For the past 35 years, large American cities have decentralized, losing both population and employment, while their surrounding metropolitan areas have gained people and jobs. The country has become increasingly urban in character while, at the same time, the proportion of people and jobs in large cities has declined. For example, the 94 largest central cities in the United States in 1990 contained about 20 percent of the total U.S. population. In 1960 these 94 cities comprised more than 26 percent of the total U.S. population. More than 16 percent of the U.S. population lived in the 25 largest cities in 1960, while only about 12 percent lived in these same cities in 1990. From 1960 to 1990, the United States experienced a 65-percent growth in jobs, while the top 25 cities saw only a 14-percent growth in employment.

Although the trend is clear, fluctuations around it are often both substantial and difficult to explain. Some cities have flourished, doing much better than the trend, while others have grappled with shrinking population and employment and increasing crime, poverty, and racial polarization. At one time a city may be doing well compared with other large cities. At a different time, that same city may be losing population and employment at a rate that is faster than the trend for large cities. Where and why have some cities performed well and others languished? This article will attempt to answer this question by examining the performance of various cities in different regions of the United States.

The article draws upon a new data set prepared for the U.S. Department of Housing and Urban Development (HUD) by the Center for Urban Policy Research (CUPR). CUPR amassed approximately 1,400 variables to depict the economic and social conditions in America's cities in 1980 and 1990, documenting fluctuations around the long-run urban trend. The comprehensive array of variables is divided into six categories: employment and economic development; demographic factors; housing and land use; poverty and income distribution; fiscal conditions and the public sector; and social, environmental, health, and other factors. The exhaustive list of variables helps to measure cities' performance along any number of dimensions, such as health, poverty, crime, unemployment, education, and racial integration.

One innovative feature of the data set is that for most variables it provides measures for central cities and for either their surrounding suburbs or the entire metropolitan area. This feature will help researchers at HUD, CUPR, and elsewhere to compare the performance of central cities with that of their surrounding areas. Important insights can be gleaned from such comparisons. For example, one can begin to infer reasons for the disparities between a city and its suburbs by comparing a suburb that is doing well with a neighboring central city that is deteriorating. A second innovation of the database is an index of dissimilarity -- a barometer of racial segregation -- that measures the proportion of a metropolitan area's residents that must move to achieve perfect racial integration.

This article is divided into five sections. The first section describes cities within their regional context. The second section focuses on decentralization patterns, including the "Edge City." The third, fourth, and fifth sections discuss urban trends, the distributional effects of these trends among cities, and the recent performance of cities in general.

Cities Within Their Regional Context

From 1980 to 1990, the Northeast and Midwest regions -- the Frostbelt -- were population and employment losers relative to the rest of the country. Midwestern cities lost 4.2 percent of their population during the decade, while the population in Northeastern cities grew by only 0.4 percent. In both regions suburban employment growth was more than 50 percent greater than in the cities.

Cities in the Northeast and Midwest also suffered severe population and employment losses. Pittsburgh lost 12.8 percent of its population and 9.7 percent of central-city jobs. Detroit's population and central-city employment shrunk by 14.6 percent and 15 percent, respectively. Other Frostbelt cities, such as Buffalo; Newark, New Jersey; Chicago; Cincinnati; Kansas City, Missouri; and St. Louis, suffered significant losses in population and/or central-city employment.

On the other hand, the South and West regions -- the Sunbelt --performed very well. The population of the average city in the South and West grew by 6.1 percent and 18.6 percent, respectively. Average city employment rose by even more: 9.3 percent for the South and 58.1 percent for the West.

The Sunbelt cities also fared far better than their frozen counterparts. Five Texas cities saw their population and central-city employment increase by 10 percent or more: Austin (34.7-percent increase in population, 39.9-percent increase in central-city employment); Dallas (11.4 percent, 10.0 percent); El Paso (21.2 percent, 26.8 percent); Fort Worth (16.2 percent, 15.2 percent); and San Antonio (19.1 percent, 27.4 percent). Other cities with double-digit growth in both categories include: Charlotte, Jacksonville, Virginia Beach, Anchorage, Boise, Fresno, Sacramento, San Diego, Las Vegas, Phoenix, and Tucson.

