Farm Employment Transitions: A Markov Chain Analysis with Self-Selectivity

A stationary, first-order Markov chain model with selection bias correction for legal status is estimated by maxixmum likelihood methods using the National Agricultural Worker Survey data for 1989-2004 to evaluate the likelihood of workers staying in U.S. agriculture by legal status. Although the conditional steady state probability in US agriculture is highest for uanauthorized workers, there is little difference between legal statuses. Simulations of the estimated model indicate that a legal status change for unauthorized workers would result in only small changes in the steady state probability of being in US agriculture, particularly after 2001.