The returns to homeownership: an MSA level analysis.

Brown, Christopher L. ; Chhachhi, Indudeep S.

ABSTRACT

This paper examines the returns to homeowners across 208 Metropolitan Statistical Areas (MSAs) over the period 1989 to 1999. We find a significant difference in returns to homeowners across MSAs, with the highest returns in the North Central United States and the lowest returns in New England and the Middle Atlantic states. We also find income growth and the percentage of renters in the MSA impact the returns to homeowners.

INTRODUCTION

Single-family housing--by virtue of its sheer size--represents one of the most significant individual investment categories for the U.S. economy. Its overall importance has been amply demonstrated recently as the housing sector has come across as a rare bright spot in our current economic downturn. Approximately two thirds of U.S. households participate in this sector. For most homeowners, the housing investment is their largest single investment. Therefore, the importance of the housing sector to individuals cannot be overstated.

Previous research in this area has focused on risk and return in the housing market. Articles in the popular press have analyzed the rent versus buy decision in a general framework (Fortune, 1994). Using the example of a hypothetical $100,000 home and a number of assumptions with regard to rate of price appreciation, holding periods and tax treatment, Peach (1988) demonstrates that most homeowners can expect a relatively good return on their investment over the years. A number of studies have utilized nationwide data and a number of others have focused on regional market data. Using data for four large metropolitan areas, Case and Shiller (1990) demonstrate that price changes are a function of factors such as construction costs and changes in adult population. Crone and Voith (1999) examine the risk-return relationship in a single market using a very comprehensive database that extends over twenty years. In an effort to fully comprehend all segments of this market, Pollakowski, Stegman, and Rohe (1991) concentrate their efforts on the low-medium end of the income scale. Nationwide and/or regional studies have given us very valuable preliminary insights on returns to homeownership and how they compare with other asset classes.

Almost all studies in this area indicate a strong need for us to better understand the variations across different markets. Returns on housing investments vary considerably across cities. Chinloy and Cho (1997), for example, find the correlation between returns on housing in different cities can be very low or negative. Each Metropolitan Statistical Area (MSA) has unique characteristics that impact the returns to homeowners. Factors like the MSA's population growth, the growth of the labor market in the MSA, the supply of rental property, and the level of new housing construction vary across different markets. Additional factors that influence the price levels and appreciation rates of homes in an area, including property and income tax rates, also have distinctive local or statewide characteristics. These significant differences in economic and demographic characteristics and the low correlation of housing returns across cities point to the need for analysis at the MSA level. A recent study by Jud and Winkler (2002) uses MSA level data to investigate the factors that impact real housing price appreciation. They find population growth, real changes in income, construction costs, and interest rates influence real housing price appreciation.

This study extends the work of Jud and Winkler (2002) by measuring the returns to homeownership for 208 MSA's in the United States from 1989 to 1999. One factor that influences the returns to homeowners is home price appreciation. Jud and Winker (2002) show that home price appreciation varies greatly across MSA's. Other factors that influence returns to homeownership include real estate taxes, maintenance and insurance costs, state and local income taxes (through the use of itemized deductions), and the difference between homeownership costs and the cost to rent. The next section of the paper discusses the data and methodology. The final section presents some preliminary findings.

DATA AND METHODOLOGY FOR RETURN CALCULATIONS

The median home price for 1989 for each MSA and the home price index published by the Office of Federal Housing Enterprise Oversight (OFHEO) are used to estimate changes in the market value of the median residence over the ten year period. The analysis assumes a home is purchased at the median home price with a 20 percent downpayment and a 30 year mortgage. The initial interest rate on the mortgage is 10.13 percent. This is the average 30 year conventional loan rate for 1989. The analysis assumes the loan is refinanced in January 1993 at a rate of 8.022 percent. Refinancing the loan at that time is rational given the interest rate changes and assumed holding period. We estimate real estate taxes based on the property tax rate for each MSA. Annual property insurance and maintenance costs are assumed to be 1.5 percent of the market value of the property. The median rent in 1989 and the fair market rent for a three bedroom dwelling in 1999 are used to construct the average rent variable over the ten year period.

