Anda di halaman 1dari 6

American Economic Association

Health, Human Capital, and Life Cycle Labor Supply

Author(s): Charles Hokayem and James P. Ziliak
Source: The American Economic Review, Vol. 104, No. 5, PAPERS AND PROCEEDINGS OF
One Hundred Twenty-Sixth Annual Meeting OF THE AMERICAN ECONOMIC ASSOCIATION
(MAY 2014), pp. 127-131
Published by: American Economic Association
Stable URL:
Accessed: 15-12-2017 22:01 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

American Economic Association is collaborating with JSTOR to digitize, preserve and extend
access to The American Economic Review

This content downloaded from on Fri, 15 Dec 2017 22:01:38 UTC
All use subject to
American Economic Review: Papers & Proceedings 2014, 104(5): 127-131

Health, Human Capital, and Life Cycle Labor Supply1

By Charles Hokayem and James P. Ziliak*

The fundamental importance of human and to 1999 the only consistent consumption mea-
health capital on labor market outcomes is well sure was food spending. Since then the PSID
known (Grossman 1972; Weiss 1972). Higher has collected rich data on consumption expen-
levels of human capital, both through formal ditures, enabling us to isolate medical out-of-
schooling and on-the-job experience, lead to pocket spending from nonmedical spending.
higher wages and income, while good health Likewise, research on health in the PSID had
increases the number of healthy days to consume been limited to the standard self-rated index of
goods and leisure. Perhaps surprisingly, most global health, with a paucity of possible instru-
human capital papers have ignored the role of ments. The PSID now collects other health con-
health, and most health papers have ignored its ditions, including height and weight used to
effect on endogenous human capital (Heckman construct body mass index (BMI). The rise of
1976; Shaw 1989; Sickles and Yazbeck 1998; obesity in the United States has been linked to a
Imai and Keane 2004; French 2005; Blundell et number of poor health conditions such as type II
al. 2013). However, if health affects how much diabetes, high blood pressure, heart disease, and
an individual works, and how much an indi- certain forms of cancer. Many of the obesity-
vidual works determines the skills acquired on related conditions are not permanently disabling
the job, then variations in health will influence in the sense of requiring labor-force exit but do
human capital and, subsequently, future labor make the individual more susceptible to illness
supply choices and income. This suggests an that may require periods out of work, thus inter-
interaction between health and human capital rupting the human capital process.
that has not been addressed, and yet ignoring it In Figure 1 we depict average levels of BMI
could lead to a biased assessment of how health Prime (BMI relative to 25, the cut off for over-
and wages, and attendant public policies such asweight) at each age among men in our sample.
taxes and transfers, affect labor supply over the It is clear that the average man in the PSID is
life cycle. overweight since BMI Prime exceeds 1, but
We estimate a structural model of life cycle unhealthy men (those with self-rated health of
labor supply of men that combines health pro- good, fair, or poor) are much heavier on average
duction with learning-by-doing using newly than healthy men (those with self-rated health
available data on consumption and health in the of very good or excellent), suggesting that BMI
Panel Study of Income Dynamics (PSID). Prior might be a good instrument for overall health.

I. Modeling Labor Supply

* Hokayem: US Census Bureau, SEHSD, HQ-7H168,
4600 Silver Hill Rd, Washington, DC 20233-8500 (e-mail:; Ziliak: Department of The individual chooses hours of work (Nt),
Economics, University of Kentucky, Lexington, KY nonmedical consumption (C), and medical
40506-0034 (e-mail: We thank Richard spending (Mt) to maximize the present dis-
Blundell, Kyle Caswell, David Johnson, Charles Nelson, counted value of uncertain and time-separable
Trudi Renwick, Karl Scholz, and seminar participants at the
10th World Congress of the Econometric Society, and the
utility defined over leisure (L), nonmedical
New Directions in Consumption Research with the PSID consumption, and the stock of health (. Ht ),
Conference for helpful comments on an earlier version. U(Lt9Ct,Ht ):
All errors are our own. The views expressed in this paper
are those of the authors and not necessarily those of the US
Census Bureau.
Go to 1257/aer. 104.5. 127 to visit the
(1) V(A K Ht) = U(L C H,)
article page for additional materials and author disclosure
statement(s). + Et[V(Al+l,Kl+l,Hl+1 )],

This content downloaded from on Fri, 15 Dec 2017 22:01:38 UTC
All use subject to

time to exercise, and purchasing medical ser-

vices, y(Mt, , L), evolving as

(4) Ht+l = (1 - Su)Ht + y(Mt,Lt).

