The booming economy of the 1990s was characterized in the Dallas Region by the rapid rise of many new high-tech industries; robust growth in the financial services sector in tandem with increases in personal wealth and income; burgeoning international trade resulting from the implementation of the North American Free Trade Agreement; and an almost insatiable demand for residential, commercial, and industrial space that drove increasing levels of construction activity. Supporting this economic growth was a substantial in-migration of businesses and residents seeking to benefit from the Region's competitive advantages, including lower housing costs and strong job growth. Table 1 shows that most metropolitan statistical areas (MSAs) in the Dallas Region continued to outperform the nation in employment, personal income, and population growth. Today, however, these competitive advantages have eroded over time, caused by the same rapid growth that made the Dallas Region boom initially, and now threaten to constrain future growth in some areas. In particular, the rapid increase in housing costs in a few MSAs is creating concern among local officials and residents about housing affordability and economic growth.
The relationship between employment growth and home appreciation values reflects the strength of individual housing markets. We examined 12 MSAs in the Dallas Region over a five-year period (1995-2000). A strong positive correlation exists between employment growth and rising home prices1 (see Chart 1).
1 The following linear regression model developed by the Dallas Region office was estimated: Y = , where Y is the dependent variable Conventional Mortgage Home Price, X1 is the independent variable Payroll Employment Growth Rate (MSA), and are unknown parameters, and is the error term. This model proved to be statistically significant, with the independent variable explaining 41 percent (R2) of the variation in the dependent variable. The model home price appreciation can be calculated by manipulating the equation Y = 0.9265x + 0.0169.
The MSAs of Austin-San Marcos, Boulder-Longmont, Colorado Springs,and Denver were among the leaders in employment growth and home price appreciation and are the focus of this section as we explore whether housing costs are indeed getting out of hand. We will look first at the factors that drive housing demand and supply and determine housing prices.
Factors Influencing Housing Demand
The demand for housing is strongly influenced by at least three factors: (1) demographic shifts (population growth, net migration, and the expansion and contraction of households); (2) rising and falling employment; and (3) wealth and income. These factors, interacting with changes in the supply of housing and economic conditions (e.g., interest rates), influence the pace of housing construction, and ultimately home prices.
New housing demand is fueled by population growth, particularly households migrating directly into the area. All four MSAs were among the fastest-growing in the United States during the 1990s. Austin-San Marcos, ranked sixth in the nation in population growth, grew at an annual rate of 3.4 percent, followed by Colorado Springs (18th) with an annual growth rate of 2.6 percent. Denver (34th) and Boulder-Longmont (38th) each posted annual growth rates of 2.2 percent. In comparison, U.S. population growth averaged 0.9 percent annually throughout the 1990s. The Austin-San Marcos MSA added nearly 300,000 new residents during the 1990s, and the combined MSAs of Denver, Boulder-Longmont, and Colorado Springs added over half a million residents.2
Net domestic in-migration accounted for much of this population growth (see Chart 2) and represented 42 to 57 percent of the total gains in population experienced by these MSAs from 1991 through 1999.3 Population growth generated by natural increase (births minus deaths) does not necessarily lead to immediate demands for new housing units. However, significant increases in domestic in-migration will likely strengthen demand for housing in the short term, as these households must establish new residences.
Attracting these new migrants were strong underlying economies characterized by rapid job growth, low unemployment rates, and high median family incomes. The four MSAs ranked among the top 10 percent of more than 300 U.S. MSAs in nonfarm employment growth since 1994 (see Chart 3). Moreover, they have experienced very low levels of unemployment--from 2.1 percent to 3.3 percent as of July 2000--compared with the U.S. unemployment rate of 4 percent.4 Strong job growth and low unemployment are two major reasons why a large influx of new residents moved to these MSAs during the 1990s. The in-migration of new residents created an immediate demand for new housing as opposed to population growth generated by natural increase alone.
