Factors impacting price for retail space in Houston.

Hanna, Michael E. ; Caples, Stephen C. ; Smith, Charles A. 等

ABSTRACT

This paper investigates the demand for retail space in the Houston area from 1981 to 2006. Economic factors such as the employment level, new retail space constructed, vacancy rate, and average price per square foot are presented. Relationships between these variables are studied to determine which variables might be most helpful in forecasting future price for retail space.

INTRODUCTION

This paper investigates the demand for retail space in the Houston area from 1981 to 2006. Data is presented on economic factors such as the employment level, new retail space constructed, change in retail space absorbed into the market, vacancy rate, and average price per square foot. Relationships between these variables are studied to determine which variables might be most helpful in forecasting future price for retail space.

LITERATURE REVIEW

Estimating future price for retail space has historically presented problems to practitioners in the field. Supply and demand are constantly at work in the market place. The creation of new jobs in an area increases the demand for retail space, which would normally result in higher prices for the space. This higher price spurs additional construction of retail space, which increases the supply of retail space. The increased supply would normally provide downward pressure on the price. There is a constantly changing dynamic economy.

Malizia (1991) recognized that long-term demand-side forecasting models needed to include economic development variables in forecasting demand for retail space. Wheaton and Torto (1990) linked job growth to industrial supply and demand. There is a plethora of empirical data linking employment to various factors influencing demand for real estate, or methods and models to forecast one aspect of real estate or another. Valente, Wu, Gelfand and Sirmans' (2005) present a spatial model for predicting apartment rents. Lentz and Tse (1999) present models to relate the performance and needs of the goods market to the demand for commercial real estate.

To effectively forecast retail space demand and the price for that space, a relationship needs to be established between readily available employment information and retail demand forecasts. Lentz and Tse (p. 231) noted, "The commercial real estate market is frequently observed to be in an extended state of disequilibrium." Since there is a time lag between the beginning of the construction cycle and the time when the finished space is available for rent, it can be difficult to make an accurate estimate of future space demands. It is common to overbuild or fail to build enough space simply because the market changed at some point during the construction cycle. The decision to build new retail space should be made after weighing expectations of future demand, retail space under construction, absorption rates and the amount of vacant space already in the market. Lentz and Tse (p. 248) observed," With future demand uncertain, the supply (quantity) of space and the realized demand for space may not match. If the supply is less than the realized demand, the space producer will be able to lease out all the new space. On the other hand, if the supply is greater than the realized demand, the excess supply will cost the space producer holding costs on the vacant units." With this background, we investigate the market for retail space in the Houston area from 1981 to 2006.

THE DATA

Employment data for the Houston area was collected from the Texas Employment Commission, and retail space market data was provided by REVAC, Inc. All data was located either online or in print form. The Texas Workforce Commission publishes quarterly and annual economic statistics on their website, separated by city and type of employment. This information was used to determine overall Houston non-agricultural employment, changes in the Consumer Purchasing Index, and the percentages of goods producing and manufacturing jobs. These data provide a record of historical growth, and are helpful in making estimates of future economic growth. Data relating to the retail space market was also collected. The most critical for our purposes is the absorption of retail space--the difference between space built and space leased. Additional variables include market vacancy rates, average rent per square foot and the amount of new space constructed.

Table 1 contains historical employment data for the Houston metropolitan area. These data are available online, and similar data are available in most major cities. These data are usually updated several times per year. A practitioner in the field can use these to analyze and draw relationships between the variables. Table 1 presents employment data for the Houston area from 1975 to the present. While total employment over the years has fluctuated, there has been a trend of overall growth in the economy since the recession of the 1980's. Since 1988, the Houston economy has grown at a rate of 2.47% per year. This is an average increase of almost 45,000 new jobs per year. While overall employment is up, employment in goods producing and manufacturing jobs has seen a decline over the last decade. A similar pattern has existed for the rest of the United States because of a shift to a more service oriented and knowledge-based economy.

Table 1 indicates moderate to strong economic growth in the Houston job market. This employment growth should cause demand for existing retail space to increase. The question is--by how much? Table 2 contains retail space market data which can be used to determine the connection between job growth and retail space demand. Some relationships become obvious once the data are assimilated. For instance, there is a relationship between vacancy rate and the percentage change in the market rent. Table 2 also shows that overbuilding has occurred in the last several years, since construction has outpaced absorption. The excess retail space in the market has begun to cause a slow down in the increase in average market rent. Meanwhile, the vacancy rate has been increasing yearly despite substantial job growth in the market.

There will always be some vacant space in the market. This is sometimes called the natural vacancy rate. A vacancy rate of about 14 percent in the Houston area since the mid-80s is observed in the data. The average annual retail space constructed per year in Houston has averaged 3.2 million square feet, while the absorption rate has averaged only 2.9 million square feet per year.

