a tension between government efforts to ensure educational opportunities for all and the ability of the educational market to put pressure on underperforming schools.

The Costs of Special Education In the type of voucher system described here, each child would be worth a voucher amount that represents the average cost of educating a child in that town in that grade, but all children do not cost the same to educate. Children with diagnosed disabilities, for example, have much higher costs associated with their need for special education, and programs for educating disabled children require extra resources (such as trained teachers, smaller classes, or special equipment). In the United States, 6.43 million students aged 3–21 are provided with special education services, and the average student with a disability costs about $12,700 a year to educate, more than twice the cost of educating a student without a disability. The United States spends about $50 billion a year on special education, or 14% of total elementary and secondary education spending.14

The higher cost of special education students raises problems in the context of a voucher system because schools will have an incentive to avoid special education students. These students bring vouchers of the same amount, yet they cost much more to educate. Schools will want to take only the students who can be educated effectively for the voucher amount and will shun the highest-cost special education students. This student selection by schools will reduce the options available to special education students. In principle, the gov- ernment could use antidiscrimination regulations to deal with this problem, but, in practice, schools may have many subtle ways of deterring applications from such students. They might, for example, institute a very low-quality special education program that would deter special education students from applying.

The government could address this problem by making the voucher amount for any child match the cost of educating that child. Children with special education needs who cost more to educate could receive larger voucher amounts to offset the extra costs associated with educating them. Because it is very hard to adjust voucher amounts for the specific educational needs of each child, however, this potential problem with vouchers will remain.

special education Programs to educate disabled children.

14 New America Foundation (2014).

320 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

11.3 Evidence on Competition in Education Markets

In the previous section, we discussed the theory of how vouchers may or may not improve the efficiency with which education markets function in the United States. There is substantial uncertainty about the ultimate effects of vouchers. In this section, we review the evidence on the effects of competi- tion in education markets in an effort to understand what impact the wide- spread use of vouchers might have in the United States and other nations.

Direct Experience with Vouchers Several small-scale voucher programs have been put in place in the United States in recent years. Probably the most studied program has been the one used in Milwaukee. Starting in 1990, the state of Wisconsin allowed families with income no more than 175% of the poverty line to apply for a voucher worth about $3,200 that could be used for tuition at any nonsectarian (not religiously affiliated) private school. Studies of this program, reviewed in the Empirical Evidence box, provide some support for the notion that vouch- ers can allow students to improve the quality of their education. Studies of larger-scale voucher programs in other nations have suggested even stronger effects. The effects might be much larger with widespread adoption of vouch- ers, which would put competitive pressure on all schools to improve their performance. There is a growing body of evidence from around the world that suggests that school choice and vouchers that encourage private school enroll- ment do lead to better public school performance.15

Experience with Public School Choice Some school districts have not offered vouchers for private schools but have instead allowed students to choose freely among public schools. In some cases, students are allowed to choose any local school, not just the one nearest them. Evidence on these programs comes from the fact that the best schools are oversubscribed when there is choice, so they must use a lottery to allocate slots. This process allows researchers to assess differences in outcomes between comparable lottery winners (who attend better schools) and lottery losers (who are otherwise identical but do not get to attend such schools). Results from these programs are mixed, suggesting that the effects may vary depending on the nature of the local public school geography and quality differences.16

Other possible choices for public school include magnet schools, special public schools set up to attract talented students or students interested in a particular subject or teaching style, and charter schools, small, independent

magnet schools Special public schools set up to attract talented students or students interested in a particular subject or teaching style.

15 See, for example, Figlio and Hart (2014) for the United States; Chan and McMillan (2009) for Canada; and Neilson (2013) for Chile. For opposing views, see Card, Dooley, and Payne (2008) for Canada; and Hsieh and Urquiola (2003) for Chile. 16 See Cullen, Berry, Jacob, and Levitt (2003) and Deming et al. (2014).

E D U C A T I O N ■ C H A P T E R 1 1 321

public schools that are not subject to many of the regulations imposed on traditional public schools, including restrictions on teacher qualifications. A larger literature has emerged suggesting that charter schools in particular may be effective at improving student performance in both New York City (Dobbie and Fryer, 2013; Hoxby and Murarka, 2009) and Massachusetts (Angrist et al., 2013).

Estimating the Effects of Voucher Programs

A number of studies, both in the United States and abroad, have attempted to estimate the impact of voucher programs on student achievement. In the United States, Rouse (1998) studied the effect of the Milwaukee voucher program on the achievement of students who used their vouchers to finance a move to private schools.A She noted that one cannot directly compare students who do and do not use vouchers because they may differ along many dimensions; for exam- ple, students who take advantage of a voucher program may be more motivated than those who do not. This selective use of vouchers would bias any comparison between the groups. An important feature of the Milwaukee program, however, is that participating private schools had to accept all students who applied unless the school was oversubscribed (too many applicants for the available slots). Oversubscribed schools had to select randomly from all applicants, using a lottery.

