Education Policy and Intergenerational Transfers in Equilibrium B. - - PowerPoint PPT Presentation

Education Policy and Intergenerational Transfers in Equilibrium

Education Policy and Intergenerational Transfers in Equilibrium

• B. Abbott, G. Gallipoli, C. Meghir, G. Violante

Education Policy and Intergenerational Transfers in Equilibrium Introduction

◮ Government interventions in markets for higher education are large

and complex.

◮ We want to understand how these programs have affected

long-run outcomes, i.e. efficiency and mobility.

◮ e.g. what has been the long-run impact of federal college

subsidy/grant programs on GDP

◮ There are many papers (some very good) estimating short-run

effects of policy on education attainment, but none answer the key policy question of the impact of large scale policy changes on

• utput, inequality and welfare.

◮ Through the lens of a general equilibrium framework we are able to

address these questions.

Education Policy and Intergenerational Transfers in Equilibrium Introduction

Possible reasons for government intervention:

◮ Credit constraints ◮ Income taxation ◮ Imperfect insurance markets ◮ Equity

Education Policy and Intergenerational Transfers in Equilibrium Introduction

Our model captures:

◮ The dependence of college attainment decisions on:

◮ Parental transfers, which are possibly conditional ◮ The complexity of the existing student finance system with public and

private loans and grants that depend on parental wealth/income

◮ Liquidity Constraints ◮ Both cognitive and non-cognitive skills ◮ Gender

◮ Intergenerational Transmission:

◮ Wealth (parents are altruistic and paternalistic) ◮ Cognitive skills (exogenous process) ◮ Non-cognitive skills (endogenous because of dependence on

parental education)

Education Policy and Intergenerational Transfers in Equilibrium Introduction

More on the model:

◮ It is a steady-state overlapping generations model with

intergenerational links.

◮ There are seven factors of production (three levels of education

attainment × two genders, plus capital).

◮ The six human capital aggregates are imperfectly substitutable. ◮ Each factor has its own equilibrium price (large economy). ◮ idiosyncratic productivity shocks: wage risk varies by gender and

attainment.

Education Policy and Intergenerational Transfers in Equilibrium Introduction

Empirical Approach

◮ We first estimate several components of the model separately:

◮ The wage processes ◮ The aggregate production function

◮ We specify some parameters in advance, e.g. intertemporal

elasticity of substitution, Frisch elasticities of labour supply.

◮ Given these specifications we then estimate the remaining

parameters using a method of moments approach.

◮ Effects of cognitive and non-cognitive skills on non-pecuniary

(‘psychic’) costs of education.

◮ Extents of altruism and paternalism

Education Policy and Intergenerational Transfers in Equilibrium Introduction

Highlighted Results

◮ Current financial aid programs improve welfare and removing them

would reduce GDP by over 4% in the long-run.

◮ Further expansions of financial aid programs would have only

modest effects:

◮ Every additional dollar crowds out 60-70 cents of private investment

through reduced parental transfers and student labour supply.

◮ Eliminating any credit constraints induces gains through improved

sorting, not increased enrollment.

◮ A small group of high-ability children from poor families, especially

girls, would greatly benefit.

Education Policy and Intergenerational Transfers in Equilibrium Introduction

The rest of the talk:

• 1. Describe the life-cycle
• 2. Present and discuss Method of Moments Estimation
• 3. Present counterfactual analyses

Education Policy and Intergenerational Transfers in Equilibrium Model

The Life Cycle

Education Policy and Intergenerational Transfers in Equilibrium Model Life cycle: Education

The life-cycle - education

◮ At age j = 0 an individual receives a transfer from their parents.

From then on she/he is independent.

