Credit Constraints and Capital Misallocation in Agriculture: Theory and Evidence from Uganda

2017-01-13

2018-02-19

The unit of analysis is the household throughout. Survey and willingness responses are provided by the household head wherever possible, and crop yields are measured by us at the plots. We measure
1. The empirical distribution of willingness-to-pay (WTP) for 1 sack of DAP and 1 sack of CAN fertilizer (worth approximately 200,000 UGX in total), and how this distribution changes when farmers receive a 200,000 UGX cash transfer shortly prior to WTP elicitation.
2. The average return to randomly receiving fertilizer (measured by differences in quantity of maize, value of crop output, or profit, respectively), and how this changes when farmers receive a 200,000 UGX cash transfer shortly prior to WTP elicitation.
3. Linear and nonlinear estimates of the relationship between returns to fertilizer and WTP, and how it changes when farmers receive a 200,000 UGX cash transfer shortly prior to WTP elicitation.

Farmers are given the opportunity to purchase fertilizer (a bundle of one 50 Kilogram sack of DAP and one 50 Kilogram sack of CAN fertilizer) through a Becker-DeGroot-Marschak mechanism. The mechanism elicits their maximum willingness-to-pay (WTP) for the fertilizer (between 0 and 200,000 UGX, in increments of 20,000) and then charges a random price. If WTP exceeds the price, the farmer purchases the fertilizer at the random price, otherwise they do not purchase. This allows us to estimate returns by comparing output or profits of those who do and do not receive the fertilizer. A treatment group receives a cash transfer of 200,000 UGX shortly prior to WTP elicitation (the control group receive a 5,000 UGX consolation prize).

We run the study in two six-month seasons, beginning January 2017 and January 2018 respectively.

Assignment to receive the 200,000 UGX cash transfer is randomly assigned by computer, stratified (as closely as possible) by village.
Assignment to fertilizer depends on bidding in the Becker-DeGroot-Marschak WTP elicitation. Respondents receive the fertilizer if their bid exceeds their individual price. Prices are pre-assigned randomly by computer at the individual level, stratified by lottery assignment and (as closely as possible) by village, from a distribution that assigns 41% of the mass each to prices of 0 or 200,000 UGX and the remaining 18% uniformly across the range {20,000, 40,000, …, 180,000}.

The unit of randomization is the household

No

1232 households in 154 villages. Of which 412 households in 51 villages in Season 1, and 820 households in 103 villages in Season 2.

1232 households, of which 412 in season 1 and 820 in season 2.

769 Individuals were assigned to the “consolation” cash transfer (5,000 UGX) and 463 were assigned to the full cash transfer (200,000 UGX). Before Willingness To Pay elicitation survey, individual price assignment was {(501,0); (25, 20,000); (25, 40,000); (25, 60,000); (24, 80,000); (24, 100,000); (24,120,000); (25, 140,000); (24, 160,000); (23, 180,000); (512, 200,000)}, where for each pair, the first number represents the number of individuals assigned to one of the prices, and the second number is the pre-assigned individual price. At the end of the Willingness To Pay elicitation survey, 592 Individuals purchased the fertilizer and 636 did not.