This administrative solution has the benefit of approximat- ing the type of randomized trial that is the gold standard in empirical research. The randomized lottery allowed Rouse to form a control group (students who applied to oversub- scribed schools but were randomly rejected) and a treatment group (students who applied to the same schools and were randomly accepted). These groups should be comparable, except that the treatments go to the private schools rather than remaining in the public schools like the controls. Rouse found that the treatment group saw an increase in academic performance: there was a rise in math test scores of 1 to 2% per year relative to the control group, although there was no difference in reading scores across the two groups.B

In the United States, about 10% of students are enrolled in private schools, a proportion that doubles or triples in the low-income developing world, where public schools may be of particularly low quality. Introducing a voucher program may,

therefore, have a great effect in developing countries, where private schools are a closer substitute for public schools, than in developed countries. Angrist et al. (2002) studied a Colombian voucher program called PACES that gave more than 125,000 pupils vouchers that covered somewhat more than half the costs of private secondary school. Many of the vouchers were distributed by lottery, thus allowing Angrist and colleagues to compare the randomly selected lottery win- ners (the treatment group that received vouchers) and losers (the control group that did not receive vouchers).

The study found that students who won vouchers were 10% more likely than lottery losers to finish eighth grade, primarily because they didn’t repeat as many grades before the age of school leaving. The study also found that lottery winners scored significantly higher on standardized achieve- ment tests than did losers. Winners were also less likely to be married or cohabiting and worked 1.2 fewer hours per week, suggesting an increased focus on schooling among lottery winners. The study concluded that the vouchers cost the government $24 per winner, yet the improved schooling attainment and quality increased the wages earned by this group by between $36 and $300 per year, making this an enormously successful program.

Recent results from other nations support this view. Neilson (2013) studied the effects of a widespread voucher program in Chile, and also found very large positive effects for poor children who had access to this program; he esti- mated that school vouchers reduced the achievement gap between poor and non-poor students by about one-third. Muralidharan and Sundararaman (2013) studied the lottery- based allocation of students to private schools in India and found modest improvements in test scores at much lower educational cost.

EMPIRICAL E V I D E N C E

A Her work builds on earlier conflicting analyses by Greene et al. (1996) and Witte (1997). B The other major experience with vouchers in the United States has been several privately financed “experiments” in which low- income public school children were given scholarships to attend private schools. Recent studies of such a program in New York have provided only mixed evidence on its success, however, in contrast to the positive evidence in Milwaukee.

RO BE

RT D

O DG

E/ E+

/G ET

TY IM

AG ES

322 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

Experience with Public School Incentives Although the United States has limited experience with vouchers and school choice, it has a much larger experience with another aspect of educational reform: school accountability. Any move to an increase in school choice in the United States would bring with it an increased use of testing to ensure that schools are meeting educational standards, and the country has a large body of experience with accountability measures. As of 2002, 25 states explicitly linked student promotion or graduation to performance on state or local assessment tests, 18 states rewarded teachers and administrators on the basis of successful student performance on exams, and 20 states penalized teachers and administrators on the basis of subpar student exam performance.17 This approach to school accountability was codified in federal law through the No Child Left Behind Act of 2001.

Making schools accountable for student performance can provide incentives for schools to increase the quality of the education they offer. By some measures, accountability requirements have had this intended effect. Several studies have found that implementing strong accountability programs have led to improvements in test scores.18 Deming et al. (2014) used a long- run follow-up of Texas students to show that students in schools at risk of facing sanctions for low performance were more likely to have attended and completed college and to have obtained a four-year degree.

At the same time, accountability programs can have two unintended effects. First, they can lead schools and teachers to “teach to the test”— that is, to narrowly focus their teaching on enabling students to perform well on the test that determines school accountability, not on a broadly improved education. Jacob (2005) and Dee and Jacob (2011) found evidence of students performing better on accountability tests per se than on broader measures of student ability. More generally, Reback et al. (2014) found that schools distort their resources toward specialists and targeted teachers and away from whole-class instruction, leading to less satisfied teachers, but still resulting in higher test scores, even on more general examinations.

Second, schools can manipulate the pool of test takers and the conditions under which they take tests to maximize success. For example, Jacob (2005) and Figlio and Getzler (2002) found that the introduction of accountability in Chicago and Florida led schools to reclassify low-skilled students as special education or disabled students (and thus exempt from testing) to raise average school scores. Figlio and Winicki (2005) found that schools even manipulated their cafeteria menus around testing time, increasing calories to improve student energy levels and test scores! Teachers may even cheat to improve test scores; Jacob and Levitt (2003) found that a teacher is more likely to provide the answers to standardized tests to students if the teacher has more at stake (through accountability regimes).

