The Face of Texas

Jobs People Business Change

FEDERAL RESERVE BANK OF DALLAS • OCTOBER 2005 The Face of Texas Table of Contents 1 Introduction Jobs, People, Business, Change Mine Yücel

3 Texas in the Most Recent Recession and Recovery Mine Yücel The only constant is change itself. A Greek philosopher wrote those words more than two millennia ago, at a time when events probably didn’t 6 Why Did Texas Have a move as rapidly as they do today. Accepting the inevitability of change still Jobless Recovery? leaves us the task of understanding the often bewildering world around us. Pia M. Orrenius, Jason L. Saving and Priscilla Caputo Most of us need help in deciphering the meaning of the changes we can see and identifying the changes we can’t see. That is the purpose of this collection 12 Industry Clusters in Texas of essays—to give our readers solid, useful perspectives that will assist in Laila Assanie and Mine Yücel understanding the changing Texas economy. 17 Economic Progress in The new century finds Texas in transition—no longer booming on the Texas Economy high oil prices, finally rebounding from the late 1990s technology bust and Robert W. Gilmer still looking for the next economic driver. Our times are complex. Texas’ economy is being reshaped not only by what’s happening in the state and 24 Texas Border Benefits from Retail Sales to Mexican Nationals nation but also by a globalizing world, an important part of which lies just Keith R. Phillips and over the Rio Grande. Roberto Coronado These essays explain the recent past, give a textured picture of the current landscape and offer a glimpse of what changes may be over the 27 Texas Border Employment and Maquiladora Growth horizon. Our goal is straightforward: We want to promote a fuller Jesus Cañas, Roberto Coronado understanding of Texas’ evolving economy, so the state’s citizens will be better and Robert W. Gilmer equipped to take advantage of opportunities that lie ahead. 33 Do Higher Oil Prices Still Benefit Texas? Stephen P. A. Brown and Mine Yücel

37 The Changing Face of Texas: Population Projections Harvey Rosenblum and Implications Executive Vice President and Director of Research D’Ann Petersen and Laila Assanie Federal Reserve Bank of Dallas this recession was a major factor in the state’s prolonged downturn. Pia Orrenius, Jason Saving and Introduction Priscilla Caputo survey the weak jobless Mine Yücel recovery after the most recent recession and suggest that it may be caused by structural change in the Texas labor market. They note that structural change is not new to Texas. The state went through structural change in the 1980s after the oil bust and may be going through another one now. They show that the high-tech and apparel industries are undergoing structural losses, while the health care, education and government sectors are undergoing structural gains. But, just as the oil industry decline paved the way for the diversification and growth of the Texas economy a decade later, the structural change going on today will pave the way for a more dynamic and prosperous Texas.

Oil’s Impact The oil industry has been undergoing change for the past 20 years, shrinking while other sectors of the Texas economy have grown. The Texas economy’s diversification away from energy and the energy sector’s declining importance prompt Stephen Brown and Mine Yücel to ask whether high oil prices are still a benefit to the Texas economy. They show that higher The economic landscape of Texas is While the drivers of the economy may energy prices still benefit the state—even changing. The state lost more than change, one constant is the close relation- though it is by less than in the boom years 200,000 jobs during the tech bust and ship the state has with the Mexican econ- of the 1970s and early ’80s. They also find recentT recession. A majority of these jobs omy. The interconnection is crucial to the evidence that the Texas economy has were in the high-tech sector, which was border economies and is a big factor in become less sensitive to fluctuations in oil the main driver of the Texas economy in the changing demographics of Texas. prices than it was in the ’70s and ’80s. the 1990s. With the decimation of the tech First, oilfield activity has become less sen- sector, which industry will be the driver of Structural Change sitive to fluctuations in energy prices. Sec- the Texas economy in the future? This The articles in this publication dis- ond, the energy industry makes up a monograph doesn’t attempt to answer cuss some of the changes in the economic smaller share of the Texas economy than it that question, but we explore some of the landscape of our state. Mine Yücel looks at used to. Together these factors mean that ways the Texas economy has been chang- the Texas economy’s performance during Texas output is about 15 percent as sensi- ing and some of the current issues facing the most recent recession and explains tive to oil price fluctuations as it was from it. why it was different from previous reces- 1970 to 1988. Texas employment no The state has gone through boom and sions. She argues that unlike previous longer seems to be positively affected by bust cycles before, but each downturn has recessions, the most recent recession was oil price fluctuations. been followed by a stronger and more primarily due to a high-tech bust rather diverse economy. The oil bust was partic- than an oil price shock. Although oil Business Mix ular to Texas and hurt the state’s economy, prices were relatively high during the Laila Assanie and Mine Yücel outline while low oil prices helped the rest of the recession, they did not benefit Texas as the importance of industry agglomeration nation. The tech bust, on the other hand, much as in the past because the state has to an economic growth. They highlight the was experienced similarly in Texas and diversified away from oil. In addition, she key clusters in Texas and its six major met- the nation. Texas bore a larger brunt shows that the high-tech sector grew very ropolitan areas through economic base because it had a higher share of high-tech fast in Texas in the 1990s, to a share higher analysis. They find that oil and gas extrac- manufacturing and service industries than the national average. Texas’ higher tion and its support activities, pipelines, than the nation. share of industries that were hit hard in natural gas distribution, refining and oil-

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 1 field-machinery manufacturing are still because retail sales strength varied closely agglomerated in Texas. However, high- with peso strength, especially in Laredo tech and transportation industries have and McAllen. been added to this mix. Computers, Another perspective on border cities telecommunication services, semicon- is presented by Jesus Cañas, Roberto ductors and air transportation firms now Coronado and Bill Gilmer. They show how have a larger presence in Texas than in the expansions and contractions of the ma- nation as a whole. quiladora industry have affected Texas Bill Gilmer analyzes per capita in- border cities. NAFTA’s passage and the come growth in various regions of the peso devaluation in the early ’90s led to Texas economy. He shows that the state maquiladora growth and the relocation of economy has been growing rapidly since component parts and material suppliers Texas’ population is 1969, either matching or exceeding the to Texas cities along the border. Texas bor- nation’s growth. But the Texas Triangle der cities developed rapidly in the ’90s as faster growing, cities of Houston, Dallas/Fort Worth, part of this supply chain. Cañas, Coron- Austin and San Antonio grew faster than ado and Gilmer observe that proximity to younger and more average. Outside the Texas Triangle, the U.S. market becomes a crucial advan- income growth was much slower, tage for the maquiladoras when there is a ethnically diverse although population growth was not. short inventory cycle, when the weight- than the nation’s. Especially after 1989, the Texas Triangle to-value ratio of goods is high, when there cities contributed three-fourths of the is frequent retooling, when quality is more The Hispanic state’s income growth. Gilmer notes that important than price and when intellec- the Mexican border area represents a tual property rights are critical. However, population will challenge to state economic development they note that the state is unlikely to because the border cities’ average per repeat the banner performance of the ’90s be the dominant capita income is only 50 to 60 percent of as foreign competition slows the growth force in Texas the national average. The border saw of Texas border-city suppliers to the explosive gains in the ’90s following the maquiladora industry. Hence, Texas may by 2020. passage of NAFTA and the growth of the see less stimulus in the future from maquiladora industry, but high popula- maquiladora expansion. tion growth and high in-migration rates kept income per capita low in this area. Population Shift The article also explains that the growth in Finally, D’Ann Petersen and Laila wages and salaries after 1989 came Assanie discuss Texas’ changing demo- through a change in industry mix as the graphic makeup and how it will shape the economy shed low-wage jobs and economy. Texas’ population is faster replaced them with better-paying ones. growing, younger and more ethnically diverse than the nation’s. The Hispanic Border Influence population will be the dominant force in Texas border cities are a unique blend Texas by 2020. The authors demonstrate of U.S. and Mexican cultures, languages that there are large disparities between and customs and follow the ups and ethnic groups in income and education. downs of the Mexican and U.S. Such disparities may imply a decline in economies. Keith Phillips and Roberto real income and a lower-skilled and less- Coronado look at how border cities on the educated labor force in Texas compared Texas side benefit from cross-border traf- with the nation. On the other hand, the fic by consumers from their sister cities on young and fast-growing population also the Mexican side. They estimate retail means Texas’ housing market may con- sales in four metro areas along the tinue to be vibrant even as the baby Texas–Mexico border. They find that in boomers age. Texas’ challenge is to reduce 2001, retail sales to Mexican nationals the disparities and make our differences accounted for nearly 20 percent of retail work for us. sales in border metros. Laredo had the highest share, with 41 percent of its retail Yücel is a senior economist and vice presi- sales going to consumers from across the dent in the Research Department of the border. Phillips and Coronado also show Federal Reserve Bank of Dallas. that unexpected changes in the peso’s real value affected these border metros

2 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 Texas in the Most Recent Recession and Recovery Mine Yücel

After decades of faring better than the picked up in Texas in 2004. But while the rest of the country, Texas’ economic 1.7 percent increase put Texas on par with growth has lagged both the nation’s and the nation, it still left the state below its What are the its own past performance for almost three historical pace. What are the reasons for Ayears. Texas’ prolonged downturn? Why did the reasons for The most recent U.S. recession was state lose its edge? short-lived, beginning in March 2001 and Texas’ prolonged ending that November, according to the Past Performance National Bureau of Economic Research. It Texas employment growth, on aver- downturn? Why took Texas another 20 months—until July age, exceeded the nation’s from 1970 did the state 2003—to bottom out, based on the Texas through 2004, with a 2.8 percent rate to Coincident Index.1 Employment growth the country’s 1.8 percent (Chart 1). The lose its edge? Chart 1 Texas Bests U.S. Employment Growth for More Than 30 Years

Percent 8 7 6

5 Texas 4 Texas average 3 2 U.S. average 1 0 –1 United States –2 ’71 ’73 ’75 ’77 ’79 ’81 ’83 ’85 ’87 ’89 ’91 ’93 ’95 ’97 ’99 ’01 ’03 ’05

NOTES: Data are year-over-year, seasonally adjusted, annualized rates. Shaded bars signify national recessions. SOURCES: Bureau of Labor Statistics; Federal Reserve Bank of Dallas; National Bureau of Economic Research.

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 3 Chart 2 Texas Economy Follows Oil Prices

Detrended Texas employment 2005 U.S. dollars per barrel 8 90 80 6 Employment cycles 70 4 60 After oil prices crashed, 2 50 Texas diversified and 0 40 30 –2 the industry became a 20 –4 Oil price 10 much smaller share of

–6 0 ’73 ’75 ’77 ’79 ’81 ’83 ’85 ’87 ’89 ’91 ’93 ’95 ’97 ’99 ’01 ’03 ’05 the state’s economy.

NOTE: Employment data are seasonally adjusted and have had the time trend removed. SOURCES: Bureau of Labor Statistics; Federal Reserve Bank of Dallas; Department of Energy.

state’s ability to dodge national recessions estate boom cratered. Low oil prices ben- sector losing 280,000 jobs nationwide is one reason Texas has done so much bet- efited the national economy but sent from March 2001 to July 2003. These job ter. Eight of the 10 post–World War II Texas into a steep decline. However, Texas losses plus those in high tech constituted recessions followed oil price shocks. And skirted the national recession again in 50 percent of the total U.S. employment unlike the nation as a whole, Texas bene- 1990, when West Texas Intermediate decline. Texas lost 21,200 transportation fited from high oil prices, especially in the crude spiked to $45 per barrel with the jobs, which, combined with its high-tech 1970s through early 1990s. As can be seen Iraqi invasion of Kuwait (Chart 3d). losses, accounted for 62 percent of the in Chart 2, the Texas economy followed state’s total.2 changes in oil prices fairly closely, with This Time Around Chart 5 illustrates Texas’ high-tech employment rising and falling with the oil Texas looked much more like the roller coaster. California is included as a price. The Texas employment cycle started nation in the 2001 recession than it did in comparison, along with the United States. diverging from oil-price movements in past downturns, for two reasons (Chart 4). High tech grew very fast in the 1990s but the 1990s as the economy diversified away First, although oil prices were high, this came back down just as fast. High-tech from oil and gas. recession was primarily due to a high-tech production in Texas grew six times as fast High oil prices were a boon to the bust, not an oil price shock. Second, high as the state’s overall output. During the Texas economy and helped it grow, even oil prices do not help the Texas economy recession, Texas high-tech manufacturing during national recessions, as seen in as much as they have in the past. lost 107,400 jobs, nearly a third of its Chart 3. Oil prices that nearly tripled from The collapse of high tech in the employment. Even though California $4 to above $10 per barrel (refiners’ acqui- recent recession was greatly felt in Texas. started with a higher base and therefore sition cost) sent the United States into The state had a larger share of high-tech grew less in percentage terms, more jobs recession in December 1973 but boosted employment than the U.S. average, so job were created in Texas. During the build- output and employment in Texas (Chart losses in those industries were relatively up, total high-tech manufacturing jobs 3a). Oil prices started creeping up again higher. From March 2001 through July increased by 47,000 in Texas, while they in the late 1970s and rose from around $12 2003 (the Texas recession), 39 percent of rose by only 17,000 in California. In semi- per barrel in 1978 to almost $30 when Iraq the jobs lost nationwide were in high conductors, for example, California added invaded Iran in September 1980. The U.S. tech—426,800 of them in manufacturing 22,000 jobs, while Texas added 35,000. economy went into recession—again, and 610,000 in services. Fifty-one percent Texas also grew faster than the nation in without Texas (Chart 3b). of the 208,900 jobs lost in Texas were in telecom services, adding 50,000 jobs dur- But just as high oil prices helped high tech—51,900 of them in manufac- ing the ’90s, then losing 30,000 during the Texas, low ones hurt it. The nation went turing and 55,500 in services. recession. into recession again in August 1981, and The events of September 11 also con- Elsewhere in this publication, “Do Texas followed 10 months later, the result tributed to Texas’ steep downturn. The Higher Oil Prices Still Benefit Texas?” dis- of oil prices that began falling from record transportation industry is important to cusses how the relationship between oil highs in March 1982 and the pull of the the state’s economy and has a larger share and the Texas economy has evolved. national downturn (Chart 3c). of total employment than in the nation. When the industry was a larger share of Texas had its own recession in 1986, Transportation was especially hard-hit by the Texas economy, higher oil prices were when oil prices collapsed and the real fallout from the terrorist attacks, with the always a net benefit to the state. That

4 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 Chart 3 changed in the late 1980s, when volatile Chart 4 Texas and U.S. Economies in Texas Looks More Like the Nation energy prices helped erode the promi- Previous Recessions (Total nonfarm employment) nence of energy-intensive and energy- a. Texas Skirts Recession in the 1970s… producing industries. Index, April 2001 = 100 (Total nonfarm employment) After oil prices crashed, Texas diversi- 102 Index, December 1973 = 100 fied and the industry became a much U.S. recession 101 April – November 2001 110 smaller share of the state’s economy. For United States 108 100 U.S. recession example, oil and gas output, which 106 Texas December 1973 – March 1975 accounted for nearly 20 percent of total 99 104 Texas output in 1981, accounts for only 102 98 Texas United States 100 about 6 percent today. Similarly, oil and 97 98 gas jobs account for only 2 percent of Oct. Mar. Aug. Jan. June Nov. Apr. Sep. Feb. July Dec. 96 Texas employment, down from a high of ’00 ’01 ’01 ’02 ’02 ’02 ’03 ’03 ’04 ’04 ’04 94 about 5 percent in 1982. The upshot is SOURCES: Bureau of Labor Statistics; Federal Reserve Bank of June Sept Dec. Mar. June Sept. Dec. Mar. June Sept. Dallas; National Bureau of Economic Research. ’73 ’73 ’73 ’74 ’74 ’74 ’74 ’75 ’75 ’75 that rising oil prices benefit Texas much less now than they did in the past. b. …and Again in 1980 Texas is still a large producer and (Total nonfarm employment) exporter of oil and gas, and when prices Chart 5 go up, it helps producers, royalty owners The Boom and Bust of High-Tech Index, February 1980 = 100 and the state through increased severance Manufacturing… 110 108 U.S. taxes. So, unlike the rest of the country, recession Texas Jobs: Index, January 1990 = 100 106 February – Texas gets an offset. But that offset is 160 104 July 1980 much less now than it was 25 years ago. Texas 102 United States In sum, Texas’ economic perform- 140 100 ance has been below par the past three 120 United States 98 years. Unlike other downturns, the 2001 96 recession was primarily due to a high-tech 100 94 Aug. Oct. Dec. Feb. Apr. June Aug. Oct. Dec. Feb. Apr. June bust, not an oil price shock. And although 80 California ’79 ’79 ’79 ’80 ’80 ’80 ’80 ’80 ’80 ’81 ’81 ’81 oil prices were relatively high, they did not benefit Texas as much as in the past 60 c. Texas Follows Nation into Recession in 1981… ’90 ’91 ’92 ’93 ’94 ’95 ’96 ’97 ’98 ’99 ’00 ’01 ’02 ’03 ’04 ’05 (Total nonfarm employment) because the state economy has diversified. In addition, high tech grew very fast Index, August 1981 = 100 …and Telecom Services in Texas in the 1990s, to a share that was 110 higher than the national average. Texas’ Jobs: Index, January 1990 = 100 108 U.S. recession higher share of industries that were hard- 210 106 Texas August 1981 – hit in the recent recession was a major 190 104 November 1982 102 factor in the state’s prolonged downturn. 170 100 Texas 150 United States 98 Yücel is a senior economist and vice presi- 96 United States 130 94 dent in the Research Department of the 110 Feb. May Aug. Nov. Feb. May Aug. Nov. Feb. May Aug. Nov Federal Reserve Bank of Dallas. California ’81 ’81 ’81 ’81 ’82 ’82 ’82 ’82 ’83 ’83 ’83 ’83 90 70 Notes ’90 ’91 ’92 ’93 ’94 ’95 ’96 ’97 ’98 ’99 ’00 ’01 ’02 ’03 ’04 ’05 1 d. …but Dodges the 1990 Downturn The Texas Coincident Index aggregates the SOURCES: Bureau of Labor Statistics; Federal Reserve Bank of (Total nonfarm employment) movements of key regional indicators— Dallas. employment growth, the unemployment rate Index, August 1990 = 100 and gross state product—to gauge the state’s 104 overall economic direction.

102 2 Texas One point to note is that both high-tech and 100 transportation employment were falling even before the onset of the recession. The two sec- 98 U.S. United States recession tors were responsible for 73 percent of all job 96 August 1990 – losses in Texas from December 2000 to July March 1991 2003. 94 Feb. May Aug. Nov. Feb. May Aug. Nov. Feb. ’90 ’90 ’90 ’90 ’91 ’91 ’91 ’91 ’92 SOURCES: Bureau of Labor Statistics; Federal Reserve Bank of Dallas; National Bureau of Economic Research.

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 5 Why Did Texas Have a Jobless Recovery? Pia M. Orrenius, Jason L. Saving and Priscilla Caputo

In early 2001, the U.S. and Texas structural change in the economy. While ers on firm payrolls. However, the house- economies fell into recession. While the much has been written on the nation’s hold survey has a much smaller sample National Bureau of Economic Research experience during this period, there is lit- size and depends on population estimates IBusiness Cycle Dating Committee tle information on what caused the job- that are not always reliable, mainly declared the U.S. recession over in less recovery in Texas. For this reason, it’s because of uncertainty about immigra- November of that year, job growth did not important to examine these explanations tion rates. Given these weaknesses and resume until June 2003. Texas job growth to see which of them can shed light on the the adoption of a statistical method to broke into positive territory two months state’s experience. compensate for missed job growth in later, and there is evidence that, like the start-up firms, most experts—and the nation, economic activity in the state Employment Statistics? BLS—consider the payroll survey the bet- picked up long before that. Two Bureau of Labor Statistics (BLS) ter gauge of employment. Following a typical recession, surveys are the primary source for employment begins to rise at about the national and state employment data. The Productivity Growth same time output does. But in the two establishment, or payroll, measure—offi- or Uncertainty? years after the 2001 recession, U.S. real cially, Current Employment Statistics— If the data are sound and the country output growth averaged 2.5 percent, while surveys about 400,000 work sites each did experience a jobless recovery in 2002 employment growth was essentially zero. month. Critics contend this survey under- and 2003, could high productivity growth The divergence between output and states job creation at economic turning or substantial uncertainty have been the employment was even more pronounced points because it misses employment in cause? in Texas, where real output—as measured the new firms created during a recovery’s U.S. productivity growth averaged 4.3 by gross state product—grew faster than initial stages. The alternative, household- percent during this period, and some the nation’s, but employment fell at based Current Population Survey contacts experts believe that increase—well above an average annual rate of 0.2 percent. individuals directly about their employ- the post–World War II average of about Clearly, something was different this time. ment status. According to this survey, 2 percent—enabled companies to step Many explanations have been offered there has been little jobless about the up production without hiring more work- for the unusually weak labor market per- recovery: Jobs have grown each year since ers. Others believe the uncertain environ- formance, including problems with meas- the 2001 recession.1 ment that followed various corporate uring employment, high productivity The household survey might seem accounting scandals and the 9/11 attacks growth, widespread uncertainty in the sounder than the payroll survey because led to a wait-and-see approach by em- wake of 9/11 and corporate scandals, and it is not limited to wage and salary work- ployers.