Medical spending is isolated from nonmedi-

cal spending as the former is a direct input into
the production of good health - higher out-of-
pocket medical spending is treated as an invest-
ment in good health. Leisure time is also an
input into the production of good health in that
leisure (or at least some portion) is spent in exer-
cise and other health-promoting activities, creat-
Figure 1. BMI Prime over Life Cycle
ing an implicit tradeoff between human capital
production and health capital production. The
potential amount of leisure time spent in health
production, and likewise the hours of work spent
where V(At, Kt, Ht) is the value function deter-
in human capital production, are governed by
mined by the stocks of assets (A,), human capital
(Kt), and health capital (//,); = 1/(1 + p) is
"healthy time" (htt), Lt + Nt = htt. Total time
the discount factor based on rate of time prefer-
in a period ( T ) is the sum of healthy time and
ence (p); and Et is the time t expectations opera-
sick time (st), htt + st = T.
tor reflecting that there is uncertainty over future human and health capital production
earnings and health. functions generate nonseparabilities in the
Income comes from interest on prior period intertemporal budget constraint because stocks
assets (rtAt) and labor earnings (wtNt), wheredepreciate
rt over time. The forcing variable is the
is a time t interest rate on composite assets assumption
and of exogenous health shocks that have
wt is the before-tax hourly wage rate. Incomea direct effect on sick time. The basic idea is that
can be spent on nonmedical consumption with a poor health shock reduces future wages via two
a normalized price of 1, on medical services at
channels - a direct effect on future productivity
the price p, or can be saved and carried toand thean indirect effect of lowering labor supply
next period. The resulting asset accumulationin the current period, which, in turn, reduces the
constraint is amount of human capital gained on the job.
We combine the envelope conditions for the
state variables
(2) At+i = ( 1 + rt)(At + wtNt -Ct- p?Mt). with the first-order conditions to
solve the optimization problem. This yields a
Following Shaw (1989), we specify standard the intertemporal Euler equation for non-
medical consumption, along with time-nonsep-
observed wage (wt) as the product of the human
capital stock and the unobserved rental rate arable
on Euler equations for hours of work and
human capital (Rt)9 wt = Rt Kt. In each period an spending. The resulting system is highly
individual inherits a stock of human capital nonlinear,
that and thus we follow Shaw (1989) to
depreciates at rate SK. New investment occursimplement
on a two-stage GMM procedure. In
stage one we estimate health and human capital
the job through learning-by-doing that depends
on hours worked and human and health capital,production functions, and in stage two we treat
x{Nt, Kt, Ht): the latter as known parameters to estimate util-
ity preferences. We specify human capital (i.e.,
(3) Kt+X = (1 - Sk)K, + x(NKHt). wages) as a quadratic function of past wages,
hours, health, and interactions among the three
That is, hours of work determine not only factors.
cur- Likewise, we specify current health as
rent period utility, but also future utility avia
function of lagged health, leisure, medical
wages because of learning on the job. spending, and interactions of the three factors.
Likewise, in each period an individual inher-
The utility function is direct translog, which
its a stock of health that depreciates at rateadmits
H. nonseparability among current consump-
Health can be replenished by devoting leisure tion, leisure, and health.

This content downloaded from on Fri, 15 Dec 2017 22:01:38 UTC
All use subject to

Figure 2. Real Hourly Wages over Life Cycle Figure 3. Annual Hours of Work over Life Cycle

II. Data

Men are drawn from the Survey Research

Center subsample of the PSID from 1999-2009
who meet the following criteria: (i) head of
household; (ii) work at least three years; (iii)
between the ages of 25 and 60; (iv) real net
hourly wages between $2 and $300; (v) real
annual food spending greater than $500 and less
than $80,000; (vi) real total spending between
the 0.5 and the 99.5 percentiles; and (vii) real
consumption growth between -80 and 300 per-
cent. The sample contains 1,654 men and 9,052
person-years. Figure 4. Real Net Worth over Life Cycle

