Investigating Agricultural Production Relations Across Bangladesh, India and Pakistan Using Vector Error Correction and Markov-Switching Models

This paper has investigated the short- and long-term relationships as well as regime switching behaviour between the per-capita agricultural production of Bangladesh, India and Pakistan using vector error correction model (VECM) and Markov-switching VECM model (MS-VECM). These countries were selected due to their high agricultural economic importance to the South Asian region. The study has used FAO's statistical dataset and the study period was 1961-2010. The residual diagnostics have validated the models since there is no deficiency in normality, autocorrelation or heteroskedasticity. The empirical results have confirmed the existence of two long-term cointegrating vectors between the variables and have demonstrated that an unexpected shock to the respective log per-capita agricultural production of India and Pakistan cause transitory impacts. On the other hand, an unexpected shock to the log per-capita agricultural production of Bangladesh causes a permanent disequilibrium in all variables. Finally, MS-VECM model has shown two volatility regimes (i.e. low and high volatility). The Markov-switching impulse responses have indicated that, one, agricultural production adjustments in the first regime are smoother than those in the second regime, and two, these adjustments are faster in the case of a shock to the agricultural production of Pakistan than to the agricultural production of either Bangladesh or India