6 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 These factors likely played an impor- Chart 1 tant role in the jobless recovery. But job Texas Employment Tracks the Nation’s in 2001 and After growth in 2002–03 was far below what Index, 2000 = 100 Texas and the nation saw in earlier peri- 110 ods of relatively high productivity growth, such as the late 1990s, and substantial 100 uncertainty, such as the late 1970s. So 90 United States there is more to the story. 80 Texas Structural Change? 70 A widely read article from the Federal 60 Reserve Bank of New York offers another explanation for the jobless recovery.2 Erica 50 Groshen and Simon Potter consider two 40 types of effects that could shake up labor ’78 ’80 ’82 ’84 ’86 ’88 ’90 ’92 ’94 ’96 ’98 ’00 ’02 ’04 markets: (1) short-term cyclical adjust- NOTE: Data show total nonfarm employment. ments that vary with the business cycle, SOURCES: Bureau of Labor Statistics; Federal Reserve Bank of Dallas. and (2) longer term structural changes, in which some industries decline while others grow. ance, is largely designed around the needs the state level. In a jobless recovery, how- The economists contend that an of the cyclically unemployed—people ever, the traditional relationship between unusually large amount of structural who need short-term help with income employment growth and overall eco- change in the labor market, as opposed to sustenance while they search for a job. nomic activity breaks down. This means temporary cyclical adjustments, hindered The system is generally not designed to payroll employment data may not have the resumption of employment growth in provide longer term retraining for dis- accurately reflected the state’s overall eco- 2002 and 2003. When jobs shift across placed workers whose sectors perma- nomic health during 2002–03, making it industries, new positions have to be cre- nently shrink. Public job-training pro- impossible to date the Texas recession ated and filled, which takes far more time grams are becoming more common, how- using those numbers. than simply recalling workers to their ever. Lawmakers recognized the effects of The NBER solves this conundrum for jobs, as might occur with cyclical change. structural change in the labor market in the nation by using several variables in So if structural change is on the rise, it passing such bills as the Workforce Invest- addition to employment—such as indus- could explain the jobless recovery. ment Act of 1998 and the Trade Adjust- trial production and, especially, gross The kind of structural change ment Assistance Reform Act of 2002. domestic product—to date U.S. business Groshen and Potter consider can result Assuming structural change has cycles.5 Most of these numbers are not from a myriad of factors that cause some accelerated at the national level, can the available in a timely fashion at the state industries to decline as others grow. These same be said for Texas? Taking the level, and they are not available at all on a factors include technological and demo- Groshen–Potter approach, we compare quarterly or monthly basis, which would graphic change, reorganization of pro- recent patterns to earlier recessions to see be needed to date the Texas recession. duction, trade and outsourcing—any one if structural change has increased in Texas We use the national dates for a base- of which can permanently alter a state’s or and, if so, whether it helps explain the line analysis of Texas. After all, Texas em- nation’s industrial mix. Cyclical job losses, state’s recent experience. ployment closely tracked the nation’s in by contrast, move with the business cycle. 2001 and thereafter, suggesting that simi- As the economy enters a recession, jobs Measuring Structural Change lar factors drove both economies into are temporarily lost in response to soften- To measure structural job change, recession (Chart 1). Texas output also ing demand. They are added back as the Groshen and Potter compare employ- tracked the nation’s reasonably well in economy picks up again. ment growth in the recession and the 2001 and 2002 (Chart 2). That said, esti- Looking at job growth by industry, recovery.3 They make this comparison for mates of real output at the state level are Groshen and Potter find that structural each major industry over the length of the subject to a higher degree of uncertainty factors played a much greater role in the recession as designated by the National than at the national level, and there is United States during 2001–02 than in Bureau of Economic Research (NBER)— anecdotal evidence Texas emerged from earlier U.S. recoveries. They attribute this March 2001 to November 2001.4 The recov- the recession after the nation. To check to a changing labor market in which cycli- ery is defined as the 12 months following the validity of our findings, we repeat the cal job losses have been minimized and the business cycle’s trough in November. analysis using an end date of March 2003 structural changes are more pervasive. Pinpointing recession dates for Texas rather than November 2001.6 This conclusion has important im- is more complicated. Economic analysts Chart 3 shows how Texas job growth plications for public policy. The tradi- often look to payroll employment growth to fared during the 2001 recession and the tional safety net in the United States, with date state recessions because this is the 12-month recovery for each one-digit such elements as unemployment insur- most timely and reliable data available at industry, the broadest category in the

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 7 Chart 2 ness cycle. Industries that grew slower Texas Real Output Tracks the Nation’s in 2001 and After than the statewide average during the Index, 2000 = 100 recession but faster during the recovery 110 are in the procyclical flows quadrant because they move with changes in the 100 business cycle. Industries that grew faster 90 than the statewide average during the

80 United States recession but slower during the recovery are in the countercyclical flows quadrant Texas 70 because they tend to add jobs when the

60 rest of the economy declines but lose jobs when the rest of the economy does well. 50 The size of each industry’s bubble on 40 the chart represents its share of total Texas ’78 ’80 ’82 ’84 ’86 ’88 ’90 ’92 ’94 ’96 ’98 ’00 ’02 ’04 employment in March 2001, when the NOTE: Data show real U.S. gross domestic product and Texas gross state product. recession began. The larger the bubble, SOURCES: Bureau of Economic Analysis; Federal Reserve Bank of Dallas. the larger the industry’s share of the state’s workforce at that time. The results suggest that the recent business cycle has been dominated by Chart 3 structural gains and losses, as most major Structural Change Prevalent Among Texas Industries in 2001–02 industries fall into the structural change Job growth in recovery (percent) quadrants in Chart 3. Manufacturing of 4 Procyclical flows Structural gains both durable and nondurable goods suf- 3 fered the largest structural losses, Government Health services whereas health services and government 2 Construction had the biggest structural gains. Overall, 1 Other services about 75 percent of March 2001 employment was concentrated in industries that 0 FIRE subsequently underwent structural –1 Nondurable change. The next section breaks down manufacturing Retail these major industries to take a closer look –2 trade Durable manufacturing Wholesale at job adjustments. –3 TCPU trade Mining Industries with Structural Loss. Indus-

–4 tries in Chart 4 are classified according to Structural losses Countercyclical flows subsectors in the North American Indus- –5 try Classification System (three-digit –6 –5 –4 –3 –2 –1 0 1 2 3 4 5 Job growth in recession (percent) NAICS codes). The southwest portion of Chart 4 includes a number of high-tech NOTES: TCPU is transportation, communications and public utilities. FIRE is finance, insurance and real estate. The recession is dated March to November 2001; the recovery is dated December 2001 to November 2002. sectors, among them computer and elec- SOURCES: Bureau of Labor Statistics; Federal Reserve Bank of Dallas. tronic product manufacturing (includes semiconductors); electrical equipment, appliance and component manufacturing; telecommunications; and Internet service Standard Industrial Classification (SIC) ery. If an industry grew slower than the providers (ISPs), search portals and data system.7 All growth rates are relative to the statewide average in each of the two peri- processing services. High tech’s presence average for total Texas employment dur- ods, it falls in the southwest portion of the in the structural loss quadrant is not suring the relevant period. For example, if an chart, labeled “structural losses” because prising, since the 2001 recession kicked industry grew 5 pecent slower than the these industries lose jobs regardless of off a prolonged retrenchment and re- Texas economy as a whole during the overall economic conditions. If an indus- structuring for the sector in Texas, a recession, its growth rate is –5 percent. try grew more rapidly than the statewide process from which the state has not fully Likewise, if an industry grew 5 percent average during both intervals, it is in the emerged. faster than the Texas economy during the northeast portion of the chart, labeled Apparel manufacturing also falls in recovery, its growth rate is 5 percent. “structural gains” because such industries the structural loss quadrant. In contrast to The horizontal axis on Chart 3 gain jobs regardless of the overall econ- high tech, the apparel industry has been measures the relative growth rate during omy. declining in the United States and Texas the recession; the vertical axis measures The remaining quadrants deal with for many years. Indeed, apparel experi- the relative growth rate during the recov- industries that rise and fall with the busi- enced the largest job losses in percentage

8 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 terms during both the recession and the accommodations. Comparing Texas Recessions recovery. It may be surprising that so few in- Is structural change a bigger factor The northeast quadrant of Chart 4 dustries fall into the cyclical category, but today than in the past? shows the industries that grew faster than it’s important to remember that we are Groshen and Potter conclude that for total Texas employment during the reces- comparing each industry to the overall the United States as a whole, it is. More in- sion and recovery. This quadrant consists state economy. If an industry’s employ- dustries in the recent recession fell into the mainly of sectors related to the provision ment fell slightly during the recession and structural-change quadrants, compared of health care and education, including rose slightly during the recovery, it’s cate- with earlier recessions. They find that 79 local government. gorized as countercyclical because its percent of U.S. employment was in indus- Given recent policy and demographic employment fell by less than the state tries affected more by structural than developments, this trend is understand- average during the recession and rose by cyclical shifts in the 2001 recession, up able. Rapid advances in medical technol- less than the average during the recovery. from about 50 percent in previous down- ogy, coupled with an aging population, Recession Dating. What if the Texas turns. are producing an increased emphasis on recession was longer than the nation’s and It’s a somewhat different story for health care, regardless of the business cycle. did not end in November 2001? If so, the Texas. Chart 6 shows job adjustments by The rise in the economic return to analysis so far biases the findings toward major industry during the recession and education, the burgeoning youth popula- structural change by attributing 2002 job recovery of the early 1980s.10 That reces- tion and renewed public attention to edu- losses to the recovery instead of to what sion was more severe than the recent one, cational quality have produced an in- may have been a continuing recession. To with several large industries—such as creased emphasis on education that check our results, we repeat the exercise durable manufacturing and mining— doesn’t ebb and flow with economic con- under the assumption that Texas emerged experiencing double-digit job losses. Nev- ditions, either. Since local government is from the recession in March 2003—much ertheless, except for government, educa- the largest provider of K–12 education, it’s later than the nation and about four tion and health services, the losses were not surprising that employment in this months before employment growth fairly concentrated in the structural cate- sector rose during the recession, as well as resumed in the state. gories. In fact, Texans were about as likely the recovery. A few industries move from one quad- to work in structural-change sectors in the Countercyclical Industries. The south- rant to another, but the overall picture is 1982–83 recession as they were in 2001. east corner of Chart 4 consists mainly of one in which structural change still domi- The share of structural job losses was industries in the energy sector. Rising nates cyclical change (Chart 5). About two- about 72 percent during the earlier energy prices were a contributing factor thirds of employment is concentrated in period, compared with 76 percent in the to the 2001 recession.8 As home to a major industries undergoing structural change, 2001 recession. While the relationship can share of the U.S. energy industry, Texas compared with three-fourths when No- be seen a bit more easily in Chart 3 than in benefits from high oil prices (although to vember 2001 is used as the end date. Chart 6, the two graphs confirm that a lesser extent than when the industry constituted a much larger part of the state’s economy). Chart 4 Since energy prices were higher dur- Structural Change Prevalent in 2001–02: A Look at More Detailed Industries ing the 2001 recession than during the Job growth in recovery (percent) 2002–03 recovery, it makes sense that 20 Procyclical flows Structural gains energy is categorized as countercyclical Warehousing & 9 15 Local govt. for this period. Natural resource and min- Rail storage ing industries in this quadrant include oil transportation Clothing Ambulatory 10 and gas extraction and mining support Transportation stores health services equipment Waste activities. 5 management One notable countercyclical industry Accommodations 0 that doesn’t fit into the natural resource Air transportation category is real estate. What high oil prices -5 Oil & gas did for natural resource industries during Real estate extraction –10 ISPs the recession, low-interest loans likely did Computer & electronic Electrical Funds, trusts for homebuyers. product mfg. equipment, Mining support activities & other –15 Telecommunications Procyclical Dating. Despite expecta- appliance & financial Apparel mfg. component mfg. vehicles tions of “normal” cyclical losses, few in- –20 Pipeline transportation dustries fall into the procyclical category Structural losses Countercyclical flows –25 during and after the 2001 recession. The –20 –15 –10 –5 0 5 10 northwest quadrant of Chart 4 consists of Job growth in recession (percent) only about 9 percent of total employment. NOTES: The recession is dated March to November 2001; the recovery is dated December 2001 to November 2002. Educational services is Among the industries in this quadrant are behind local government in the northeast quadrant. retail, transportation-related sectors and SOURCES: Bureau of Labor Statistics; Federal Reserve Bank of Dallas.

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 9 Chart 5 inventory, production and employment. Structural Change Prevalent with Alternate Recession End Date Additional contributing factors are a Job growth in recovery (percent) decline over time in the severity of energy 15 Procyclical flows Structural gains Mining Highway, street & and food supply shocks and the deregula- support bridge construction tion of financial markets. 10 activities Electronics & Ambulatory But does structural change really appliance stores health care explain the jobless recovery? Structural 5 Warehousing change, as measured here, was about as 0 prevalent in Texas in the 2001 recession Computer & Local and ensuing recovery as in the early 1980s –5 electronic government product mfg. recession and recovery. The difference –10 State government between the two periods is the severity of Telecommunications Pipeline the change. Job losses were much deeper transportation –15 in the 1982–83 downturn (and worse yet Apparel in 1986). Nevertheless, employment –20 mfg. rebounded with a short lag, and there was Structural losses Countercyclical flows nothing like the jobless recovery experi- –25 –50 –40 -30 –20 –10 0 10 20 30 enced post-2001. Job growth in recession (percent) Another possibility is that the invest-

NOTES: The recession is dated March 2001 to March 2003; the recovery is dated April 2003 to March 2004. ment bust that characterized the 2001 SOURCES: Bureau of Labor Statistics; Federal Reserve Bank of Dallas. recession and its aftermath may have driven both structural losses in the labor market and the jobless recovery. The investment bust followed the investment Chart 6 boom that had characterized certain fast- Structural Change a Major Factor in 1982–84 growing industries—led by high tech—

Job growth in recovery (percent) during the 1990s. In Texas, for example, 8 post-2001 venture capital commitments Procyclical flows Structural gains Other services fell sharply to about 20 percent of their 2000 levels. The investment bust likely 6 delayed employment growth during the recovery in the sectors that had been 4 Retail trade FIRE booming. If this was the case, sectors that were fast-growing before the recession 2 Durable mfg. would fall into the structural loss category Construction in our analysis. These industries may or 0 Educational Nondurable may not belong there, depending on services mfg. whether they will eventually resume –2 above-average job growth. Health services Mining The data suggest that the investment –4 TCPU Government Wholesale bust played an important role in Texas trade Structural losses Countercyclical flows during the recent business cycle. In fact, –6 of the state’s 16 fastest-growing industries –20 –15 –10 –5 0 5 10 15 Job growth in recession (percent) in the 1990s, 10 appear in the structural-

NOTES: TCPU is transportation, communications and public utilities. FIRE is finance, insurance and real estate. The recession is dated loss quadrant of Chart 4, meaning they March 1982 to March 1983; the recovery is dated April 1983 to March 1984. shed jobs both during and after the 2001 SOURCES: Bureau of Labor Statistics; Federal Reserve Bank of Dallas. recession. Groshen and Potter show that for the nation, seven of the 18 fastest- growing industries fall into the structural structural change, as defined by Groshen international trade. A decline in the role of loss category. and Potter, is not new to Texas. cyclical change, meanwhile, has been It is likely that as these industries’ linked to factors such as improved mone- expansion fell short of expectations, More Sectors Undergo tary policy, which appears to have less- investment dried up and employment Structural Change ened the duration and severity of U.S. declined. The industries include several Several explanations have been business cycles.11 Better supply chain man- high-tech subsectors, such as telecom- offered for the growing role of structural agement has also allowed firms to munications and ISPs, search portals and change in the U.S. economy, including respond more quickly to changes in data processing services. Not all fast- technological change and increasing demand and avoid sudden large swings in growing industries fall into the structural

10 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 loss quadrant, however. Three of Texas’ construction is at record levels, freed-up Notes fastest-growing industries in the 1990s are capital and labor are moving into sec- 1 Annual household employment was lower in 2002 in the structural gain category—ware- tors—such as education and health— than 2001, but yearly job growth is calculated housing and storage, ambulatory health where structural growth is most pro- December-over-December and was 0.26 percent services and social assistance. nounced. Exports to China are booming, in 2002. and the border economy is thriving as a 2 “Has Structural Change Contributed to a Jobless Little New About result of freer trade with Mexico. Recovery?” by Erica L. Groshen and Simon Pot- Structural Change ter, Federal Reserve Bank of New York Current Issues in Economics and Finance, August 2003. The Texas economy has undergone Summary 3 Alternative measures of structural change are fundamental restructuring as the state The aftermath of the 2001 recession is discussed at length in “Can Sectoral Realloca- has diversified away from agriculture and often described as a jobless recovery. It tion Explain the Jobless Recovery?” by Daniel energy and become more like the nation. took Texas and U.S. employment almost Aaronson, Ellen R. Rissman and Daniel G. Sulli- These trends began in earnest in the 1970s four years to reach their respective prere- van, Federal Reserve Bank of Chicago Economic and intensified in the 1980s with the drop cession levels, which they finally did in Perspectives, Second Quarter 2004. in oil prices and collapse of the banking January 2005. Many factors contributed to 4 Groshen and Potter compare employment sector. The 1990s saw tremendous growth labor market weakness in 2002 and early growth in the recession and the recovery for of the state’s high-tech industries and fur- 2003, including high productivity growth, two-digit industries as defined by the Standard ther consolidation in the energy sector. In the war on terror and corporate scandals. Industrial Classification (SIC) system. SIC codes were replaced by the North American Industry both the 1970s and again in the 1990s, In their New York Fed article, Groshen Classification System (NAICS) in 2002. Texas’ economic growth was character- and Potter highlight another potential 5 In a statement announcing the dating of ized by large inflows of workers who source of labor market weakness—struc- the 2001 recession, the NBER called real GDP brought different skills and education tural change. The economists imply that “the single best measure of ‘aggregate eco- with them and contributed to the state’s because new industries are replacing old nomic activity.’” See www.nber.com/cycles/ economic transformation. ones, jobs are being created and filled at a recessions.pdf. The decline of industries paves the slower rate than in past business cycles, in 6 Texas payroll employment began to grow in way for diversification and the growth of which workers were simply laid off and August 2003, while retail sales began to grow in new sectors as workers, capital and know- rehired by the same or similar employers. September 2002. As a compromise, we selected how are freed up to pursue better ends. Applying the Groshen–Potter meth- March 2003 as an alternative end date for the For example, at one time, 90 percent of the odology to Texas, we find that structural state recession. Eleventh District Beige Book accounts also suggest the second quarter of U.S. labor force was engaged in farming. change also dominated cyclical change in 2003 may have been the turning point for Texas. Today, that number is a mere 3 percent. the state during the last business cycle. We 7 SIC codes are used in Charts 3 and 6 so that While this transformation is clearly do not find, however, that the amount of employment by industry can be compared over beneficial in the long run, in the short to structural change has increased over time, time. The newer, three-digit NAICS codes are medium term, this type of change is not as Groshen and Potter argue is the case for used in Charts 4 and 5. without its critics. People may primarily the nation. 8 See “Do Energy Prices Threaten the Recovery?” see the negative connotations of struc- Structural change is an enduring fea- by Stephen P. A. Brown, Federal Reserve Bank of tural change, without seeing the benefits. ture of the state’s economy. But while Dallas Southwest Economy, May/June 2004. This may be because certain advances in Texas labor markets experienced struc- 9 In 2004, oil prices rose again, and they are cur- trade and technological change have large tural change in earlier recessions, they did rently higher than they were during the 2001 benefits that are spread across many peo- not experience drawn-out weakness once recession. Natural gas prices have also ple, such as all U.S. consumers, while the a recovery was under way. In other words, remained high. costs of such advances can be small but the recent jobless recovery remains a bit 10 The 1982–83 Texas recession is assumed to concentrated on a select few (such as laid- of a mystery. The investment boom and have lasted from March 1982 to March 1983. off textile workers). subsequent bust may have had something This period roughly corresponds to the downturn in both state output and employment. Texas has not been immune to the to do with it. Many of the 1990s’ fastest- 11 forces of trade and outsourcing. Semicon- growing industries ended up with the See “New Economy, New Recession?” by Evan F. Koenig, Thomas F. Siems and Mark A. Wynne, ductor production has moved out of largest relative and most persistent job Federal Reserve Bank of Dallas Southwest Econ- Austin and Dallas to Asia, for example, losses. The extent of the state’s high-tech omy, March/April 2002. and major computer companies have investment boom and subsequent bust concentrated their software development may help explain why the effect on Texas in India, outsourcing thousands of jobs employment growth was so significant there. Big retailers and national banks and lasting. continue to expand in the state, often dis- placing or absorbing local businesses in Orrenius and Saving are senior economists the process. in the Research Department of the Federal At the same time, the state’s economy Reserve Bank of Dallas. Caputo worked on has many strengths. Workers and this article while an economic analyst at investors continue to flock to Texas, home the Bank.

