The Social Benefits of Higher Education David Bloom, Matthew Hartley, and Henry Rosovsky March 4, 2004.

The Social Benefits of Higher Education

David Bloom, Matthew Hartley, and Henry Rosovsky

March 4, 2004

State Spending Per Higher Education Student(in real 2002 dollars)

0

1000

2000

3000

4000

5000

1994 1995 1996 1997 1998 1999 2000 2001 2002

Fiscal year beginning

Am

ou

nt

spen

t (2

002

$)

California Massachusetts Florida Michigan

Source: Websites of Chronicle of Higher Education and National Center for Education Statistics; student numbers in 1994, 2001, and 2002 are estimates.

The Array of Higher Education Benefits Public Private

Economic

Social

• Increased Tax Revenues

• Greater Productivity

• Increased Consumption

• Increased Workforce Flexibility

• Decreased Reliance on

Government Financial Support

• Higher Salaries and Benefits

• Employment

• Higher Savings Levels

• Improved Working Conditions

• Personal/Professional Mobility

• Reduced Crime Rates

• Increased Charitable

Giving/Community Service

• Increased Quality of Civic Life

• Social Cohesion/Appreciation of

Diversity

• Improved Ability to Adapt to and

Use Technology

• Improved Health/Life Expectancy

• Improved Quality of Life for

Offspring

• Better Consumer Decision Making

• Increased Personal Status

• More Hobbies, Leisure Activities

Source: The Institute for Higher Education Policy, “Reaping the Benefits: Defining the Public and Private Value of Going to College”, March 1998.

Rate of Return

• A formal way to compare the immediate costs and the subsequent benefits of investment in schooling.

• The costs include tuition and fees and the income forgone while in school.

• The benefits include the higher earnings that individuals expect to earn as a result of schooling.

Private vs. social, in practice

• The private rate of return reflects the direct cost of schooling to individuals (i.e., their out-of-pocket cost).

• The social rate of return reflects the full cost to society of schooling, including any public subsidies.

• The private rate of return should be based on after-tax income, whereas the social rate of return should be based on pre-tax income, but this is not always done.

Typical Estimates of Returns to Education,

based on 98 country studies during 1960-1997 (“The classic pattern of falling returns to education

by level of … education are maintained.”)

Private Social

Primary 26.6% 18.9%

Secondary 17.0% 13.1%

Higher 19.0% 10.8%

Source: G. Psacharopoulos and H. Pastrinos, “Returns to Investment in Education: A Further Update”, World Bank Policy Research Working Paper 2881, September 2002 (from Table 1).

Inferences and concerns….

(from Psacharopoulos and Patrinos 2002)

“Private returns are higher than social returns where the latter is defined on the basis of private benefits but total (private plus external) costs. This is because of the public subsidization of education and the fact that typical social rate of return estimates are not able to include social benefits. Nevertheless, the degree of public subsidization increases with the level of education, which has regressive income distribution implications.”

“There is a concern in the literature with what might be called “social” rates of return that include true social benefits, or externalities…. If one could include externalities, then social rates of return may well be higher than private rates of return to education.”

Constructing Estimates of Private and Social Rates of Return to Education

• Requires data/assumptions about the lifetime earnings trajectories of individuals with different amounts of schooling.

• Requires data/assumptions about the out-of-pocket costs of schooling to individuals and to society.

• Requires data/assumptions about any income spillovers that result from one worker’s education.

Simulation Assumptions

• Worker’s earnings with no schooling: $40,000 per year, age 20 through 60

• Worker’s earnings with 1 year of additional schooling: $44,000 per year, age 20 through 60

• Private cost of 1 yr of schooling: $5,000• Social cost of 1 yr of schooling: $20,000• Number of workers whose earnings are increased

due to one worker’s schooling increasing by 1 year: $1,000

• Amount by which their earnings are increased: $2 per year

Lifetime Earnings Trajectory: Effect of One Year of Schooling

40,00044,000

0

Age: 20 21 30 40 50 60

Annual income

Foregone income

Worker’s increased earnings

Lifetime Earnings Trajectory: Effect of One Year of Schooling,

With Direct Costs

40,00044,000

-5,000

0

Age: 20 21 30 40 50 60

Annual income

Worker’s increased earnings

Foregone income

Lifetime Earnings Trajectory: Effect of One Year of Schooling,

With Social Costs

40,00044,000

-20,000

Age: 20 21 30 40 50 60

Annual income

Worker’s increased earnings

Foregone income

Lifetime Earnings Trajectory: Effect of One Year of Schooling,

With Social Costs and Income Spillovers

40,00044,000

-20,000

46,000

Age: 20 21 30 40 50 60

Annual income

Income spilloversWorker’s increased earnings

Foregone income

Simulated Rates of Return

No schooling cost 8.5

Private schooling cost = $5,000

8.3

Social schooling cost = $20,000

8.0

Social schooling cost, and income spillover of $2,000

14.3

0.000

0.020

0.040

0.060

0.080

0.100

0.120

0.140

1962

1964

1966

1968

1970

1972

1974

1976

1978

1980

1982

1984

1986

1988

1990

1992**

1994

1996

1998

2000

2002

Year

Pe

rce

nta

ge

Male Female

Estimates of the rate of return to schooling in the US, by gender,1962 and 1963-2003