We calculate the cash flows associated with purchasing and holding the median home in each MSA over the ten year holding period. The initial investment is the 20 percent downpayment. The annual cash flows are the annual costs of homeownership minus the average rent for the MSA. The annual costs of homeownership include the principal and interest payments on the loan, real estate taxes, property insurance, and maintenance costs. We also estimate the tax benefits of owning a home in the MSA using state and local income tax data, the annual interest paid on the loan, and the estimated real estate taxes. The tax benefits are estimated for investors in the 15 percent, 28 percent and 36 percent marginal tax brackets.

The terminal cash flow is from the sale of the property. The analysis assumes the home is sold at the end of the ten year period. The net sales proceeds are estimated as the market value of the property in 1999 minus six percent selling costs.

Three internal rate of return calculations are performed for each MSA. The first calculation assumes the homeowner will receive no tax benefits. The second calculation assumes the homeowner has a marginal tax rate of 15 percent. The third and fourth calculations assume the homeowner's marginal tax rates are 28 percent and 36 percent, respectively.

STATISTICAL TESTS

After calculating the returns for the 208 MSA's in the sample, we employ an ordinary least squares (OLS) regression model to determine which factors influence housing returns. Any factors used in the computation of the return are omitted from the regression model. Our model includes five dummy variables to reflect regional differences in home prices. Previous research indicates there should be differences in home price appreciation for different regions of the country. While our research is focused on returns to the housing investment, not just home price appreciation, it is important to allow for regional differences.

The regions are divided based on regions set by the U.S. Census Bureau. Region 1 is comprised of New England and the Middle Atlantic states. Region 2 comprises the Middle Atlantic states. Region 3 includes states in the South Central U.S., while Region 4 is comprised of states in the North Central U.S. Region 5 is comprised of states in the Mountain region. Region 6 is the Pacific region. In the statistical analysis, the Pacific region is used as the default region. Therefore, the parameter estimates for each region compares the return in that region to the return in the Pacific region. Table 1 lists the states that are included in each region.

We also include the change in income from 1989 to 1999. Homebuyers in areas with significant income growth are more likely to bid up the price of homes and increase the returns to existing homeowners. This is partly due to homeowners' desire to sell their homes to purchase larger (and more expensive) homes. Significant income growth also makes it possible for families to move from renters to homeownership.

The percentage of occupied homes that are rented is also included in our model. The higher the percentage of renters, the lower the demand for owner-occupied housing. Less demand for owner-occupied housing should result in slower home price appreciation and lower returns to housing investments. Finally, we include the percentage of homes that are vacant. A higher percentage of vacant homes indicates a higher supply of housing. Therefore, higher vacancy percentages should be associated with lower returns on housing investments.

FINDINGS

Without considering the tax benefits of homeownership, the highest returns are earned by homeowners in Duluth, Minnesota (15.26 percent), followed by Salem, Oregon (14.32 percent) and the Salt Lake City-Ogden, Utah MSA (14.06 percent). Homeowners in the Connecticut and New Hampshire areas earn the lowest returns. The MSA's with the lowest returns are Hartford-New Britain-Middletown, Connecticut (-41.11 percent), New Haven-Meriden, Connecticut (-29.0 percent), and Waterbury, Connecticut (-27.16 percent).

When the tax benefits of homeownership are considered, the rankings change slightly. For taxpayers in the 36 percent marginal tax bracket, the highest returns are earned in Portland, Oregon (16.99 percent), Salem, Oregon (16.78 percent) and Eugene-Springfield Oregon (16.37 percent). The MSA's with the lowest tax-adjusted returns are Hartford-New Britain-Middletown, Connecticut (-30.33 percent), New Haven, Meriden, Connecticut (-20.67 percent) and Waterbury, Connecticut (-20.44 percent).

The effect of the tax deductibility of interest expense on the returns varied widely across MSA's. Homeowners in 17 of the MSA's in the sample received no tax benefits from homeownership. This is largely due to the modest home prices in these MSAs. However, it is also due to lower than average state and local income taxes. Nine of these MSA's are in Texas and three are in Tennessee. Neither of these states has a state income tax.

The results of our regression model are shown in Table 2. The parameter estimates on the region dummy variables represents the difference in returns between the region in question and the Pacific region. There is a significant difference in returns to housing investments across geographic regions. Returns on housing are significantly lower in the New England and Middle Atlantic states than in any other region of the country. The highest returns are in the North Central region of the U.S. Interestingly, the closest returns to the Pacific region are in the Middle Atlantic region. The Middle Atlantic region is the only region with an insignificant dummy variable, indicating that returns in the Middle Atlantic region are not significantly different from returns in the Pacific region. The other regional dummy variables are significant at the .05 level or below.