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 11 Industry Clusters in Texas Laila Assanie and Mine Yücel

Advances in technology have dramat- An agglomeration economy, also ically reduced transportation and com- known as a cluster, is defined as a geo- munication costs. Access to distant goods, graphically concentrated group of indus- services and even labor has become much tries related by technology or skills, with easier.A Increased access to markets has close linkages among buyers and suppli- also brought increased competition, pres- ers. Clusters are important because they suring firms to reduce costs to maintain provide their participants with easy and profitability. In this age of globalization, as lucrative access to knowledge and special- Michael Porter notes, the importance of ized resources required to operate effi- generalized urban economies diminishes, ciently. This enhances participants’ pro- and agglomeration economies become ductivity and spurs innovation. Clusters much more important.1 also attract new business and investment

12 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 to the region. It is this increased efficiency Chart 1 Texas Compared with the United States and the ability to innovate and attract outside investment that give cluster partici- Location pants a competitive advantage.2 A good quotients 6 example of an industry cluster is the Dal- Oil and gas extraction Support activities las telecom corridor that attracted hun- for mining 5 dreds of high-tech manufacturing and Petroleum refining

services firms to the metro during the 4 high-tech boom in the 1990s. Pipeline transportation Industry clusters lead and shape the 3 Chemicals Oilfield Petroleum and economic growth of a region. One simple Natural gas mfg. machinery mfg. products distribution 2 way of determining industry clusters is Electronic components Wired telecom and products mfg. Computer and peripheral through economic base analysis. The eco- Air transportation carriers equipment mfg. Home health Services incidental 1 Aircraft care services nomic base of a region is defined as Communications Other Computer systems Truck to transportation and parts equipment mfg. telecom design and related services transportation industries whose external demand gener- mfg. services ates outside revenues and stimulates local 0 economic growth. The assumption is that SOURCE: 2000 Census IPUMS data; authors’ calculations. nontraded goods and services tend to be uniformly distributed, do not bring outside income into the region, and therefore, do not form the region’s economic agglomeration with both Texas and the com services 21 percent greater and elec- base. United States. tronic components manufacturing, which To determine which goods and serv- includes semiconductor manufacturing, ices produced in Texas and its major met- How Texas Compares is 44 percent higher than in the nation. ropolitan areas are basic, or exportable, With the United States Finally, the nation’s employment share in we use location quotients (LQ), a tool Historically, Texas has been known for air transportation is approximately 60 commonly used to analyze the economic oil, cotton and cattle. But in recent percent less than that of Texas. Moreover, base of a region. Location quotients com- decades the state’s image has changed the high-tech, oil and gas, and air trans- pare the local economy with a reference substantially. Today, the economic base is portation industries are the largest em- economy (for example, the Dallas econ- more diverse and includes transportation, ployers in Texas, confirming their impor- omy with the U.S. economy) to identify computer, semiconductor and telecom- tance to the state’s economic base. areas of specialization. The quotients are munication firms. Since higher concentration indicates computed as follows: Chart 1 plots the location quotients of the presence of clusters, or specialization, Texas with the United States as the refer- these industries are the central drivers of ence region.6 The chart shows that even the Texas economy. The prominence local employment in industry i/ though Texas has diversified, the energy of these industries—high tech, telecom- U.S. employment in industry i LQi = industry is still a large part of its economic munication services and air transporta- total local employment/ base. Oil and gas extraction and its sup- tion—helps to explain why the state’s total U.S. employment port activities, pipelines, natural gas dis- economy fared worse than the nation’s tribution, refining and oilfield-machinery during the recent downturn. Despite the Location quotients higher than 1 manufacturing are still agglomerated in high-tech bust, high-tech manufacturing indicate that the regional concentration Texas. However, high-tech and trans- and service sectors remain clustered in of these industries is greater than their portation industries have been added to Texas.7 national concentration and so they are this mix. Computers, telecommunication Based on location quotients, nearly 35 likely to be part of the economic base of services, semiconductors and air trans- percent of Texas’ employment is in indus- the region. The greater the location quo- portation firms now have a larger pres- tries that can be classified as “basic” or ex- tient, the higher the concentration and ence in Texas than in the nation. portable. In these basic industries, 17 per- the more certain we are of the basic Looking first at energy-related indus- cent of Texas’ workforce produces goods nature of the industry.3 tries, the state’s share of employment in and services that satisfy nonlocal demand. Our location quotients are computed oil and gas extraction and mining is nearly using 2000 census employment data from six times the national share. Much less Texas Major Metros Integrated Public Use Microdata Series dramatic, yet significant, are the location The major metropolitan areas in (IPUMS) files.4 We first analyze Texas’ eco- quotients for high-tech manufacturing Texas account for more than two-thirds of nomic base and compare the state’s geo- and services. The share of computer and the state’s employment, so the sectors that graphic dispersion of industries with that peripheral equipment manufacturing in lead these metro economies determine of the nation.5 We then look at the basic Texas employment is 78 percent higher the state’s overall economic base. The activities of the six major metropolitan than in the nation, wired telecommunica- composition of economic activity varies areas in Texas and compare the degree of tion services 50 percent higher, other tele- significantly among Texas’ major metros.

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 13 Each metro area specializes in a unique Chart 2 set of industries, diversifying and strength- Dallas Compared with the United States ening the state’s economy.8 Location Dallas. Location quotients for Dallas quotients confirm the metro is the high-tech mecca, 5 Communications, audio and transportation hub and telecommunica- video equipment mfg. tion nexus of Texas. Dallas’ share of work- 4 Other telecom Wired telecom ers in other telecom services, air trans- Oil and gas Air transportation services services Computer extraction Electronic components systems portation services, communications 3 and products mfg. designs and Structural clay related services equipment manufacturing and computer products mfg. Radio, TV and systems design and services is twice or 2 computer sales more when compared with the overall Data Nondepository Aircraft and Railroad processing credit and Management of state. This concentration of high tech parts mfg. mfg. related Computer and peripheral services companies and 1 activities becomes even more striking when the equipment mfg. enterprises metro is compared with the United States 0 (Chart 2). Employment shares of commu-

SOURCE: 2000 Census IPUMS data; authors’ calculations. nications equipment manufacturing are four times greater than in the nation, while those of telecom services (both wired and other) and air transportation exceed the national shares by three times. Chart 3 Moreover, although Dallas doesn’t have Austin Compared with the United States much energy industry concentration Location compared with the state, oil and gas ex- quotients traction is—surprisingly—Dallas’ fifth- 12 Computer and peripheral most-concentrated industry in compari- equipment mfg. 10 son with the nation. Despite the recent downturn, high-tech manufacturing and 8 services firms remain key contributors of Electronic components and products mfg. the metro’s economic base. 6 Austin. High tech and state govern- Public finance Adminstration of Radio, TV and activities ment compose Austin’s economic base. 4 Other information Computer system designs human resource computer stores services and related services programs This is evident from the metro’s location 2 Executive offices quotients, which surpass the state’s and legislative bodies Data processing Legal services Colleges and employment shares in computer and services universities 0 peripheral manufacturing (6.4 times), SOURCE: 2000 Census IPUMS data; authors’ calculations. electronic components manufacturing (4.8 times), public administration of environmental quality programs (3.8 times), public finance activities (3.6 times), and executive and legislative bodies (3.3 Chart 4 times). Compared with the nation, Houston Compared with the United States Austin’s larger presence of computer and Location chip makers is even more pronounced quotients (Chart 3), and the metro’s share of work- 14 ers employed in these industries signifi- Support activities 12 for mining cantly exceeds the national share (11.5 and 6.9 times, respectively). Although the 10 Pipeline transportation Oil and gas Petroleum recent high-tech bust hit Austin hard, 8 extraction refining semiconductor and computer manufac- Oilfield machinery mfg. Petroleum and petroleum turing industries remain key elements of 6 products, wholesale the metro’s economic base. Metals and minerals, except Management of Chemicals mfg. 4 Natural gas petroleum, wholesale Water companies and Houston. Houston is Texas’ oil and gas distribution transportation enterprises Air capital and home to the sixth largest sea- 2 Computer and Petroleum and transportation Architectural and peripheral equipment mfg. port in the world. Not surprisingly, the coal mfg. engineering services 0 location quotients convey a similar story. SOURCE: 2000 Census IPUMS data; authors’ calculations. Five of the 10 most geographically concentrated industries compared with the

14 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 state as well as the nation are related to oil Chart 5 San Antonio Compared with the United States and gas production, drilling and oil services (Chart 4). Given that the oil and gas Location industry is more prevalent in Texas than in quotients 6 the United States, Houston exhibits much National security stronger oil- and gas-related industry and international 5 affairs clusters in reference to the nation than to Footwear mfg. Texas. Location quotients for both 4 upstream and downstream oil and gas Not specified Savings institutions Business 3 utilities General support Restaurants activities such as extraction, support merchandise including credit unions Natural gas services Home and other Textile product stores Wired telecom activities for mining, pipeline transporta- distribution health care food services mills carriers 2 Electric, gas, services tion, petroleum refining and wholesaling Petroleum Scientific and other utilities Insurance refining Electronic R&D carriers and of petroleum products more than double shopping and services Traveler 1 related activities when the base region changes from Texas mail-order houses accommodation to the United States. Downstream activi- 0

ties such as petroleum refining and chem- SOURCE: 2000 Census IPUMS data; authors’ calculations. icals manufacturing also bolster port activity. Hence, the share of water transportation employment is twice as high in Houston when it is compared with the rest Chart 6 of Texas (2.6 times) as well as the nation Fort Worth Compared with the United States (2.4 times). Also, because Texas has a high share of computer manufacturing, Hous- Location ton’s edge over the nation in computer quotients 7 and peripheral equipment manufacturing Aircraft and parts mfg. (2.5 times higher) is larger than its edge 6 over Texas (1.4 times higher). 5 Aerospace San Antonio. San Antonio’s economic Railroad rolling stock products and 4 Communications, audio base thrives on tourism and the presence parts mfg. Rail and video equipment mfg. transportation of large insurance firms, electric and gas Oil and gas Air transportation Travel arrangements 3 extraction production and distribution firms, and Paperboard boxes Warehousing and reservations and containers mfg. and storage services four military bases. Also, the metro has a 2 Oilfield machinery mfg. Wired significant presence of health care organi- 1 Support activities Machinery mfg. Computer systems telecomm for mining Electronic components design and related zations and recently has become home to carriers and mfg. services several telemarketing companies (Chart 0 5). The metro exhibits similar employ- SOURCE: 2000 Census IPUMS data; authors’ calculations. ment share ratios when compared with the state or nation. First, San Antonio’s share of employment in national security and international affairs is more than five national share by 112 percent and the times that of the nation as well as the state share by merely 38 percent. Job state, largely because of the strong mili- losses during the recent downturn were Exportable goods and tary presence in the metro. Second, there mitigated in San Antonio because of the services are important is sizable specialization of insurance metro’s low concentration of high-tech providers and electric and gas producers industries. Thus, the concentration of the because they generate and distributors. Third, San Antonio has metro’s base industries has held steady. more than twice the share of its aggregate Fort Worth–Arlington. Fort Worth– out-of-state revenue and labor force employed in scientific re- Arlington is a major air and rail trans- search and development compared with portation hub in Texas with historic ties to stimulate state growth. the country and Texas. Last, specialization oil, aircraft and aerospace product manu- in industries related to tourist activity— facturers. Fort Worth’s employment shares general merchandise stores, restaurants in aircraft, aerospace products and parts and traveler accommodation services—is manufacturing, communications equip- at least one and a half times higher in the ment manufacturing as well as air and rail metro than in the state and nation. The transportation are two to four times higher only vivid difference is the concentration than the state’s shares. This agglomeration of workers in wired telecom services; the becomes more prominent when com- metro’s share of these workers exceeds the pared with the nation (Chart 6). Fort

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 15 Chart 7 Michael E. Porter, Economic Development Quar- El Paso Compared with the United States terly, vol. 14, February 2000, pp. 15–34. 3 Although the use of location quotients is preva- Location lent, this measure of the economic base can quotients 12 have shortcomings. Some of the pitfalls of Footwear mfg. regional analysis using location quotients are 10 underestimating the degree of geographic concentration if the reference region is a net Household Cut and sew apparel mfg. 8 appliance mfg. exporter of the good or service and overestimat- ing the degree of geographic concentration if the 6 Nonferrous metal reference region is a net importer of the good or Restaurants production and processing Savings National service. and other institutions security and Plastic food 4 4 Warehousing including international Integrated Public Use Microdata Series: Version products mfg. Agricultural Truck services and storage credit unions affairs implements mfg. transportation 3.0, by Steven Ruggles, Matthew Sobek, Trent Leather tanning Radio, 2 Natural gas and products mfg. Electronic TV and cable Alexander, Catherine A. Fitch, Ronald Goeken, distribution components Department Services incidental broadcasting Justice and public Patricia Kelly Hall, Miriam King and Chad Ron- and products mfg. to transportation services order activities 0 stores nander, Minneapolis, Minn.: Minnesota Popula- SOURCE: 2000 Census IPUMS data; authors’ calculations. tion Center, 2004. For more information, see www.ipums.org. 5 The economic base is computed by adding surplus service-export employment (individuals Worth–Arlington’s employment shares in Conclusion employed in producing services in excess of these same industries is more than three The Texas economy thrives on a local demand) to total manufacturing and mining employment. Therefore, the share of Texas to six times those of the nation. Since the diverse mix of industries. Once known as employment included in the export base is recent recession, the composition of Fort the land of oil, cotton and cattle, Texas has total manufacturing employment + total Worth’s leading industries has remained developed into a high-tech hub. High- mining employment + {sum of [(LQ–1/LQ) × unchanged. tech and energy sectors are the state’s employment] for all service-providing sectors El Paso. El Paso specializes in manu- densest clusters, but Texas has many with LQ above 1.1} facturing, trade and transportation other industries whose shares in the state total Texas employment. because of its close ties with Mexico are higher than their shares in the nation 6 For Texas and all its major metropolitan areas, (Chart 7). More recently, the metro has and thus contribute to the state’s eco- industries with high location quotients, which seen substantial growth in its high-wage nomic base. Nearly 35 percent of Texas’ have been referenced in the text as key contrib- manufacturing and service sector. Gener- employment is in industries that can be utors of their respective economic bases, were ally, the trade sector displays limited spe- classified as basic, or exportable. In these also the largest employers in the state and its cialization. As a result of El Paso’s location basic industries, slightly less than half the major metros, unless otherwise noted. along the U.S.–Mexico border, however, workers are engaged in producing goods 7 Evidence from the 2000 and 2004 Bureau of the metro’s employment share in ware- and services that satisfy nonlocal Labor Statistics data shows that the recession housing and storage is twice that of Texas demand. Exportable goods and services did not change the ordering of Texas’ basic industries. The location quotients of several as well as the nation. The metro’s employ- are important because they generate out- high-tech industries declined slightly, but their ment shares in footwear, cut and sew of-state revenue and stimulate state shares are still higher than U.S. shares. apparel, textile and household appliance growth. Moreover, because clusters 8 For an in-depth analysis of the attributes of the manufacturing are more than 10 times improve efficiency and innovation, the Texas major metros and how each of them grew higher than the state’s employment formation and growth of clusters are during the 1990s, see, “Economic Recovery shares. These shares are also high when important for Texas to maintain its com- Under Way in Major Texas Metros,” by D’Ann compared with the nation. Cut and sew petitive edge in this era of globalization. Petersen and Priscilla Caputo, Federal Reserve apparel, footwear and household appli- Bank of Dallas Southwest Economy, March/ ance manufacturing exceed national Assanie is an assistant economist and April 2004. shares by eight times. Despite the high Yücel is a senior economist and vice presi- shares, the passage of NAFTA has led to dent in the Research Department of the much of this manufacturing going across Federal Reserve Bank of Dallas. the border. Thus, the number employed in these industries makes up only 4 percent Notes of El Paso’s total employment. The largest 1 See “Competitive Advantage, Agglomeration employers in El Paso today are still closely Economies and Regional Policy,” by Michael E. tied to the maquiladora industry across Porter, International Regional Science Review, the border but are a different set of indus- vol. 19, no. 1 & 2, 1996, pp. 85–94. tries, including plastic products manufac- 2 For more information on what clusters are and turers, electronic component and product how they affect competition and innovation, see “Location, Competition and Economic Develop- manufacturers, department stores, truck- ment: Local Clusters in a Global Economy,” by ing, warehousing and storage firms.

16 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 Since 1969, the Texas economy has grown rapidly, consistently matching or Economic Progress exceeding the growth of the national economyS from one decade to the next. Real personal income growth rates in in the Texas Economy Texas matched the U.S. rates even during the oil bust years of 1979–89 and Robert W. Gilmer exceeded U.S. rates in 1969–79 and 1989–2001 (Table 1). Measured by total population, growth in Texas was substantially greater in all periods. The state’s largest metropolitan areas—Dallas–Fort Worth, Houston, Austin and San Antonio, which together make up what is known as the Texas Tri- angle—have contributed the largest part of this growth, especially since 1979. Out- side the Texas Triangle cities, real income growth has failed to match U.S. growth since 1979, although population has expanded somewhat faster. This growth has improved Texas’ economic position relative to the rest of the United States. Texas moved from the nation’s fourth most populous state in 1969 to second in 2001, trailing California but ahead of New York and Florida. In terms of personal income, Texas has moved from the sixth largest state economy in 1969 to the third largest today, behind California and New York. The state’s large metropolitan areas Table 1 have similarly moved up the ranking of Growth of Population and Personal Income in Texas and the United States the nation’s largest cities.1 Dallas–Fort (Average percent per year) Worth, Houston and San Antonio made Population most of their climb through these rank- 1969–1979 1979–1989 1989–2001 United States 1.1 1.0 1.2 Texas 2.3 1.9 2.0 Dallas–Fort Worth 2.4 3.0 2.6 Houston 3.4 1.9 2.3 Austin 4.1 3.8 3.9 In terms of personal San Antonio 1.9 2.1 1.8 Texas Triangle 2.8 2.5 2.5 income, Texas has Rest of Texas 1.7 1.2 1.3 moved from the sixth Personal Income 1969–1979 1979–1989 1989–2001 largest state economy United States 3.7 3.0 2.9 in 1969 to the third Texas 6.0 3.0 4.3 Dallas–Fort Worth 5.4 4.7 4.7 largest today, behind Houston 8.0 2.4 5.1 Austin 2.3 5.6 8.3 California and New San Antonio 4.7 4.2 4.0 Texas Triangle 6.5 3.8 5.0 York. Rest of Texas 5.2 1.7 2.9

NOTE: Based on 1999 metropolitan area definitions of the Office of Management and Budget. Dallas–Fort Worth and Houston use the consolidated metro area definition. SOURCES: Bureau of Economic Analysis; author’s calculations.