◮ She then chooses whether to continue with high school until age

jHS or enter the labor market permanently

Education Policy and Intergenerational Transfers in Equilibrium Model Life cycle: Education

High-school decision:

V ∗

g0 (ˆ

a, θ, q, κǫ) = max

• Vg0 (ˆ

a, θ, q, κǫ) − κHS(g, θ, κǫ), Ez[V LH

g0 (ˆ

a0, θ, z0)]

• ,

ˆ a = (ˆ a0, ˆ aCL).

Utility (psychic) costs of high school:

κHS(g, θ, κǫ) = ςHS + ςHS

1 1{g=f} + ςHS 2

log(θnon) + ςHS

3

log(θcog) + ςHS

4 κǫ.

◮ We will estimate ςHS by fitting model to the distribution of high

school drop-out decisions.

◮ We find non-cognitive skills are very important at this stage.

Education Policy and Intergenerational Transfers in Equilibrium Model Life cycle: Education

◮ Those who choose not to complete college enter the workforce as

singles (will get married later), solve the following problem: V e

gj

• aj, θ, zj
• =

max

cj,ℓj,aj+1 ug

• cj, ℓj
• + βEz
• V e

g,j+1

• aj+1, θcog, zj+1
• s.t.

(1 + τc)cj + aj+1 = (1 − τw) we

g εe gj

• θcog, zj

1 − ℓj

• +ψ + [1 + r (1 − τk)] aj

aj+1 ≥ 0, cj ≥ 0, ℓj ∈ [0, 1] zj+1 ∼ Γ e

gz

• zj+1 | zj
• .

Education Policy and Intergenerational Transfers in Equilibrium Model Life cycle: Education

College Completion decision:

V ∗

g1

• a1, ˆ

aCL, θ, q, κǫ

• =

max

• Vg1
• a1 + ˆ

aCL, θ, q

• − κCL(g, θ, κǫ), Ez[V HS

g1 (a1, θ, z1)]

• .

Utility (psychic) costs of college:

κCL(g, θ, κǫ) = ςCL + ςCL

1 1{g=f} + ςCL 2 log(θnon) + ςCL 3 log(θcog) + ςCL 4 κǫ.

◮ We will estimate ςCL by fitting model to the distribution of college

graduation rates.

◮ We find cognitive skills and preference shocks are relatively more

important than at the HS decisions, but non-cognitive still matters.

Education Policy and Intergenerational Transfers in Equilibrium Model Life cycle: Education

College financing

◮ sources of funding:

• 1. grants
• 2. subsidized loans up to a limit
• 3. unsubsidized loans underwritten by the government but more

expensive than commercial loans

• 4. Commercial loans (unavailable to poorer students)
• 5. working
• 6. Parental transfers

◮ means testing based on parental wealth and income, which determine the

state variable q q =    1 − student qualifies for subsidized + unsubsidized loan 2 − student qualifies for unsubsidized loan 3 − student qualifies for commercial + unsubsidized loan

Education Policy and Intergenerational Transfers in Equilibrium Model Life cycle: Education

◮ A student with wealthy parents (q = 3) has the option to borrow privately

and faces the following budget constraint: (1 + τc)cj + aj+1 − (1 − τw) wg,HSεg,HS

j

(θ, zj = 0)

• 1 − ¯

t − ℓj

• + φ (q, θ) =

= [1 + r (1 − τk)] aj if aj ≥ 0, (1 + r p) aj

• therwise

aj+1 ≥ −ap

◮ Private borrowing is less costly than from the government, so

wealth students with access choose this route.

◮ φ (q, θ) is net tuition (tuition minus grants).

Education Policy and Intergenerational Transfers in Equilibrium Model Life cycle: Education

◮ A student who qualifies only for unsubsidized government loans (q = 2)

faces the budget constraint: (1 + τc)cj + aj+1 + bj+1 − (1 − τw) wg,HSεg,HS

j

(θ, zj = 0)

• 1 − ¯

t − ℓj

• =

= φ (q, θ) + [1 + r (1 − τk)] aj if aj ≥ 0, bj = 0 (1 + r u) bj if aj = 0, bj < 0 aj+1 ≥ 0 bj+1 ≥ −b

◮ In the time period target (year 2000) government loans could not exceed

$23,000, and carried an interest rate prime + 2.6%.