A key similarity among the four MSAs is their strong reliance on high-tech industries (see third quarter 2000 Dallas Regional Perspectives). Households in these MSAs also share an above-average reliance on stock market gains (often used to finance home purchases), another reason why housing demand has performed so strongly in these markets. According to an article written by RFA/Economy.com's chief economist Mark Zandi entitled "Judging Regional Economies," differences in regional economic performance can be explained, in part, by the "new economy"; however, downside risks also exist in these MSAs, as described in the following quote:
Those regional economies with large and thriving high-tech industries are thus the regional economies that are reaping the largest gains in income and wealth. Colorado, California, Massachusetts, Minnesota, Texas, Utah and Washington are the prototypical new regional economies...[however] the high-flying new regional economies may ultimately price themselves out of the new economy--witness ...the spiking house price gains in the Bay Area of California or in New York... 5
In addition, a U.S. Department of Housing and Urban Development report entitled "State of the Cities 2000" found that while the technology-driven expansion has raised wages, the resulting booming economy has also caused home prices to rise at twice the rate of inflation over the 1997-1999 period.6
In addition to demographic and employment trends, housing demand is also heavily influenced by household income patterns, which have implications for the quantity and quality of housing that households can afford. Two sources of data are indicative of households' ability to afford housing: per capita personal income and median family income. Once again, all four MSAs exhibited more rapid rates of per capita personal and median family income growth than the national averages (see Chart 4). The demand for housing during the past decade clearly has been influenced by the strong gains in income.
New residential construction, reflected by building permits issued, is cyclical in nature, and is generally correlated with underlying economic conditions in each MSA. Although building permits are not a perfect indicator of additions to the housing stock, they are a good indicator of changes in housing supply.
The strong expansion in residential homebuilding in the 1990s (particularly in recent years) is prompting concerns about overbuilding in these markets, drawing comparisons with the real estate crisis of the mid-1980s. However, residential building permit data do not support this argument. Chart 5 shows the trend in residential building permits per 1,000 residents for the four MSAs during the past two decades. Even though total permits at the close of the decade were above the lows of the late 1980s and early 1990s, the number remains far below the overbuilt highs of the early 1980s, which were one contributing factor to the real estate crisis of the 1990s.
Housing Affordability Decreases as Housing Costs Outpace Median Family Income
Housing costs in these four MSAs have outpaced gains in median family income (see Chart 6). Consequently, affordability has suffered in these areas, particularly in recent years. The Housing Opportunity Index compiled by the National Association of Homebuilders (see Table 2) shows the four MSAs ranked between 123rd and 134th in housing affordability out of 173 MSAs7 surveyed in second quarter 2000. Families earning the median income in these MSAs could afford only slightly more than half of all homes available for sale, well below the percentage for other U.S. families.8
7 The MSAs were ranked from most affordable (1) to least affordable (173).
The acceleration of housing prices in recent years has been also the result of spiraling costs for land and building materials and the growing scarcity of skilled construction workers. Complicated and lengthy development procedures (e.g., delays in the permitting process, local government regulations, zoning, inspections, etc.), in addition to rising fees and building restrictions, are also driving up construction costs. The economic boom of the 1990s, which has contributed to the high cost of housing, is pushing residents to the fringes of these four MSAs, resulting in urban sprawl and exacerbating traffic congestion.
Despite several years of strong gains in homebuilding activity, all four MSAs continue to experience strong housing demand in excess of supply, causing an increase in housing prices that has accelerated in recent years. We now examine more closely what factors have contributed to the escalation in housing prices in each of these MSAs.
Austin-San Marcos MSA
A thriving economy caused largely by an expanding high-tech industry and influx of new residents (approximately 2,000 each month) is contributing to a hot housing market. Employment growth during the 1990s helped absorb much of the excess supply of housing from the 1980s. The tremendous run-up in stock prices and the addition of many high-paying jobs in the second half of the 1990s have contributed mightily to the rapid appreciation of new and existing homes, particularly expensive homes for the nouveau riche.