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 17 Table 2 success of the Texas economy not by its Rank of Texas Triangle Metro Areas in United States by Population and Personal Income size, growth rates or ranking, but by the Population state’s ability to improve the welfare of its 1969 1979 1989 2001 citizens. In particular, we will look at the Dallas–Fort Worth 12 9 9 8 state’s ability to raise its per capita income Houston 13 10 10 9 levels to those of the nation—to join and Austin 75 63 63 39 perhaps outperform the nation’s main- San Antonio 37 33 33 32 Texas Triangle 4443 stream. Income per person presents a number of flaws as a measure of general Personal Income welfare, but it serves here as a widely rec- 1969 1979 1989 2001 ognized and useful summary of the stan- 3 Dallas–Fort Worth 13 10 9 8 dard of living. Houston 16 9 10 9 Austin 86 69 55 37 Texas Per Capita Income San Antonio 45 39 38 35 In 1969, per capita income in Texas Texas Triangle 9433 was $3,373, or 87.7 percent of the U.S. SOURCES: Bureau of Economic Analysis; author’s calculations. level. Fueled by the oil boom after 1973, Texas’ per capita income grew rapidly to briefly exceed that of the United States by Table 3 1981–82 (Chart 1). The 1980s oil, banking Contribution to Texas Personal Income Growth and real estate bust quickly erased these Percent gains, and by the end of the decade, state 1969–1979 1979–1989 1989–1999 1989–2000 1989–2001 per capita income had returned to 87.9 Texas 100.0 100.0 100.0 100.0 100.0 percent of the U.S. level. Dallas–Fort Worth 23.6 31.9 30.7 31.3 30.6 The 1990s brought new advances rel- Houston 28.5 23.8 28.2 28.6 29.7 ative to the nation as oil, high tech and a Austin 4.1 6.0 8.7 8.6 8.3 free trade- and maquiladora-inspired San Antonio 6.3 8.1 7.1 6.9 7.0 boom along the Texas–Mexico border Texas Triangle 62.6 69.8 74.6 75.4 75.5 produced another burst of Texas eco- Rest of Texas 37.4 30.2 25.4 24.6 24.5 nomic growth. By 1998, Texas per capita SOURCES: Bureau of Economic Analysis; author’s calculations. income returned to 94.4 percent of U.S. levels and made no further progress through 2001. ings between 1969 and 1979 (Table 2).2 8.3 percent. The combined metro areas, We can examine Texas per capita Since 1979, Dallas–Fort Worth and Hous- collectively designated the Texas Triangle income growth both geographically and ton have shared the eighth through tenth in the table, accounted for three-fourths by the components of income—wages spots in population and personal income, of the state’s income growth between 1989 and salaries, proprietor’s income, prop- while San Antonio moved slowly upward and 2001. erty income, transfers and other sources. to 32nd in population and 35th in per- In this article, we will measure the By component, the most interesting sonal income. Austin, however, made steady and dramatic gains. In 1969, at No. 75 in pop- Chart 1 ulation, Austin was the size of Convergence of Texas to U.S. Per Capita Income Levels Canton, Ohio, or Fort Wayne, Ind. But by Index: U.S. per capita income = 100 2001, at 39th, Austin’s population com- 120 pared favorably with that of Nashville or New Orleans. During the same period, 110 Triangle cities Austin surged from 86th to 37th in per- 100 sonal income. Table 3 summarizes the contribution 90 Texas of these different metro areas to Texas’ 80 personal income growth. Except for the 70 oil bust years, Houston contributed Rest of Texas nearly 30 percent of growth, and Dallas– 60 Fort Worth’s growth exceeded Houston’s 50 by the late 1970s. San Antonio’s growth ’69 ’73 ’77 ’81 ’85 ’89 ’93 ’97 ’01 contribution held steady at 6 to 8 percent, while Austin’s doubled from 4.1 percent to SOURCE: Bureau of Economic Analysis.

18 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 Table 4 results come from the growth of wages Growth Rate of Real Per Capita Personal Income and Factors Contributing to Its Growth and salaries and proprietor’s income. The (Average percent per year) geographic designation focuses largely on the Texas Triangle cities, which have Component Percentage Point Contribution Per Capita fueled both the state’s growth and most of Personal Nonfarm wages Proprietor’s Property Transfer Other its recent convergence to U.S. per capita income and salaries income income payments income 1969–1979 income levels. United States 2.6 1.6 –.1 .3 .4 .4 Texas 3.6 3.0 –.2 .2 .2 .4 Framework for Analysis Dallas–Fort Worth 2.9 2.1 .1 .2 .2 .4 The general framework used here is Houston 4.4 4.2 –.1 –.1 .1 .3 shown in Table 4, which summarizes per Austin 3.4 2.7 –.1 .3 .1 .5 capita income growth in Texas by compo- San Antonio 2.7 1.3 .1 .3 .4 .6 nent of income, geographic area and time El Paso 1.4 .9 .2 .3 .6 –.6 Texas Triangle 3.5 2.9 0 .1 .1 .4 period from 1969 to 2001.4 The data are Rest of Texas 3.4 2.8 –.5 .4 .3 .4 presented as percentage point contributions to average annual real per capita 1979–1989 income growth in each region and time United States 2.0 1.4 0 .7 .1 –.1 Texas 1.1 .2 0 .7 .2 0 period.5 Dallas–Fort Worth 1.6 1.1 .1 .4 0 –.1 For example, the growth of per capita Houston .5 –.8 .5 .6 .2 0 income in Texas from 1969 to 1979 aver- Austin 1.9 1.7 –.6 .6 0 .1 aged 3.6 percent per year, with most of the San Antonio 2.0 1.1 .1 .7 .1 .1 growth (3 percent per year) coming from El Paso 1.7 .2 –.2 .7 .1 .9 wages and salaries per capita and smaller Texas Triangle 1.2 .3 .2 .6 .1 0 contributions from property income (0.2 Rest of Texas .6 –.6 –.5 1.1 .4 .2 percent), transfer payments (0.2) and 1989–2001 other per capita income (0.4). Proprietor’s United States 1.7 1.8 .1 –.1 .2 –.2 income per capita grew more slowly than Texas 2.2 2.4 .4 –.4 .1 –.3 other components, reducing the growth Dallas–Fort Worth 2.0 2.4 .2 –.3 .1 –.3 Houston 2.7 2.3 1.0 –.6 .1 –.2 rate by 0.2 percent. Austin 2.9 4.2 .1 –.7 –.1 –.6 The components of income defini- San Antonio 2.2 2.1 .7 –.3 .2 –.5 tions follow standard conventions for El Paso 1.7 .9 .8 –.2 .5 –.3 accounting for personal income in the Texas Triangle 2.4 2.5 .5 –.4 .1 –.3 national income and product accounts. Rest of Texas 1.6 1.5 0 –.4 .4 0 The definitions are fairly obvious: non- SOURCES: Bureau of Economic Analysis; author’s calculations. farm wages and salaries; farm and nonfarm proprietor’s income earned by sole proprietorships, partnerships and tax- Chart 2 shows the path of the four in fact, it may be this very proximity that exempt corporations; property income cities since 1969 in terms of income guarantees their different personalities. from dividends, rent and interest; and growth relative to the nation’s. The gains Because no pair of cities in the Texas Tri- transfer payments for no current services and losses of the boom and bust in oil and angle is more than 240 miles apart, each rendered. The “other income” category is real estate are visible in all four cities, but has assumed a role in the state economy a residual made up mainly of benefits most notably in Houston and Austin. All that sets it apart and makes it distinct paid to wage and salary workers, but it cities made gains in the 1990s, especially from the others.6 also includes a residence adjustment for Austin. San Antonio made the least Dallas–Fort Worth. Dallas–Fort Worth workers who live and work in different progress, despite beginning from the low- is a major inland transportation hub and areas. est per capita base. The two high-tech distribution center for Texas, Louisiana, The rationale for the geographic focus metros began losing ground in relation to Arkansas and Oklahoma and claims the on the Texas Triangle has partly been dis- the United States well before the national world’s fifth busiest airport. Following the cussed above, primarily because three- recession began in 2001, with Austin oil bust, Dallas emerged as the state’s fourths of the region’s personal income peaking at 110 percent of U.S. levels in banking and financial center. Dallas and growth came from these metro areas after 1999 and Dallas–Fort Worth at 112 per- Fort Worth also have a significant presence 1989. Also, most of the forces driving cent in 2000. Houston reached 115 per- of oil-related activity, notable on any income convergence have come from the cent of U.S. per capita income in 2001. standard except that set by Houston. Triangle cities. While per capita income San Antonio stood at 88 percent. High-technology industries, especially levels were, on average, well above The fact that the four cities have such telecommunications, became a major national norms and rising through the different income levels and very different center of growth in the 1990s. 1990s within the Triangle, they were behavior over time might seem surprising Houston. Houston’s bread and butter falling back to near 70 percent outside of it. in light of their geographic proximity. But, remains oil and natural gas, with oil pro-

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 19 Chart 2 industry, with such features as Fiesta Convergence of Texas Triangle Cities to U.S. Per Capita Income Levels Texas, SeaWorld, the River Walk, El Mer- Index: U.S. per capita income = 100 cado and others. Lackland Air Force Base, 140 Fort Sam Houston and Randolph Air Force Base provide a major military pres- 130 Houston ence. 120 Dallas–Fort Worth One could speculate that if Texas’ geography had been only slightly differ- 110 ent—with navigable rivers or a saltwater Austin 100 inlet that cut into the heart of the state— the four cities could easily have been one. 90 San Antonio The port, the inland distribution point 80 and the political capital would all have 70 been colocated. Because the four Triangle ’69 ’73 ’77 ’81 ’85 ’89 ’93 ’97 ’01 cities play such different economic roles, SOURCE: Bureau of Economic Analysis. adding up their current populations produces a not far-fetched approximation of what might have been a single metro area. ducers, oil services and machinery com- The combined ranking of the Triangle panies, refineries and petrochemicals cities (bottom of Table 2) shows that such directly or indirectly accounting for half a combination would rank third among all the metro area’s jobs. The Texas Medical U.S. consolidated metro areas—behind Center and Johnson Space Center, along New York and Los Angeles but ahead of with companies such as Continental Air- Chicago—in both personal income and lines, American General Insurance and population in the 1990s. HP/Compaq, help define the non-oil part It is difficult to generalize about the of Houston’s economy. Houston is the area outside the Triangle, or to easily char- state’s major deepwater port—the second acterize an area that includes cities as dif- largest in the country based on tonnage— ferent as El Paso, Amarillo, Texarkana and and home to the state’s international busi- Beaumont. The decline of agriculture ness community. throughout the second half of the 20th The fact that the four Austin. Because it is the state capital century played a large role in the region’s Texas Triangle cities and site of the University of Texas’ main poor performance. campus, Austin’s major strength has his- In addition, the Texas–Mexico border have such different torically been a robust government sector. acts as a drag on any measure of eco- Beginning in the late 1960s, Austin began nomic progress or welfare in the state, income levels and developing a significant presence in high including per capita income. Gilmer, technology: IBM in 1967, Texas Instru- Gurch and Wang have already examined very different behavior ments in 1969 and Motorola in 1974. The the Texas border cities using the same over time might arrival of chipmaker-consortium Semat- framework employed here.7 The border ech in 1988 provided the momentum for cities’ average per capita income is only seem surprising the 1990s. Today, about 120,000 employ- 50 to 60 percent of the national average ees—25 to 30 percent of the local work- and has only occasionally matched or in light of their force—are tied to technology industries, exceeded the state’s overall growth rate and Dell Inc. has emerged as the city’s (such as Laredo in the 1990s). El Paso, by geographic proximity. most important technology employer. far the largest Texas–Mexico border city, Austin is also renowned for its music saw its per capita income fall from 73 per- industry. Billed as the “Live Music Capital cent of the U.S. average in 1969 to 63 per- of the World,” the city sponsors a number cent in 2001. Although the border saw of festivals and conventions based on gains in income and jobs in the 1990s, music. rapid population growth due to high San Antonio. San Antonio’s historic birthrates and in-migration meant living role has been as the distribution point for standards did not improve nearly as much South Texas and northern Mexico, a role as overall growth statistics might indicate. that has grown with the rapid expansion of the maquiladora industry and the How Income Grew in Texas implementation of the North American Except for the oil bust years, Texas’ Free Trade Agreement. Tourism is a major per capita income outgrew the nation’s by

20 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 Table 5 a significant margin (see Table 4). The dif- Impact on Per Capita Income of Industry Mix, Differential Regional Earnings ference was a full percentage point from and Jobs Per Capita 1969 to 1979 (3.6 versus 2.6) and by half a percentage point from 1989 to 2001 (2.2 Percentage Point Contribution to Annual Growth Rate versus 1.7). With the oil bust and recovery Wages and salaries Industry Differential Jobs per factored in, however, the difference in per worker mix regional earnings capita 1969–1979 favor of Texas narrows to 0.2 percent (2.3 Texas 1.5 1.3 .2 1.5 versus 2.1 over the 32-year period), and Dallas–Fort Worth .8 1.1 –.2 1.3 per capita income rises from 88 percent to Houston 1.9 1.1 .8 2.3 94 percent of the national average. Austin 1.2 1.2 0 1.4 Also except for the oil bust years, San Antonio 1.2 1.5 –.3 .2 most of the growth in Texas’ real per Texas Triangle 1.4 1.2 .2 1.5 capita income came from increases in real Rest of Texas 1.5 1.5 0 1.2 wages and salaries per capita—83 percent 1979–1989 from 1969 to 1979 and 109 percent from Texas .3 .8 –.5 –.1 1989 to 2001. Only during the years of the Dallas–Fort Worth .9 .9 .1 .2 oil and banking crisis did real wages and Houston –.2 .7 –.8 –.6 Austin 1.2 1.2 0 .5 salaries fail to contribute strongly to San Antonio .5 .8 –.3 .6 income growth; only 17 percent of growth Texas Triangle .4 .8 –.4 –.1 came from that source from 1979 to 1989. Rest of Texas –.3 .8 –1.0 –.4 Growth in property income (most proba- 1989–2000* bly in the first half of the 1980s) was the Texas 1.8 1.4 .4 .8 major factor contributing to income Dallas–Fort Worth 2.1 1.3 .8 .7 growth during the decade of the down- Houston 1.9 1.5 .3 .5 turn. Austin 3.7 1.3 2.4 1.5 Proprietor’s income makes its largest San Antonio 1.2 1.4 –.2 1.1 contribution from 1989 to 2001. Houston Texas Triangle 2.1 1.4 .6 .8 has the strongest contribution from the Rest of Texas .7 1.0 –.3 .8 self-employed in this period (1 percent) * Data extend only to 2000 due to a change in the distribution of jobs from the Standard Industrial Classification to North American Industry and during the previous period as well Classification System in 2000, making it impossible to compare 1989 with 2001. (0.5 percent). In 16 cities in Texas and NOTE: Differences due to rounding error. Louisiana, all with strong ties to oil, the SOURCES: Bureau of Economic Analysis; author’s calculations. first result of the oil bust was a large number of new “proprietors,” presumably new 1979 and negative from 1989 to 2001. resources, labor supply, infrastructure or businesses started by people unemployed Other income per capita makes its other local factors. This region-specific by the downturn.8 This forced entrepre- largest contribution from 1969 to 1979, is advantage is called differential regional neurship was followed in the late 1980s neglible from 1979 to 1989 and turns earnings.9 and early 1990s by rapidly growing propri- slightly negative in the most recent WS/P = WS/E × E/P = industry mix × etor’s income, the fruit of the businesses period. differential regional earnings × E/P that succeeded. The often-used analogy of a forest fire leaving behind the seeds for A Closer Look at Wage Table 5 summarizes the contribution the forest’s regeneration seems to apply to and Salary Growth of each of these elements to real per capita Texas in recent years, with entrepreneur- Because wages and salary growth per income.10 The first column is wages and ship sowing the seeds. On average, propri- capita account for such a large share of salaries per worker; the second and third etor’s income contributed 0.5 percent to Texas per capita income, we will examine columns divide this category into two per capita income growth in Texas Trian- it more closely. We can divide wages and parts. The fourth column is the employ- gle cities in the 1990s. salaries per capita (WS/P) into two parts: ment population ratio, or jobs per capita. Property income (dividends, rent and wages and salaries per employee (WS/E) Industry mix was a significant factor interest) was the biggest contributor to and the employment population ratio in all areas and in every period. Texas was per capita income growth during the oil (E/P). clearly shedding low-wage jobs and bust and recovery years. The 1980s saw a replacing them with better-paying jobs WS/P = WS/E × E/P large run-up in property values, which fell throughout the entire period. back slowly late in the decade but drove Further, we can offer two reasons for the We also see gains from differential up rental values, and a sharp hike in inter- growth of wages and salaries per regional earnings in the two periods of est rates due to inflation and tight mone- employee: (1) improvements in the indus- rapid growth. In the 1990s the Texas tary policy increased income from inter- try mix that allow more workers to move Triangle cities added 0.6 percent per year est-earning sources. The contribution of into higher-paying industries, or (2) spe- to per capita income thanks to these property income is small from 1969 to cific advantages the region offers in advantages. The measure highlights the

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 21 state’s booms and busts: Houston added 0.8 percent per year from 1969 to 1979, which turned to –0.8 percent the following decade. Large regional differentials in Austin (2.4 percent) and Dallas–Fort Worth (0.8 percent) mark the 1990s tech boom. A look back at Chart 2 shows that these cities were already giving back some of their tech gains by 2001. During the two decades of strong growth, During the two decades of strong Texas generated jobs faster than the rate growth, the state generated jobs faster than the rate of population growth, of population growth, despite rapid despite rapid in-migration (Table 6). Per capita job growth has occurred inside and in-migration. outside the Triangle cities despite the fact, as mentioned above, that the border cities were unable to attain job growth much faster than population growth. This contributed 1.5 percent per year to Texas per capita income growth (as seen in column 4 of Table 5) from 1969 to 1979 and 0.8 percent from 1989 to 2000. The slight decline in the 1980s (–0.1 percent) was primarily due to slower job growth Table 6 in Houston and areas outside the Triangle. Employment and Population Growth, 1969–2001 Job Growth* Summary and Conclusions 1969–1979 1979–1989 1989–2001 Measured by standards of population, United States 2.2 1.8 1.5 employment and income growth, the Texas 3.8 1.8 2.7 Texas economy has outperformed the Dallas–Fort Worth 3.6 3.2 3.0 U.S. economy since 1969. As shown in Houston 5.8 1.3 2.7 Table 7, by 2001 the state as a whole had Austin 5.5 4.4 4.9 raised its per capita income to 94 percent San Antonio 2.0 2.6 2.7 of the national average, up from 88 per- Texas Triangle 4.3 2.5 3.0 Rest of Texas 3.0 .8 2.0 cent in 1969. Over the same period, the average annual growth rate of per capita Population Growth* income was 2.3 percent for Texas versus 1969–1979 1979–1989 1989–2001 2.1 percent for the United States. United States 1.1 .9 1.2 Economic progress has been uneven Texas 2.3 1.9 2.0 over time. The oil boom briefly pushed Dallas–Fort Worth 2.3 3.0 2.6 Texas per capita income above the Houston 3.4 1.9 2.3 nation’s in 1981–82. In the subsequent Austin 4.1 3.8 3.9 collapse of oil, banking and real estate, San Antonio 1.9 2.1 1.8 Texas fell back to almost its 1969 position Texas Triangle 2.8 2.5 2.5 Rest of Texas 1.7 1.2 1.3 relative to the United States. Most subsequent progress has come since 1989, and Jobs Per Capita* it primarily can be attributed to more jobs 1969–1979 1979–1989 1989–2001 available to the general population and United States 1.1 .8 .3 an improving mix of jobs with higher Texas 1.5 –.1 .7 salaries. Dallas–Fort Worth 1.3 .2 .4 Table 7 also indicates the uneven Houston 2.3 –.6 .4 geographic progress. In fact, the forces of Austin 1.4 .5 1.0 convergence to U.S. levels have mostly San Antonio .2 .6 .9 come from the Texas Triangle metropoli- Texas Triangle 1.5 –.1 .5 Rest of Texas 1.2 –.4 .7 tan areas of Dallas–Fort Worth, Houston, Austin and San Antonio. All these cities * Annualized growth rates. have outperformed the United States SOURCES: Bureau of Economic Analysis; author’s calculations. since 1969, with the most dramatic gains

22 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 coming out of Austin. The addition of Table 7 a large high-technology workforce to a Performance of Regions of the Texas Economy stable, if less-well-paid, government and 2001 per Annual growth rate university base fueled both rapid growth capita income Percent of 1969–2001 and rising per capita income in the state (dollars) U.S. level (percent per year) capital. Except for San Antonio, all the United States 30,413 100 2.1 cities enjoy living standards above the Texas 28,472 94 2.3 U.S. average. Dallas–Fort Worth 33,247 109 2.2 The uneven nature of Texas’ eco- Houston 34,916 115 2.5 nomic history makes it difficult to predict Austin 31,511 104 2.8 future progress. The geographic concen- San Antonio 26,887 88 2.3 Texas Triangle 32,897 108 2.4 tration of growth seems unlikely to Rest of Texas 21,357 70 1.8 change, but the state’s advantages relative to the rest of the nation (as measured SOURCES: Bureau of Economic Analysis; author’s calculations. by differential regional earnings) were dominated by the oil boom from 1969 to 1979 and to some extent by the high- growth and the locational advantages of the to the North American Industry Classification tech expansion of 1989–2001. Advantages state’s largest metro areas. System, beginning in 2001. This made it impos- were concentrated first in Houston, 4 The framework was developed by Daniel H. Gar- sible to compare the distribution of jobs and then in Austin and Dallas–Fort Worth. nick. See “Accounting for Regional Differences income by industry in 1989 and 2001. Predicting the source or location of the in Per Capita Personal Income Growth, next great round of expansion is impossi- 1929–79,” by Daniel H. Garnick, Survey of Cur- ble. rent Business, vol. 62, September 1982, pp. However, since 1969 Texas’ cost 24–34, and “Accounting for Regional Differ- advantages, tax advantages, climate and ences in Per Capita Income Growth: An Update and an Extension,” by Daniel H. Garnick and lifestyle have prepared the ground for fur- Howard L. Friedenberg, Survey of Current Busi- ther growth and development, including ness, vol. 70, January 1990, pp. 29–40. periodic excesses. These Sunbelt advan- 5 Constant dollars are obtained by deflating with tages should persist, making renewed the personal consumption expenditure deflator economic expansion in Texas and contin- (1996 = 100) for all areas. ued progress in raising the state’s living 6 “The Simple Economics of the Texas Triangle” standards simply a matter of time. (January 2004) and “The Texas Triangle as Megalopolis” (April 2004), both by Robert W. Gilmer is a vice president at the Federal Gilmer, in Houston Business, Federal Reserve Reserve Bank of Dallas. Bank of Dallas. 7 “Texas Border Cities: An Income Growth Per- Notes spective,” by Robert W. Gilmer, Matthew Gurch and Thomas Wang, The Border Economy, Fed- 1 The statistics for Dallas–Fort Worth and Hous- eral Reserve Bank of Dallas, June 2001, pp. 2–5. ton use their consolidated metropolitan statistical area definition throughout this article. The 8 “Finding New Ways to Grow: Recovery in the Oil ranking of metro areas includes consolidated Patch,” by Robert W. Gilmer, Houston Business, metropolitan statistical areas (CMSAs) but then Federal Reserve Bank of Dallas, July 1996. excludes all the parts of these CMSAs (metro- 9 The actual calculation of industry mix and differ- politan and primary metropolitan statistical ential regional earnings is spelled out carefully in areas) in the subsequent ranking process. Garnick and Friedenberg (1990). The calculation 2 The end years used here—1969,1979,1989 and depends on the definition of hypothetical income 2001— are all peak years in the U.S. business (H), total wages and salaries that would have cycle. Although Texas and its metro areas did been earned in Texas if compensation were paid not always follow the U.S. cycle, particularly in at the national rate in each industry. Hypotheti- the 1980s, these years were typically times of cal income was calculated using the wage and economic expansion for Texas, making compar- salary employment categories in the Bureau of isons to the U.S. economy appropriate. Economic Analysis’s Regional Economic Infor- mation System, essentially a one-digit definition 3 The most notable flaw in the use of per capita in the Standard Industrial Classification. Using income as a measure of welfare is that it tells us this definition, nothing about the size distribution of income among the population. However, this article WS/P = industry mix × differential regional divides per capita income into enough cate- earnings × E/P = H/E × WS/H × E/P. gories by component and geography to give 10 The data in Table 5 extend only to 2000 because some insight into how income growth is affected of the change in the industrial classification sys- by regional wage levels, job growth, population tem from the Standard Industrial Classification