(based on standard human capital earnings functions fit to Current Population Survey data)

Notes on estimates of the rate of return to schooling in the US

In keeping with standard practice among labor economists, the rate of return to schooling is estimated as the coefficient on years of schooling in an ordinary least squares regression in which the dependent variable is the natural logarithm of the previous years' annual earnings. In addition to years of schooling, the regressors include years of potential work experience (defined as age minus years of schooling minus 6) and its square, and dummy variables for nonwhites and for currently married individuals. Rate of return estimates were constructed separately for males and females for 1962, and for each year during 1964-2003. (The education variable is not usable for 1963.) The data are restricted to individuals aged 20-64 who were employed full time (35 or more hours per week) and year round (50-52 weeks per year). Top-coded earnings figures are adjusted upward by 15 percent of the relevant top code. In addition, we follow the algorithm proposed by Jaeger (1997) for dealing with the change in the CPS education variable that was initiated in 1992 (David A. Jaeger. "Reconciling the old and new census bureau education questions: recommendations for researchers." Journal of Business & Economic Statistics. July 1997. Vol. 15, No. 3). The rate of return estimates may be interpreted as the percentage increase in wages that is associated, cet. par., with each additional year of schooling.

Some limitations of the human capital approach

• Evidence of a causal link that runs from schooling to income

• Narrow conception of the benefits of schooling – Neglects the effect of schooling on an individual’s welfare, above

and beyond any increment to their earnings– Neglects the effect of schooling on social welfare, beyond the

income gains to the individuals who receive the schooling

Higher Education and Good Governance(cross-country correlations with tertiary enrollment rates; all governance indicators are

scored on 6 or 10 point scales with higher values reflecting better ratings)

Governance Indicator 1990

(114 countries)

1995

(121 countries)

Corruption in government

Positive and Significant Positive and Significant

Rule of law Positive and Significant Positive and Significant

Bureaucratic quality Positive and Significant Positive and Significant

Ethnic tensions Positive and Significant Positive and Significant

Repudiation of contracts by government

Positive and Significant Positive and Significant

Risk of expropriation Positive and Significant Positive and Significant

Source of governance data: International Country Risk Guide (ICRG) data, as compiled and processed by Stephen Knack and the University of Maryland.

Variable Definitions

(Source: International Country Risk Guide) • Corruption in Government -- Lower scores indicate "high government officials are likely to

demand special payments" and that "illegal payments are generally expected throughout lower levels of government" in the form of "bribes connected with import and export licenses, exchange controls, tax assessment, police protection, or loans."

• Rule of Law -- This variable "reflects the degree to which the citizens of a country are willing to accept the established institutions to make and implement laws and adjudicate disputes." Higher scores indicate: "sound political institutions, a strong court system, and provisions for an orderly succession of power." Lower scores indicate: "a tradition of depending on physical force or illegal means to settle claims." Upon changes in government new leaders "may be less likely to accept the obligations of the previous regime."

• Quality of the Bureaucracy -- High scores indicate "an established mechanism for recruitment and training," "autonomy from political pressure," and "strength and expertise to govern without drastic changes in policy or interruptions in government services" when governments change.

• Ethnic Tensions -- This variable “measures the degree of tension within a country attributable to racial, nationality, or language divisions. Lower ratings are given to countries where racial and nationality tensions are high because opposing groups are intolerant and unwilling to compromise. Higher ratings are given to countries where tensions are minimal, even though such differences may still exist.”

• Risk of Repudiation of Contracts by Government -- This indicator addresses the possibility that foreign businesses, contractors, and consultants face the risk of a modification in a contract taking the form of a repudiation, postponement, or scaling down" due to "an income drop, budget cutbacks, indigenization pressure, a change in government, or a change in government economic and social priorities." Lower scores signify "a greater likelihood that a country will modify or repudiate a contract with a foreign business."

• Risk of Expropriation of Private Investment -- This variables evaluates the risk "outright confiscation and forced nationalization" of property. Lower ratings "are given to countries where expropriation of private foreign investment is a likely event."

Higher Education andEntrepreneurial Activity

• The Total Entrepreneurship Activity (TEA) Index represents represents the share of adults involved in new firms or start-up activities.