17 Hanushek and Raymond (2004). 18 Hanushek and Raymond (2004); Dee and Jacob (2010); Ahn and Vigdor (2014).

E D U C A T I O N ■ C H A P T E R 1 1 323

Bottom Line on Vouchers and School Choice While the evidence reviewed in this section is mixed, it suggests that more school accountability, more school choice (in particular, through charter schools), and more use of vouchers may improve student outcomes. Yet voucher systems raise serious concerns about equitable treatment of the “worst” students, who might get left behind as their higher-ability, higher- motivation friends move on to better schools. Some sort of guarantee of educational access must be provided to ensure that every student has the option of at least one educational alternative, even if this reduces the pressure of competition on schools that will not be allowed to fail.

11.4 Measuring the Returns to Education

Regardless of the use of public education or private education, the government must still make some decision about the share of its budget to devote to education. For the government to decide how much to invest in education, it must undertake the type of cost-benefit analysis discussed in Chapter 8. Measuring the costs associated with education is fairly straightforward by using the techniques of opportunity cost introduced in Chapter 8. Measuring the benefits, however, is much trickier. There is an enormous economics literature devoted to measuring the returns to education, that is, the benefits that accrue to society when individuals get more schooling or when they get schooling from a higher-quality environment (such as one with better- qualified teachers or smaller class sizes).

Effects of Education Levels on Productivity The topic that has received the most attention from economists studying education is the effect of education on worker productivity. In a competitive labor market, workers’ wages equal their marginal product, so wages are typically used as a proxy for productivity. The idea of these studies is to let the market reveal whether education has raised productivity: if individuals are more productive as a result of being more highly educated, then firms should be willing to pay more to employ them.

There is a large literature that shows that more education leads to higher wages in the labor market. A typical estimate, which comes from comparing the earnings of those with more and less education, is that each year of education raises earnings by about 7%. There is little controversy over the question of whether those with more education earn more. There is substantial controversy, however, over the implications of this correlation. Two very different interpretations have been offered for this result.

Education as Human Capital Accumulation The typical view of education is that it raises productivity by improving worker skills. Just as firms invest in physical capital, education is the individual’s means of investing in human capital. More education raises a worker’s stock of skills and allows the individual to earn more in the labor market.

returns to education The benefits that accrue to society when students get more schooling or when they get schooling from a higher-quality environment.

human capital A person’s stock of skills, which may be increased by further education.

324 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

Education as a Screening Device An alternative view is also consistent with the correlation between higher levels of education and higher levels of earnings. In the screening model, education acts only to provide a means of separating high-ability people from low-ability people and does not actually improve skills. In this model, more highly educated workers would be more productive and have higher wages, but it would not be because education has improved their human capital. Rather, it would be because only those who turn out to be the most productive workers have the ability to pursue higher levels of education, so the very fact of having more education has signaled their high ability (and productivity). The school system in this model is not adding any value in terms of raising productivity; its only value is in screening for the most able and productive workers, who can obtain the most education.

Thus, in the screening model, employers pay more to more highly educated workers not because education has raised their productivity, but because education is serving as a signal of underlying motivation by screening out unmotivated workers. In the human capital model, more educated workers earn more because education has raised their marginal product; in the screening model, more educated workers earn more because their education has signaled high ability.

Policy Implications The human capital and screening models may have the same prediction for the correlation between wages and education, but they result in very different recommendations for government policy. Under the human capital model, government would want to support education or at least provide loans to individuals so that they can get more education and raise their productivity. Under the screening model, however, the government would not want to support more education for any given individual. In this model,

the returns to education are purely private, not social. Higher education serves as a signal that a person is more productive, but it does not improve social productivity at all. In fact, by getting more education, a given worker exerts a negative externality on all other educated workers by lowering the value of their education in the labor market. In the following cartoon, the King’s declaration would lower the signaling ability of a degree because all of the productive workers who worked hard to actually earn a degree would suffer when unproductive workers were able to raise their education level.

At the same time, education does play a valuable social role as a screening device in the screening model, allowing the labor market to recognize and reward the most able workers. Thus, the appropriate government policy in this model would be to support the establishment of educational institutions, if they are the best screening device, but not to subsidize an individual to get the education because this has no social return and simply lowers the value of education to others.

screening A model that sug- gests that education provides only a means of separating high-ability individuals from low-ability individuals and does not actually improve skills.

“It is my wish that this be the most educated country in the world, and toward that end I hereby ordain that each and every one of

my people be given a diploma.”

© T

he N

ew Y

or ke

r C ol

le ct

io n

19 72

J .B

. H an

de ls

m an

fr om

ca

rt oo

nb an

k. co

m . A

ll Ri

gh ts

R es

er ve

d.

