Premium
Technology and Technical Efficiency Change: Evidence from a Difference in Differences Selectivity Corrected Stochastic Production Frontier Model
Author(s) -
BravoUreta Boris E.,
GonzálezFlores Mario,
Greene William,
Solís Daniel
Publication year - 2021
Publication title -
american journal of agricultural economics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.949
H-Index - 111
eISSN - 1467-8276
pISSN - 0002-9092
DOI - 10.1111/ajae.12112
Subject(s) - propensity score matching , baseline (sea) , productivity , difference in differences , matching (statistics) , production (economics) , frontier , impact evaluation , significant difference , stochastic frontier analysis , control (management) , econometrics , economics , average treatment effect , treatment and control groups , statistics , economic growth , mathematics , geography , microeconomics , political science , archaeology , management , law
This study presents a new approach to evaluate the economic impact of development projects for cases where baseline and endline data for comparable treated and control samples are available. An important contribution is to bring together propensity score matching (PSM) and difference‐in‐differences (DID) techniques, commonly used in impact evaluation studies, with stochastic production frontiers (SPF) that have become well‐established in the productivity literature. To illustrate our proposed framework, we use baseline and endline data from a rural environmental development program implemented in Nicaragua between 2012 and 2016. The results support the use of the approach proposed and reveal that selectivity from unobservables can be significant. The analysis shows that a severe drought over the period analyzed had significant negative effects on both control and treated farms. However, project beneficiaries enjoyed significantly better results, attributable to the project, compared to the control group. Overall, the results exhibit relatively low levels of technical efficiency with no significant variation across models, time, and treatment status.