Education Policy and Intergenerational Transfers in Equilibrium Model Life cycle: Education

◮ A less advantaged student who qualifies for a subsidized government loan

(q = 1) faces the budget constraint: (1 + τc)cj + aj+1 + bj+1 − (1 − τw) wg,HSεg,HS

j

(θ, 0)

• 1 − ¯

t − ℓj

• + φ (q, θ) =

=    [1 + r (1 − τk)] aj if aj ≥ 0, bj = 0 bj if aj = 0, 0 > bj ≥ −bs −bs + (1 + r u)

• bj + bs

if aj = 0, bj < −bs (1) aj+1 ≥ 0 bj+1 ≥ −b

◮ Subsidized loans are available up to $17,250 do not accrue interest during

• college. Another $5750 of unbsubsidized loans up to $23,000 also

available.

Education Policy and Intergenerational Transfers in Equilibrium Model Life cycle: Education

Marriage

◮ If individual does not complete college or high school they work as

singles until period jcol + 1

◮ At that point all individuals draw a match from the empirical

distribution of the opposite sex to reproduces matches in the data

◮ All characteristics random conditional on education

Distribution of Husband-Wife Education Matches (CPS Data - 2000) Wife’s Edu HSD HSG CLG Hus- HSD 0.107 0.030 0.002 band’s HSG 0.027 0.498 0.042 Edu CLG 0.002 0.056 0.236

Education Policy and Intergenerational Transfers in Equilibrium Model Life cycle: The Family

The Family

The life-cycle - work, transfers and retirement

◮ After matching, collective model with full commitment and equal

weights.

◮ There are economies of scale in consumption. ◮ Each family has two identical boys or girls, born at age 30. ◮ At age 46 the family decides on transfers to their children. ◮ Child ability and educational preferences are known at that point. ◮ Couples retire at the same time and live to a max age of 100. ◮ Perfect annuity markets in retirement, pension income.

Education Policy and Intergenerational Transfers in Equilibrium Model Life cycle: The Family

Intergenerational Links (cognitive)

◮ There is a known probability that a person of ability θcog has a child

• f ability ˆ

θcog

◮ We estimate this probability matrix from the NLSY ◮ We use the AFQT for mothers and the PIAT Maths for children

Transition Matrix (cognitive skill bins) Children Mothers 1 2 3 4 5 1 0.455 0.238 0.197 0.065 0.047 2 0.258 0.242 0.242 0.157 0.110 3 0.160 0.223 0.271 0.190 0.157 4 0.114 0.171 0.257 0.209 0.249 5 0.072 0.076 0.195 0.242 0.415

Education Policy and Intergenerational Transfers in Equilibrium Model Life cycle: The Family

Intergenerational Links (non-cognitive)

◮ The non-cognitive skills of the child also depend on the education of the

mother.

◮ Thus, the distribution is endogenous and varies with the distribution of

education attainment

Conditional Probabilities of Non-Cognitive Tercile 1 Child’s Cognitive Quintile Mother’s Edu 1 2 3 4 5 HSD 0.585 0.453 0.350 0.311 0.189 HSG 0.527 0.418 0.266 0.235 0.178 CLG 0.578 0.388 0.289 0.201 0.139 Conditional Probabilities of Non-Cognitive Tercile 3 Child’s Cognitive Quintile Mother’s Edu 1 2 3 4 5 HSD 0.118 0.180 0.311 0.372 0.464 HSG 0.111 0.252 0.348 0.432 0.504 CLG 0.139 0.270 0.358 0.443 0.525

Education Policy and Intergenerational Transfers in Equilibrium Model Life cycle: The Family

Intergenerational Links (resource transfers)

◮ The transfer decisions of parents are as follows: ◮ (Brace yourself for notation!)