Home prices have risen faster in the past two years. The median price of a new home in Austin-San Marcos rose 7.3 percent in 1999, to $126,500. However, by June 2000, the median price was $151,100, an increase of 18.2 percent from a year earlier. As recently as May 1997, Austin-San Marcos had more than 6,000 homes listed for sale, or a 6.8-month supply. In February 2000, however, there were slightly fewer than 3,000 active listings, or a two-month supply.9 The average single-family home in the MSA is on the market only 30 to 35 days before it is sold.
At the same time, residential building permits in the Austin-San Marcos MSA (January through July 2000) registered double-digit gains in both single-family and multifamily permits, and are on pace to surpass last year's peak in homebuilding activity during this expansion. Yet the inventory of new and existing homes remains very low because strong job growth and rapid net in-migration have homebuilders struggling to keep up. The results are rising home prices and lengthening homebuilding schedules.
Denver MSA
Denver MSA job growth remains robust, expanding at a rate of 3.3 percent for the 12-month period ending June 2000.10 Job and population growth is outpacing housing production, driving housing costs upward. The price of a single-family home in Denver appreciated an average of 12.1 percent in 1999. The average price of an existing single-family home in the Denver MSA rose to a record $250,000 in August 2000, up 19.3 percent from a year earlier.11 Meanwhile, the median price--believed to be a better measure of affordability--broke the $200,000 barrier for the first time. Homes priced under $200,000 are increasingly hard to find. Resales are spending an average of 38 days on the market in 2000, compared with a 44-day average in 1999. Permits for new, single-family homes for the first seven months of 2000 dropped 13.4 percent from the same period a year ago.12 Demand for housing is strong, and builders are unable to replenish the existing stock at a rate that is commensurate with demand.
Employment figures for Boulder-Longmont reveal a 3 percent growth rate through the first seven months of 2000. The MSA's unemployment rate of 2.6 percent in July 2000 was well below the national rate,13 requiring a steady influx of in-migrants to sustain its strong employment growth. However, strong employment conditions, coupled with rising land and housing costs, are eroding the cost advantage that local residents enjoyed at the beginning of the 1990s. According to the Federal Home Loan Mortgage Corporation's (FHLMC) repeat-.purchase house price index, the price of existing homes in Boulder-Longmont increased more than 15 percent in second quarter 2000 from a year ago.14
Robust economic growth is creating a strong demand for housing. This demand, in addition to the current shortfall in supply, is causing home prices to soar. Severe growth pressures (e.g., traffic congestion, demands on the infrastructure) are mobilizing the public to demand that local governments restrict further development. Tight labor markets, the shortage of affordable housing, and mounting antigrowth sentiment may limit expansion and constrain future growth.
Colorado Springs MSA
High-tech manufacturing, national defense, tourism, and a large cadre of military retirees drive the Colorado Springs economy. This MSA is experiencing moderate job growth (3.2 percent, January through July 2000),15 very low levels of unemployment, relatively strong in-migration, and rapid income growth. In recent years, Colorado Springs has attracted a large number of businesses because its cost of doing business is lower than the neighboring Denver and Boulder-Longmont MSAs. A study conducted by Coldwell Banker ranked it as the most affordable housing market in Colorado.16 Nonetheless, growth pressures, such as urban sprawl and congestion, are emerging problems.
16 The 2000 Coldwell Banker Home Price Comparison Index is based on a single-family dwelling model with approximately 2,200 sq. ft., 4 bedrooms, 21/2 baths, family room (or equivalent), and 2-car garage. Surveyed homes and neighborhoods are typical for corporate middle-management transferees. http://www.coldwellbanker.com/FRAMES/TopFrame.asp?Item=hpci.
According to the Colorado Legislative Council, an advisory council to the state legislature, Colorado Springs continues to enjoy a "sustained real estate boom."17 According to the FHLMC repeat-purchase house price index, the price of existing homes in Colorado Springs increased 7.4 percent during second quarter 2000 from a year ago.18 However, home prices have not accelerated as rapidly in Colorado Springs as they have in neighboring Denver and Boulder-Longmont. One reason may be that although employment growth has been strong in all three MSAs, median family income has not grown as rapidly in Colorado Springs.