As can be seen in Figure 1, construction lags behind absorption, and it seems to react to changes in absorption. Perhaps a better understanding of forecasting retail space demand would benefit the market as a whole. This might bring about a decrease in the market vacancy rate, and the average rents may increase.

Figure 2 provide the changes in price per square foot and the retail space vacancy rate for the years 1982-2006. The 1980s were difficult years for the Houston economy. Employment fell, the vacancy rate increase, and prices dropped as providers of retail space offered price incentives to keep their space occupied.

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

THE MODEL

To predict the price of commercial retail space, the use of a regression model was investigated. The following variables which are candidates for predicting the price include the change in employment, the change in square footage of space absorbed, the vacancy rate, and the change in space available each year. These same variables are also lagged one year to see if there is a lag in the impact. The variables are defined as follows:

P = Price per square foot

E = Change in employment

[E.sub.1] = Change in employment lagged 1 year

F = Change in square footage of space absorbed

[F.sub.1] = Change in square footage of space absorbed lagged one year ([F.sub.1])

V = Vacancy rate (V)

[V.sub.1] = Vacancy rate lagged one year ([V.sub.1])

S = Change in retail space available (S)

[S.sub.1] = Change in retail space available lagged 1 year ([S.sub.1])

With these variables, we have

P = f(E, [E.sub.1], F, [F.sub.1], V, [V.sub.1], S, [S.sub.1])

While price is a function of all of these variables, several of these independent variables are correlated with other independent variables. Minitab was used to analyze the data, and a Best Subsets stepwise regression model was run on these data to determine which of these variables were significantly contributing to the price of retail space. The overall best model included E, F, [F.sub.1], V, and [V.sub.1]. The equation is

P = 26.2 + 0.0414 E - 0.631 F - 0.435 [F.sub.1] - 0.375 V - 0.276 [V.sub.1]

The coefficient of determination is 0.53. The positive coefficient for E is expected as additional jobs would typically result in an increase in demand for retail space. The negative coefficients for the other variables are also to be expected. As a decrease in price is usually associated with an increase in absorption (number of square feet occupied), we would expect the coefficients for F and [F.sub.1] to be negative. Similarly, as the vacancy rate (V, and [V.sub.1]) increases, the price would normally decrease.

SUMMARY AND CONCLUSIONS

If the relationship between the economic variables in the Houston real estate market can be better understood, perhaps overbuilding could be avoided. This reduction in overbuilding will have positive effects on the retail space market as a whole, as the surplus of vacant retail space will be absorbed and the vacancy rate will decline, raising the average rent commanded by the market. However, some builders may still choose to overbuild, as vacant land generates no revenue. These builders feel that they are better served by building the retail space and having it partially vacant as opposed to building less space or holding vacant land. While this could be a profitable choice by the individual producers of retail space, the overall market may be hurt as rents may drop and vacancy rates may rise.

There are many factors that impact the price of retail space in the Houston market. The most important variables found in this study are change in employment for the current year, change in square footage of space absorbed for the current and previous year, and change in vacancy rate for the current and previous year. However, this model should not be expected to forecast with complete accuracy. With a coefficient of determination of 53%, the unexplained variability in price for retail space is 47%. While this model should help in predicting the price for retail space, further study needs to be performed to identify other variables that would generate better predictions.

REFERENCES

Benjamin, John D., Glenn W. Boyle & C. F. Sirmans (1990). Retail Leasing: The Determinants of Shopping Center Rents. AREUEA Journal 18(3), 302-312.

Lentz, George H. and K.S. Maurice Tse (1999). Supply Adjustments to Demand Shocks in the Commercial Real Estate Market. Real Estate Economics 27, 231-262.

Malizia, Emil E. (1991). Forecasting Demand for Commercial Real Estate Based on the Economic Fundamentals of U. S. Metro Markets. The Journal of Real Estate Research 6(3), 251-265.

Valente, James, ShanShan Wu, Alan Gelfand, and C.F. Sirmans (2005). Apartment Rent Prediction Using Spatial Modeling. The Journal of Real Estate Research 27(1), 105-136.

Wheaton, William C. and Raymond G. Torto (1990). An Investment Model of the Demand and Supply For Industrial Real Estate. AREUEA Journal 18(4), 530-547.

