development i firisk, financial markets, and …cru2/econ731_files/presentations/klee.pdffirisk,...
TRANSCRIPT
Development I
�Risk, Financial Markets, and HumanCapital in a Developing Country�, by
Jacoby and Skou�asMark Klee
12/11/06
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 2
1 Introduction
Goals
� Examine response of human capital investment to �uctuations in family income�What are effects of �nancial market constraints in developing countries?� Estimate structural model of human capital investment to quantify cost of using childlabor as self-insurance
� Distinguish effects of anticipated, unanticipated income shocks
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 3
Results
� Child labor plays signi�cant self-insurance role for poor families in rural India, thoughnot costly in the long-run
� Reject intra-village complete markets� Only small-farm households are inadequately insured ex ante
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 4
1. Introduction2. Theoretical Framework3. Data4. Estimation Strategy5. Results6. Conclusion
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 5
2 Theoretical Framework
Value of Education
� Average grade attainment in 6 ICRISAT villages rose steadily by cohort for bothsexes, suggests education valuable� Education as route to more lucrative employment outside agriculture� Education to enhance farm management ability� Estimate conditional farm pro�t function with household �xed effects, with educationof head shifting pro�t level and interacting with use of high-yielding seed varieties� Signi�cantly positive returns to education increasing with fraction of land underHYVs
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 6
Model
� Child from household i is eligible for school for 0 � t � T consecutive agriculturalseasons
� Sit, school attendance given cumulative history of shocks at time t, augments begin-ning of period human capital stock Hit according to Hit = g(Sit; Hit; �it)
� Functional form of g determines cost of child labor as insurance (ex. git = Hitf (Sit; �it),f log-concave)
� @g@S ,
@g@H > 0
� �it is education productivity shifter, may re�ect illness or aggregate effects like schoolclosings
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 7
Household0s Problem
� Consumers solve
maxfcit;Sitg
E0
"TXt=0
�tU(ct) + �(HiT+1; BiT+1)
#BiT+1 = end of schooling period bequest of �nancial assetscit = household consumption� Dynamics of school attendance governed by
Et�1
��it
�Wt
Wt�1
�zit
�= 1 (1)
Wt = child wage determined in child labor market, independent of human capital stockwhile still in schoolzit = gH(t)
gs(t�1)gs(t)
= Marginal Rate of Transformation between school attendance inadjacent periods�it = shadow price of date t consumption relative to date t-1 consumption, determinedby �nancial market structure
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 8
Impact of Income Fluctuations on School Attendance
� Assume:
g(Sit; Hit; �it) = (1� �)Hitf (Sit; �it)
f (Sit; �it) = exp
� � exp
��1 [Sit � �it]
��, > 0
where f (0; 0) = 1, f 0 > 0; 8 and f is chosen so that log zit = (�Sit���it)
.� To focus on income uncertainty, if all changes in �it are anticipated, taking logs andrearranging equation (1) yields
�Sit = � [log �it +� logWit � log(1 + �it)] + ��it (2)
where �it is non-zero error in forecasting �it�Wt
Wt�1
�zit
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 9
Role of Financial Markets: Complete Markets
� If households can trade in state-contingent consumption claims, can reallocate re-sources across time and states at �xed prices (lifetime budget constraint)
�With complete markets, �it constant across households in the same market at ratioof consumption claim in state "t to that in state "t�1, scaled by �
� Separation of household's human capital investment and consumption decisions� If markets are complete across villages, �it = �t for each village. In particular,�t =
1
1+rft� Income shocks (either aggregate or idiosyncratic) do not affect school attendance� If markets are only complete within villages, �it = �t in the village. In particular, �t isthe intertemporal MRS in consumption� Aggregate income shocks (to Fit) transmitted to school attendance by increasingshadow price of consumption
� Idiosyncratic income shocks have no effect on school attendance
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 10
Role of Financial Markets : Incomplete Markets
� If consumption is not ex ante insurable, separation breaks down, extent to whichdepending on household's ability to transfer resources across time
� Relation between assets at beginning of t and end of t-1 given by A�it = Rt(Ait�1)� Assume MC of storage is increasing, so R00t < 0 for Ait�1 > 0, relevant for villageIndia
� Household budget constraint is now Cit + Ait = Rt(Ait) + Fit �WtSit
� �it is household speci�c intertemporal MRS in consumption (scaled by �), governedby Et�1 [�itR0t] = 1, or equivalently log �it = � logR0t(At�1) + log(1 + !it) where ! ismean 0 forecast error in �itR0t from unanticipated income shocks
� Presence of �it in equation 2 means that school attendance depends on unantici-pated idiosyncratic income shocks, not merely unlike complete markets
� Functional form of Rt determines how school attendance responds to anticipatedincome shocks
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 11
�With increasing MC of storage, R0t is affected by anticipated income shocks, con-sequently school attendance
� If R000 < 0, magnitude of response of school attendance to increase in income islarger than response to decrease in income
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 12
3 Data� ICRISAT Village Level Studies survey, data on time allocation of children 5-18 from40 households in 6 villages� Aurepalle and Dokur: red soil with limited water storage capacity, highly unpre-dictable rainfall
� Shirapur and Kalman: black soil with better water storage capacity, level and timingof rainfall erratic
� Kanzara and Kinkheda: more reliable rainfall� Data based on one day recall from June 1975 to the end of 1978� No appreciable summer cultivation, so two crop seasons: rainy (Kharif) and post-rainy (Rabi)
� Focus on response of school attendance to income �uctuations during season changessince full income only well de�ned over complete production cycle
� Two seasonal averages of school hours (including travel)� S1 averages all child's monthly observations in a season, including zeros (despite
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 13
