This paper develops an approach to estimating the effect of government employment and training programs on measured unemployment. The theoretical aspects of the method draws heavily on earlier work on labor market flow equilibrium (see note). Previous estimates of the direct or statistical impact of government programs on the unemployment rate have not been based on a general equilibrium model and no attempt was made in these studies to constrain the labor market to be in equilibrium. This paper describes an empirical procedure based on Markov flow equilibrium and then implements it with data on youth employment and training programs. Models with both constant and variable transition probabilities are analyzed. The results are contrasted with results of the alternative methodology used in previous studies. (Author)