E D U C A T I O N ■ C H A P T E R 1 1 325

Differentiating the Theories While these theories have radically dif- ferent policy prescriptions, in practice, it is hard to tell the theories apart. An enormous literature in labor economics has proposed a wide variety of approaches to differentiating the theories, and the conclusion is very clear: most of the returns to education reflect accumulation of human capital, although there may be some screening value to obtaining a high school or higher education degree. The details of these studies are reviewed in the Empirical Evidence box.

Effect of Education Levels on Other Outcomes As discussed earlier in this chapter, a major motivation for government inter- vention in education is the externality generated by more education. In recent years, a number of studies have assessed the impact of increased education on external benefits. Key findings include the following:19

■ Higher levels of education are associated with an increased likelihood of participation in the political process and more awareness of current policy debates (Milligan et al., 2004; Dee, 2004).

■ Higher levels of education are associated with a lower likelihood of criminal activity (Lochner and Moretti, 2004).

■ Higher levels of education are associated with improved health of the people who received more education and of their children (Currie and Moretti, 2004; Lleras-Muney, 2005; Chou et al., 2010).20

■ Higher levels of education of parents are associated with higher levels of education of their children (Oreopoulos et al., 2003).

■ Higher levels of education among workers are associated with higher rates of productivity of their coworkers (Moretti, 2004).

■ Higher levels of education lead to lower likelihood of engaging in risky behaviors such as smoking and heavy drinking ( Jensen and Lleras-Muney, 2012).

These findings, along with the findings that more education results in higher wages, suggest that there are large private and public returns to increasing human capital through increasing years of education.

The Impact of School Quality A smaller but growing literature has investigated a different question: What is the impact of higher-quality schools on the returns to education? This literature must initially grapple with the question of how to define school quality. The most common measures used are average class size (the ratio of students to teachers within a school) and school spending per student.

19 For a detailed review of this literature, see Oreopoulos and Salvanes (2009). 20 A recent paper by Clark and Royer (2010) using evidence from the United Kingdom questions the conclusion of Lleras-Muney (2005) and others that more education leads to lower mortality rates.

326 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

Estimating the Return to Education

A simple approach to estimating the return to a year of educa- tion in terms of higher wages is to compare people with more education (the treatment group) to people with less education (the control group), but this approach suffers from the type of bias problems discussed in Chapter 3: people who obtain more education may be of higher ability than people who obtain less. Thus, the estimated difference in wages between these groups can arise either from human capital accumula- tion or from the underlying ability differences in the groups.

Two methods try to control for this bias in estimating the true human capital effects of education. The first tries to con- trol directly for underlying ability in a wage regression so that any remaining effect of education represents true productivity effects. Researchers include, for example, standardized test scores of students as youths to try to control for their abil- ity. The problem with this approach is that this crude mea- sure of the differences between individuals does not take into account unobserved factors such as motivation (e.g., Dick can be less intelligent than Jane, but because he studies harder, he is still of higher ability).

The other approach to control for bias in estimating the human capital returns to education has been quasi- experimental studies that try to find treatment and control groups that are identical except for the amount of schooling they receive. One quasi-experimental approach was taken by Duflo (2004), who studied the impact of a large-scale public school construction project in Indonesia. Between 1973 and 1978, more than 61,000 new primary schools were opened in Indonesia, with more schools in some areas than in others. Duflo studied students who were of primary schooling age when schools were built. The treatment group of students lived in areas with more school construction; the control group of students lived in areas with less school construction. Because Duflo was worried that these areas might differ for other reasons, she also contrasted the young people in each area with older people in those same areas who were

educated before the school construction project to remove the effects of any differences between regions that persisted over time. She found that education rose in areas where schools were constructed much more than in areas where they were not. Years later, she found that the adult wages of people young enough to have benefited from the new schools were higher relative to their older counterparts who did not benefit from school construction in the treatment areas compared to the control areas. This study uses the “difference-in-difference” strategy discussed in Chapter 3 to show that there was a true productivity gain from “increased education.”

Another example is the use of the quasi-experiment pro- vided by the passage of mandatory schooling and child labor laws in the United States in the late 1800s and early 1900s. Before this time, there was no requirement that children attend school and no limit on child labor. These laws set up the minimum age at which children had to start school, the minimum age at which they could drop out, and the minimum number of years of education required before children could engage in full-time work. Studies have shown that mandatory schooling and child labor laws significantly increased the level of education attained by students in the United States. These studies compare individuals born in states where schooling/ child labor laws changed to require more education (the treatment group) to those born in states where laws did not change (the control group). Once again, these groups were of similar ability other than the laws that affected their manda- tory level of schooling. Later in life, however, the people who received more (mandatory) education had higher wages than the control group, showing once again that more education raised productivity relative to another group with the same level of ability.