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 23 Texas Border Benefits from Retail Sales to Mexican Nationals Keith R. Phillips and Roberto Coronado

Over the past 10 years, trade between in Mexico but also on the U.S. side due to the United States and Mexico has increasing border interlinkages. The boomed, partly because of the significant maquiladora industry has been the main Oreduction in tariffs from NAFTA and the economic growth driver along the strong growth in the maquiladora indus- Texas–Mexico border. try. Along with the expansion in trade, Several studies have addressed the there has been strong population growth issue of cross-border retail trade as part of along the northern border of Mexico. a larger question of the maquiladora in- Generally, the population in Mexican bor- dustry’s impact on the regional economies der cities is significantly larger than in the of U.S. border cities. The first studies on the corresponding U.S. sister cities. Moreover, subject date back to the early 1970s and the South Texas border metros are a short indicate that a significant portion of drive from the industrial city of Monter- maquiladora salaries was spent on the U.S. rey, which had a population of 3.8 million side of the border, mainly on food and in 2000. The large and growing population clothing. More specifically, one study esti- on the Mexican side of the border repre- mates that a 10 percent increase in sents an important consumer base for maquiladora employment translates into a retail stores in U.S. border towns. 23 percent increase in retail sales in While commercial trade between the Brownsville, a 13 percent increase in United States and Mexico is well docu- Laredo, an 11 percent increase in El Paso mented, less is known about the size of and a 7 percent increase in McAllen.2 the nations’ cross-border retail trade. Perhaps the first researcher to study Though small in comparison with com- the impact of the maquiladoras along the mercial trade, this retail trade is a signifi- Texas border in a comprehensive manner cant part of many border city economies. was J. Michael Patrick.3 His main conclu- In 2003 alone, there were more than 38 sion regarding cross-border retail trade million noncommercial crossings at the activity is that growth in the maquiladora El Paso Convention & Visitors Bureau bridges along the Texas–Mexico border. industry in Mexico stimulates U.S. border Many of these individuals were coming to job growth mostly in the retail and service purchase goods to take back to their home areas having the biggest share of their sectors, not in the manufacturing sector country. Due to differences in national retail sales going to Mexican nationals will as commonly perceived. policies such as environmental laws, taxes be impacted the most by large swings in One of the first studies to quantify the and consumer safety regulations, people the value of the peso. We thus check that impact of Mexican nationals on retail cross daily to purchase goods and services our estimates are consistent with the trade on the U.S. side of the border was on both sides of the border. effects on local retail sales of movements done by the San Diego Chamber of Com- Since most of the retail trade con- in the real dollar/peso exchange rate. merce in 1979.4 Through surveys, the ducted on the U.S. side of the border is study estimated that 7.5 percent of San done in cash, it is difficult to document the Previous Research on Border Retail Diego’s retail sales ($407 million) could be share of retail spending by Mexican Traditionally, the border has been a attributed to Mexican nationals. In 1993, nationals. In this article, we use a simple region of fast population and job growth according to a study by the San Diego Dia- consumption function to estimate the compared with the rest of the United States logue, about 42 percent of the people who amount of retail spending that is essen- and Mexico. The Border Industrialization crossed into San Diego were Mexican tially exported to Mexico via cross-border Program—enacted in 1965 by the Mexican nationals with the main purpose of shop- shoppers.1 Since the true amount spent by government after the United States ended ping. They accounted for $2.8 billion in Mexican nationals is not known, it is diffi- the Bracero Program—gave birth to the retail sales.5 cult to estimate the accuracy of our meas- maquiladora industry, which in turn inten- More recently, in 2002, Charney and ures. Theory tells us, however, that metro sified the border region’s growth, not only Pavlakovich-Kochi estimated the eco-

24 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 nomic impact of Mexican visitors to the aged 46 percent. For each of the four bor- Table 1 Border Exported Retail Sales, economy of Arizona. They found that der MSAs, they multiplied 0.46 by total 1978–2001 Mexican visitors spent $962 million, with personal income to get an estimate of the vast majority in department stores (41 retail sales purchased by the local popula- Average share (percent) percent) and grocery stores (25 percent), tion and then subtracted sales to locals Brownsville 25.7 mostly in border counties.6 Similarly, on from total sales to get net exported retail El Paso 11.3 the Texas–Mexico border, the Center for sales. If the value of net exported retail Laredo 51.1 Border Economic Studies at the Univer- sales is negative, that means more local McAllen 35.6 sity of Texas–Pan American estimated income is spent outside the local econ- SOURCE: Authors’ calculations. that total expenditures by Mexican visitors omy than income spent by outsiders in in the lower Rio Grande Valley amounted the local community. While it is evident to $1.4 billion in 2003.7 that many Mexican nationals cross the Other studies have focused on the border to shop, U.S. citizens also cross We use the results from the model to impact of exchange rate fluctuations on into Mexico to dine at restaurants and to estimate exported retail sales for the four U.S. border retail sales. For instance, Diehl buy local handicrafts, medicines, liquor, Texas border MSAs. Table 1 reports the concludes that the 1982 Mexican eco- dental services and other products and average share of exported retail sales for nomic crisis that triggered peso devalua- services. Border residents also vacation these four areas during our estimation tion stunned South Texas retailers by cut- and shop at other destinations in the period. According to our results, in 2001, ting retail sales as much as 80 to 90 per- United States. Remittances to family Mexican shoppers accounted for more cent in many border businesses.8 Simi- members in Mexico can also reduce the than $2 billion in retail sales, representing larly, Patrick and Renforth estimate, amount of local income spent on local 0.75 percent of total retail sales in Texas. In through the use of almost 4,000 surveys, retail goods and thus reduce net exported 2001, McAllen was the biggest net ex- that the 1994 peso devaluation resulted in retail sales. porter of retail sales to Mexicans, with a strong 41.8 percent decline in retail Using a constant fraction of local per- almost $1 billion in sales, representing 33 sales, but the results varied by city, store sonal income to estimate the amount that percent of its total local retail trade activ- type, distance from the border and rela- locals spend on retail—and using this ity. Laredo came in second with $540 mil- tive domestic market size.9 Gerber docu- amount to estimate net exported retail— lion in exported retail sales, or 39 percent ments the relationship between peso produces reasonable results. However, we of total retail sales. Brownsville registered value fluctuations and total taxable sales can further refine the model by decom- $256 million (16 percent of total retail in San Diego and Imperial counties, posing personal income into three com- sales), while El Paso, the biggest of the where he finds that an unanticipated 10 ponents, allowing the coefficient on each four cities in terms of population, ex- percent decline in the value of the peso component to differ. The border region ported only $215 million (6 percent) to depresses total taxable sales by approxi- has a low employment-to-population Mexican nationals. El Paso’s figure is well mately 1 percent in San Diego County and ratio due to its young labor force and high below its average exported retail sales of 2.22 percent in Imperial County.10 unemployment rates. It also has persist- 11.3 percent and is primarily due to the Many of the studies, however, are ently low per capita personal income yet contracted maquiladora activity south of region- and time-specific, making com- strong job growth rates. If these factors the border. Ciudad Juárez registered its parisons across regions and over time dif- play differing roles in retail spending, it is worst maquiladora performance in 2001 ficult. Also, many of the studies were done important to separate them out. We divide and 2002, with employment declining using time-consuming, labor-intensive, personal income (Y) as follows: almost 25 percent.

and thus expensive, survey techniques On average over the 1978–2001 ( that would be difficult to perform consis- Y =××Y ( POP EMP ,, period, Mexican nationals accounted for tently over time and across regions. To ((POP EMP 1.6 percent of Texas retail sales, or $5.1 overcome these limitations, we use a sim- million on a daily basis. Chart 1 shows ple consumption function approach that where POP is population and thus Y/POP is that over time Laredo has the highest produces a consistent annual time series per capita income, POP/EMP is the inverse share of exported retail sales to actual of exported retail sales for the four metro- of the employment-to-population ratio and total sales, followed by McAllen, Browns- politan statistical areas (MSAs) on the EMP is total employment. We then try to ville and El Paso. Texas–Mexico border. estimate the impact of the three components of personal income on retail sales Sensitivity to Exchange Rate Swings Using a Different Approach across the 23 non-border Texas MSAs. We Although there is no straightforward Phillips and Manzanares propose a use quarterly retail sales data at the metro way to determine the accuracy of our simple model in which it is assumed that level from 1978 to 2001, available from the results, retail sales from Mexican nation- individuals spend a fixed proportion of Texas comptroller’s office. Annual personal als should be sensitive to swings in the their income on consumption, or in this income for metro areas (less contributions value of the peso. These swings represent case, retail sales.11 For instance, they find for social insurance) from 1978 to 2001 is price shocks for Mexican nationals shop- that from 1986 to 1998 retail sales as a available from the Commerce Depart- ping on the U.S. side, and border retailers fraction of personal income in Texas aver- ment’s Bureau of Economic Analysis. know that sharp declines in the peso’s

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 25 Chart 1 Notes Exported Retail Sales Shares 1 For a more detailed version of this article, please Percent of total retail sales Real exchange rate see “Exported Retail Sales along the Texas–Mex- ico Border,” by Keith R. Phillips and Roberto 70 .020 Coronado, Federal Reserve Bank of Dallas work- Laredo 60 ing paper (forthcoming). Real exchange rate .016 50 2 See “Maquiladoras along the Texas–Mexico Bor- der: An Econometric Evaluation of Employment 40 .012 and Retail Sales Effect on Four Texas Border 30 SMSAs,” by Richard J. Holden, Texas Depart- McAllen ment of Community Affairs, Regional Economic 20 Brownsville .008 Development Division, February 1984. El Paso 10 3 .004 See “The Economic Impact of Maquiladoras on 0 Border Development: A Rio Grande Valley Case Study— Some Preliminary Findings,” by J. -10 0 ’78 ’79 ’80 ’81 ’82 ’83 ’84 ’85 ’86 ’87 ’88 ’89 ’90 ’91 ’92 ’93 ’94 ’95 ’96 ’97 ’98 ’99 ’00 ’01 Michael Patrick and Roland S. Arriola, paper presented at the Western Social Science Associ- SOURCES: Authors’ calculations; Federal Reserve Bank of Dallas. ation meeting, El Paso, Texas, 1987; “The Employment Impact of Maquiladoras along the Chart 2 U.S. Border,” by J. Michael Patrick, in The Border Retail Sales Are Closely Related to Real Exchange Rate Maquiladora Industry: Economic Solution or Problem?, ed. Khosrow Fatemi, New York: Total retail sales index, 1978 = 100 Real exchange rate Praeger Publishers, 1990, pp. 31–35; and “The 650 .020 Impact of NAFTA on Border Maquiladora and McAllen Industrial Activity,” by J. Michael Patrick, Tech- nical Report, Center for Entrepreneurship and 550 .016 Real exchange rate Economic Development, University of Texas– 450 Pan American, 1991. .012 4 “Mexican Impacts on Retail Sales,” San Diego 350 Chamber of Commerce, San Diego Economic Brownsville .008 Bulletin, vol. 27, no. 6, 1979. Laredo El Paso 250 5 “Who Crosses the Border: A View of the San .004 Diego/Tijuana Metropolitan Region,” Technical 150 Report, San Diego Dialogue, University of Cali- fornia, San Diego, 1993. 50 0 ’78 ’79 ’80 ’81 ’82 ’83 ’84 ’85 ’86 ’87 ’88 ’89 ’90 ’91 ’92 ’93 ’94 ’95 ’96 ’97 ’98 ’99 ’00 ’01 6 “The Economic Impacts of Mexican Visitors to Arizona: 2001,” by Alberta H. Charney and Vera SOURCES: Authors’ calculations with data from Texas Comptroller of Public Accounts and Banco de México. K. Pavlakovich-Kochi, Research Studies, Eco- nomic and Business Research Program, Univer- sity of Arizona, July 2002. 7 “The Economic Impact of Mexican Visitors to value result in a sharp drop in Mexican Outlook the Lower Rio Grande Valley 2003,” by Suad shoppers. Under our model, exported In mid-2005 the real value of the peso Ghaddar, Chad Richardson and Cynthia J. retail sales seem to be responsive to was above its 20-year average and the Brown, Technical Report, Center for Border Eco- changes in exchange rate (see Chart 1). maquiladora industry was continuing to nomic Studies, University of Texas–Pan Ameri- If exported retail sales represents a bounce back from its downturn in can, 2004. significant portion of total retail sales, 2001–03. Both of these factors should con- 8 “The Effect of the Peso Devaluation on Texas changes in the value of the peso should tinue to stimulate growth along the Texas Border Cities,” by Philip N. Diehl, Texas Busi- have statistically significant impacts on side of the border. Looking to 2006, Mex- ness Review, vol. 57, May/June 1983, pp. total retail sales. To asses this, we perform ico is hoping to have its second consecu- 120–25. some statistical tests on the sensitivity of tive presidential election without a peso 9 “The Effects of the Peso Devaluation on Cross- overall retail sales to changes in the value devaluation. The Texas border commu- Border Retailing,” by J. Michael Patrick and William Renforth, Journal of Borderlands Stud- of the peso. Results show that, in all MSAs nity is hoping for the same, as its economy ies, vol. 11, Spring 1996, pp. 25–41. but El Paso, changes in the real exchange ebbs and flows with the movements in the 10 “The Effects of a Depreciation of the Peso on rate have statistically significant impacts value of the peso and the accompanying Cross Border Retail Sales in San Diego and on total local retail sales. The magnitude waves of Mexican shoppers. Imperial Counties,” by James Gerber, working of the impact was the largest in Laredo. paper, San Diego State University, June 1999. Since our results show that El Paso had the Phillips is a senior economist and policy 11 “Transportation Infrastructure and the Border smallest share of its retail sales going to advisor at the San Antonio Branch and Economy,” by Keith R. Phillips and Carlos Man- Mexican nationals and Laredo had the Coronado is an assistant economist at the zanares, The Border Economy, Federal Reserve largest, these results are consistent with El Paso Branch of the Federal Reserve Bank Bank of Dallas, June 2001. our previous findings (Chart 2). of Dallas.

26 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 Texas Border Employment and Maquiladora Growth Jesus Cañas, Roberto Coronado and Robert W. Gilmer magazine

Now Courtesy of Mexico

In the 1990s, the Texas economy increased retail sales in U.S. border cities; the industry lose 290,000 jobs between exceeded even the remarkable perform- and rapid expansion of the maquiladora October 2000 and July 2003, many ance of its U.S. counterpart. State job industry. observers are questioning the industry’s Igrowth averaged 2.9 percent per year from Maquiladora expansion came on the future. Recession, rising wages in Mexico, 1990 to 2000, well ahead of the 1.8 percent heels of NAFTA implementation and the low-wage competition from countries annual increases in the United States. 1994–95 peso devaluation. In recent such as China and Mexico’s inability to Three engines drove the Texas economy years, however, this part of the border deal with growing problems in its compet- forward in the 1990s: the oil sector, high boom has turned to bust. After watching itive environment have all contributed to tech (especially in Austin and Dallas) and a boom in border-city employment. Table 1 Percent Job Growth Along the Texas–Mexico Border Employment growth in the four largest Texas border cities topped that of the Texas El Paso Laredo Brownsville McAllen nation, and the three south Texas cities 1991 .7 2.3 4.3 2.0 1.5 outperformed the state by a wide margin 1992 1.9 3.6 8.7 4.8 5.3 1993 3.3 2.2 3.6 4.4 4.5 (Table 1).1 1994 4.3 3.9 7.2 6.0 7.3 The accelerated job growth along the 1995 2.9 0 –5.5 .4 2.7 Texas–Mexico border was the result of 1996 3.2 2.0 5.1 2.8 3.3 several factors: a quick Mexican recovery 1997 4.4 2.7 7.6 3.0 3.8 after the 1994–95 financial crisis; tight 1998 3.6 1.1 2.7 2.3 5.6 labor markets in the United States that 1999 2.2 2.0 5.3 5.8 6.5 attracted employers to the border in 2000 2.8 1.5 3.0 5.1 5.1 1990–2000 2.9 2.1 4.1 3.7 4.6 search of the region’s surplus labor; a strong peso for much of the period, which SOURCE: Federal Reserve Bank of Dallas, El Paso Branch, with data from the Texas Workforce Commission.

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 27 Chart 1 2). Throughout the latest recession and Maquiladora Employment Growth slow recovery, manufacturing was the Thousands, seasonally adjusted data hardest hit part of the U.S. economy, and 1,400 maquiladora output and employment 1,300 1,200 generally followed the lead of U.S. indus- 1,100 trial production. In mid-2003, however, 1,000 strong U.S. industrial growth finally 900 800 returned, and as Table 2 shows, maqui- 700 ladora employment has returned to re- 600 covery as well. Job growth remains 500 400 uneven among the Mexican border cities, 300 however, with Ciudad Acuña, Matamoros 200 and Piedras Negras recovering more 100 0 slowly. ’80 ’81 ’82 ’83 ’84 ’85 ’86 ’87 ’88 ’89 ’90 ’91 ’92 ’93 ’94 ’95 ’96 ’97 ’98 ’99 ’00 ’01 ’02 ’03 ’04 ’05

SOURCES: Federal Reserve Bank of Dallas, El Paso Branch, with data from INEGI. How Do Maquladoras Affect the Texas Border Economy? The original vision for maquiladoras the recent downturn. Tyco, General Electric and ITT. was the “twin plant,” with capital-inten- This article looks at the maquiladora’s The maquiladora industry has been sive operations located a few miles inside role in today’s Texas economy, especially highly cyclical since its inception, falling the U.S. border and low-wage, labor- how it affects Texas border cities. We also into its first recession in 1974 with an 11.5 intensive operations close by on the Mex- assess the industry’s future and the percent decline in employment. Table 2 ican side. However, the bulk of U.S. man- prospects for the maquiladora to again be shows the uneven effects of the latest ufacturing was already established in the a significant factor in job growth in Texas maquiladora downturn on Mexican bor- Midwest, and trucking deregulation and Mexico. der cities. Maquiladora employment in would make transportation links between Ciudad Juárez was higher than in all the the border and the Midwest both easier Growth and Decline other cities combined when the recession and cheaper in the 1970s and ’80s. The The maquiladora industry began in began, and it has sustained the largest twin-plant vision was never realized along 1965 and experienced slow but steady percentage losses from peak to trough the border. Instead, the maquiladora sup- growth under the Border Industrialization (27.7 percent). Piedras Negras, Nuevo ply chain remained concentrated in states Program. The canceled Bracero Program Laredo and Matamoros also suffered large such as Illinois, Michigan and Ohio. had used Mexican labor in agriculture, percentage losses, all in excess of 24 per- What economic impact would a new and the replacement maquiladora was cent. Ciudad Acuña and Reynosa were maquiladora in Mexico have on a neigh- designed to relieve the resulting high exceptions to the deep recession, with boring U.S. city? The list might run as fol- unemployment rates in northern Mexico. Ciudad Acuña declining only 10.6 percent lows. To select and develop a site, U.S. The new program used low-wage Mexican and Reynosa continuing to grow through- legal, engineering and financial assistance labor as a lure to draw U.S. manufacturing out the downturn. Newer plants, a better would be used. Once established, the new to the region, allowing companies to industry mix and a business-friendly plant would rely on U.S.-based businesses move production machinery and environment account for their better perfor customs, brokerage, warehousing and unassembled parts into Mexico without formance. transportation services. The plant would tariff consequences, as long as the assem- The cyclical nature of the maquila- also purchase a variety of office, packag- bled product was returned to the United dora industry is not surprising, given ing and industrial supplies. Corporate States for final sale. its close ties to U.S. manufacturing (Chart management, engineers and quality spe- Chart 1 shows the elevenfold increase in maquiladora employment between Table 2 1980 and its peak in 2000, from 120,000 Texas–Mexico Maquiladora Border Employment workers to 1.3 million. In 1980, about 94 Peak Trough percent of maquiladora employment was Jobs Date Jobs Date April 2005 in the border states of northern Mexico.2 Ciudad Juárez 262,550 October 2000 189,930 June 2003 213,389 Today, the share has slipped to 76 percent, Ciudad Acuña 37,512 November 2002 33,541 February 2005 33,674 but the northern states still dominate. In Piedras Negras 15,222 February 2000 10,939 December 2004 11,187 2004, 2,810 operating plants accounted Nuevo Laredo 22,915 February 2000 17,171 April 2003 22,233 for about 9 percent of formal employment Reynosa 86,925 April 2005 N/A N/A N/A in Mexico, or 3 percent of the total labor Matamoros 68,413 October 2000 51,900 August 2003 53,002 force. The companies operating under the NOTES: Seasonally adjusted data; border twin cities are as follows: Ciudad Juárez–El Paso, Ciudad Acuña–Del Rio, Piedras Negras–Eagle maquiladora program are a who’s who of Pass, Nuevo Laredo–Laredo, Reynosa–McAllen and Matamoros–Brownsville. U.S. industry, including Delphi, Mattel, SOURCE: Federal Reserve Bank of Dallas, El Paso Branch, with data from INEGI.