• Individuals with higher levels of education have higher levels of entrepreneurial activity.

• This result holds true in a large number of countries (though the data almost entirely cover only developed countries).

Total Entrepreneurship Activity Index by Education Level

5.26.6

8.18.9 9.6

0

2

4

6

8

10

12

Education Level

TE

A

Correlation Coefficient: TEA and Education Level, by Country

0

5

10

15

20

25

30

Education Level

TE

A

Some Secondary

Secondary Degree

P ost Secondary

College/University

GraduateExperience

Proportion of WWII veterans among men by age in 1964

(Source: tabulated from the Current Population Survey)

0

10

20

30

40

50

60

70

80

20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64

Age

Perc

en

tag

e

Proportion of Korean War veterans among men by age in 1964

(Source: tabulated from the Current Population Survey)

0

10

20

30

40

50

60

20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64

Age

Pe

rce

nta

ge

Proportion of men with college or higher degrees by age in 1964(Source: tabulated from the Current Population Survey)

0

5

10

15

20

25

20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64

Age

Pe

rce

nta

ge

Proportion of men and women with college or higher degrees by age in 1964

(Source: tabulated from the March 1964 Current Population Survey)

0

5

10

15

20

25

20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64

Age

Perc

en

tag

e

Men Women

Income Spillovers Analysis

Study of earnings variation among US workers who are (statistically) comparable in terms of their: – Educational attainment– Years of work experience– Gender– Race– Marital status– Industry– Occupation– State of residence– Full-time, year-round status

Main Result of the Income Spillovers Analysis

• Other things equal, workers earn more when they are located in states that have higher proportions of college graduates.

• Thus, not only do college graduates have relatively high productivity and earnings, they also appear to enhance the productivity and earnings of those with whom they

work.

Income Spillovers Analysis -- Descriptive Statistics

Variable Definition Mean

WholeSample

Male Female

Hourly Wage Weekly earnings divided by usual hours worked per week 16.32 18.10 13.98

Age Age in years 38.59 38.59 38.60

White Non-Hispanic white .754 .762 .745

Black Non-Hispanic black .109 .092 .129

Hispanic Hispanic .088 .099 .074

Other Other race/ethnicity .049 .048 .051

Married Married .630 .680 .568

Education Years of education 13.14 13.02 13.28

Less than high school Less than 12th grade of education .124 .142 .102

High school High school graduate .365 .371 .358

Some college Some college without Bachelor’s degree .259 .240 .281

College or higher Bachelor’s degree or higher .252 .247 .259

Percentage with college or higher degrees by state Percentage of workers with college or higher degrees in the state .279 .278 .280

Percentage with graduate degrees by state Percentage of workers with graduate degrees in the state .097 .097 .097

1982 Percentage of sample from the 1982 CPS .180 .186 .173

1992 Percentage of sample from the 1992 CPS .352 .356 .346

2002 Percentage of sample from the 2002 CPS .468 .458 .480

Number of observations 35,531 19,512 16,019

NOTE: All observations have been drawn from the 1982, 1992, and 2002 March CPS.Sample includes full-time workers between the ages of 18 and 64. Weekly earnings are in 2002 current dollars, deflated by the CPI.

Industry and Occupation Categories

Industry Categories• 1 "Agriculture & Forestry"• 2 "Mining"• 3 "Construction"• 4 "Manufacturing-Durables"• 5 "Manufacturing-Nondurables"• 6 "Transportation"• 7 "Communication"• 8 "Utilities"• 9 "Wholesale Trade"• 10 "Retail Trade"• 11 "Finance, Insurance & Real Estate"• 12 "Private Household Service"• 13 "Business & Repair"• 14 "Personal Service"• 15 "Entertainment Service"• 16 "Professional Service"• 17 "Public Administration"

Occupation Categories• 1 "Managers & Administrators"• 2 "Professional & Technical"• 3 "Sales"• 4 "Clerical"• 5 "Private Household Service"• 6 "Service exc Private Household

Service"• 7 "Craftsmen"• 8 "Operatives exc Transport"• 9 "Transport Equipment Operatives"• 10 "Laborers"• 11 "Farmers"

Income Spillovers Analysis: Log Wage Regression Results with Percentage with College or Higher Degrees by State, Whole Sample

Dependent variable: log (weekly earnings).