Although all of these approaches have some limitations, the result of the analysis is surprisingly consistent: each year of education raises wages by 7 to 10%. This is strong evidence for the human capital model of educational attainment.A

EMPIRICAL E V I D E N C E

A There is mixed evidence for one particular type of screening, often called the “sheepskin effect”: getting a degree from high school, college, or graduate school. Some studies (Tyler, Murnane, and Willet, 2000; Jaeger and Page, 1996) find that students earning a degree relative to students with similar test scores and years of schooling earn 10 to 25% more. But more recent analysis by Martorell and Clark (2010) compares students who just barely pass high school completion exams to those who just barely fail and finds that obtaining the high school degree itself has little impact on subsequent earnings.

RO BE

RT D

O DG

E/ E+

/G ET

TY IM

AG ES

E D U C A T I O N ■ C H A P T E R 1 1 327

As reviewed in the Empirical Evidence box, a number of approaches have been taken to estimate the impact of school quality on student test scores. Experimental evidence from Tennessee suggests that smaller class sizes lead to much higher student test scores. Yet an attempt to dramatically reduce class sizes in California did not have the expected positive effects, perhaps because the associated rapid rise in the number of classes required led the state to hire under- qualified teachers. These findings suggest that the outcomes of efforts to improve school quality can be very dependent on the approach taken to improvements.

Recent studies of effective charter schools have suggested looking beyond measures such as class size. In particular, analyses of what makes charter schools most effective suggest that longer school days and a longer school year, teachers who receive frequent feedback, and instruction driven by data on student performance are the keys to more successful schools. Fryer (2011) discusses recent attempts to take these lessons to traditional public schools in Houston with promising first-year results. Other recent studies have suggested additional ways to improve educational quality. For example, Collins and Gan (2013) studied variation across schools in student “tracking” and found that schools that sort students more aggressively improve the test scores of both high- and low-performing students.

11.5 The Role of the Government in Higher Education

The focus of our discussion thus far has primarily been on elementary and secondary education, yet there is an enormous higher education sector in the United States, which comprises 198 universities, 2,968 four-year colleges, and 1,738 two-year degree-granting institutions. Institutions of higher education spend about $496 billion per year, almost 43% of total educational spending.21 Interestingly, in contrast to other levels of education, the higher education system in the United States is viewed as an enormous success. U.S. research universities are consistently rated as the best in the world. The clear market evi- dence for the success of higher education in the United States is the vast inflow of foreign students to U.S. institutions of higher education: 886,052 foreign students each year spend more than $22.1 billion to enroll in American col- leges and universities. The number of foreign students studying here has risen by 161% in the past 20 years and now represents 4.2% of all higher-education enrollment in the United States. This compares to only 8,862 American stu- dents who are studying abroad for more than one semester in any given year.22

The major difference between higher education and primary/secondary education in the United States is the degree of private provision and competition. Only 10% of students are enrolled in private elementary/ secondary schools, and public schools typically have a local monopoly.23 In higher education, 28% of students attend private institutions, and students

21 National Center for Education Statistics (2015). 22 Institute of International Education (2015). 23 National Center for Education Statistics (2015).

328 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

Estimating the Effects of School Quality

A major focus of research in labor economics is estimating the impact of school quality on student outcomes. Studies in this area have recognized that we cannot simply compare school districts with better and worse schools and look at the result- ing implications for students. Districts with better schools (the treatments) differ in many ways from districts with worse schools (the controls). For example, residents in the treatment districts are likely to be the ones who provide a better home environment for their children. Therefore, it is necessary to find an approach that allows researchers to identify the effects of school quality alone on educational outcomes.

Two approaches have been used to address this issue. The first is using experimental data. The state of Tennes- see implemented Project STAR in 1985–1986, randomly assigning 11,000 students (grades K–3) to small classes (13–17 students), regular classes (22–25 students), or regular classes with teacher’s aides. Krueger (1999) analyzed the data from this experiment and found that there was a large improvement on standardized test scores for the first year and a slight improvement for each year thereafter in a small class. These effects were largest for poor and minority students. Krueger and Whitmore (2001) found that small class size effects persisted later in life; that is, being in a small class for those four years increased test scores in middle school and

increased the likelihood of taking a college entrance exam. Overall, their estimates imply that the real rate of return to smaller class sizes (doing a standard cost-benefit analysis of the experiment) is roughly 5.5% per year.

