Parental income pooling on educational expenses: Evidence from Tanzania

How to cite

Schürz, Simon. 2018. "Parental income pooling on educational expenses: Evidence from Tanzania." AEA RCT Registry. AEARCTR-0002672. https://doi.org/10.1257/rct.2672-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Simon Schürz*

* Primary investigator

Registry ID
AEARCTR-0002672
DOI
https://doi.org/10.1257/rct.2672-1.0
First registered
Last updated
Status
In development
Trial dates
2018-01-22 – 2018-02-23
Countries
Private
JEL codes
O1, D1, D7
Registry record
http://www.socialscienceregistry.org/trials/2672

Abstract

In developing countries women’s avoidance of bargaining with their spouse about the allocation of household resources has been suggested as a source of inefficiency. In particular, women seem to shield resources for educational investment for children from their husbands. Using a lab-in-the-field experiment this project studies whether parental preference heterogeneity and female empowerment in the household cause such behavior consistent with a non-cooperative model. Besides a direct cost, shielding resources from the spouse implies an efficiency cost from the lack of income pooling, which, in developing countries, allows access to indivisible investment goods and insures certain expenses against income shocks. In the study I am able to vary and measure efficiency losses directly. In the experimental design parents are asked to choose for or against joint decision-making and income pooling on a consumption versus educational voucher choice. I set up and test a simple model of bargaining avoidance and income pooling to address the roles of preference matching, bargaining powers, taking into account the potential endogeneity of marital matching. By additionally eliciting beliefs about the spouse's preferences, the study opens up a behavioral channel to understand household efficiency. Parental decision-making is further related to child outcomes. Understanding the underlying mechanism of an important occurring inefficiency of parental decision-making will allow to discuss improvements in cash transfer policies.

Intervention

The intervention consists of a decision-making experiment for parents to measure the level of efficiency in their resource allocation.
The treatment is an experimentally varied budget size in a parental decision-making experiment. Each treatment implies a different efficiency gain for parents that pool their income to choose jointly how to allocate resources between cash payout and educational investments for their child (rather than choosing individually).
Several determinants that have been suggested by economic theory (preference heterogeneity, female empowerment in the household, beliefs) will be tested for their impact on parental efficiency.
Using the random 'efficiency treatment', it can additionally be checked how parents with particular characteristics react to the experimentally varied treatment. Eliciting choices for four treatment levels per parent, an empirical strategy using family fixed effects regression allows to deal with the endogeneity of marital matching and an estimation of causal impacts of the above named efficiency determinants.

Experimental design

The experimental session is conducted with families (mother, father, one child). All subjects will participate in a patience experiment. Fathers and mothers take part in a decision-making experiment. Mothers additionally attend an experiment designed to measure their empowerment in the household. In the following the designs of all three parts are described in detail:
Patience Elicitation: Time preferences are elicited with choice lists with early and delayed payoffs or time-investment exercises. The choice lists used in the experiment additionally allow for a measure of present-biasedness. Subjects will select their preferred choice in a list of 10 decisions per decision sheet. For instance choices are made between upfront payoff of 5000 TSh and delayed payoffs between 5.000
and 10.000 TSh (1 USD ~ 2180 TSh). Starting with equal payoffs the delayed option increases monotonically. Two decision sheets are presented to the subjects in random order. After all choices are made, one list and one decision are randomly selected for payout. Particular focus is placed on trust issues related to the delayed payment option. Using the observed switching point from early to delayed payoff, the future equivalent (FE) is calculated at the midpoint of the two delayed payoffs around the switching point. Normalizing it by the early payoff results in a comparable indicator for patience. The higher the normalized future equivalent, the more impatient is the individual.
Measure for female empowerment in the household: To measure decision-weights in the household I will apply two approaches. The first part consists of a set of question on the decision-making process across various subcategories, such as food, educational expenses and financial decisions. For these questions tie-breaker questions are included to avoid bunching at the joint decision option. Additionally
women’s empowerment, i.e. their decision weight, in the household is measured using an incentivized experiment. Women’s willingness to pay to receive a cash transfer themselves rather than their spouse is directly measured. The design follows a choice list type representation of pre-specified values, for which the mother chooses that the amount is paid either to the father or to herself. The amount for the father remains constant, while the mother’s amount decreases monotonically. Full information on payouts, but not on actual experimental choices, is given to both the woman and the spouse, thereby excluding hiding motives. By randomly assigning starting values, it can be assured that spouses cannot learn about women’s choices from the payouts.
Decision-making experiment: Each parent is asked to make an individual decisions, separated from the partner. In the first stage, the individual decision is elicited. The parent is asked to indicate his or her preferred division of a 10.000 TSh budget (around 4,5 USD) between a consumption good and an educational voucher for the child. The voucher represents a commitment device for child investment. The value of the voucher relative to the consumption option is multiplied by a factor two, such that any the monetary value dedicated to the voucher is twice the value of the cash option. The vouchers, which can be used to purchase high-priced stationary such as text books, exercise
books and mathematical sets can be used in the stationary show in direct vicinity to the school. The revealed preference for educational investment is measured by the budget share allocated to the voucher. Subjects are additionally asked to mark their belief about the spouse’s preference (incentivized).
In the second stage, income pooling decisions are collected for several treatment levels: Each parent is asked to indicate whether to stay with the individual decision from stage one or to switch to joint decision-making with the spouse for a number of ’joint premium’ levels that are reflected in increased budgets to be divided between consumption and voucher. This means the choice in the joint option will differ from
the individual stage only in its pay-off structure. For each first stage choice there are four second stage decisions, where the joint budgets are treated with -20, 0, 20 or 40 % excess payoff. Depending on the choice in the second stage the final payout will be determined. One of the final choices will then be randomly selected for payout. If a parent has chosen the individual option for that choice, then the payout is determined from the decision in the first stage. If a parent has chosen the joint option for that choice, the decision will be made together with the spouse.
Within the data collection process, subjects in the experimental session also participate in a distributional preference experiment for an additional research project. For the same reason children take part in patience and distributional preferences experiments.

Design details

Randomization method
Within subject treatment
Randomization unit
Individual
Clusters
10-15 school classes (30-40 children + their parents) from 5 schools
Observations
400 families, 800 parents
Arms
400 families, 800 parents: within subject treatment
Primary outcomes
Efficiency loss in any decision in the decision-making experiment (dummy variable, at parent and family level) Efficiency loss in monetary terms (continuous variable, at parent and family level)
Secondary outcomes
Child outcomes (school and behavioral grades) Parental educational investment for child (yearly, monetary terms)

Cited by

No citation data available for this registration yet.

Keywords