Variable (1) (2) (3) (4) (5) (6)

Constant 1.304**(.029)

1.011**(.031)

1.165**(.043)

1.379**(.052)

1.497**(.049)

1.696**(.056)

High school .198**(.008)

.195**(.008)

.092**(.037)

.079**(.036)

.077**(.034)

.062*(.034)

Some college .346**(.009)

.339**(.009)

.183**(.041)

.152**(.040)

.125**(.038)

.095**(.038)

College or higher .660**(.009)

.638**(.009)

.314**(.041)

.256**(.041)

.210**(.039)

.149**(.039)

Percentage with college or higher degrees by state 1.363**(.051)

.733**(.124)

.315*(.172)

.879**(.116)

.480**(.161)

Percentage with college or higher degrees by state*High school

.402**(.137)

.450**(.135)

.236*(.128)

.288**(.127)

Percentage with college or higher degrees by state*Some college

.593**(.148)

.678**(.147)

.368**(.138)

.450**(.137)

Percentage with college or higher degrees by state*College or higher

1.155**(.147)

1.348**(.147)

.786**(.138)

.981**(.137)

State dummies No No No Yes No Yes

Industry and occupation dummies No No No No Yes Yes

Adjusted R-squared .314 .327 .329 .345 .414 .430

NOTE: All reported regressions also include the following independent variables: age and its square, three race/ethnic dummies, a marriage dummy, a female dummy, and two year dummies. Standard errors are in parentheses. Columns 4 and 6 include fifty state dummies. Columns 5 and 6 include 16 industry and 10 occupation dummies. * Statistically significant at the .10 level. ** Statistically significant at the .05 level.

Variable (1) (2) (3) (4) (5) (6)

Constant 1.253**(.040)

.975**(.043)

1.069**(.057)

1.222**(.070)

1.396**(.065)

1.557**(.075)

High school .193**(.011)

.189**(.011)

.182**(.049)

.161**(.048)

.166**(.046)

.141**(.045)

Some college .319**(.012)

.311**(.012)

.220**(.055)

.180**(.054)

.180**(.052)

.136**(.051)

College or higher .618**(.012)

.597**(.012)

.275**(.056)

.211**(.056)

.198**(.053)

.132**(.053)

Percentage with college or higher degrees by state 1.299**(.071)

.887**(.162)

.743**(.232)

1.040**(.153)

.837**(.217)

Percentage with college or higher degrees by state*High school

.038(.181)

.118(.179)

-.081(.170)

.010(.168)

Percentage with college or higher degrees by state*Some college

.344*(.200)

.455**(.199)

.134(.188)

.254(.186)

Percentage with college or higher degrees by state*College or higher

1.126**(.199)

1.331**(.198)

.760**(.187)

.969**(.186)

State dummies No No No Yes No Yes

Industry and occupation dummies No No No No Yes Yes

Adjusted R-squared .291 .303 .305 .321 .388 .404

Income Spillovers Analysis: Log Wage Regression Results with Percentage with College or Higher Degrees by State, 18-64 year old males

Dependent variable: log (weekly earnings).

NOTE: All reported regressions also include the following independent variables: age and its square, three race/ethnic dummies, a marriage dummy, and two year dummies. Standard errors are in parentheses. Columns 4 and 6 include fifty state dummies. Columns 5 and 6 include 16 industry and 10 occupation dummies.* Statistically significant at the .10 level. ** Statistically significant at the .05 level.

Income Spillovers Analysis: Log Wage Regression Results with Percentage with College or Higher Degrees by State, 18-64 year old females

Dependent variable: log (weekly earnings)

Variable

(1)

(2)

(3)

(4)

(5)

(6)

Constant 1.095** (.042)

.788** (.044)

.989** (.064)

1.280** (.076)

1.397** (.077)

1.656** (.086)

High school .213** (.013)

.211** (.012)

.016 (.056)

.014 (.055)

-.020 (.052)

-.022 (.051)

Some college .379** (.013)

.373** (.013)

.172** (.060)

.157** (.060)

.063 (.056)

.049 (.056)

College or higher .707** (.014)

.683** (.014)

.392** (.061)

.341** (.061)

.221** (.057)

.166** (.057)

Percentage with college or higher degrees by state 1.425** (.070)

.634** (.191)

-.092** (.255)

.720** (.178)

.078 (.237)

Percentage with college or higher degrees by state *High school

.753** (.207)

.764** (.205)

.558** (.193)

.570** (.191)

Percentage with college or higher degrees by state *Some college

.768** (.219)

.813** (.217)

.590** (.204)

.629** (.202)

Percentage with college or higher degrees by state *College or higher

1.073** (.218)

1.247** (.217)

.768** (.203)

.951** (.202)

State dummies

No No No Yes No Yes

Industry and occupation dummies

No No No No Yes Yes

Adjusted R-squared

.284 .301 .302 .321 .397 .414

NOTE: All reported regressions also include the following independent variables: age and its square, three race/ethnic dummies, a marriage dummy, and two year dummies. Standard errors are in parentheses. Columns 4 and 6 include fifty state dummies. Columns 5 and 6 include 16 industry and 10 occupation dummies.* Statistically significant at the .10 level. ** Statistically significant at the .05 level.