The other approach is a quasi-experimental analysis of changes in school resources. An interesting example is California, which by the mid-1990s had the largest class sizes in the nation (29 students per class on average). The California state government in 1996 provided strong financial incentives for schools to reduce their class size to 20 students per class in grades K–3, at a cost of more than $1 billion per year. Bohrnstedt and Stecher (2002) reviewed the evidence on the impacts of this major reform, using variation across schools in the rate at which they implemented smaller class sizes. Schools that implemented smaller class sizes quickly were the treatment group, while schools that went more slowly were the controls. They found that there was little beneficial impact of smaller classes on student outcomes, perhaps because the state hired underqualified teachers to fill the extra classes or perhaps because the state was forced into educa- tionally unproductive approaches such as combining different grades in one class. Thus, there remains some controversy about the returns to increased public-sector investments in school inputs.

EMPIRICAL E V I D E N C E

RO BE

RT D

O DG

E/ E+

/G ET

TY IM

AG ES

have free choice over the entire nation of where to go to college.24 The relative success of higher education, where the United States is the world leader, and primary/secondary education, where the United States performs relatively poorly, provides some evidence for the power of competition to improve educational performance. As noted in our discussion of privatization, even with a minority of students enrolled in private schools, the competition from the private schools can lead to efficiency in the public sector.

Current Government Role The U.S. government currently intervenes in the higher education sector through four channels.

State Provision The primary form of government financing of higher education is direct provision of higher education through locally and state- supported colleges and universities. These institutions offer subsidized low tuition for in-state students and somewhat less subsidized costs for out-of-state

24 National Center for Education Statistics (2015).

E D U C A T I O N ■ C H A P T E R 1 1 329

■ ■ ■ ■ FIGURE 11-4

State and local funding for colleges

and universities ($252 billion)

Pell Grants ($32 billion) Tax breaks ($19.5 billion)

Government Spending on Higher Of the roughly $304 billion

the government spends annually on higher education, 83% is in the form of state and local funding for colleges and universities. The remainder is split among Pell Grants and tax breaks; student loans, while large in volume, are not a net cost to the government.

Data from: National Center for Educational Statistics (2015); U.S. Department of Education (2014); Office of Management and Budget (2012).

students. Currently, state and local governments spend about $242 billion per year on their institutions of higher education.

Pell Grants The Pell Grant program is a subsidy to higher education administered by the federal government that provides grants to low-income families to pay for their educational expenditures. For a student from a family with annual income below $15,000, the Pell Grant program provides a grant of $5,775. For a some- what higher-income student, the grant amount is reduced according to parental income and assets and student income and assets. The Pell Grant program cur- rently provides $32 billion per year in grants to about 9 million students.

Loans The federal government also makes loans available to students for high- er education expenditures through the direct student loan program. For stu- dents who qualify on income and asset grounds, the government subsidizes the loan cost to students by (1) guaranteeing a low interest rate (the 2014–2015 rate for the 10-year loan was 4.66%, compared to 15-year home mortgage rates, the cost to the private sector of borrowing, of about 3.3% over that period) and (2) allowing students to defer repayment of the loan until they have graduated. Stu- dents who do not qualify can still receive loans at the same low interest rate but must start repaying them immediately rather than deferring them until their education is complete. A dependent undergraduate can borrow up to $31,000 ($23,000 in subsidized loans) per degree program, an independent undergradu- ate can borrow up to $57,500 ($23,000 in subsidized loans), and a graduate or professional student up to $138,500 ($65,500 subsidized). The total amount of loans made each year under the direct loan program is $104 billion.25

direct student loans Loans taken directly from the U.S. Department of Education.

25 Congressional Budget Office (2015), Table 2, http://www.cbo.gov/publication/43054. This is in sharp contrast to a long-standing previous program, the Guaranteed Student Loan Program, under which the federal government subsidized private banks to make student loans; this program was, on net, a large cost to the government before ending in 2010.

330 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

Tax Relief The final way in which the government finances higher education is through a series of tax breaks for college-goers and their families. The largest of these are the Lifetime Learning Tax Credit (LLTC), put into place in 1998, and the American Opportunity Tax Credit (AOTC), an expansion of the HOPE credit since 2009. The LLTC provides tax credits to lower- and middle-income families of up to $2,000 per year per person for the costs of higher education, and the AOTC provides a credit of up to $2,500. Alternatively, individuals can deduct from their taxable income up to $4,000 per year in higher education expenses. Interest paid on student loans is also tax deductible, as is some scholar- ship and fellowship income, and there are tax-free savings accounts for higher education as well. These tax breaks add up to about $19.5 billion per year in forgone government revenue.

What Is the Market Failure, and How Should It Be Addressed? The arguments discussed earlier to motivate public intervention in education markets, such as provision of a common set of values, apply much less strongly in the context of higher education, where a larger share of the returns are private. Some of the recent studies cited show public returns to college education (in terms of improved health or productivity spillovers), but these benefits have not yet been shown to be large relative to government expenditures on higher education.