Regional characteristics, such as cheap labor in the South, inexpensive energy in the Southwest, and desirable climates in both the South and West, have contributed to the recent prosperity of cities in these areas.

Certain cities have flourished in declining regions, just as others have declined in prosperous regions. Despite being in the Northeast, Boston's population and central-city employment grew by 2 percent and 12.8 percent, respectively, from 1980 to 1990. New Orleans lost nearly 11 percent of its population and 15 percent of its central-city jobs, notwithstanding its southern location. Boston was certainly helped by the development of high-technology industries in and around the city, particularly along Route 128. New Orleans, on the other hand, never recovered from the oil-based recession of the 1970s. It failed to develop a replacement engine to drive economic growth and encourage people to remain in or relocate to the area.

Patterns Of Urban Decentralization

Prior to World War II, and for nearly three decades thereafter, U.S. industry enjoyed an insular domestic market and dominance abroad. However, foreign industrial competition grew increasingly stiff during the 1970s. The automobile industry is a perfect illustration. Abetted by the OPEC oil shocks, foreign automakers were able to sell their cheaper, more reliable, and more fuel-efficient vehicles to American consumers. By the 1990s, one of every four vehicles in the United States was produced abroad. Foreign manufacturers made similar inroads in the steel, textiles, and consumer electronics industries. To meet the decreased demand for domestic products, American firms downsized their domestic operations, displacing hundreds of thousands of workers.

Many of the losses in American manufacturing are attributable to the cost advantage of foreign producers by the ready supply of cheaper, nonunionized, low- or semiskilled labor. In their efforts to be more cost competitive, domestic producers have begun to build plants or subcontract manufacturing overseas, exacerbating the loss of less skilled jobs in the United States, particularly in its cities.

Manufacturing jobs have decentralized, moving out of central cities. Services, which by their nature cannot be imported, and high-skilled professional jobs, in which the United States has a comparative advantage, have moved in. Services and high-skilled, high-technology jobs have also grown in the suburbs, far removed from the cities that experienced the greatest manufacturing job losses.

As jobs and population left the central city, retailers followed, filling shopping malls and creating "Edge Cities" -- highly developed retailing centers located far from traditional downtown areas. With merchandise and other urban amenities now closer to where they live, suburban shoppers abandoned many centrally located stores. Inevitably, urban retailers either curtailed their operations or went out of business, further depleting the withering urban job base.

Reasons For Observed Structural Changes

Residential suburbanization has been an ongoing process in the United States since World War II. Indeed, there is a "natural evolution theory" of suburbanization, which postulates: As incomes grow, people will be willing to spend more on spacious homes, larger yards, and the concomitant higher commuting costs. So, over time, one would expect to observe larger numbers of people residing in suburbia, independent of other socioeconomic phenomena. There is presently a debate in the literature as to whether urban problems (such as crime and poverty) have accelerated the rate of suburbanization, but the evidence is far from conclusive. The only certainty is that the United States is becoming increasingly suburban and has been moving in that direction for decades.

Manufacturing and service jobs have also been moving to the suburbs over the past 25 years. As discussed above, a major reason is the availability of skilled labor: As well-trained workers increasingly reside in the suburbs, businesses have located their operations there. In addition, suburban land is often cheaper, regulations are less strict, and transportation is less costly and time consuming. Another advantage is that firms that locate in the suburbs risk no potential environmental liability left by a previous polluter.

Global factor-price equalization is the cause of many of the job losses in Northern and Midwestern cities. In other words, American manufacturers cannot pay real wages substantially greater than those being paid by foreign firms and remain price competitive. One wage-reducing response to foreign competition is to relocate production to lower wage/less unionized areas. Domestically, this has been accomplished by the movement of manufacturing facilities, particularly to the sparsely unionized South. Internationally, producers have moved their production to low-wage countries. Both Ford and General Motors, for example, now have plants in Mexico.