Housing construction activity in the Dallas Region likely will remain strong into 2001, although off slightly from its 1999 peak. The decline since 1999 can be attributed to a softening in demand for homes, albeit from already high levels; comparatively higher mortgage rates; and weaker growth in capital gains. One of the most significant downsides to the Region's booming economy is that, at this stage of the cycle, incomes are failing to keep up with escalating housing prices.
Several supply constraints have contributed to this growing gap between income and housing prices. Lack of skilled construction labor, a shortage of developable lots, and increasing antigrowth sentiment are major impediments to new home construction. The tight housing market is putting a squeeze on potential homebuyers and renters as they face rising home prices and rents resulting from the limited supply of housing. This trend is making it more difficult for employers to attract entry- and mid-level workers. Housing availability is a key factor in the continued economic strength of these MSAs.
Throughout the 1990s, housing production in these MSAs has played catch-up, with demand for housing stimulated largely by strong job growth. Economic theory suggests that when housing grows scarce and less affordable, households will respond by deferring homeownership, by "doubling and tripling up" occupancy, or by commuting longer distances from where housing may be more affordable. Short of an economic slowdown in these areas, demand for housing is likely to remain strong, with home prices continuing to rise rapidly. Unless mortgage interest rates, which have come down 50 basis points since peaking at 8.52 percent in May 2000 (monthly average),19 start to climb again, healthy employment conditions and favorable consumer attitudes will continue to generate robust housing demand. In the long term, the Austin-San Marcos, Boulder-Longmont, Colorado Springs, and Denver MSAs must try to balance continued strong economic growth with intensifying antigrowth sentiment.
19 FHLMC, Federal Reserve Board (Haver Analytics).
Insured financial institutions in the Dallas Region have increased their relative lending volume to the highest level in the past 14 years. This loan growth has risen rapidly over the most recent 21/2-year period. The Region's insured institutions reported a loan-to-asset (LTA) ratio as low as 43.1 percent for year-end 1990,20 compared with 60.2 percent as of June 30, 2000. This section examines what factors have contributed to the increase in loan levels, looks at insured institutions experiencing the most rapid rate of loan growth, and considers the implications for banks and thrifts from these shifts in asset composition. Emerging risks include the potential for increasing levels of credit risk, expanding loan exposure at a time when the economy could be slowing, and an allowance for loan and lease losses (ALLL) that has lagged the Region's rapid level of loan growth and that is already the lowest among all FDIC Regions.
20 Excludes NationsBank--Texas, which consolidated with NationsBank in June 1998.
Factors contributing to why Dallas Region banks and thrifts have increased loan levels include relatively low LTA levels and a strong demand for loans generated by the robust Dallas Region economy. A discussion of the recent trend in loan growth is not complete, however, without a look back at why the Region's insured institutions have held a greater proportion of securities than loans since the late 1980s.
Background
Dallas Region insured institutions have placed greater emphasis on securities than have banks and thrifts elsewhere in the country since the banking crisis of the late 1980s and early 1990s. Reasons may stem from the memories of large credit losses of the 1980s,21 favorable funding sources,22 or both. In any case, Dallas Region insured institutions have performed well compared with banks and thrifts elsewhere in the country despite this difference in asset allocation.
21 See Regional Outlook, fourth quarter 1999, pp. 24-25.
22 See Regional Outlook, second quarter 2000, p. 9.
The Dallas Region's ability to hold a high percentage of securities and maintain profitability was helped by a favorable interest rate environment. However, the shape of the yield curve from 1997 through 1998 provided incentives for insured institutions to extend maturity distributions in order to attain higher yields. For example, mortgage-backed securities (MBS) with maturities in excess of 15 years held by the Region's insured institutions stood at 63 percent of total MBS as of year-end 1999, up from 40 percent two years prior. In comparison, 48 percent of MBS portfolios nationwide were held in long-term securities, up from 33 percent over the same period. This strategy was successful while interest rates were steady; however, the extended maturity distribution also elevated the potential for greater interest rate and market risk.
