The Role of Real Estate Investment in Insurance Company Portfolios

Abstract

[Excerpt] An area of ongoing debate in the area of long-term investment has been what is the appropriate role of real estate in an institutional portfolio, Numerous studies using a mean-variance setting have suggested that pension funds should allocate more of their capital to real estate with a 20 percent maximum suggested. However, these conclusions are tenuous because of the severe difficulties in determining the risk and return on real estate that realistically captures the illiquidity and other aspects of real estate pricing. Perhaps, because of these problems, real estate investment by major insurance companies has been relatively small. While real estate makes up about 17.4 percent of the affiliated investments of property-casualty insurance organizations on average, only 1.3 percent or $7.9 billion of their 1991 portfolio of nonaffiliated investments consisted of real estate. Of the amount invested in nonaffiliated real estate, over 77.9 percent of funds assigned to real estate were invested in mortgages while only 22.1 percent of real estate investment consisted of investment properties. The purpose of the current study is to explore the relative benefits of equity and debt real estate with respect to the portfolio of a representative property-casualty insurance company. In investigating the role of real estate in the portfolio allocation process, the modified internal rate of return (MIRR) of Graaskamp (1972) is used which focuses on cash flows and directly incorporates the long term nature of real estate investment. In addition to this, this study uses two unique data sources on equity and mortgage real estate returns to try to estimate the effects of changing insurance company investment in these forms of real estate investment. The second section discusses the methodology used in the portfolio allocation process while the third section describes the data sources used in this study. The fourth section reports our findings and compares these results to the prior literature. The fifth section concludes the study.