Labor-saving technological change is yet another way that firms can reduce labor demand and labor costs. Both increases in worker productivity and automation -- the outright replacement of human workers with machines -- have been successfully introduced. For its most routine tasks, such as spot-welding chassis assemblies, the auto industry now uses robots. In banking the automated teller machine has assumed the most routine tasks that bank tellers used to perform. In short, low- and semiskilled labor is increasingly being replaced by lower cost machinery in both manufacturing and services.

Distributional Effects Of Structural Changes

Foreign competition and labor-saving technological change have combined to slow the demand for labor and with it, the growth in real wages. Real average hourly earnings exhibited no significant growth from 1966 to 1994. In fact, measured in constant 1982 dollars, the $7.52 average hourly wage of 1966 is slightly higher than its $7.42 counterpart of 1995.² For urban workers who depend on hourly wage jobs for a living, they have not increased their earning power. Measured in 1994 dollars, real median family income, which now incorporates more two-earner families, has shown little increase, growing only slightly from $37,319 in 1976 to $38,782 in 1994.³

These figures conceal a stunning redistribution of income between skill categories and income classes. As unskilled labor saw its earnings decline, skilled workers reaped large increases. From 1975 to 1992, nominal average earnings doubled for high school dropouts, rose 2.5 times for high school graduates, nearly tripled for holders of bachelor's degrees, and tripled for holders of advanced degrees. During the same period, the consumer price index increased 2.5 times, meaning only those with college degrees and beyond increased their real earnings. In 1992 the holder of a bachelor's degree earned an average annual wage of $32,269, 72 percent more than a high school graduate. Over a worklife the annual difference translates to $600,000 more for a college graduate (for example, $1.421 million vs. $821,000).

According to income categories, between 1979 and 1989, the average real incomes of families in the lowest quintile fell by 2.1 percent, while the incomes of families in the top quintile rose by 13.9 percent.

The redistribution of income away from those with less education and wealth has exacerbated urban poverty. American cities harbor a disproportionate share of the Nation's poor and poorly educated citizens. Therefore, when earnings of the poor and undereducated slip, cities bear the brunt. A comparison of the proportions of city and suburban populations in poverty confirms this fact. In 1990 in the Northeast, 19.9 percent of the average city's population lived in poverty. The average city in the Midwest, South, and West had 20.8 percent, 19.7 percent, and 15.6 percent, respectively, of its population living in poverty. The corresponding numbers for each region's suburbs were: 6.8 percent for the Northeast, 6.3 percent for the Midwest, 9.3 percent for the South, and 9.3 percent for the West.

While well-paying manufacturing jobs have disappeared from cities, affordable housing remains concentrated in older urban cores. The location of affordable housing away from centers of suburban job growth has trapped the poor and minorities in central cities, removed from economic opportunity. These trends have also fostered income and racial polarization: Middle and upper income Americans work and reside in suburbia, while a predominantly African-American and Hispanic underclass crowds the inner city. According to the 1990 Index of Dissimilarity calculated in HUD's new data set, for the average large metropolis of the Northeast and the Midwest, two-thirds of the population would have to be moved to achieve perfect racial integration, 73.4 percent and 69 percent, respectively. For the South 64.4 percent and for the West 50.3 percent of the population would have to move to achieve racial integration. More importantly, for the Northeast and the Midwest, the percentages for 1990 exceeded their 1970 counterparts, confirming an increase in racial polarization.

Summary

Large cities as a group continue a longrun trend of relative population and employment loss. As population and employment centers, large cities are less dominant than they were in the 1950s and 1960s. While decentralization is the overall trend, the performance of individual cities fluctuates around the trend line. The CUPR data set highlights the movements for individual large cities. As the data suggest, poverty, unemployment, and isolation are huge problems for many American cities, particularly those in the Northeast and Midwest. At the same time, some cities are much healthier, both economically and socially; the majority of these cities are in the South and West.