This paper provides a background for the interpretation of empirical evidence of the net savings effects of tax based incentive programs. The degree to which savings placed in tax sheltered accounts crowds out savings in other forms plays a critical role in assessing how well such programs achieve their goals, and in measuring their fiscal cost. Empirical evidence has been conflicting. However, recent work has indicated that some of each dollar contributed represents new savings. Findings of substantial 'created' savings trouble many economists because of the absence of a theoretical basis for such a result. 'Mental accounting' and 'learning process' explanations are explored, but are not found to offer clear directions for policy.Final Draft, August 1998: PDF.