We also find that income growth is positively related to returns to housing investments. The parameter estimate on income growth is positive and significant at the .0001 level. This indicates that MSA's experiencing positive income growth are more likely to experience positive returns to housing. In areas with strong income growth, the demand for housing is higher, leading to higher returns for existing homeowners. As income grows, more renters become homeowners and existing homeowners look to move into higher quality residences.

The percent of properties that are vacant does not provide any additional insight into the returns on housing investments. The percent of properties that are renter occupied is negatively related to the returns to housing investments (significant at the .0001 level). As stated earlier, a higher concentration of rental properties provide a larger inventory of properties that can be acquired without purchasing a home. In this case, it is more likely that individuals will choose to rent rather than own their own homes.

CONCLUSIONS

Our findings are consistent with the findings of Chinloy and Cho (1997) and Jud and Winkler (2002). Returns to housing investments vary greatly across MSAs. We also find returns are correlated within particular regions of the country, with the lowest returns in the New England and the Middle Atlantic states and the highest returns in the North Central U.S. Demographic factors such as the percentage of renters and the income growth in an MSA influence the returns to homeowners.

REFERENCES

Case, K. E. & R. J. Shiller (1990). Forecasting prices and excess returns in the housing market. AREUEA Journal, 18(3), 253-273.

Chinloy, P. & M. Cho (1997). Housing returns and restrictions on diversification. Real Estate Finance, Fall, 57-63.

Crone, T. M. & R. P. Voith (1999). Risk and return within the single-family housing market. Real Estate Economics, 27(1), 63-78.

Federal Reserve Economic Database, Federal Reserve Bank of St. Louis (2000). Thirty year conventional mortgage rates. Retrieved January 2001 from the World Wide Web: http://www.stls.frb.org/fred/data/irates/mortg.

Jud, G. D. & D. T. Winkler (2002). The dynamics of metropolitan housing prices. The Journal of Real Estate Research, 23(1-2), 29-45.

National Low Income Housing Coalition (NHLIC), Fair market rents for 1999. Retrieved December 2000 from the World Wide Web: http://www.nlihc.org/oor99.

Office of Federal Housing Enterprise Oversight (OFHEO), House Price Index, Second Quarter 2000. Retrieved September 14, 2000 from the World Wide Web: www.ofheo.gov.

U.S. Census Bureau, (1990) Census. Retrieved January 2001 from the World Wide Web: http://www.venus.census.gov/cdrom/lookup.

Why owning is still a better deal than renting. Fortune, 78, November 14, 1994.

Yahoo Real Estate: City Profiles. Retrieved January 2001 from the World Wide Web: http://verticals.yahoo.com/cities/profile.html.

Christopher L. Brown, Western Kentucky University

Indudeep S. Chhachhi, Western Kentucky University

Table 1: States Included in Each Region

Region States

 1 Connecticut, Massachusetts, Maine, New Hampshire, New
 Jersey, New York, Pennsylvania, Rhode Island, Vermont

 2 Delaware, Florida, Georgia, Maryland, North Carolina,
 South Carolina, Virginia, West Virginia

 3 Alabama, Arkansas, Kentucky, Louisiana, Mississippi,
 Oklahoma, Tennessee, Texas

 4 Iowa, Illinois, Indiana, Kansas, Michigan, Minnesota,
 Missouri, North Dakota, Nebraska, Ohio, South Dakota,
 Wisconsin

 5 Arizona, Colorado, Idaho, Montana, New Mexico, Nevada,
 Utah, Wyoming

 6 Alaska, California, Hawaii, Oregon, Washington

Table 2: Regression Results

Variable Parameter t-Statistic P-value
 Estimate

Intercept -0.79405 -0.20 0.8454
Region 1 -10.31105 -6.28 <.0001
Region 2 1.13345 0.75 0.4562
Region 3 3.71512 2.56 0.0112
Region 4 4.12398 2.85 0.0048
Region 5 3.74611 2.04 0.0424
Income Growth 0.31719 5.84 <.0001
Percent Renters -0.32417 -4.43 <.0001
Percent Vacant 0.02022 0.16 0.8715