Education Policy and Intergenerational Transfers in Equilibrium Model Life cycle: The Family

Wj(aj, zf

j , zm j , θf cog, θm cog, ef , em; ˆ

g, ˆ θ, ˆ κǫ) = max

cm j ,cf j ,ℓm j ,ℓf j ,ˆ a0,ˆ aCL,aj+1

• u(cm

j , cf j , ℓm j , ℓf j )

+βEzf ,zm

• Wj+1(aj+1, zf

j+1, zm j+1, θf cog, θm cog, ef , em)

• +2ω(ˆ

g)V ∗

0 (ˆ

a, ˆ g, ˆ θ, ˆ q, ˆ κǫ) + 2ξ · 1{ˆ

e=CL}

• s.t.

(1 + τc)cj + aj+1 + 2ˆ a + 2 ˆ aCL 1 + r = (1 − τw) wf,eεf,e

j

• θf , zf

j

1 − ℓf

j

• + (1 − τw) wm,eεm,e

j

• θm, zm

j

1 − ℓm

j

• +ψ + [1 + r (1 − τk)] aj

cj = [(cm

j ) ˜ ρ + (cf j ) ˜ ρ] 1 ˜ ρ

ˆ q =          1 if aj ≤ a∗ and max

• wm,eεm,e

j

, wf,eεf,e

j

• ≤ w∗

2 if aj ≤ a∗ and max

• wm,eεm,e

j

, wf,eεf,e

j

• > w∗

2 if a∗ < aj ≤ a∗∗ 3 if aj > a∗∗ ˆ θcog ∼ Γθcog

• ˆ

θcog|θf

cog

• ,

ˆ θnon ∼ Γθnon

• ˆ

θnon|ˆ θcog, ef

Education Policy and Intergenerational Transfers in Equilibrium Model Life cycle: Retirement

Retirement

◮ We describe the problem starting backwards. ◮ At retirement there is a constant stream of social security and own

asset income

◮ Annuity markets are perfect. ◮ When she retires he annuitizes savings and receives 16.4% of

average earnings within the education group in Social Security. W R

j (aj, ef, em)

= max

cj,aj+1

• u
• cj, 1, 1
• + βζj+1W R

j+1(aj+1, ef, em)

• s.t.

cj + aj+1 = pe,f + pe,m + ψ + ζ−1

j+1[1 + r (1 − τk)]aj

aj+1 ≥ 0, cj ≥ 0.

Education Policy and Intergenerational Transfers in Equilibrium Estimation/Parameterization

Estimation/Calibration

Education Policy and Intergenerational Transfers in Equilibrium Estimation/Parameterization Wages

Wages and shocks

◮ There is a separate wage process for each education and gender group ◮ These are specified as (e denotes an education group and θ ability)

lnwit = ln(wg,e

t

) + f g,e(age) + λg,e(θcog,i) + ze

it + mit

◮ Note that there is a deterministic return to age (known to the individual).

We do not allow for endogenous accumulation of experience.

◮ wg,e

t

is the aggregate (endogenous) price of human capital of type e (6 prices).

Education Policy and Intergenerational Transfers in Equilibrium Estimation/Parameterization Wages

The ability Gradient λg,e

Education group Male Gradient Female Gradient Less than HS 0.428 (0.054) 0.184 (0.057) HS Graduate 0.516 (0.030) 0.601 (0.036) College Ggraduate 0.797 (0.109) 0.766 (0.099)

Table: Estimated ability gradient λe (NLSY79)

Education Policy and Intergenerational Transfers in Equilibrium Estimation/Parameterization Wages

Productivity Processes zi,j

zi,j = ρg,ezi,j−1 + ηi,j, ηi,j ∼ (0, σ2

η,g,e)

Less than HS HS Graduates College graduates ̺m 0.955 (0.010) ̺m 0.952 (0.005) ̺m 0.966 (0.015) σ2