28 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 Chart 2 research suggests that U.S. border warehousing employment accelerated Maquiladora Ties to states—with the exception of Arizona, quickly. Business service employment, U.S. Industrial Sector where job losses ranged from negligible to especially personnel supply services, Index, January 2000 = 100, seasonally adjusted small—gained jobs as a result of growth in computer programming and data pro- 110 the maquiladora industry.8 cessing, grew 45 percent from 1990 to U.S. industrial 105 A more recent development has been 2004. El Paso’s maquiladora-related busi- production the arrival of component parts and mate- nesses rely heavily on temporary staffing 100 rial suppliers in U.S. border cities. Specific agencies to hire additional personnel to 95 examples can be found in El Paso, neigh- meet rising demand. Computer program- 90 Maquiladora bor to Ciudad Juárez, which is home to the ming and data service workers help mini- employment largest number of maquiladora employ- mize the burden of paperwork required by 85 ees along the U.S. border. Over the past customs agencies to export or import 80 2000 2001 2002 2003 2004 2005 decade, an increasing number of rubber components. Legal employment grew 20 SOURCES: Federal Reserve Bank of Dallas, El Paso Branch, with data and plastics, electronics and electrical percent over the same period. Similar from INEGI; Federal Reserve Board. equipment, and metal fabricating plants results can be found up and down the U.S. have begun to operate in El Paso to serve border. cialists would be drawn to the border to as suppliers to the maquiladora industry The definitive study on the linkages visit this plant, and they would spend (Chart 3). between maquiladoras and the border money on food and lodging.3 Maquiladora Components supplied include com- economy, by Gordon Hanson, takes all employees draw their salary in Mexico but puter housings, electrical wiring har- these factors into account.9 Hanson esti- do a significant share of their shopping in nesses, special dies and tools, and electri- mates that a 10 percent increase in the United States, stimulating employ- cal switches. About 26 plastic-injection maquiladora output in a Mexican border ment in local retail and service sectors. molding plants can be identified, 31 metal city will increase employment in its U.S. These impacts on the U.S. border stamping companies, and 12 electric- and city pair by 1.1 to 2 percent. He provides have been recognized for some time, and electronic-related companies. Together, more specifics by estimating that this a number of studies were conducted in these companies employed 4,000 workers same 10 percent increase in output would the 1970s and ’80s to quantify them. For in 2004. The manufacturing sectors that increase wholesale trade employment in instance, in 1972, Ladman and Poulsen supply the maquiladoras paid about 40 the U.S. city by 2.1 to 2.7 percent, trans- found that Agua Prieta, Sonora, maquila- percent more in hourly wages than the portation services by 1.7 to 2.7 percent, dora workers spent 40 percent of their low-wage apparel, textile and leather in- manufacturing by 1.2 to 2.1 percent and wages in Arizona.4 Ayer and Layton esti- dustries that traditionally operated in retail trade by 1 to 1.8 percent. mated the maquiladoras’ impact on value El Paso. added and population, using an input– Maquila manufacturing in Mexico The Role of Recession output model for the Arizona–Mexico also positively influences El Paso’s The recent recession has played an border economy. They concluded that employment in transportation, real important role in the latest downturn of Mexicans’ expenditures due to the growing estate, and legal and accounting services the highly cyclical maquiladora industry. presence of twin plants increased value (Chart 4). Given the rapid increase in At the same time, maquiladoras have long added by 14 percent and population by 11 trade flows after 1993, transportation and served as a low-wage platform for U.S. percent on the U.S. side of the border.5 In a 1984 study of the Texas border, Holden estimated that maquiladora Chart 3 Maquiladora Suppliers in El Paso employment had a large impact on employment in the border communities Index, 1990 = 100 of El Paso, Laredo, McAllen and Browns- 350 ville. For instance, a 10 percent increase in 300 maquiladora payroll results in a 2 to 3 per- Apparel cent increase in employment in El Paso 250 and McAllen as well as a 3 to 4 percent Plastics 200 increase in Laredo and Brownsville.6 In another study, Sprinkle found that 150 Metalworking during the early 1980s Ciudad Juárez maquiladoras accounted for one of five 100 jobs created in El Paso, and these new jobs 50 Electrical equipment were concentrated in the service sector.7 0 Silvers and Pavlakovich assessed the rela- 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 tive magnitude of employment gains and NOTE: North American Industry Classification System (NAICS) codes used in the chart were 3327, 3329, 3328, 3332, 3335, 3159, losses across U.S. border regions due to 3261 and 3344. maquiladora industry activity. Their SOURCE: Federal Reserve Bank of Dallas, El Paso Branch, with data from Bureau of Labor Statistics.

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 29 Chart 4 detailed results are reported elsewhere.11 Service Employment in El Paso To examine the future of the industry Index, 1990 = 100 under various assumptions, we simulated 350 maquiladora employment following its second quarter 2000 peak. The base case 300 was the actual outcome through 2002, a 250 Transportation and decline of 14.5 percent for the industry as transportation services a whole. Scenario 1 (S1) assumed no 200 recession and that the U.S. unemploy- Real estate 150 ment rate held firm at a historically low 4 percent through the end of the period. 100 Legal and accounting services Real relative wages rose in this scenario, 50 just as in the base case. Scenario 2 (S2)

0 assumed the recession occurred but that 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 real relative maquiladora wages fell 6.1 NOTE: North American Industry Classification System (NAICS) codes used in the chart were 4889, 5311, 5312, 5313, 5411 and 5412. percent after second quarter 2000 instead SOURCE: Federal Reserve Bank of Dallas, El Paso Branch, with data from Bureau of Labor Statistics. of rising 16.8 percent. And scenario 3 (S3) assumed the best of both worlds for maquiladora managers, falling real rela- manufacturing, and the rise of new low- tion and Development and World Trade tive wages and no recession. wage alternatives such as China, India Organization rules. Chart 5 shows the results for all and Vietnam has broadened U.S. options To focus on the question of how maquiladoras combined. This can be for manufacturing. These very low-wage maquiladoras will respond to economic computed two ways: as the result of a sin- competitors, plus rising real Mexican recovery and which sectors would gle estimate based on the sum of all wages, have become a factor in pushing benefit, we developed some econometric maquiladora employment or as the sum some maquiladora activity abroad. Mex- estimates. As in other models, our of the simulation results for 10 maqui- ico generally has looked at the loss of the methodology confirmed that past maquiladora sectors. Fortunately, they agree lowest wage jobs as an inevitable price of ladora employment has primarily been quite closely. Eliminating the U.S. reces- progress, because increasing domestic driven by the business cycle and relative sion in S1 would provide an increase of wage levels must be seen as a positive real wages.10 Trends and dummy shift approximately 20 percent in employment aspect of economic development. The variables were included to account for in the simulation period, replacing a government has expressed reluctance to structural change, particularly testing for decline of about 14.5 percent in the base enter into subsidy programs to retain or breaks with the 1994 implementation of case. The percentage turnaround for S2 is attract these industries, considering such NAFTA and the 1994–95 financial crisis in similar, and the combined effect in S3 is a action as poor fiscal policy and a violation Mexico. The general methodology follows 31 percent increase. of Organization for Economic Coopera- several papers by Branson and Love, and Four individual sectors do not

Chart 5 Simulation Results

Total Maquiladoras Sum of Maquiladora Sectors

Scenario 3 31.3 Scenario 3 31

Scenario 2 21.3 Scenario 2 17.7

Scenario 1 21.3 Scenario 1 20.1

–14.5 Base -14.6 Base

–20 –16 –12 –8 –4 0 4 8 12 16 20 24 28 32 36 –20 –16 –12 –8 –4 0 4 8 12 16 20 24 28 32 36 Percent change in jobs Percent change in jobs

NOTE: The base cases in the two calculations are slightly different because the chemicals sector was excluded from the sum of regression results. There was a break in the data for this sector. SOURCE: Authors’ calculations.

30 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 Table 3 the rise of foreign competition means Throughout the 1990s, the United Maquiladora Sectors That Are Unresponsive even sectors returning to positive growth States supplied the vast majority of to a U.S. Economic Rebound with economic recovery may experience maquiladora industry inputs. In 2000, 90 (Percent change in jobs) slower job growth than in the recent past, percent of maquiladora inputs were from Not as some products within the sector are the United States and 9 percent were from Leather Toys Furniture classified commoditized. Asia, with China contributing only 1 per- Base –25.6 –31.5 –8.7 –4.9 S1 –7.8 –21.0 –8.9 –2.1 In assessing Mexico’s competitive cent (Chart 6). By 2004, 59 percent came S2 –6.8 –34.7 17.3 –22.1 prospects, the nation retains crucial from the United States and 35.7 percent S3 –7.4 –5.1 4.4 5.8 advantages over the rest of the world, from Asia, including 11.1 percent from even as domestic wages rise. The most China. The United States remains the SOURCE: Authors’ calculations. important factor is proximity to the U.S. majority supplier, but this rapidly moving market. For example, bulky items that trend continued to run in favor of Asia respond to an upturn in the U.S. econ- have a high ratio of weight to value, such into 2005. omy: leather, toys, furniture and a group as large-screen televisions or major appli- The vehicle for entry of foreign inputs of other, unclassified maquiladoras. Their ances, will remain competitive. Proximity to Mexico is 20 sectoral promotion pro- simulation results are summarized in also matters if the inventory cycle is short, grams, or PROSECs, created by the Mexi- Table 3. Combined, these maquiladoras if there are constant design changes or if can government in December 2000. They accounted for 226,782 jobs at the second there must be frequent retooling. Mexico were created in response to implementa- quarter 2000 peak, or 18.1 percent of the will also be competitive when quality is tion of NAFTA Article 303, which in Janu- total. These sectors are unlikely to return more important than price, such as with ary 2001 eliminated duty-free imports of to growth with U.S. economic recovery. medical equipment or when intellectual maquiladora inputs from non-NAFTA The three largest maquiladora sec- property rights are critical.12 countries. The PROSECs protect the entry tors, together accounting for 76.1 percent to Mexico of non-NAFTA components of the peak employment, all respond pos- Texas-Based Suppliers that are not readily available in the itively to economic recovery in the simu- The maquiladoras’ contribution to domestic market, allowing them to enter lations. In S1, electrical machinery U.S. border city growth in the 1990s under reduced tariffs of zero to 5 percent. records an 18.2 percent increase, in place stemmed from (1) the spillovers from Despite the paperwork and the need to of a 26.1 percent decline. Textiles turn rapid maquiladora expansion in neigh- track the origin of thousands of parts to around to record a 63.2 percent gain in S1, boring Mexican cities and (2) the shift of comply with PROSECs, maquiladoras and transportation equipment (which did many maquiladora suppliers to border have apparently fully embraced the pro- not decline after second quarter 2000) cities from their base in the Midwest. We grams. grows by another 4.5 percent in this sce- have already shown how foreign competi- Data are not available on exactly nario. tion and rising real wages in Mexico have which inputs are being displaced, making In conclusion, less than 20 percent of reduced the prospects for maquiladora it difficult to assess the impact on Texas maquiladora employment is in sectors growth, but foreign competition is also border communities. For example, if the that are unresponsive to economic recov- making significant inroads into the 1990s shift of suppliers to the border from ery in the United States, and overall maquiladora supply chain. This raises the the Midwest was based on just-in-time growth seems likely to continue. However, possibility of slowing, or even reversing, inventory needs, it may be difficult for even those sectors that continue to grow the increase of U.S. border-city suppliers Asian suppliers to take their place. How- in simulations are going to be influenced to the maquiladora industry. ever, given the extent and pace at which by foreign competition. The effect of foreign competition is often couched in Chart 6 terms of a product cycle, in which product Mexican Maquiladora Imports by Country of Origin development and testing occur in the Percent United States, initial long production runs 100 take place in Mexico and ultimately prod- 90 2000 uct commoditization happens in China or 80 2001 another low-wage competitor. The more 2002 quickly and easily a product is commodi- 70 2003 60 2004 tized, the quicker it will move to China. 2005:Q1 Leather, toy and furniture sectors are 50 often cited as no longer competitive in 40 Mexico. But even within the most 30 advanced sectors, we may find individual 20 products susceptible to being lost to lower 10 wage countries in exactly the same way— 0 United States Asia without China China Other computers, cell phones, modems, print- ers and disk drives, for example. Hence, SOURCE: Federal Reserve Bank of Dallas, El Paso Branch, with data from Banco de México.

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 31 ernment has argued that this must be 5 “The Border Industry Program and the Impact of seen as a highly desirable result of suc- Expenditures on a U.S. Border Community,” by Mexico retains cessful economic development and Mex- Harry Ayer and Ross Layton, Annals of Regional Science, vol. 8, 1974, pp. 105–17. ico’s move up the product cycle. The next 6 important generation of maquiladoras should not be “Maquiladoras Along the Texas–Mexico Border: An Econometric Evaluation of Employment and judged by the ability to generate low-wage competitive Retail Sales Effect on Four Texas Border jobs, but by productivity, value added or SMSAs,” by Richard J. Holden, Texas Depart- advantages over rising wages. Critics, at the same time, ment of Community Affairs, Regional Economic claim Mexico simply has not done an ade- Development Division, February 1984. many of its low-wage quate job of preparing the way for more 7 “Project Link: An Investigation of Employment sophisticated manufacturing. To illustrate Linkages Between Cd. Juárez and El Paso,” by rivals, based on this point, many observers cite the failure Richard Sprinkle, University of Texas at El Paso, (so far) of proposed reforms in energy, December 1986. proximity to the 8 labor law, taxes and telecommunications. “Maquila Industry Impacts on the Spatial Redis- United States, These and other reforms are badly needed tribution of Employment,” by Arthur L. Silvers to prepare Mexico for a fine market econ- and Vera K. Pavlakovich, Journal of Borderlands Studies, vol. 9, December 1994, pp. 47–64. political and omy. 9 Finally, it is not just the maquiladora “U.S.–Mexico Integration and Regional Economies: Evidence from Border-City Pairs,” financial stability, industry that is affected by foreign com- by Gordon Hanson, Journal of Urban Econom- and the rule of law. petition, but the U.S.-based supply chain ics, vol. 50, September 2001, pp. 259–87. as well. In 2000, 90 percent of inputs to the 10 For more information, see the papers from the maquiladoras came from the United Dallas Fed conference “Maquiladora Downturn: States, and four years later that number Structural Change or Cyclical Factors?” available at was only 59 percent. Texas border cities in www.dallasfed.org/news/research/2003/ the 1990s developed rapidly as a critical, 03maquiladora.html. In particular, see the pre- Asian suppliers have taken market share, new part of this supply chain, with suppli- sentations by Everardo Elizondo Almaguer, it would be hard to argue that the ers shifting from the Midwest to the Banco de México; William C. Gruben, Federal Reserve Bank of Dallas; James Gerber, San maquiladora market share of Texas-based U.S.–Mexico border. We lack industry Diego State University; and Ernesto Acevedo suppliers has not been reduced. Future detail to know exactly how the recent suc- Fernández, Ministry of Finance and Public expansion of Texas-based suppliers is cess of foreign suppliers is affecting Texas Credit. likely to slow as well. border cities, but again, declining eco- 11 “Maquiladora Downturn: Structural Change or nomic stimulus from maquiladora expan- Cyclical Factors?” by Jesus Cañas, Roberto Conclusion sion would seem to be the rule. Coronado and Robert W. Gilmer, International Mexico’s maquiladora jobs are grow- Business and Economics Research Journal, vol. ing once more, beginning with the Cañas and Coronado are assistant 3, August 2004; “Dollar Appreciation and Manu- resumption of U.S. industrial expansion economists at the El Paso Branch of the facturing Employment and Output,” by William H. Branson and James P. Love, National Bureau in mid-2003. Mexico retains important Federal Reserve Bank of Dallas. Gilmer is a of Economic Research, Working Paper No. competitive advantages over many of its vice president at the Federal Reserve Bank 1972, July 1986; “The Real Exchange Rate and low-wage rivals, based on proximity to the of Dallas. Employment in U.S. Manufacturing: State and United States, political and financial sta- Regional Results,” by William H. Branson and bility, and the rule of law. The maqui- James P. Love, National Bureau of Economic Notes Research, Working Paper No. 2435, November ladora industry is stable, competitive and 1 The four border cities contributed 6.8 percent of growing again. 1987; “The Real Exchange Rate, Employment the 2.3 million jobs generated in Texas from and Output in Manufacturing in the U.S. and It is unlikely, however, to repeat the 1990 to 2000, while making up 6.1 percent of Japan,” by William H. Branson and James P. banner performance of the 1990s, at least Texas employment in 1990. Love, National Bureau of Economic Research, not in the near future. There were ele- 2 Mexican border states include Baja California, Working Paper No. 2491, February 1988. ments of unique, one-time stimulus in the Sonora, Chihuahua, Coahuila and Tamaulipas, 12 See “Maquiladora Downturn: Structural Change 1990s, with the collapse of the peso in excluding Nuevo León. or Cyclical Factors?” by Jesus Cañas, Roberto 1994–95 and the implementation of 3 “The Employment Impact of Maquiladoras Coronado and Robert W. Gilmer, Federal NAFTA in 1994. Further, foreign competi- Along the U.S. Border,” by J. Michael Patrick, in Reserve Bank of Dallas Business Frontier, Issue tion appears to have taken away the The Maquiladora Industry: Economic Solution or 2, 2004. Problem?, ed. Khosrow Fatemi, New York: potential for any growth in several low- Praeger Publishers, 1990, pp. 31–35. wage sectors and probably has reduced 4 “Economic Impact of the Mexican Border Indus- the growth potential of a number of other trialization Program: Agua Prieta, Sonora,” by sectors as well. Jerry R. Ladman and Mark O. Poulsen, Arizona Rising real wages in Mexico have State University, Center for Latin American Stud- accelerated the transfer of low-wage jobs ies, May 1972. to other countries, and the Mexican gov-

32 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 Do Higher Oil Prices Still Benefit Texas? Stephen P. A. Brown and Mine Yücel

Texas and oil. These two words have gone hand in hand since 1889, when the state started producing oil. Since then, the Texas economy has often been driven by volatileT energy prices—suffering with low oil prices and benefiting with high oil prices. The effects of energy prices on the Texas economy were particularly evident during the 1970s and 1980s (Chart 1). As energy prices rose, the Texas economy expanded at a rapid pace, with strong employment and income growth. Although the Texas economy continued to expand until 1986, the oil and gas sector began to slip as energy prices slid from their 1981 heights. The oil price collapse in July 1986 touched off a statewide recession and significant job losses. Since the early 1980s, however, the Texas energy industry has shrunk and other sectors of the Texas economy have grown. Despite these changes, Texas remains the top oil and natural gas producer in the United States and exports most of its production of these two com- modities to other states. Consequently, the energy industry remains an important driver of the state economy. The diversification of the Texas economy away from energy and this sector’s continuing importance to the state prompt us to consider: How much do swings in energy prices affect the Texas economy today? How much has that relationship changed since the energy boom years of the 1970s and 1980s?