The key labor supply variable for our analysis

is annual hours of work on all jobs. Healthy time
is 8,760 less sick hours, defined as the amount but by age 35 the hourly wage begins to diverge
of work missed due to own illness or others' quite dramatically. Likewise, healthy men work
illness. Self-rated health, which ranges fromabout 1 94 hours more per year, or more than two
(poor) to 5 (excellent), serves as the health stock
weeks at full time, in part because they report 24
measure. hours less annual sick time. However, Figure 3
Nonmedical spending is constructed shows that
bythe hours of work difference is less
ming food at home and away (including food dispersed across the life cycle than wages. On
stamps and home delivered meals), thegasoline
other hand,and
Figure 4 shows that the lifetime
other transportation, utilities, education of lowerand
wages and hours of work translates into
care, homeowner insurance, and rent substantially lower net worth, averaging nearly
paid, which
includes payments by renters as well $150,000 less for unhealthy men. Again, this
as imputed
rent for homeowners calculated as divergence
6 percentemergesofaround age 35, suggesting
the house value. Medical out-of-pocket that health shocks early in the human capital
includes surgical and dental, hospitalizations, accumulation process transmit into a future of
health insurance premiums, and prescription
lower wages and wealth.
drugs. Medical and nonmedical spending are
adjusted by the square root of family size.
III. Results

Financial data are deflated by the personal con-

Table 1 records results for the human and
sumption expenditure deflator for 2006.
Figure 2 depicts the age-specific average
health capitalreal
production models, along with the
wage by health status. Overall, within-period
healthy men translog utility parameters. The
instruments consist
earn $6.22 more per hour than unhealthy men, of the time t values of the

This content downloaded from on Fri, 15 Dec 2017 22:01:38 UTC
All use subject to

Table 1

Human capital production Health capital production Utility function

Parameter Estimate SE Parameter Estimate SE Parameter Estimate SE

wt 0.714*** (0.142) Ht 0.690*** (0.137) InL, 1.00

wj -0.047*** (0.006) Hj 0.001 (0.010) In C, 0.062*** (0.019)
w,N, 0.078** (0.032) H,L, -0.008 (0.016) In L, In C, -0.036*** (0.005)
N, 0.186 (0.172) L, 0.130 (0.136) (InL,)2 -0.270*** (0.005)
Nj -0.046** (0.023) I2, -0.009 (0.010) (In C,)2 -0.001 (0.001)
H, -0.100 (0.083) M, 0.014 (0.045) In H, 0.071 (1.700)
Hj 0.007 (0.011) Mj -0.001 (0.001) In//, InL, 0.053*** (0.007)
H,N, -0.007 (0.028) H,M, 0.004 (0.005) In H, In C, 0.013 (0.009)
w,H, 0.067** (0.029) L,M, -0.001 (0.006) (In H,)2 -0.046 (0.675)

Sargan 49.95 Sargan 158.4 Sargan 66.90

test [df] [8] test [df] [8] test [df] [88]
p-value 0.00 p-value 0.00 p-value 0.95

Notes: All models control for time effects. Standard errors

*** Significant at the 1 percent level.
** Significant at the 5 percent level.
* Significant at the 10 percent level.

regressors, along with

Here we
see a one-unit increase in health
and (e.g.,state
economic conditions moving
andfrom policies.
good to very good, roughly a stan-
The large effect of dard
the deviation
increase) increases
wage futureand
wages its
by 1.3 percent,
square indicates strong or an elasticity of in
persistence 0.05. This
effect is small, and
with future wages increasing atsolely
a driven by the inter- ra
in the current wage, action
i.e.,between wages and health. Comparing
diminishing margin
healthy to unhealthy
productivity. The interaction workers we findwages
between in Table 2 an
hours worked suggeststhat thethat hours
marginal effect worked
of an extra hour of work and
has a larger payoff for unhealthy
human capital are complements, workers, while wit
the opposite is the case
learning-by-doing technology. At forthe
an extrasame
improve- tim
ment in health.
the interaction between wages and health sug
In the middle
gests that human capital andofhealthTable 1 we see very high ar
complements. persistence in health capital. Time and money
seem effect
To explore further the to have little to
of no effect
currenton health, human
is consistent
and health capital on futurewith "flat of the curve" medicine capital
in Table 2 we present since the RAND
partial health experiment
effects of hoursof
and health on wages.theThe
1970s. marginal product of
hours on wages evaluated at
The final columns the
of Table mean
1 present the util- valu
of the variables is ity
equal to 0.10.
function estimates, which point This
to importantmean
an increase in annual hours in by preferences.
500 The estimatesunder
a standard deviation) that leisure and nonmedical
increases the wage spending nex
period by 2.6 percent, or
are direct an elasticity
substitutes of
in utility, or that hours of 0.12
We note that this is much smaller than the 8 work and nonmedical spending are comple-
percent effect reported in Keane (2011) using
ments, consistent with Shaw (1989) and Ziliak
and Kniesner (2005). Although health has no
Shaw's (1989) estimates. We believe this comes
from our inclusion of health in the human capi-
direct effect on utility, the interactions between
leisure and health and consumption and health
tal process, which dampens learning-by-doing.
In the bottom panel of Table 2 we show thesuggest that leisure and nonmedical consump-
marginal product of health on human capital.tion are each complements with good health.