The major motivation for government intervention in higher education is not to produce positive externalities but rather to correct the failure in the credit market for student loans. As noted at the start of this chapter, it is much harder to get a loan to finance education than it is to obtain a loan to finance the purchase of a car or a home because there is no collateral for banks to repossess if the loan is not repaid. As a result, in the absence of government intervention, banks may be unwilling to loan money to finance higher education. Government intervention is motivated by the need to ensure credit to students for higher education so that they can obtain higher education if it is productive for them to do so.

The major source of government expenditure on higher education is not through loans (see Figure 11-4), however, and the rationale for other types of government intervention is less clear. Indeed, work by Marx and Turner (2015) suggested that Pell Grants largely crowd out borrowing for college so that on net there is little increase in funding available to pay for education.

At the same time, other studies have found that having grants rather than loans can affect the nature of education. For example, Field (2006) studied a program at New York University law school where admitted students were randomly assigned either loans or grants of the same financial value. Students who were randomly assigned grants were twice as likely to enroll at that university as those assigned loans, and students assigned grants were also about 40% more likely to take low-paying public interest jobs after graduation rather than higher-paid private-sector legal work. Another study of a selective

E D U C A T I O N ■ C H A P T E R 1 1 331

undergraduate institution that replaced loans with grants found that students were more likely to turn down higher-salaried jobs in favor of low-paid “public interest” jobs (Rothstein and Rouse, 2007). These findings suggest that moving from grants to loans can have real effects on behavior.

At the same time, there is no real rationale for providing subsidies only to the higher-income individuals who benefit from tax deductions. Indeed, Bulman and Hoxby (2015) find little effect of these tax credits on actual college attendance. And there is an uncertain case for the state education provision, which, as Figure 11-4 shows, is by far the largest source of public spending on higher education. Presumably, states provide public education to improve the skill level of their workforce. This goal is undercut, however, by the mobility of college graduates to other states. A study by Bound et al. (2004) found that for every ten students educated by state schools, only three of the state school graduates remain in the state in the long run. Given that the major market failure for higher education is in credit markets, shifting state resources away from direct provision and toward loans may improve efficiency.

On the other hand, large cuts in public education financing over the past decade have led to an ever-growing burden of student interest debt, which has recently surpassed $1 trillion. In 2015, President Obama proposed a number of initiatives to try to address this problem, including reducing federal aid and loan availability for schools that do not deliver educational “value” to their students, defined as better career outcomes at more affordable tuition prices. The President also proposed limiting student loan payments to 10% of income and forgiving debt after 20 years of repayment. While this could assist former students with crushing loan costs, it also raises government spending on the student loan program.26

11.6 Conclusion

The provision of education, an impure public good, is one of the most important governmental functions in the United States and around the world. Because of external returns, market failures, or redistribution, governments have traditionally decided to be the majority providers of educational services. In this chapter, we learned that one cost of the government role can be a reduction in the level of educational attainment of children. Voucher systems can address this problem, but they raise a host of additional issues about segregation and the feasibility of private educational markets. The optimal amount of government intervention in education markets depends on the extent of market failures in private provision of education and on the public returns to education. A large literature suggests sizeable private returns to education, with some evidence of public returns as well.

26 Matthews (2013).

332 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

e

Q U E S T I O N S A N D P R O B L E M S

1. State and federal governments actively support education at the primary, secondary, and collegiate levels. But they mandate education at the primary and secondary levels, while merely providing subsidies and loan guarantees at the collegiate level. Of the key rationales for public provision of education described in Section 11.1 of the text, which do you think underpins this differential treatment?

2. Consider two metropolitan areas, one that has many small school districts and one that has only a few large school districts. How are the efficiency and equity effects of introducing a voucher system likely to differ across these two areas?

3. Suppose that a family with one child has $20,000 per year to spend on private goods and education, and further suppose that all education is privately provided. Draw this family’s budget constraint. Suppose now that an option of free public education with spending of $4,000 per pupil is introduced to this family. Draw three different indifference curves corresponding to the following three situations: (a) a free public education would increase the amount of money that is spent on the child’s education; (b) a free

public education would decrease the amount of money that is spent on the child’s education; (c) a free public education would not affect the amount of money spent on the child’s education.

4. Empirical evidence suggests that better-educated adults donate more to charity than do less- educated adults with similar income levels. Why might this evidence justify public subsidization of education? What potential biases may make it difficult to interpret this empirical relationship?

5. Some have argued that introducing a voucher system would be particularly good for two groups of students: those who are the worst off under the current system, and many of the students who are the best off under the current system. Why might this be the case?

6. Several researchers have found evidence of sheepskin effects in which the labor market return to twelfth grade is higher than the return to eleventh grade and the return to the fourth year of college is higher than the return to the third year of college. Why does this evidence of sheepskin effects bolster the screening explanation for the relationship between education and earnings?

