health shocks, household consumption, and child nutrition aida galiano (university of zaragoza)...
TRANSCRIPT
Health Shocks, Household Consumption, and Child Nutrition
Aida Galiano(University of Zaragoza)
&Marcos Vera-Hernández
(UCL & IFS)
Motivation
• In developing countries:• Lack of well functioning credit/insurance markets• Risk translates into welfare loss, inefficient
production choices, and deficient human capital accumulation
• Important question: • How well insured are households against health
shocks?– Important for policy: the role of health insurance
Objective
• In this paper we look at how households react to health shocks– How their consumption, debts, transfers and savings
change with health shocks– Who gets affected within the household
• Three waves of a panel of very poor households in relatively small towns of Colombia– Detailed data on household consumption– Children anthropometrics
Connection with the literature:
• This paper is related to two branches of the literature:
• A policy oriented literature that analyzes the role of health shocks and health insurance in developing countries
Connection with the literature:• A more methodological literature:
– Standard test: If household is fully insured against idiosyncratic shocks, consumption growth should be uncorrelated with shocks (or changes in income) once aggregate shocks are controlled for
– The consumption smoothing literature has usually summarized shocks by measuring income changes
– Little attention to the nature of the shock
Connection with the literature:• A more methodological literature:
One could claim that some elements of household consumption are part of the health production function
• Health care expenditures• Food• Fuel (indoor pollution is an important problem)• Transport (health facilities are distant)• Clothes (blankets, shoes to keep the person warm)
Connection with the literature:• A more methodological literature:
– In a poor environment, household might need to increase some consumption items (i.e. food) in order to improve the health of the individual that has fallen ill
– Under those circumstances, if the households are fully insured then household consumption should increase with health shocks
Connection with the literature:• A more methodological literature:
– It is much harder to think how to do a test of full insurance
– Null correlation between health shocks and household consumption would usually imply full insurance
– However, this is unclear if consumption is part of the health production function
– A positive correlation between consumption and health shocks does not imply that the household is fully insured
Preview of the results:• Following a health shock to an adult working
men, the household increases:– Total household consumption– Health care expenditure– Food consumption– Non-health care consumption– A few more items such as fuel and transportation
• However, children’s weight decreases• Which is evidence against full insurance despite
the increase in household consumption• It is very important to understand intrahousehold
behaviour when testing for full insurance
Data
• Familias en Accion:– Conditional Cash Transfer programme in Colombia– Mothers receive a cash transfer every two months– The cash transfer is the sum of the nutritional subsidy
and the educational subsidy– The mother receives the nutritional subsidy is she has
at least one child under 6 and all her children under 6 are up to date with preventive visits
– The nutritional subsidy is independent of the number of children
– The educational subsidy is per child that regularly attends school
Data
• Familias en Accion (FeA) Survey:– Survey collected to evaluate the FeA programme– Three waves of data collected:– Wave 1: Summer of 2002– Wave 2 : July to November 2003– Wave 3 : December 2005 to March 2006
– Municipalities smaller than 100,000 inhabitants and could not be capital of a “department”
– Sample of eligible households (the poorest of the poorest)
Data
• Very poor households:
• The average family size is 7
• Average consumption is about 114 US dollars per month
• Share of food consumption in total consumption is 73%.
• 20% of children are chronically disnourished
Data
• Health shocks• We consider health shocks of individuals that
have worked for paid at least once in their lives• Shock=1 if the individual has been unable to
perform his/her daily activities in the last 15 days due to health reasons
• Moreover, we differentiate the shocks according to gender and age of the person that suffers the shock:– Male/Female & 12-17, or 18-64, 65 or older.
Data• Dependent variables• Household consumption in US$, 2003 prices• Children anthropometric variables:
– Height per Age z-score– Weight per Age z-score– Weight per Height z-score
• Weight per Age z-score for child i = (Wi – Mi)/SDiWi = Weight of child iMi = Median weight amongst children of the reference
population with the same age and sex as child iSDi = Standard deviation of weight of those children of
the reference population with the same age and sex as child i
Empirical specification for child’s anthropometrics
3 ,2 ,3
1
3
1
tXssy ihmtmtihmtj
Fhmt
Fj
j
Mhmt
Mjihmt
i=individual, h=household, m=municipality, t=time
SjM = health shock suffered by male in age group j
SjF = health shock suffered by female in age group j
Θmt = municipality – time fixed effect (to control aggregate shocks)
X = household composition, variables that reflect eligibility to FeA, child’s age
Standard errors are clustered at the municipality level to consider autocorrelation
The empirical specification for household consumption is the same, but without the subscript I
Results: Household Consumption (US$)
Results: Anthropometrics children 0-10
• Household consumption increases after a health shock of a 18-64 old person
• Food consumption increases for males (usually the person that brings more income)
• Results are consistent with the household trying to improve the health of the breadwinner
• Is the household fully insured against health shocks?
• Exactly for the same type of shocks (male 18-64), we see that children’s weight decreases
• But food consumption in the household increases
• There must be an intrahousehold reallocation of resources
Results: specific items
Conclusions• When a male 18-64 suffers a health shock
• The household increases consumption (including food) but children’s weight decreases
• Evidence that the household is not fully insured and there is an intrahousehold relocation of resources
Conclusions• The paper gives a warning about using the
null correlation between consumption growth and shocks as evidence of full insurance
• Risk associated with health shocks is not fully insured by health insurance as there are other expenses (food, transportation, fuel…) that might need to increase with health shocks