This content downloaded from on Fri, 15 Dec 2017 22:01:38 UTC
All use subject to

Table 2 - The Effect of Work and Health on Human Capital

Human capital partial effect Full sample Healthy Unhealthy

Marginal product of Nt on Wt+ ' 0. 10 0.10 0.10

Percent change in Nt (500 hours) 22.1% 21.8% 22.8%
Percent change in Wt+' 2.6% 2.5% 2.9%

Marginal product of Ht on Wt+' 0.07 0.09 0.04

Percent change in Ht ( 1 unit) 25.7% 22.7% 36.1%
Percent change in Wt+' 1.3% 1.5% 0.7%

Note: Calculations based on production parameters in Table 1 and me

IV. Discussion French, Eric. 2005. "The Effects of Health,

Wealth, and Wages on Labour Supply and
The descriptive evidence presented here showsRetirement Behaviour." Review of Economic
that starting around age 35 unhealthy men have Studies 72 (2): 395-427.
Grossman, Michael. 1972. "On the Concept of
significantly lower life cycle profiles of wages
and assets, which is preceded by an increase Health Capital and the Demand for Health."
in obesity around age 30. Our model estimates Journal of Political Economy 80 (2): 223-55.
show strong evidence of learning by doing,Heckman,
with James J. 1976. "A Life-Cycle Model of
Earnings, Learning, and Consumption." Jour-
an elasticity of future wages with respect to cur-
nal of Political Economy 84 (4): SI 1-44.
rent hours of about 0. 12. While the effect of good
health on future wages is smaller, it seems toImai,
haveSusumu, and Michael P. Keane. 2004.
"Intertemporal Labor Supply and Human Cap-
a larger effect on the margin for already healthy
ital Accumulation." International Economic
men, suggesting increasing returns to good
health. In ongoing research we are introducing Review 45 (2): 601-41.
taxation to explore more comprehensively Keane, the Michael P. 2011. "Labor Supply and
role of tax policy and health policy on labor-mar- Taxes: A Survey." Journal of Economic Litera-
ket outcomes over the life cycle via simulation ture 49 (4): 961-1075.
of the utility function parameters. Heretofore Shaw, Kathryn L. 1989. "Life-Cycle Labor
such an analysis was not possible with prime-age Supply with Human Capital Accumula-
workers in the United States for lack of credible tion." International Economic Review 30 (2):
data on hours, incomes, consumption, health, 431-56.
and assets in a single data source. This rich new Sickles, Robin C., and Abdo Yazbeck. 1998. "On
PSID data offers many future opportunities for the Dynamics of Demand for Leisure and the
policy relevant research. Production of Health." Journal of Business and
Economic Statistics 16 (2): 187-97.
REFERENCES Weiss, Yoram. 1972. "On the Optimal Lifetime
Pattern of Labour Supply." Economic Journal
Blundell, Richard, Monica Costa Dias, Costas 82 (328): 1293-1315.
Meghir, and Jonathan M. Shaw. 2013. "Female Ziliak, James P., and Thomas J. Kniesner. 2005.
Labour Supply, Human Capital, and Wel- "The Effect of Income Taxation on Consump-
fare Reform." National Bureau of Economic tion and Labor Supply." Journal of Labor Eco-
Research Working Paper 19007. nomics 23 (4): 769-96.

This content downloaded from on Fri, 15 Dec 2017 22:01:38 UTC
All use subject to