H I G H L I G H T S

■ Education is primarily provided by state and local governments in the United States, and only a small share of students go to private schools.

■ The rationales for public intervention in educa- tion include positive externalities, failures in credit markets, failures of family utility maximization, and redistribution.

■ Publicly provided free education may crowd out the educational attainment of those who want to choose higher levels of education but don’t want to forgo the free public good.

■ Vouchers might solve this crowd-out problem by allowing people to choose the optimal level of education for themselves, as well as interjecting competition into the education market.

■ At the same time, vouchers may lead to increased educational stratification, and the education

market may face difficulties in implementing competition.

■ Existing evidence suggests that private school choice through vouchers can move students to better schools, but a much richer evaluation of the total social effects of vouchers is needed before policy conclusions can be drawn.

■ There is a sizeable private return to additional schooling that appears to reflect increased human capital accumulation rather than screening. There is also some evidence of public returns in terms of outcomes such as increased voting and better health.

■ The government supports higher education through direct spending, grants, loans, and tax breaks. The rationale for nonloan interventions, and particularly for public universities, is unclear.

E D U C A T I O N ■ C H A P T E R 1 1 333

e e e 7. What are the advantages of comparing twins to

investigate the relationship between education and earnings? What are the drawbacks of doing so?

8. Suppose you want to evaluate the effectiveness of vouchers in improving educational attainment by offering a voucher to any student in a particular town who asks for one. What is wrong with simply comparing the educational performance of the students receiving vouchers with those who do not receive vouchers? What would be a better way to study the effectiveness of vouchers?

9. Seven in ten students attending publicly fund- ed universities leave the state after graduation, indicating that a very large fraction of states’ investments in human capital bears fruit else- where. Why, then, might states still play such a large role in higher education financing?

10. The U.S. Department of Education regularly con- ducts the National Assessment of Educational Prog- ress (also known as the “Nation’s Report Card”) to monitor student achievement in subjects such as reading, writing, and mathematics. Visit their data website at www.nces.ed.gov/nationsreportcard /naepdata/ (or start at the main website at http:// www.nces.ed.gov/nationsreportcard/). Compare the progress from 1998 to 2003 of students in your state with the progress of students in one other state. In which subject areas or grade levels has your state compared most favorably with the other state?

The e icon indicates a question that requires students to apply the empirical economics principles discussed in Chapter 3 and the Empirical Evidence boxes.

A D VA N C E D Q U E S T I O N S

11. Many state constitutions explicitly require that the state provide an “adequate” level of school fund- ing. How might raising this level of “adequacy” actually lead to reduced overall levels of educa- tional spending?

12. Epple and Romano (2002) describe theoretical evidence that school vouchers will lead to “cream-skimming,” where private schools will pick off the better students and leave public schools with lower-ability average students. They propose targeted vouchers, in which different- sized vouchers go to different groups of students, to combat this potential concern. How would you design a targeted voucher system that would lead to a reduced level of cream-skimming?

13. The town of Greenville has three families, each with one child, and each of which earns $20,000 per year (pre-tax). Each family is taxed $4,000 per year to finance the public school system in the town, which any family can then freely attend. Education spending is $6,000 per student in the public schools. The three families differ in their preferences for education. Though families A and B both send their children to the public school, family B places a greater value on education than family A. Family C places the

greatest relative value on education and sends their child to private school.

a. Graph the budget constraints facing each of the three families, and draw a possible indifference curve that could correspond to the choice each family makes. The town is considering replacing its current system with a voucher system. Under the new system, each family would receive a $6,000 voucher for education, and families would still be able to send their children to the same public school. Since this would be more costly than the current system, they would also raise taxes to $6,000 per household to pay for it.

b. Draw the budget constraint the families would face under this system. Suppose that when the new system is introduced, family A continues to send their child to public school, but family B now sends their child to private school (along with family C’s child).

c. Explain how you know that family C is made better off and family A is made worse off by the voucher policy.

d. Show, using diagrams, that family B could be made better or worse off by the voucher policy.

334 P A R T 2 ■ E X T E R N A L I T I E S A N D P U B L I C G O O D S

e14. Lazear (2001) noted that when one simply compares the performance of students in small and large class sizes, there is little difference, despite the presumption (and experimental evidence) that smaller class sizes improve perfor- mance. He argued that one reason for the lack of an observed relationship between class size and student outcomes is that schools may put more disruptive children in smaller classes. How would

this practice bias the estimated effect of class size on student outcomes?

15. One way to structure a student loan repayment plan is to make it income-contingent—that is, to relate the amount that a student would have to repay in any given month to how much income he or she earns. How might the existence of such a plan alter a student’s choice of college major?

Order your essay today and save 10% with the discount code ESSAYHELP