Oil Production in Texas: A Brief History1 The first economically significant oil in Texas was discovered in Corsicana in 1894. Discoveries in Navarro County followed. By 1901 the Spindletop oil field was producing 75,000 barrels per day and had contributed to the first Texas oil boom. In the early 1900s, Texas produced relatively little oil and gas—crude oil pro-

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 33 Chart 1 duction was only about 1.3 percent of Texas Employment Tracks Oil Prices in 1970s and 1980s total U.S. production, and natural gas was Employment (in thousands) 2004 dollars per barrel 0.1 percent of U.S. production. By 1952, 800 90 Texas’ shares of total U.S. crude oil and 80 600 natural gas production peaked at 45 and Detrended Texas 70 52.2 percent, respectively. Crude oil and 400 employment 60 natural gas production continued to increase in the state, with the peak for 200 50 both coming in 1972. 40 0 As oil and gas production increased in 30 –200 Texas, so did their importance to the state 20 economy. The creation of OPEC in 1960 –400 Refiners' acquisition cost 10 and subsequent oil price increases in the –600 0 1970s and early 1980s gave rise to a boom ’74 ’77 ’80 ’83 ’86 ’89 ’92 ’95 ’98 ’01 in the Texas economy. Oil and gas output SOURCES: Bureau of Labor Statistics; Federal Reserve Bank of Dallas; Energy Information Administration; authors’ estimates. became an increasing share of Texas output (Chart 2). In 1981, at the height of world oil prices, oil and gas extraction was about 20 percent of total Texas gross state Chart 2 product. Oil and Gas Extraction’s Share of Texas Output Peaks in 1981 After reaching $38 per barrel in 1981, oil prices began softening. Gradually slid- Percent of Texas GSP 20 ing during the next few years, prices finally collapsed to $11.82 per barrel in July 1986. This led to a recession in Texas 15 that lasted 17 months and had a devastat- ing effect on state employment. The number employed in the Texas 10 mining industry (which is mostly oil and gas extraction) rose from about 7,000 in 5 1900—0.7 percent of total state employment—to 90,000 by 1950—a 3.1 percent share. At the oil and gas industry’s peak in 0 1981, Texas employment in oil and gas ’63 ’65 ’67 ’69 ’71 ’73 ’75 ’77 ’79 ’81 ’83 ’85 ’87 ’89 ’91 ’93 ’95 ’97 ’99 ’01 extraction and oilfield machinery reached SOURCE: Federal Reserve Bank of Dallas. 366,200—6 percent of total nonfarm employment in the state (Chart 3). By the time the oil industry bottomed out in 1987, 175,000 jobs had been lost in the oil Chart 3 and gas extraction and oilfield machinery Energy Sector Employment Declines After Early 1980s sectors. Share of Texas nonfarm employment (percent) 5 Refining and Petrochemicals 4.5 Oil and gas extraction After the first Texas refinery opened in Chemicals and allied products 4 Oilfield machinery the Corsicana oil field in 1898, the petro- 3.5 Petroleum and coal products leum refining and petrochemical indus- 3 tries flourished in the state. In 1939 (the 2.5 earliest data available from the U.S. Cen- 2 sus of Manufacturers), the chemical 1.5 industry employed about 6,800 produc- 1 tion workers, and the petroleum refining industry employed 19,000 (accounting for .5 5.5 and 15 percent of total manufacturing 0 ’67 ’69 ’71 ’73 ’75 ’77 ’79 ’81 ’83 ’85 ’87 ’89 ’91 ’93 ’95 ’97 ’99 ’01 employment, respectively). Refining’s share of state output was highest in 1939 SOURCES: Bureau of Labor Statistics; Federal Reserve Bank of Dallas. at 28 percent of total manufactured goods. By 1958, the Texas petroleum refining industry reached its zenith with 43,000

34 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 employees. Chart 4 Texas Economy Diversifies Away from Mining After Mid-1980s Today, the refining industry con- tributes about 11 percent of Texas manu- GSP index, 1971 = 100 facturing output and 1.5 percent of total 600 Trade Total Texas output. Employment has also TCPU Agriculture 500 steadily declined to less than 0.3 percent Services Construction Manufacturing Government of total Texas employment (Chart 3). The 400 FIRE Mining petrochemical industry provides about 12 percent of Texas manufacturing output, 300 1.6 percent of total Texas output and less than 0.9 percent of total Texas employ- 200 ment. 100 The refining and petrochemical industries provide some counterbalance 0 to the effects of changing energy prices on ’71 ’73 ’75 ’77 ’79 ’81 ’83 ’85 ’87 ’89 ’91 ’93 ’95 ’97 ’99 ’01 the Texas economy. These two industries NOTE: TCPU is transportation, communications and public utilities; FIRE is finance, insurance and real estate. generally are hurt by rising oil and natural SOURCES: Bureau of Labor Statistics; Federal Reserve Bank of Dallas. gas prices.

Diversification of the share of the Texas economy today than in determine just how this relationship dif- Texas Economy the early 1980s. fered across the two periods, we analyze As output in the Texas mining indus- To examine in more detail how the the data in two different ways. We exam- try shrank, output in other Texas indus- Texas economy’s diversification away ine how much of the actual fluctuation in tries continued to grow after the mid- from energy-producing industries has Texas output and employment arose from 1980s. Texas saw output gains in manu- affected its response to volatile energy oil price shocks and other causes in each facturing, construction, agriculture and prices, we developed an econometric of the two periods. We also estimate and the service-producing sectors—whole- model that captures the effects of oil price compare by how much Texas output and sale and retail trade; transportation, com- shocks on the Texas economy for the employment would have responded to a munications and public utilities (TCPU); period 1970–2002.2 We find that the rela- 10 percent oil price shock in each of the services; finance, insurance and real tionship between oil prices and the Texas two periods. estate (FIRE); and government (Chart 4). economy is considerably different today We find changes in oil prices Growing at a faster rate than total Texas than it was during the oil boom and bust accounted for a much higher percentage gross state product, manufacturing, trade, years of the 1970s and 1980s. of fluctuations in the Texas economy in TCPU, services and FIRE accounted for Our analysis reveals that the relation- 1970–87 than in 1988–2002. In the earlier increasing shares of Texas output. In con- ship between oil prices and the Texas period, nearly half the fluctuation in Texas trast, agriculture, construction and gov- economy breaks between 1987 and 1988, output (46 percent) arose from changing ernment posted decreasing shares. which indicates that the effects of chang- oil prices. In the latter period, however, A similar picture emerges for Texas ing oil prices on the economy were differ- less than 10 percent of Texas output fluc- employment since the mid-1980s. Ser- ent in 1970–87 than in 1988–2002. To tuations arose from oil price shocks. In vices, construction and trade grew faster than total employment and accounted for Chart 5 increasing shares of Texas nonfarm Texas Employment Shifts Away from Mining After Early 1980s employment (Chart 5). Employment shares for TCPU and FIRE remained rela- Texas Employment Index, 1970 = 100 475 tively constant, while those for manufac- Services Construction turing and government decreased along 425 Trade TCPU FIRE Mining with mining. 375 Total Manufacturing Government Oil and the Texas Economy 325 Even without a rigorous analysis, it’s 275 evident the relationship between energy 225 prices and the Texas economy has 175 changed since the 1980s. Oil and gas production accounted for 19.4 percent of 125 Texas output in 1981 and only 6 percent in 75 ’70 ’72 ’74 ’76 ’78 ’80 ’82 ’84 ’86 ’88 ’90 ’92 ’94 ’96 ’98 ’00 ’02 2002. Similarly, output and employment in energy-related industries, such as oil NOTE: TCPU is transportation, communications and public utilities; FIRE is finance, insurance and real estate. and gas field machinery, claim a smaller SOURCES: Bureau of Labor Statistics; Federal Reserve Bank of Dallas.

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 35 contrast, the fluctuations in U.S. GDP Our estimates confirm gross state product by 2.6 percent and accounted for about 40 percent of the employment by 1 percent. During the fluctuations in Texas output in the latter the Texas economy has 1988–2002 period, a 10 percent increase period. in oil prices would have raised Texas gross become less sensitive state product by 0.4 percent with no sig- The Response to to oil price fluctuations, nificant net effect on employment. Oil Price Shocks We find evidence for two ways in The Texas economy’s response to an but it still responds which the Texas economy has become oil price shock is significantly different in less sensitive to fluctuations in oil prices the two periods (Table 1). For 1970–87, we favorably to higher than it was in the 1970s and 1980s. The estimate that an oil price increase would first is that oilfield activity has become have led to sustained gains in both output energy prices. less sensitive to fluctuations in energy and employment. In particular, a 10 per- prices. The second is that the energy cent increase in oil prices would have led industry makes up a smaller share of the to a 2.6 percent increase in Texas gross in 1988–2002 would have yielded only a Texas economy than it used to. Together state product and about a 1 percent 6.6 percent increase in the rig count. these factors have meant that Texas out- increase in employment.3 An oil price Similarly, oil and gas employment put is about 15 percent as sensitive to oil increase of 10 percent also would have showed a much smaller response in the price fluctuations as it was from 1970 to temporarily boosted the growth rate of second period. We estimate that a 10 per- 1987. Texas nonfarm employment no the Texas economy, with output growing 1 cent increase in oil prices would have gen- longer seems to be affected by oil price percent faster during the next few quar- erated a 9.5 percent increase in Texas oil fluctuations. ters and employment growing 0.1 percent and gas employment for 1970–87 but only faster over the next three to four months, a 1.1 percent employment increase in Brown is a senior economist and assistant then a little slower thereafter. 1988–2002. vice president and Yücel is a senior The economy was much less respon- One reason for the weaker response economist and vice president in the sive to oil prices in the period 1988–2002, in the rig count and employment may be Research Department of the Federal and the nature of the response was differ- changes in technology. After the 1986 Reserve Bank of Dallas. ent. In the second period, a 10 percent crash in oil prices, companies improved increase in oil prices would have led to oilfield technology and produced more oil Notes only about a 0.4 percent gain in gross state with fewer rigs. Therefore, the same rise in This article was previously published under the product. The net response of employment oil prices brings forth fewer rigs and oil- title “The Effect of High Oil Prices on Today’s Texas Economy,” in Federal Reserve Bank of to a rise in oil prices is basically nil. The field workers in the latter period. In addi- Dallas Southwest Economy, September/October negligible result in employment may arise tion, contacts in the industry say there are 2004. from the energy sector’s greatly muted fewer prospects for new drilling in Texas, 1 See “Oil and Gas Industry,” The Handbook of response to oil price fluctuations in the and companies are increasingly shifting Texas Online, www.tsha.utexas.edu/handbook/ 4 latter period and the inability or reluc- their drilling overseas. online. tance of oil companies to hire new 2 We use a vector-autoregressive model with oil employees as energy prices rose. Oil Price Effects on the prices, U.S. GDP, Texas gross state product, To further examine the channels Texas Economy Texas nonfarm employment, Texas employment through which oil price shocks affect the Over the past 20 years, the Texas in oil and gas extraction, and the Texas rig count Texas economy, we examined the effects energy industry has shrunk while other as variables. of oil price shocks on the rig count and oil sectors of the Texas economy have grown. 3 These results are similar to those found in and gas employment in both periods. We Nonetheless, Texas produces more oil and “Energy Prices and State Economic Perfor- found that the rig count responded much gas than any other state in the nation. mance,” by Stephen P. A. Brown and Mine K. Yücel, Federal Reserve Bank of Dallas Economic more strongly to oil price increases in the Texas accounts for 20 percent of crude oil Review, Second Quarter 1995. Using input–out- first period than in the second. For and 26 percent of natural gas production put analysis, Brown and Yücel estimate that a 10 1970–87, we estimate that a 10 percent in the United States (excluding federal off- percent increase in oil prices would have increase in oil prices would have boosted shore). Texas also exports oil and natural boosted Texas employment by 1.37 percent in the rig count by 20 percent. In contrast, gas to the rest of the nation. Conse- 1982 and by 0.3 percent in 2000. the same percentage increase in oil prices quently, higher energy prices still benefit 4 Drilling has shifted toward natural gas in the the state—even if it is by less than in the United States and Texas, but because natural Table 1 boom years of the 1970s and early 1980s. gas prices generally moved with oil prices during the estimation periods, the shift may not Effect of a 10 Percent Increase Our estimates confirm the Texas alter the rig count’s weakening response to oil in Oil Prices on Texas Economy economy has become less sensitive to oil prices. Texas Oil price fluctuations, but it still responds Texas nonfarm Rig and gas favorably to higher energy prices. During GSP employment count employment the 1970–87 period, a 10 percent increase 1970–1987 +2.6% +1.0% +20% +9.5% 1988–2002 +0.4% 0 +6.6% +1.1% in oil prices would have boosted Texas

36 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 The Changing Face of Texas Population Projections and Implications D’Ann Petersen and Laila Assanie

Rich natural resources, abundant land, a central location within the United States and a business-friendly environment have long attracted both immigrants and RU.S. natives to Texas. As a result, the state’s population is faster growing, younger and more diverse than the nation’s. These rapid demographic changes present challenges for the future. As the state’s baby boomer population ages, more demands will be placed on housing, health care and social services. Hispanics, already a dominant force in Texas, are expected to become the majority population group by 2020. The significant increase in this population (both immigrant and native) has far-reaching implications for education, housing and the labor force. The key issue facing Texas will be to reduce the economic and educational disparities prevalent among the state’s ethnic groups as the population continues to grow and evolve. This article looks at population growth and demographic changes of recent decades. Then, with projections from the Texas State Data Center, we examine some sectors of the economy that will be challenged by these demographic forces in the coming decades.

Texas: Big and Getting Bigger Since the early 1900s, Texas has grown faster than the nation. However, during the Texas oil boom, the state’s population growth accelerated. From 1970 to 1980, as oil prices spiraled upward and people flocked to Texas, its population grew by The key issue facing Texas will be to 2.71 percent per year, while the nation’s reduce the economic and educational increased at a 1.14 percent pace (Chart 1). Even during the 1980s, which witnessed disparities prevalent among the state’s an oil and real estate bust, Texas almost doubled the nation’s population growth. ethnic groups as the population During the 1990s, Texas again exceeded expectations and grew by its continues to grow and evolve. largest amount yet, adding almost 3.9 million residents and surpassing New York as the second most populous state. Many immigrants and residents from other

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 37 Chart 1 Table 1 Texas and U.S. Population Growth, Total Population and Components of Population Change in Texas, 1950–2003 1970–2003 Percent change due to Annual growth rate (percent) Total Natural Net Percent Natural Net 3 Texas Population increase increase migration change increase migration United States 2.5 1950 7,711,194 2 1960 9,579,677 1,868,483 1,754,652 113,831 24.23 93.91 6.09 1970 11,196,730 1,617,053 1,402,683 214,370 16.88 86.74 13.26 1.5 1980 14,229,191 3,032,461 1,260,794 1,771,667 27.08 41.58 58.42 1 1990 16,986,510 2,757,319 1,815,670 941,649 19.38 65.85 34.15 .5 2000 20,851,820 3,865,310 1,919,281 1,946,029 22.76 49.65 50.35 2003* 22,103,374 1,259,945 699,685 560,260 6.04 55.53 44.47 0 1970s 1980s 1990s 2003 *Through July 2003. SOURCE: Census Bureau. SOURCE:Census Bureau. states were drawn to Texas’ strong econ- higher-than-average birthrate. This is a larger share of the state’s population omy and rapidly expanding high-tech partly a result of the state’s Hispanic her- growth than natural increase (Table 1). centers, such as Austin and Dallas’ tele- itage and its ties to Mexico, where total Interestingly, even with the state’s reces- com corridor. fertility rates were 2.5 percent in 2004, sion in 2001–03, net migration remained Even with the drastic economic quite a bit higher than the United States’ relatively high, thanks to strong interna- downturn of 2001, which hit Texas much 2.1 percent.1 In 2000, Texas was second in tional immigration, accounting for 44.5 harder than most other areas of the the country (behind Utah) in state rank- percent of Texas’ population increase. nation, the state gained an additional 1.26 ings for birth/fertility rates. Because birth- million residents from 2000 through 2003, rates change slowly over time, Texas will How Has Immigration Changed for a total of 22 million, again growing probably continue to see large natural the Face of Texas? twice as fast as the nation. Although increases in its population despite The healthy pace of Texas’ population domestic in-migration—people moving changes in economic conditions or immi- growth that began in the 1990s is due in to Texas from other states within the gration policies. large part to strong international immi- United States—slowed during Texas’ hard Perhaps the most important factor gration, which surpassed domestic in- economic times, the state’s high birthrate behind Texas’ more recent population migration as a contributor to population and a strong pace of immigration kept growth is the strong pace of net migration. growth in six of the nine years during the population growing at a healthy speed. Historically, people have been drawn to 1990s.2 Immigration reached historic pro- The combination of these factors— Texas because of its abundant land and portions as the number of foreign-born in higher international immigration, a high natural resources. In more recent years, Texas increased by approximately 1.38 Hispanic birthrate and less domestic people and businesses were drawn by million. In addition, immigrants kept migration—resulted in Texas’ Anglo pop- Texas’ robust economy and favorable Texas population growing during the ulation dipping below the majority level business climate. Net migration, which recent economic downturn and tepid of 50 percent in 2003 for the first time includes both domestic in-migration and recovery. From April 2000 to July 2003, since the 1800s. international immigration, was highest Texas net migration totaled 560,260, during periods of greatest economic including 430,048 (77 percent) interna- Why the Rapid Growth? expansion—the 1970s oil boom (58.4 per- tional immigrants (Table 2). Two major factors are spurring Texas’ cent) and the 1990s high-tech/telecom Texas is one of the most popular im- rapid population growth. One is the state’s boom (50.4 percent)—and accounted for migrant gateways to the United States.

Table 2 Total Population and Components of Population Change in United States and Texas

Natural Natural increase Net migration increase Net migration Total population Net Net change Net Net Total international internal Geographic (April 2000– international internal Total change migration migration area July 2003) Births Deaths Total migration migration Total (percent) (percent) (percent) (percent) United States 9,364,374 13,098,788 7,843,040 5,255,748 4,108,626 0 4,108,626 56.12 43.88 43.88 0 Texas 1,259,945 1,189,400 489,715 699,685 430,048 130,212 560,260 55.53 44.47 34.13 10.33

SOURCE:Census Bureau.

