The objective of this research proposal is to use the Health and Retirement Survey (HRS) and the companion Pension Provider Survey (PPS) to study the evolution of household's economic resources during the immediate pre-retirement period. Previous has shown that pensions are critical to two of the most fundamental decisions households make during this time: when to retire and how much to save support themselves during retirement. Because the government provides insurance for household's well-being in retirement through the Social Security system, a clear understanding of the choices households make to affect their own well- being is critical to making informed policy choices. The combination of the HRS and PPS represents the best dataset that has ever been available to study these topics. This proposal is comprised of four related projects. The first provides documentation and output of the software that has been developed to compute entitlements in the PPS. Special software is required due to the vast differently complexity and structure of pension formulas in the economy. This project will make this unique dataset accessible to a broad range of users with varying computer and research experiences. The second project provides a comprehensive summary of the level and distribution of pension and Social Security entitlements. This project is intended to be a reference for later work. The third project uses the variation in the accrual of pension and Social Security wealth as an individual continues to work to estimate a structural model of the timing of retirement. This model will then be used to simulate the consequences of policy changes on the cost of providing retirement income support. The last project examines the extent to which households with higher pension and Social Security wealth approach retirement with lower levels of other saving. A high rate of offset suggest that government and employer efforts to promote retirement saving ar either misguided or ineffective and has implications of understanding saving behavior more generally.
