In this project innovative computational econometric approaches that employ simulation are adapted in order to handle parametric estimation of joint discrete/continuous decisions with long panels. The methods are then applied to a family of long- standing econometric estimation problems. The underlying econometric models follow from dynamic optimization by economic units (such as individuals or firms) in the presence of liquidity and other similar quantity constraints. The proposed research makes these simulation estimation methods a flexible and powerful tool for the estimation of dynamic optimization problems that involve discrete/continuous decisions with switching among several regimes. Stringent parametric specification assumptions are relaxed by means of semiparametric estimation techniques. Starting from recent developments in the field of parametric simulation estimation, estimation techniques are investigated that combine semiparametric methods with simulation. Such methods are particularly important for the discrete/continuous econometric models naturally arising from the theoretical framework. The methods developed are applied to several important dynamic constrained problems. The firms application involves the problem of consumption and labor supply decisions augmented with a study of housing decision, using data from the Panel Study of Income Dynamics, and cross-checking the results with data from the Survey of Consumer Finances. The second major application uses data from COMPUSTAT and studies firms' decisions, where modes of investment finance define distinct qualitative regimes and where attrition may be due to bankruptcy and other types of exit.