38 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 Chart 2 shows the percentage growth of Chart 2 Growth of Foreign-Born, 1990–2000 the foreign-born population in the United States, Texas and the state’s six major met- Percent ros during the 1990s. The foreign-born 180 population share in Texas rose significant- 160 ly during the decade and in 2000 com- 140 posed 14 percent of the population com- 120 pared with 11 percent at the national level. 100 In recent years, growth of the foreign- 80 born has been even more rapid in Texas’ 60 major metros than in its border metros. 40 Between 1990 and 2000, the number of 20 foreign-born in the major metros more 0 than doubled (112 percent increase), U.S. Texas Austin Dallas El Paso Fort Worth- Houston San Antonio Arlington while that of the border metros increased SOURCE: Texas State Data Center. 51.6 percent, well below the state average of 90.2 percent.3 Of Texas’ major metros, only El Paso Chart 3 (31.5 percent) and San Antonio (54.3 per- Share of the Foreign-Born cent) recorded foreign-born growth rates Percent below the U.S. average (57.4 percent), 30 mostly because many of the immigrants 25 in these metros entered the state in earlier 1990 2000 years and their second-generation chil- 20 dren now reside there. Austin witnessed the strongest growth in the foreign-born 15 during the 1990s (172 percent), likely due 10 to the booming tech economy there. The share of the foreign-born in Dallas, Fort 5

Worth and Houston grew by 152 percent, 0 131 percent and 94 percent, respectively. U.S. Texas Austin Dallas El Paso Fort Worth- Houston San Antonio Shares of the foreign-born in the major Arlington metros are shown in Chart 3. SOURCE: Census Bureau. This increase in immigration has brought rapid change in the state’s ethnic Chart 4 composition. Because of Texas’ proximity Change in Ethnicity/Race for Texas to Mexico, many of the state’s immigrants are of Hispanic origin. Hispanics are by far Percent 70 the fastest growing segment of the popu- Anglo lation. During the 1990s, Texas’ Hispanic 60 Black population grew at a pace of 54 percent, 50 Hispanic adding more than 2.3 million people. As a Other result, Hispanics now make up 35 percent 40 of the state’s population, compared with 30 roughly 14 percent at the national level.4 Among states, Texas has the country’s sec- 20 ond-highest Hispanic population, behind 10 only California. Texas’ population has changed in 0 1980 1990 2000 2003 other ways as well. Anglos’ share of the total population has fallen—no longer SOURCES: Census Bureau; Texas State Data Center. above 50 percent—as their rate of growth slowed in the ’90s and the first three years of this decade, while blacks still account In recent years, growth of the foreign-born has for about 11 percent of the state’s population (Chart 4). The number of people been even more rapid in Texas’ major metros included in the “other” category has dou- than in its border metros. bled since the 1990s.5

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 39 Chart 5 its ties to Mexican heritage, will move Projected Proportion of Texas Population by Race/Ethnicity from 50 percent Hispanic (in 2000) to 61.1 Percent percent in 2040. Even Austin, where His- 60 Anglo Black Hispanic Other panics make up only 26 percent of the total today, is expected to see a major 50 increase in its Hispanic population by 40 2040—up to 44 percent. Currently, large disparities mark 30 socioeconomic conditions among Texas’ ethnic groups. Compared with their Anglo 20 counterparts, Texas’ Hispanics tend to 10 have lower levels of education, have lower wages and depend more on state services. 0 This is partly a result of immigration— 2010 2020 2030 2040 Mexican immigrants tend to have average 7 NOTE: Assuming net migration rate is half that of 1990–2000. wages 40 percent below those of natives. SOURCE: Texas State Data Center. These wage differences reflect that the immigrants are young, have scant job experience and speak little English. Chart 6 While some of the difference between U.S. Population by Age in 2003 immigrants’ and natives’ wages is made Age group, in years up after substantial time in the United 85 and over States, disparities between groups 75 to 84 remain. Without changes in socioeco- 65 to 74 nomic conditions, this implies that Texas’ 55 to 64 Baby future population could be less educated, boomers 45 to 54 (39 to 57 less competitive, poorer and more in need 35 to 44 years) of state services such as health care and 25 to 34 welfare. Texas’ challenge is to reduce these 15 to 24 Echo boomers socioeconomic differences through in- (8 to 23 years) 5 to 14 creased educational attainment and Under 5 training, so Texas can compete in the 0 5 10 15 20 25 30 35 40 45 50 nation’s workforce in coming decades. Millions Age. Texas’ overall population, like SOURCE: Census Bureau, American Community Survey 2003. the nation’s, is growing older. This aging is a result of the maturing of the baby boom generation, which makes up the largest The dramatic rise in Texas’ Hispanic major ways over the next several decades: segment of our population. In 2003, the population (both immigrant and native) in diversity and in age. baby boomers spanned the ages 39 to 57 has far-reaching implications. Hispanics’ Diversity. The Texas State Data Center (Chart 6). The youngest of the baby boom- higher-than-average birthrate suggests projects that by 2020, Hispanics will make ers will turn 60 by 2024. As they retire, the that this demographic segment will con- up the majority of Texas’ population, baby boomers will put large demands on tinue to grow at a more rapid pace than while Anglos will fall to the second-most- the Social Security system and other gov- that of Anglos and blacks, even assuming populous ethnicity (Chart 5). By the year ernment programs for the elderly, such as no immigration. In addition, Hispanics, 2040, Hispanics will account for over 50 Medicare. In addition, the boomers may on average, are younger, which has ramifi- percent of all Texans, while one-third of drive housing demand toward move-up cations for housing, education and the the population will be Anglo. Blacks are or second homes as well as houses more labor force. In 2000, the median age of expected to make up 9.5 percent of Texas’ popular with older adults or combined Hispanics in Texas was 25.5 versus 38 for population in 2040, and other races (not families. Texas Anglos. This compares with the Anglo, black or Hispanic) are expected to One factor that may mitigate Texas’ median age for all Texans of 32.3 and for grow to almost 6 percent of the popula- aging population is that the fast-growing the United States of 35.3. Currently, tion.6 Hispanic population has a different age because of its Hispanic heritage, Texas is For Texas’ border cities, which already structure than the Anglo population. As the second youngest state in the nation, have large Hispanic populations, the Chart 7 shows, in 2000 the population in behind Utah. changes could be even more dramatic. age groups over 35 was predominantly For instance, El Paso, 78.2 percent His- Anglo. For example, in 2000, 66 percent of Population Projections panic now, will likely increase to 90.3 per- Texans aged 55–59 were Anglo compared Texas’ population will change in two cent by 2040. Similarly, San Antonio, with with 20 percent that were Hispanic. Con-

40 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 versely, of Texans aged 5 and under, 44 Chart 7 Texas Population by Age and Ethnicity, 2000 percent were of Hispanic heritage, compared with 39 percent Anglo. Percent If expectations of rapid growth hold 80 70 true for Texas’ Hispanic population, His- Anglo panics will make up a much higher per- 60 Hispanic centage of most age groups by the year 50 2040, with only those over 65 being predominantly Anglo (Chart 8).8 The age dif- 40 ferential between the Hispanic and Anglo 30 populations has important implications 20 for education, housing and state services. 10

0 Demographics and Poverty Under 5 5 to 9 10 to 14 15 to 19 20 to 24 25 to 29 30 to 34 35 to 39 40 to 44 45 to 49 50 to 54 55 to 59 60 to 64 65+ Texas Becoming Poorer? Texas’ econ- Age omy grew faster than the nation’s during SOURCE: Texas State Data Center. the 1990s, and all sectors added jobs. Employment in Texas during this period grew at an annualized average rate of 3.3 percent, above the nation’s 2.2 percent. Chart 8 Despite this phenomenal growth in Texas Population by Age and Ethnicity, 2040 employment, Texas has the eighth highest Percent poverty rate in the country and has not yet 80 Anglo Hispanic achieved per capita income parity with 70 the nation. 60 During the 1990s, Texas per capita income grew rapidly—at an annual aver- 50 age rate of 7.2 percent, which exceeded 40 the nation’s 5.7 percent. Consequently, 30 Texas, which began the decade at 89 per- 20 cent of U.S. per capita income, edged up 10 to 95 percent of the U.S. average by 2000. 0 Moreover, poverty rates in the state Under 5 5 to 9 10 to 14 15 to 19 20 to 24 25 to 29 30 to 34 35 to 39 40 to 44 45 to 49 50 to 54 55 to 59 60 to 64 65+ declined—from 18.1 percent in 1989 to Age 15.4 percent in 1999—thanks to a strong NOTE: Assuming net migration rate to the state is equal to that of 1990–2000. economy. SOURCE: Texas State Data Center. Although Texans’ incomes improved during the ’90s, succeeding years have seen a reversal of this phenomenon. Hispanics in Texas were poor.10 Their paying jobs. In addition, because the non- According to 2003 data, the Texas poverty median household income was $29,873, Anglo population in Texas is far younger rate rose to 16.3 percent and Texas nomi- far below the Texas average of $39,927. than the Anglo population, a large per- nal per capita income fell to 93 percent This is an alarming number, given the centage of non-Anglos are in their early ($29,372) of the U.S. average ($31,632) as importance of this segment to Texas’ earning years, have scant work experience the Texas economy slumped into the future. and thus are more likely to have lower recession that started in 2001 and lasted Blacks had the second-highest poverty incomes. until mid-2003. The state’s higher concen- rate (23.4 percent) with a median income tration of high-tech and transportation less than that of Hispanics. Anglos fared Implications industries, which were the hardest hit, best, with the lowest poverty rate (7.8 per- If the income differential between Ang- intensified the recession’s impact. Hence, cent) and the highest median household los and non-Anglos persists, a larger share these industries shed a substantial num- income ($47,162 in 1999) in Texas. of Texans could be drawn into poverty in ber of high-paying jobs, pushing down the The disparity among ethnicities when the future. According to the Texas State state’s per capita income more so than the it comes to income and poverty is not sur- Data Center, the share of households with U.S. average. Also, Texas’ recovery from prising. Natives (predominantly Anglo) annual incomes of $25,000 or less will in- the recession has been unusually weak.9 are far more likely to have a high school crease from 30.7 percent (in 2000) to 37.5 Ethnic Disparities. Among ethnic diploma and some college education than percent by 2040. Moreover, the percentage groups, Hispanics are undoubtedly the immigrants (predominantly Hispanic).11 of families with earnings exceeding $100,000 largest segment in poverty in Texas. In Less-educated individuals tend to be will fall from 11.5 percent to 8.5 percent. 1999, more than 1.6 million (25.4 percent) lower-skilled workers employed in low- The net impact could be a decline in real

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 41 San Antonio) ranked in the bottom third Where the Poor Reside in Texas among major U.S. cities in shares of high school graduates.12 The poor live all over the state, but the living in poverty. Again, the statistics vary by race. For border metros fare worst, with the highest In contrast, poverty levels in the major instance, Anglos in Texas are more likely to be high school graduates (87.2 percent poverty rates (see table). Although poverty metros have rarely been above the state in 2000) than their non-Anglo counter- rates declined in the border metros during average (see table). However, they have been parts, especially Hispanics. In 2000, more the 1990s as the economy boomed, the higher than the U.S. average in some major than half the Hispanic population in Texas share of the population below poverty level metros. For instance, since 1989, both San did not have a high school diploma. Ang- los are also more likely to attain higher remained well above the state average of Antonio and Houston have recorded poverty levels of education than non-Anglos, 15.4 percent in 1999. McAllen, Brownsville rates slightly higher than the U.S. average. In excluding Asians. According to the Pew and Laredo had more than 30 percent of their fact, Houston is home to the highest number Hispanic Center, Hispanics are half as population in poverty, while almost one- of poor Texans (623,493). Dallas traditionally likely as Anglos to graduate from college with a bachelor’s degree by age 26 (23.2 fourth of those living in El Paso were poor. has posted lower poverty rates than the percent for Hispanics versus 47.3 percent The picture for the border metros has nation, but the recent economic downturn for Anglos). Much of the disparity is due not improved much since 1999. According to pushed its rate slightly above the U.S. aver- to rapid Hispanic immigration into the 2003 census data, Hidalgo County (McAllen age. The higher poverty rates in the Texas state: immigrants’ wages and education levels tend to be much lower than MSA), Cameron County (Brownsville MSA) border metros and some major metros may natives’.13 and El Paso County (El Paso MSA) rank be a result of their above-average shares of Hispanics are expected to make up among the top four counties in the United international immigrants. the majority of the labor force in Texas by States with the highest share of individuals 2040. If this disparity between Anglo and non-Anglo high school and college gradu- ation rates continues, the Texas economy Poverty Characteristics of United States, Texas and Its Major and Border Metros could face several important challenges. First, according to the Texas State Individuals below poverty Percent below poverty Data Center, by 2040 approximately 30.1 Place 1989 1999 2003 1989 1999 2003 percent of the labor force will not have a United States 31,742,864 33,899,812 35,846,289 13.1 12.4 12.7 high school diploma, up from 18.8 per- Texas 3,000,515 3,117,609 3,508,230 18.1 15.4 16.3 cent in 2000.14 If that occurs, a higher Austin 129,942 134,589 171,373 15.9 11.1 12.8 share of Texas’ workforce would be less Brownsville 101,362 109,288 130,733 39.7 33.1 36.5 educated and low skilled, possibly making Dallas 322,604 384,146 488,602 12.3 11.1 13.0 El Paso 155,298 158,722 189,596 26.8 23.8 27.4 the Texas economy less competitive. Fort Worth/Arlington 147,177 171,930 193,427 11.0 10.3 10.7 Second, empirical studies show that Houston 494,457 572,410 623,493 15.1 13.9 14.1 low education levels are associated with Laredo 50,116 59,339 n.a. 38.2 31.2 n.a. lower income levels; therefore, failure to McAllen 159,216 201,865 238,333 41.9 35.9 38.0 complete high school or college nega- San Antonio 252,301 234,478 266,248 19.5 15.1 16.2 tively impacts average earnings.15 Earn- ings data from the Census Bureau demon- NOTE: 1999 poverty data are the latest available for Laredo MSA. strate this point (Chart 9). An increasing SOURCES: Census Bureau; Texas State Data Center. number of less-educated laborers would reduce the average income of Texans and in turn decrease tax revenues collected by income, reduced tax revenue per house- invested in Texas’ higher education sys- the state. hold and increased burden on the state tem returns $5 or more to the Texas econ- Third, overall enrollment in public government to pay for welfare services in omy. Hence, it is essential that the educa- schools is estimated to climb rapidly, Texas. As the state is likely to depend pro- tion system keep up with the state’s growing at about half the state’s popula- gressively more on non-Anglo Texans for changing demographics. tion growth rate, according to the State future tax revenues, it is important to Texas’ education record is nothing to Data Center. Most of this increase in stu- lessen the existing wage gap and educa- brag about. Texas ranks second to last dent enrollment—Hispanics by almost tion differential between ethnic groups. among the 50 states in its share of the 100 percent and the “other” category by Education and the Labor Force. One population 25 years or older with a high 71 percent—is expected to result from way to reduce the wage gap is through school diploma (only 77.8 percent). Fur- growth in the non-Anglo population education and training. In fact, according thermore, in 2003 several Texas cities because of its younger age structure. to the Texas comptroller, every dollar (Dallas, El Paso, Fort Worth, Houston and Thus, state expenditure on public

42 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 Chart 9 U.S. Mean Earnings in 2002 by Educational Attainment

$120,000

$100,000

$80,000

$60,000

$40,000

$20,000

0 No high Graduate, Some Associate Bachelor's Master’s Doctorate Professional Home ownership school diploma including GED college degree degree degree degree degree is expected to SOURCE: Census Bureau, Current Population Survey, 2003. increase dramatically education as well as the number of stu- forces affecting the housing market and for minority dents requiring financial assistance could home ownership in the coming decades. expand rapidly unless socioeconomic dif- It remains to be seen what boomers’ pref- and foreign-born ferences between races are reduced. Ris- erences will be—whether they remain in ing education costs coupled with slow their current homes, trade up or purchase households in growth in tax revenues would adversely vacation homes. Most boomers are enter- impact the state’s financial situation. ing the stage of life when earnings peak— the coming However, it is naive to assume that thus, they may choose more affluent decades, especially the current income differential between homes or ones featuring amenities more Anglos and non-Anglos will persist popular with empty nesters. in areas that unchecked. Empirical research shows that The demographic shift of the baby second and third generations of immi- boom generation leaves fewer households have experienced grants are more likely than their forefa- headed by those in the starter home mar- thers to have access to higher level educa- ket, ages 25 to 34, which could mean a high levels of tion and, therefore, are better equipped slowdown in starter home construction. immigration, with skills required for higher paying jobs. However, immigrants and minorities, who Hence, the wage gap between non-Anglos have had historically lower home-owner- like Texas. and Anglos is likely to be reduced in the ship rates than Anglos, will likely take up future.16 some of the slack. Home ownership is For the Texas economy to remain expected to increase dramatically for robust, it is essential that the state’s edu- minority and foreign-born households in cation system make progress on at least the coming decades, especially in areas two fronts: (1) investing in resources to that have experienced high levels of improve overall student achievement, immigration, like Texas. Because Texas’ and (2) developing programs that help Hispanic population is younger and faster bridge the educational attainment gap growing than the overall population, between racial and ethnic groups. many Hispanic-headed households will Housing. What does the future hold move into the prime home-buying age for the housing industry as Texas’ popula- groups in the coming decades, which tion changes over the next several could give Texas homebuilders a boost. decades? The aging of the overall popula- This has important implications for tion, along with the baby boomers, will the apartment market in the short run as certainly impact the housing industry in well, with Hispanics currently more likely Texas as well as every other state. The to rent than own. According to census youngest baby boomers turn 40 this year, data, in 2002 the U.S. home-ownership and boomers are turning 50 at the rate of rate for Hispanics was 48.2 percent versus seven every minute and will continue to 71 percent for Anglos. Thus, Hispanics do so through 2013 (see Chart 6). This seg- have the potential to become a much ment of the population, along with aging larger segment of the home-buying mar- seniors, will be among the most potent ket.

OCTOBER 2005 | FEDERAL RESERVE BANK OF DALLAS 43 Chart 10 tion and will likely make up the majority 9 Also in 2003, Texas’ median household income Between 2000 and 2040, Texas’ Health Care by the year 2020. Disparities in income ($40,674) was below the national average of Costs Could Grow Faster Than Its Population and education between Hispanics and $43,564, putting Texas 32nd in terms of median household income among the states. Percent other ethnic groups may be a challenge to 10 300 Texas and its resources. The state could The Census Bureau uses a threshold updated every year for inflation to determine the poverty reduce such socioeconomic differences 250 level. If an individual’s or family’s income before through increased educational attain- 200 taxes and excluding capital gains or losses falls ment and training so that in coming below the applicable threshold, the individual or 150 decades, the state’s workforce will con- family is considered poor. See the Census 100 tinue to be one of the most competitive in Bureau’s web site (www.census.gov) for poverty the nation. threshold schedule. 50 11 See Orrenius and Viard, 2000. 0 Petersen is an associate economist and 12 Population Physician Days of Nursing home American Community Survey 2003, Census contacts hospitalization residents Assanie is an assistant economist in the Bureau. Research Department of the Federal 13 See “Immigrant Assimilation: Is the U.S. Still a NOTE: Assuming net migration rate to the state is equal to that of Reserve Bank of Dallas. Melting Pot?” by Pia Orrenius, Federal Reserve 1990–2000. Bank of Dallas Southwest Economy, May/June SOURCE: Texas State Data Center. Notes 2004. The data used in this article come from 14 Projection provided by Murdock et al., 2002, two main sources, the Census Bureau and assuming net migration rate to the state is equal Texas’ housing market stands to ben- the Texas State Data Center. The two sources to that of 1990–2000. differ somewhat in terminology regarding efit from its rapidly growing and diverse 15 “Educational Attainment and Border Income race/ethnicity. Thus, in an attempt to keep population and its strong pace of interna- Performance,” by Thomas Fullerton, Federal the information consistent within the article, Reserve Bank of Dallas Economic and Financial tional migration. Real estate firms of the the authors use the terminology provided by Review, Third Quarter 2001. future will be wise to market to both the the Texas State Data Center. For more informa- increasingly older Anglo population and tion regarding the definitions of race/ethni- 16 See Orrenius, 2004. the younger Hispanic population. Addi- city, see http://txsdc.utsa.edu/txdata/redistrict/ tionally, while domestic migration re-report.php and http://www.census.gov/ dropped off during the recent economic population/www/socdemo/compraceho.html. downturn, a pickup in that segment of the 1 Census Bureau, International Database. For a population would benefit Texas housing. definition of total fertility rates, see www. census.gov/ipc/prod/wp02/appE.pdf. Health Care. The aging of the Texas 2 population plus a rapidly growing popula- See “The Second Great Migration: Economic and Policy Implications,” by Pia Orrenius and tion segment with different socioeco- Alan Viard, Federal Reserve Bank of Dallas nomic characteristics than the previous Southwest Economy, May/June, 2000. Anglo majority will dramatically affect the 3 Major metros exclude El Paso. The number for health care industry in Texas. The number El Paso has been included with the other border of instances of diseases and disorders is metros. expected to increase in Texas. Trips to the 4 American Community Survey 2003, Census doctor, days in the hospital and the num- Bureau, www.census.gov/acs/www/index.html. ber of people in nursing care facilities are 5 The term Anglos refers to non-Hispanic whites all expected to rise at rates faster than the only. The term blacks refers to non-Hispanic population growth rate (Chart 10). The blacks of African as well as non-African origin. health care industry is currently one of the The “Other” category includes all people who are fastest growing sectors of the Texas econ- not Anglos, not Hispanics and not blacks. Native omy and will likely remain so as the need Americans, Asians and multiracial people are grouped in this category. increases for long-term care facilities and 6 doctors who treat the elderly and a more All projections provided by “The Texas Challenge in the Twenty-First Century: Implications of Pop- diverse population. ulation Change for the Future of Texas,” by Steve Murdock et al., The Center for Demographic and Outlook Socioeconomic Research and Education, During the 1990s, Texas grew even December 2002. Projections used in this article faster than expected, becoming the sec- assume population growth due to net migration ond-largest state in the nation. Along with is half that of 1990–2000 unless specified oth- this growth, the population has become erwise. See www.txsdc.utsa.edu. older and increasingly diverse, and today 7 See Orrenius and Viard, 2000. it is no longer dominated by an Anglo 8 Projections are based on the assumption that majority. Hispanics account for the the net migration rate to the state is equal to that fastest growing segment of Texas’ popula- of 1990–2000.

44 FEDERAL RESERVE BANK OF DALLAS | OCTOBER 2005 The Face of Texas: Jobs, People, Business, Change is published by the Federal Reserve Bank of Dallas. The views expressed are those of the authors and should not be attributed to the Federal Reserve Bank of Dallas or the Federal Reserve System.

Articles may be reprinted on the condition that the source is credited and a copy is provided to the Research Department, Federal Reserve Bank of Dallas, P.O. Box 655906, Dallas, TX 75265-5906. This publication is available on the Internet at www.dallasfed.org.

Richard W. Fisher President and Chief Executive Officer

Harvey Rosenblum Executive Vice President and Director of Research

Managing Editor Mine Yücel, Vice President Research Department

Editors Jennifer Afflerbach Kay Champagne Monica Reeves

Design and Production Gene Autry Ellah Piña FEDERAL RESERVE BANK OF DALLAS P.O. BOX 655906 DALLAS, TX 75265–5906