possible measurement error)� S2 averages only positive observations, capturing adjustments in school day lengthrather than attendance
� After deleting entries which don't allow study of school attendance in consecutiveseasons, 1256 observations on 258 children
� 84 cases of zero school attendance for an entire season coming between two sea-sons of positive attendance
� 20% child labor market participation based on original sample
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 14
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 15
Measures of Income
F = (wmNm + wfNf + wcNc) + � + VY = wmLm + wfLf + wcLc + � + VF = full incomeY = conventional incomeNi = number of type i people residing in the household = 156 for each season� = seasonal pro�t from crop cultivationV = income from homemade handicrafts, sale of livestock products, land rental
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 16
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 17
4 Estimation Strategy
�Sivt = �0 + �1�flogFivt + �2�logFvt � � logWvt + "ivt
�Sivt = change in school attendance of child i, in village v, between season-years tand t� 1�flogFivt = deviation of change in log full income from village-season-year mean change�logFvt = village-season-year mean log full income change"ivt captures school productivity shocks (��it) and is assumed orthogonal to changesin full income and child wages�1 = �2 = 0 implies complete markets both within and across villages�1 > 0 suggests incomplete markets within village, though no inference on cause ofmarket failure
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 18
Separating In uence of Credit and Insurance Markets
�Fivt = �0vt +X0ivt�1�1 + (Xivt�1 DRvt)0�2 + uivt
�0vt is a season-village-year �xed effectXivt�1 is a vector of farm characteristics lagged one seasonDRvt is deviation of season-year t rainfall from long-run village mean�2 identi�ed by assumption that effect of rain shocks varies across farms
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 19
�clogF aivt = X 0ivt�1
b�1�clogF uivt = (Xivt�1 DRvt)0b�2�Sivt = �0vt + �1a�clogF aivt + �1u�clogF uivt + �ivt
�0vt = village-season-year �xed effect to purge effect of aggregate shock, focus onidiosyncratic� Provides separate test of perfect intravillage credit markets (�1a = 0) and perfectinsurance markets (�1u = 0)
� Increasing MC of storage consumption testable by estimating
�Sivt = �0vt + �+1aD
+ivt�
clogF aivt + ��1a(1�D+ivt)�
clogF aivt + �1u�clogF uivt + �ivtwhere D+
ivt = 1 if �clogF aivt > 0
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 20
5 Results� Both coef�cients signi�cant, have expected sign using both trimmed (removing zerosfrom beginning and end of each child's time series) and untrimmed data
�Weaker wage effect and insigni�cant income effect using S2, suggests meaningfulvariation comes in days of attendance rather than length of school day
� Replacing full income with conventional, income effect is much smaller due to expost adjustments
�Wage effect still strongly signi�cant, negative after including adult wage: child wagedoesn't re�ect other village shocks
� Smaller wage effect, insigni�cant income effect including village-season dummies� Larger wage effect including child effects (accounts for self-selectivity� IV estimates of � logWvt (due to correlation with �) similar to row 7� Instruments: lagged child wage, lagged rainfall, village-season dummies
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 21
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 22
Financial Markets Structure
� Aggregate income changes affect school attendance more than do idiosyncratic (i.e.�2 > �1), though not after controlling for seasonal effects� Either aggregate income shocks are fully insured though idiosyncratic are not, oraggregate seasonal �uctuations are predictable (collinear w/ dummies)
� Insigni�cant effect of idiosyncratic shocks on school attendance in Kanzara may rep-resent highest, least erratic rainfall, relative abundance of formal credit institutions
� Responses to aggregate shocks differ across villages� Reject intra-village complete markets after adding village-season-year �xed effects� After instrumenting income, small farm households worse insured than large
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 23
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 24
Anticipated v: Unanticipated
� Instrument income with lagged farm characteristics, interactions with current devia-tions of quarterly rainfall, deviations squared, onset day of monsoon
� Both anticipated and unanticipated idiosyncratic income shocks have signi�cant ef-fect on school attendance for small farms� not well insured ex ante, nor have perfect access to seasonal borrowing, lendingopportunities
� Only anticipated idiosyncratic income shocks matter for large farms, less effect thansmall farm
� �+1a > ��1a, as predicted by increasing MC of storage, though not signi�cantly differentat 5%
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 25
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 26
Impact of Market Incompleteness
� Attendance of uninsured child B differs in a given period from that of insured child A(Svt) by �eBt�̂1.�With same initial HC, learning technology, expected percentage difference in HC at
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 27
end of education is:
E log
�HAT+1
HBT+1
�= E
�exp
���1eBt
�� 1� TX
t=0
exp
��1 �Svt
�� If eBt~N(0; �2), difference in terminal HC increases with �1 and �2
� Small estimated Human Capital losses for uninsured children, regardless of how �1, determined
� Losses largest in village with most variable rainfall� Even though large farm households have larger variance of idiosyncratic incomeshocks, �̂1 estimate is half that of small farm households
� These gains represent gain from eliminating "excess" variability of attendance, ignorelikely increase in average level of school attendance
Development I
�Risk, Financial Markets, and Human Capital in a Developing Country�, by Jacoby and Skou�as 28
6 Conclusion� Reject complete markets within village� Credit market constraints, perhaps combined with limited storage, affect human cap-ital investment decisions
� Costs of variable school attendance are relatively low� Prospective policies must understand relevant economic risks, constraints� Compulsory schooling laws may lower household welfare� Policies promoting short-term credit and insurance may affect economic growth viahuman capital investment decisions
Development I