In fact, beginning June 30, 1999, the Federal Reserve began raising short-term interest rates at the same time the U.S. Treasury was buying back longer-term bonds.23 The net effect was generally rising interest rates and a yield curve that inverted at the three-month mark. Banks holding large MBS portfolios with extended maturities saw the value of these securities begin to fall, and the overall yield was sub par, particularly compared with loans.24
23 Ibid.
24 Ibid. Chart 7 shows how a rising interest rate environment affected securities available for sale.
The Region's Banks and Thrifts Have Increased Loan Levels
In an environment in which loan demand was strong and a strategy of holding a large share of securities was less effective because of rising interest rates, many insured institutions began increasing loan levels to maintain profitability. Historically, loans have returned higher yields than securities. As shown in Table 3, assets held by Dallas Region insured institutions25 grew 22 percent between year-end 1997 and June 30, 2000, a slightly higher rate than for the nation. At the same time, loan volume increased 33 percent for the Region, compared with 23 percent for the nation. While the percentage of assets dedicated to loans increased for both the Region and the nation, the Region's loan portfolio has grown more rapidly.
25 For insured institutions in the Dallas Region as of June 30, 2000, merger adjusted.
Moreover, the types of loans showing the greatest increases include real estate construction (95 percent), commercial real estate (54 percent), and commercial and industrial (33 percent), all of which historically exhibit higher levels of risk than many other loan categories.
In addition to the shift into traditionally higher-risk assets, Table 3 shows the difference between insured institutions headquartered in metropolitan and rural areas. In general, insured institutions' loan portfolios in metro areas grew more rapidly than did portfolios of rural-based institutions for all types of loans except single-family and 1- to 4-family residential loans. Particularly noteworthy is the fact that insured institutions in metropolitan areas grew real estate construction loans 24 percentage points faster than did those in rural areas. Metro-based institutions also increased commercial real estate and consumer loans more rapidly than did rural-based institutions.
Dallas Region insured institutions were able to increase loan levels in part because they were less "loaned up" than institutions elsewhere in the country. They were therefore able to grow loan portfolios without the heavy use of noncore funds experienced elsewhere. In fact, deposits fell slightly for the four quarters ending June 30, 2000, and other borrowed funds increased by only 2 percentage points (as a percentage of total assets).
The net result is that the LTA ratio for the Region's insured institutions is increasing faster than that of the nation, and the historical gap between the LTA ratio for the Region and the nation is narrowing (see Chart 7).
A strong regional economy also contributed to the increased loan volume among the Region's banks and thrifts. The Region has outpaced the nation in gross output, employment, and population growth. The gross regional product growth from 1997 through 1999 averaged 6.1 percent,26 compared with the nation's 4.3 percent gross domestic product growth. The Region's average employment growth rate was 3.4 percent, outpacing the U.S. average of 2.5 percent. The Region reported a 1.6 percent population growth, compared with the U.S. average of 0.9 percent. These indicators illustrate the Region's economic vibrancy, which likely translates into an increased demand for loans.
26 Average growth rate is an average of growth rates for 1997, 1998, and 1999.
To compare their vulnerability with that of other institutions, we analyzed banks and thrifts with assets less than $1 billion that experienced the greatest increases in loan volume from December 31, 1997, through June 30, 2000. Interest rates began to rise during this period, and the securities component of the asset portfolio continued to decline to a level closer to the U.S. average. After adjusting for mergers and acquisitions and removing de novo institutions,27 we ranked the institutions by loan growth rate. We then examined the insured institutions in the top 25 percent and compared their characteristics with those of other banks and thrifts in the Region and the nation.