ηm

0.015 (0.002) σ2

ηm

0.017 (0.001) σ2

ηm

0.017 (0.005) σ2

zm0 0.037

(0.005) σ2

zm0 0.059

(0.003) σ2

zm0 0.094

(0.009) ̺f 0.852 (0.023) ̺f 0.953 (0.003) ̺f 0.983 (0.016) σ2

ηf

0.025 (0.005) σ2

ηf

0.019 (0.001) σ2

ηf

0.018 (0.004) σ2

zf0

0.035 (0.011) σ2

zf0

0.041 (0.003) σ2

zf0

0.076 (0.007)

Education Policy and Intergenerational Transfers in Equilibrium Estimation/Parameterization Production Sector

The Production Sector

◮ We assume a representative firm supplying goods in a competitive

market and hiring labor in a competitive labor market.

◮ The Production function is

Yt = K α

t

• sLH

HLH

t

ρ + sHS HHS

t

ρ + sCL HCL

t

ρ 1−α

ρ ,

(2) where He

t =

• se,f

He,f

t

χ + (1 − se,f)

• He,m

t

χ 1

χ .

(3)

Education Policy and Intergenerational Transfers in Equilibrium Estimation/Parameterization Production Sector

◮ We can then use the first order conditions to estimate the parameters of

the production function log ̟f,e

t

̟m,e

t

= log

• sf,e

t

sm,e

t

• + χ log
• Hf,e

t

Hm,e

t

• log
• ̟CL

t /̟HS t

• = log

sCL

t

sHS

t

• + (1 − ρ) log

HHS

t

HCL

t

• ◮ ̟g,e

t

is factor payment, and Hg,e

t

is constructed using the prices estimated from PSID.

◮ We use IV and estimate in first-differences. ◮ We consider lags of the human capital aggregates and total available

numbers of individuals in each education group as instruments.

◮ We settle on χ = 0.45 and ρ = 0.7 based on the results. ◮ We do sensitivity analysis of our main results.

Education Policy and Intergenerational Transfers in Equilibrium Estimation/Parameterization Externally Specified Parameters

Preferences

◮ Individual:

ugj(c, ℓ) = c1−γ 1 − γ + ϑg

j

ℓ1−νg

j

1 − νg

j

. Frisch elasticities are 1/3, except for women with children have 2/3. (Based on estimates of Blundell et al. (2016)) Inter-temporal substitution elasticity is set to 1/2.

◮ Couple:

u(cm, cf, ℓm, ℓf) = um(cm, ℓm) + uf(cf, ℓf)

◮ Economies of Scale:

c = [(cm)˜

ρ + (cf)˜ ρ]

1 ˜ ρ .

with ˜ ρ = 1.4 based on Voena (2015).

Education Policy and Intergenerational Transfers in Equilibrium Estimation/Parameterization Externally Specified Parameters

Costs of College

◮ We consider costs of tuition, fees, books and other

‘non-consumption’ expenses, and then deduct average private grant amounts (non-progressive), to arrive at $6700/year cost.

◮ Depending on parental background we assign annual public grant

amounts:

Income q amount Low 1 $2,820 Mid 2 $668 High 3 $143

Education Policy and Intergenerational Transfers in Equilibrium Estimation/Parameterization Externally Specified Parameters

Remaining Externally Specified Parameters

Parameter Value Description γ 2.0 Determines intertemporal elasticity of substitution (0.5) νm

j

5.5 Determines avg Frisch elast. of labour supply for men and non-mother women (0.33) νf

30−45

5.7 Determines avg Frisch elast. of labour supply for mothers (0.67) aCL 1.36 Limits borrowing of CL households to $75,000 aHS 0.45 Limits borrowing of HS households to $25,000 aLH 0.27 Limits borrowing of LH households to $15,000 bs 0.312 Limits subsidized loans to $17,250 for q = 1 students b 0.416 Limits total student loans to $23,000 for q = 1 and 2 students. ap 0.416 Limits private loans to $23,000 for q = 3 students ιu 0.053 Interest premium on Stafford loans (0.026 annually) α 0.35 Capital share of GDP δ 0.07 Depreciation rate of capital τw 0.27 Labor income tax rate τc 0.05 Consumption tax rate τk 0.40 Capital income tax rate ζj varies Mortality rates for retired hh based on US Life Tables 2000.

