Panel ARDL and NARDL Models: Role of Agriculture in Economic Growth in Selected Asian Countries

In recent macroeconometric literature, the nonstationary panel models have attracted more attention. Conventional panel data approaches assume all parameters share the common value across homogeneous cross-section units and the regressor has the symmetric effect on the dependent variable. Conventional panel models might not be applicable to macro-level panel data that characterize nonstationary and heterogeneous effects. Besides, the asymmetric effect could also be present in the relationship in which the increases of a regressor might have a different magnitude of effect on the dependent variable as compared to its decreases. Taking into these issues, this chapter will introduce the panel autoregressive distributed lags model (ARDL) and its nonlinear form (NARDL) using the three estimators, namely, mean group (MG), pooled mean group (PMG), and dynamic fixed effect (DFE) to handle the heterogeneous and asymmetric effect in modeling the long panel data. The chapter will then demonstrate the application using country-level data from Southeast and South Asia. ARDL and NARDL models are estimated using EViews and Stata to examine how the global commodity prices and each sub-agriculture productivity might contribute to economic growth, and if such effects are symmetric or asymmetric. We find that the model based on PMG estimator (Pesaran et al., J Am Stat Assoc 94:621–634, 1999) outperforms other models.