Michael E. Hanna, University of Houston-Clear Lake

Stephen C. Caples, McNeese State University

Charles A. Smith, University of Houston--Downtown

Charles P. Rollins, Houston, Texas

Table 1. Houston Employment Data (In 1,000s)

Year Total Goods Goods Mfg Mfg As Change in
 Wage & Prod. Prod. Jobs % of CPI
 Salary Jobs As % Goods
 Jobs

1975 993 309 31.1% 170 55.1%
1976 1,057 332 31.4% 176 53.0%
1977 1,126 351 31.2% 182 51.8%
1978 1,229 387 31.5% 199 51.6% 9.4%
1979 1,318 415 31.5% 214 51.7% 13.2%
1980 1,399 440 31.5% 225 51.2% 12.5%
1981 1,517 496 32.7% 249 50.3% 10.0%
1982 1,541 482 31.3% 230 47.7% 6.9%
1983 1,444 402 27.9% 181 45.1% 2.8%
1984 1,476 390 26.5% 178 45.5% 2.7%
1985 1,479 368 24.9% 173 47.0% 2.1%
1986 1,410 320 22.7% 153 48.0% -1.0%
1987 1,386 294 21.2% 146 49.8% 2.5%
1988 1,448 310 21.4% 156 50.5% 2.8%
1989 1,515 328 21.7% 164 50.0% 4.1%
1990 1,605 397 22.2% 201 49.4% 5.7%
1991 1,630 401 22.5% 206 49.7% 3.7%
1992 1,631 389 22.1% 202 50.0% 3.2%
1993 1,659 385 21.1% 202 51.2% 3.3%
1994 1,704 397 21.0% 207 50.2% 3.4%
1995 1,756 411 21.1% 216 50.4% 1.4%
1996 1,981 427 21.6% 225 52.7% 2.1%
1997 2,064 443 21.5% 235 53.0% 1.9%
1998 2,167 467 21.6% 243 52.2% 1.0%
1999 2,202 457 20.8% 235 51.5% 1.3%
2000 2,254 465 20.7% 231 49.7% 3.7%
2001 2,293 477 20.8% 233 48.9% 3.0%
2002 2,288 465 20.3% 221 47.6% 0.3%
2003 2,274 448 19.7% 210 46.8% 2.8%
2004 2,289 440 19.3% 207 47.2% 3.5%
2005 2,350 453 19.3% 212 46.9% 3.7%
2006 2,477 480 19.4% 217 45.3% 0.7%

Table 2. Houston Retail Market Data

Year Total Emply- % Absorp- New Vacancy
 Wage & mnt Change tion in Square Rate
 Salary Change Square Feet Footage
 Jobs

1975 993
1976 1057 64 6.5%
1977 1126 68 6.4%
1978 1229 103 9.2%
1979 1318 89 7.2%
1980 1399 80 6.1% 3621
1981 1517 118 8.5% 4808 5402 9.5%
1982 1541 24 1.6% 6314 2317 6.0%
1983 1444 -97 -6.3% 4261 5524 2.5%
1984 1476 31 2.2% 5075 11887 12.8%
1985 1479 2 0.2% 3267 8756 16.7%
1986 1410 -69 -4.6% -1565 3404 19.4%
1987 1386 -25 -1.7% -1363 388 22.5%
1988 1448 61 4.5% 1228 1324 22.5%
1989 1515 67 4.7% 1978 884 21.6%
1990 1764 90 5.9% 967 1110 21.6%
1991 1793 24 1.5% 2883 1025 10.3%
1992 1795 1 0.1% 4251 2836 17.4%
1993 1827 27 1.7% 2670 2069 15.9%
1994 1815 45 2.7% 4835 4520 15.1%
1995 1934 52 3.1% 2751 3383 14.8%
1996 1981 39 2.3% 1945 2451 15.0%
1997 2064 82 4.2% 4091 1836 13.8%
1998 2167 103 5.0% 4090 1470 11.5%
1999 2202 34 1.6% 6701 3871 7.5%
2000 2254 52 2.4% 4845 3934 6.0%
2001 2293 39 1.7% 6294 9218 7.0%
2002 2288 -6 -0.2% -2975 4394 11.5%
2003 2274 -15 -0.6% 1976 5430 13.3%
2004 2289 15 0.7% 3568 4813 13.5%
2005 2350 61 2.7% 722 3211 14.7%
2006 2477 73 3.1% 1890 3045 15.2%

Year Rent/ %
 Sq.Ft Change

1975
1976
1977
1978
1979
1980
1981 $9.96
1982 $11.40 14.5%
1983 $12.60 10.5%
1984 $13.10 4.0%
1985 $13.14 0.3%
1986 $12.59 -4.2%
1987 $11.06 -12.2%
1988 $11.30 2.2%
1989 $11.92 5.5%
1990 $13.19 10.7%
1991 $13.70 3.9%
1992 $13.87 1.2%
1993 $14.30 3.1%
1994 $14.59 2.0%
1995 $15.50 6.2%
1996 $15.52 0.1%
1997 $17.13 10.4%
1998 $17.68 3.2%
1999 $18.45 4.4%
2000 $18.33 -0.7%
2001 $19.07 4.0%
2002 $18.33 -3.9%
2003 $19.10 4.2%
2004 $19.15 0.3%
2005 $19.38 1.2%
2006 $19.52 0.7%