27 Banks established after December 31, 1997, are not included for two reasons. First, de novo institutions tend to exhibit rapid loan growth, which may distort the analysis. Second, these institutions did not report loans or assets prior to year-end 1997.
The 324 banks and thrifts28 in the top 25 percent exhibit very distinct trends. Changes after June 30, 1999--when short-term interest rates began their 175-basis-point climb--are particularly enlightening. This group's loan volume grew much faster (31.9 percent for the four quarters ending June 30, 2000, compared with 10.6 percent for other institutions in the Dallas Region29 and 11.2 percent for institutions nationwide). Of particular interest is the increase in real estate construction loans (46 percent), commercial real estate loans (39 percent), and commercial and industrial loans (32 percent), categories that have been historically more risky. The shift into loans and traditionally higher-risk loan categories has contributed to an increase in the group's net interest margin. In contrast, insured institutions in the nation have experienced margin compression (see Table 4). However, the ALLL-to-total-loans ratio declined to 1.11 percent for the institutions exhibiting the most rapid loan growth rates, significantly lower than the ratio for either the Region or the nation.
The results of our analysis reflect the trends relating to institutions doing business in metro and rural areas. Banks and thrifts in the high-loan-growth group tended to be in the Region's metro areas, and these institutions increased lending significantly faster than did institutions in rural areas--35 percent for metro versus 26 percent for rural. As shown in Table 4, the metro high-loan-growth group also holds a higher percentage of loans and the highest net interest margin of all institutions in the Region or nation as of June 30, 2000. Banks and thrifts with a significant allocation30 of commercial and industrial loans were by far the largest subgroup within this high-loan-growth group, representing almost 60 percent of all insured institutions in the analysis.
30 Significant allocation is defined as institutions with more than 25 percent of total loans in the following categories: commercial and industrial, real estate construction, multifamily, and commercial real estate lending.
Implications of a Significant Shift in Asset Composition
Previously, banks and thrifts were willing to accept interest rate and market rate risk because the interest rate environment was stable and funding costs were low. As interest rates increased and (realized and unrealized) securities losses mounted, insured institutions began shifting asset composition away from securities and into loans, a shift that could increase the level of credit risk. As a result, the loan review and credit administration functions have become increasingly important.
Expanding Loan Volumes May Become Risky if the Economy Slows
The Dallas Region has benefited from extraordinary economic growth. The demand for lending also has increased, providing an incentive for banks and thrifts to make loans despite intense competition from bank and nonbank sources. Should the economy slow, however, competition for the best credits may increase as demand falls overall. David Stumpf, an analyst with A.G. Edwards, wrote, "A sluggish economy not only would diminish loan demand, but also could put additional pressure on credit quality."31 In addition, the Region's rising share of loans has implications for the ALLL.
31 Stumpf, David. September 11, 2000. Credit Concerns Succeed Rate Fears on Wall Street. American Banker, p. 32.
The ALLL Has Not Kept Pace with Loan Levels
Charge-offs and past-due rates are at historically low levels for the nation and the Region. However, a lesson from past economic downturns is that today's loans could become tomorrow's problems.
The Dallas Region is indeed "loaning up." The 324 banks and thrifts in our analysis are increasing loan exposure at twice the rate of other institutions in the Region and the nation. However, insured institutions in the Dallas Region continue to report the lowest ALLL-to-total-loans ratio of any FDIC Region. Moreover, the Region's fastest-growing banks and thrifts report an ALLL ratio that has been declining and is lower than that of other insured institutions in the Region. Should the economy weaken and loans begin to deteriorate, many institutions may find themselves without adequate buffers against future losses.
, where Y is the dependent variable Conventional Mortgage Home Price, X1 is the independent variable Payroll Employment Growth Rate (MSA), 
and 
are unknown parameters, and 
is the error term. This model proved to be statistically significant, with the independent variable explaining 41 percent (R2) of the variation in the dependent variable. The model home price appreciation can be calculated by manipulating the equation Y = 0.9265x + 0.0169.
