Generalized Autoregressive Conditional Heteroskedasticity (GARCH) effects have been studied extensively in portfolio management within risk-return framework. Very few studies, however, have applied principles underlying GARCH processes to house prices and fewer have undertaken studies at the local level. Earlier GARCH studies in housing markets have focused on aggregate effects across markets or taken a macro-oriented approach. This paper estimates and develops GARCH models for each major county in California, allowing for different effects in each. Unlike previous GARCH studies, this paper employs dollar denominated house prices instead of indices to measure returns. The results indicate that there are GARCH effects in four counties in Northern California and at the state level.
(ProQuest: ... denotes formulae omitted.)
Many market participants including real estate investors, banks, non-bank financial institutions, portfolio managers, are interested in coming up with a model which accurately predicts the local housing prices. Managers of banks, Real Estate Investment Trusts (REITs), and homebuilding companies are interested in the accuracy of such housing prices predicting models, as it directly impacts their activities relating to exposure management, hedging, arbitraging, and investment and financing decisions. Policymakers frequently monitor housing prices to understand better their impact on local employment, revenue generation at county and state levels, and potential for foreclosure and bailouts. Nowadays, more attention is being focused on housing prices due to the increasing number of foreclosures and defaults on subprime mortgages.
Review of Literature
The housing sector has been studied extensively in macroeconomics. Some studies such as Stockman and Tesar (1995), Girouard and Blöndal (2001), and Lane (2001), consider housing prices in a dynamic general equilibrium model. Other studies have looked at housing prices in the context of 'bubbles', though research has shown that 'bubbles' are empirically difficult to validate. For instance, Driffill and Sola (1998) do not find any difference between bubbles and switching processes. In his study of property prices in Taiwan and stock movements, Chen (2001) was unable to conclude the presence of a bubble (Chen, 2001). Ortalo-Magne and Rady (1998, 1999, 2003a and 2003b) have done extensive research on the interactions between housing prices, housing transactions, demographic changes, changes in income distribution and aggregate economic activity. Traditional models explaining house prices have used locational factors (Do Quang et al., 1994; and Mok et al., 1995), structural variables (Do Quang et al., 1994; Do Quang and Grudnitski, 1995; and Mok, 1995) such as, floor area and median gross income (Asabere and Huffman, 1992).
According to Shiller (1998), homeowners consider investment in their homes to be one of the largest risks that they have undertaken. More than half of the assets of middle-class American families are in housing (Campbell and Cocco, 2003). Savings and portfolio choices faced by the homeowners have been extensively studied by economists1, as cost of housing is a significant portion of a household's disposable income. The housing sector affects the economy through wealth effects (Case et al., 2001) and has significant impact on related asset markets such as, the market for appliances, mortgage market, mortgage insurance, CDO, swaps and other derivative markets. According to the statistics compiled by the Federal Reserve System, households held about $14.6 tn in real estate at the end of 2003:3 which translates to about 28% of households' assets and is more than 130% of GDP. In this perspective, households held about $12.8 tn of corporate equities and mutual funds during the peak of the stock market in 2000:1. In addition, equity holdings are concentrated in the hands of the wealthy, whereas housing is the major asset for most households (Tracy et al. …