Education Policy and Intergenerational Transfers in Equilibrium Method of Moments Estimation

Method of Moments Estimation

◮ There are two blocks of moments/parameters ◮ The first relates to psychic costs and observed schooling decisions ◮ The second relates to internally ‘calibrated’ parameters and

identifying moments.

◮ Parameters are estimated by minimizing an unweighted quadratic

distance criterion between data moments and moments implied by the model.

Education Policy and Intergenerational Transfers in Equilibrium Method of Moments Estimation

Psychic Costs Estimation

◮ Although “all moments identify all parameters”, we specifically included

education attainment rates to help identify psychic cost functions. Table: Attainment Rates - NLSY97 Data

High School Drop-Out Rates (Data) Cognitive quintile 1 2 3 4 5 Non- 1 0.535 0.162 0.110 0.008 0.00 Cognitive 2 0.376 0.083 0.038 0.012 0.00 tercile 3 0.198 0.096 0.055 0.005 0.00 College Graduation Rates (Data) Cognitive quintile 1 2 3 4 5 Non- 1 0.022 0.077 0.153 0.303 0.677 Cognitive 2 0.034 0.086 0.212 0.379 0.764 tercile 3 0.034 0.162 0.255 0.473 0.772

Education Policy and Intergenerational Transfers in Equilibrium Method of Moments Estimation

Parameter High School College ςe Constant 1.697 1.872 (0.006) (0.010) ςe

1

Female Dummy

• 0.134

0.610 (0.0150) (0.006) ςe

2

log(θnon)

• 0.605
• 0.239

(0.010) (0.011) ςe

3

log(θcog)

• 0.233
• 0.779

(0.015) (0.012) ςe

4

κǫ 0.213 0.408 (0.008) (0.011)

Education Policy and Intergenerational Transfers in Equilibrium Method of Moments Estimation

College graduation decisions fit

College Graduation Rates (Data) Cognitive quintile 1 2 3 4 5 Non- 1 0.022 0.077 0.153 0.303 0.677 Cognitive 2 0.034 0.086 0.212 0.379 0.764 tercile 3 0.034 0.162 0.255 0.473 0.772 College Graduation Rates (Model) Cognitive quintile 1 2 3 4 5 Non- 1 0.087 0.114 0.166 0.298 0.648 Cognitive 2 0.094 0.125 0.177 0.334 0.725 tercile 3 0.106 0.142 0.193 0.399 0.806

Education Policy and Intergenerational Transfers in Equilibrium Method of Moments Estimation

High school drop out decisions fit

High School Drop-Out Rates (Data) Cognitive quintile 1 2 3 4 5 Non- 1 0.535 0.162 0.110 0.008 0.00 Cognitive 2 0.376 0.083 0.038 0.012 0.00 tercile 3 0.198 0.096 0.055 0.005 0.00 High School Drop-Out Rates (Model) Cognitive quintile 1 2 3 4 5 Non- 1 0.533 0.162 0.079 0.069 0.001 Cognitive 2 0.367 0.094 0.065 0.052 0.000 tercile 3 0.255 0.067 0.034 0.016 0.000

Education Policy and Intergenerational Transfers in Equilibrium Method of Moments Estimation

◮ Over 80% of the variation in psychic costs of college is explained

by skills. Over 90% for high school.

◮ For high school psychic costs non-cognitive skills are 2.5 times as

important as cognitive. For college cognitive are 3.25 times as important.

