Essays on Retirement Behavior during the Housing Boom and Bust

TitleEssays on Retirement Behavior during the Housing Boom and Bust
Publication TypeThesis
Year of Publication2018
AuthorsFu, W
Academic DepartmentEconomics
DegreePhD
Number of Pages112
UniversitySyracuse University
Thesis Typephd
ISBN Number9780438102330
Keywords0501:Economics, 0510:Labor economics, Economics, Housing wealth, Labor economics, Mortgage foreclosure, Retirement, Saving behaivor, Social Sciences
Abstract

This study contains two chapters. The first uses the Prentice-Gloeckler-Meyer proportional hazard model with individual heterogeneity to investigate the effects of loss aversion concerning the housing market and the local foreclosure rate on retirement during the housing bust periods. The second chapter creates a dynamic programming life-cycle model with the housing wealth and uses the Method of Simulated Moments to systematically study the retirement and saving behavior during the housing boom and bust (the years 2000–2014). Housing wealth is one of the biggest savings for elderly. It relates to the financial security of elderly after retirement. After the incredible growth of housing prices in the early 2000s, the housing market melted down at the end of the year 2007. A tremendous decline in property value caused a high uncertainty about the housing market. Even though elderly were not sure how severe the housing bust would be, they knew the highest value of home equity before the Great Recession. In the first chapter, we use this highest value at the year 2006 to measure the loss aversion concerning housing wealth. Higher housing equity at the year 2006 might experience more loss in the Great Recession. When there was a loss of housing wealth, it increased the uncertainty of financial resources in the future. Delaying retirement and working more years to increase savings are a reasonable plan to improve resources. For the same amount of housing wealth loss, the effect is not the same if elderly live in a different area and a different housing market. The expectation of housing market performance is also not the same. We have high-quality data on local foreclosure rates from Equifax. It provides the number of foreclosures starting in the first week of July from year 2005 to 2012 on the zip-code level. We use the local foreclosure rate to approximate the expectation of the local housing market. Coefficients of both home equity at the year 2006 and local foreclosure rate (except the year 2009) are significant and negative, meaning elderly with higher home equity at the year 2006 and elderly who live in an area with a higher foreclosure rate significantly delay their retirement. In the second chapter, we create a dynamic programming life-cycle model based on French and Jones (2011). We still take into account the risks of wage, health status, mortality and medical cost in our models. Because we study the elderly after the year 2000, the ‘Senior Citizens’ Freedom to Work Act of 2000’ that eliminates the Social Security earnings test after normal retirement age is applied. We use the re-entry state variable to control the labor force participation when there is no Social Security earnings test. New models separate the housing wealth from total wealth in the original model and consider the housing wealth through two constraints: the baseline model has an unknown proportional housing wealth in the asset accumulation equation; the modified model has a home equity borrowing constraint. New models also take into account housing wealth change in the bequest motive component. Both the baseline and modified models match the labor-force participation well and capture the high exit rate at the Medicare age. The coefficient of unknown proportional housing wealth in the baseline model indicates that elderly takes into account approximately 25 percent of their housing wealth in the asset accumulation, which, coincidently, is close to the average ratio of loan to value in the data. The modified model matches better than the baseline model in the asset quantile moments (saving behavior). Robust checks show the bequest coefficients significantly change if we do not separate housing wealth from total wealth. Surprisingly, change of bequest curvature is close to the mean of the housing wealth. Three experiments are conducted in the second chapter. We experiment with two different housing wealth projections and one tighter borrowing constraint. The results indicate that loss of housing wealth and tight borrowing constraints delay retirement. Even though we use the long-term growth rate and obtain a similar mean of labor-force participation rate, the curves significantly shift to adjust the new expectation of housing wealth change.

Notes

Copyright - Database copyright ProQuest LLC; ProQuest does not claim copyright in the individual underlying works; Last updated - 2018-07-16

URLhttps://surface.syr.edu/etd/846/
Citation Key10309