A conceptual model specified in terms of primary path relations and feedback effects is first presented. This model is subsequently used as the basis for empirical investigations which draw on data for the period 1960-79. A statistical technique (Granger-Sims test) which is capable of explicitly allowing for the time-related interdependencies within and between the series is used. The results of estimation strongly suggest a positive relation for crime rates to incarceration rates and a negative relation for incarceration rates to crime rates. The public policy implications of these results are discussed, particularly with respect to Nagel's prison moratorium proposal. (Author abstract)