Production Efficiency in Pig Farming in Thailand: A Copula-Based Frontier Model

Abstract

This study evaluates the technical efficiency of pig production in Nakhon Si Thammarat Province, Thailand, using a copula-based stochastic frontier framework that relaxes the conventional assumption of independence between stochastic noise and technical inefficiency. Farm-level data from 280 pig-farming households were analyzed, with the Gaussian copula selected as the preferred dependence specification based on model-selection criteria. The results indicate a mean technical efficiency of 0.5207, suggesting substantial scope for improving production performance under existing input conditions. Breeder pigs, labor, medicines and veterinary inputs, feed, utilities, and farm size are positively and significantly associated with pig output, whereas housing investment is not statistically significant. The estimated input elasticities sum to 1.5592, and a Wald test rejects the null hypothesis of constant returns to scale, providing evidence of increasing returns to scale. In the second-stage analysis, household income, farming experience, and participation in Good Agricultural Practices (GAP) are significantly associated with lower technical inefficiency. These associations remain broadly consistent under a fractional logit specification with bootstrapped standard errors. Additional robustness checks using alternative stochastic frontier distributions, treatment of extreme observations, and a Translog production specification indicate that the main results are broadly stable across alternative specifications. The findings indicate that technical-efficiency estimates are sensitive to model specification and are consistent with the emphasis on GAP and improved farm management in Thailand’s Swine Action Plan 2023–2027.