The study has an ultimate purpose in finding the long-run influence of total factor productivity (TFP) in agriculture to the poverty. Agricultural valued added per worker is used as the representative to total factor productivity in agriculture as Patmasiriwat and Suewattana (2010) and Nidhiprabha (2005) had initiated. Thirtle et al. (2003) recognized the possibility of time lag in adjusting TFP to the poverty incidence. In their study, they applied one-period lagged explanatory variables to Asia, Africa and Americas cases which results to be the critical factors for the poverty. This study then will establish one-period lagged TFP besides its current variable into the model for the case of Thailand. Autoregressive model is employed to be the reference model since the concept of the model matches with an expectation and the various studies which already mention above. By doing so, Ad-hoc estimation framework3 is introduced as it can be observed the prior experiment that there is one-period lagged in explanatory variable (Gujarati and Porter, 2009). However, this study observes an insignificant of the current TFP in agriculture to the poverty incidence. It means that the current TFP does not have a strong impact to the poverty in the same period. Moreover, one- period lag dependent variable is also included in order to get rid of an autocorrelation problem4, which causes the estimation to be unreliable.