◮ If variation in psychic costs is shut down, then 2/3 of the

covariance between college and cognitive skills remains. Only 23% of covariance between college and non-cognitive remains.

Education Policy and Intergenerational Transfers in Equilibrium Method of Moments Estimation

The Rest of the Parameters

Parameter Description Value (s.e.) β Time discount factor 0.944 (0.006) ϑg

j

Male and non-mother female leisure preference 0.240 (0.004) ϑf

30−45

Female with children leisure preference 0.505 (0.005) ωf Altruism towards sons 0.289 (0.008) ωm Altruism towards daughters 0.250 (0.005) ξ Paternalistic utility from a child’s college going 0.201 (0.041) a∗ Wealth upper limit for subsidized loans (group q=1) 2.07 (0.202) a∗∗ Wealth lower limit for private student loans (group q=3) 2.68 (0.499) ιp Interest premium for private student loans 0.048 (0.006) ι Basic borrowing wedge that applies to all debt 0.097 (0.010) ψ Redistributive transfer 0.44 (0.008)

Education Policy and Intergenerational Transfers in Equilibrium Method of Moments Estimation

The Rest of the Moments

Moment Matched Target Value Model Value Capital-output ratio 2.0 2.1 Average male labour supply 0.350 0.349 Average labour supply of mothers 0.210 0.211 Average IVT to female child $29,096 $29,044 Average IVT to male child $33,164 $33,012 Ratio of college grad. rate in fourth (top) quartile 1.63 1.56

• f parental wealth to grad. rate in third quartile

HS Fraction of Female Population (cross-section) 0.584 0.584 HS Fraction of Male Population (cross-section) 0.567 0.567 CL Fraction of Female Population (cross-section) 0.282 0.282 CL Fraction of Male Population (cross-section) 0.294 0.294 Fraction of students with subsidized loans 0.525 0.516 Fraction of students with any gov’t loans 0.621 0.633 Fraction of students with private loans 0.083 0.086 Fraction of workers with negative net worth 0.077 0.071 Var(log post-tax income)/Var(log pre-tax income) 0.61 0.61

Education Policy and Intergenerational Transfers in Equilibrium Un-targeted Moments

Un-targeted Moments

◮ We explore the models implications in a number of non-targeted

dimensions.

◮ e.g. intergenerational earnings mobility (Chetty et al., 2014), life-cycle

profiles of income, consumption etc. (Kaplan, 2011).

Education Policy and Intergenerational Transfers in Equilibrium Un-targeted Moments

Un-targeted Moments

◮ We explore the models implications in a number of non-targeted

dimensions.

◮ e.g. intergenerational earnings mobility (Chetty et al., 2014), life-cycle

profiles of income, consumption etc. (Kaplan, 2011).

◮ One thing that is new in the revision is the models ability to capture

attainment by ability and parental wealth

1 2 3 4 0.2 0.4 0.6 0.8 1 θcog Quartile College Attainment Rate parental net worth quartile 1 parental net worth quartile 2 parental net worth quartile 3 parental net worth quartile 4

Education Policy and Intergenerational Transfers in Equilibrium Counterfactual Experiments

Counterfactual Experiments

Education Policy and Intergenerational Transfers in Equilibrium Counterfactual Experiments

◮ Counterfactuals compare steady-state equilibria with labour

income tax rates adjusted to balance the government budget.

Education Policy and Intergenerational Transfers in Equilibrium Counterfactual Experiments

‘Unconstrained’ Economy

◮ The first experiment sets a benchmark for what policy might accomplish

regarding the effect of credit constraints on schooling.

◮ Here all debts must be paid at retirement, but otherwise agents are

unconstrained and borrow at ‘prime’.

◮ Efficiency gains are driven by improvement of selection into college by

ability, more so than increases in college enrollment.

◮ We are leaving the federal grant program in tact when we do this.

Education Policy and Intergenerational Transfers in Equilibrium Counterfactual Experiments

‘Unconstrained’ Economy

“Unconstrained” Economy Benchmark G.E. Long-run Men 0.294 0.303 College Women 0.282 0.292 Graduation Men - top 1/3 of cognitive skills 0.695 0.708 Rates Women - top 1/3 of cognitive skills 0.577 0.644 Total - top 1/3 of parental wealth 0.414 0.362 Total - bottom 1/3 of parental wealth 0.205 0.288 Crowding out of IVTs - Male –

• $3,932

Other Crowding out of IVTs - Female –

• $3,774

Statistics Student labor supply –

• 33.7%

Aggregate output – +1.16% Welfare gain – +0.42%

Crowding out is the average change in IVTs given to children who are college non-switchers (college graduates in both cases).

Education Policy and Intergenerational Transfers in Equilibrium Counterfactual Experiments

Grant Removal

◮ Next we eliminate the federal grants program. ◮ Grants are a subsidy that may be beneficial for both easing credit

constraints and undoing distortions of risk and income taxes.

◮ We find substantial effects on both selection and enrollment (particularly

for women).

Education Policy and Intergenerational Transfers in Equilibrium Counterfactual Experiments

Grant Removal

Removal of Public Tuition Grants Benchmark P .E. G.E. Short-run Long-run Men 0.294 0.242 0.271 College Women 0.282 0.201 0.222 Graduation Men - top 1/3 of cognitive skills 0.695 0.590 0.556 Rates Women - top 1/3 of cognitive skills 0.577 0.411 0.414 Total - top 1/3 of parental wealth 0.414 0.373 0.482 Total - bottom 1/3 of parental wealth 0.205 0.132 0.082 Crowding out of IVTs - Male – +$596

• $2,723

Other Crowding out of IVTs - Female – +$253

• $3,157

Statistics Student labor supply – +13.4% +4.47% Aggregate output – –

• 1.95%

Welfare gain – –

• 0.68%

Education Policy and Intergenerational Transfers in Equilibrium Counterfactual Experiments

Loan Removal

◮ Next we eliminate the federal student loans program. ◮ Except for the very richest few (q = 3), no borrowing is possible to finance

college.

◮ Again, substantial efficiency and welfare effects are mostly driven by

changes in sorting.

Education Policy and Intergenerational Transfers in Equilibrium Counterfactual Experiments

Loan Removal

Removal of Student Loans Benchmark P .E. G.E. Short-run Long-run Men 0.294 0.233 0.257 College Women 0.282 0.179 0.235 Graduation Men - top 1/3 of cognitive skills 0.695 0.548 0.497 Rates Women - top 1/3 of cognitive skills 0.577 0.366 0.367 Total - top 1/3 of parental wealth 0.414 0.363 0.576 Total - bottom 1/3 of parental wealth 0.205 0.112 0.033 Crowding out of IVTs - Male – +$2,837 +$3,740 Other Crowding out of IVTs - Female – +$2,099 +$2,199 Statistics Student labor supply – +38.3% +5.84% Aggregate output – –

• 2.95%

Welfare gain – –

• 0.65%

Education Policy and Intergenerational Transfers in Equilibrium Counterfactual Experiments

Grant Expansions

◮ We also experiment with general, means-tested and ability-tested

grant expansions.

◮ We find that means-tested performs somewhat better on efficiency

and welfare improvements.

◮ An optimal means-tested expansion (paid by labor income taxes) is

195%.

◮ Under such an expansion college is free for q = 1 (plus a bit more

money).

Education Policy and Intergenerational Transfers in Equilibrium Counterfactual Experiments

Conclusions

◮ We set up a rich policy framework for the analysis of education

policy

◮ We put particular emphasis on modeling the sources of funding for

education, both through parents and institutions

◮ existing federal grants and loans programs improve welfare and

• utput.

◮ Distortions reducing the ex-ante return are still important,

particularly for low background students. Credit constraints are less important.