Understanding US productivity trends from the bottom-up

As the U.S. economy continues to chug along towards full employment, the nation’s slowing growth in labor productivity – the amount of goods and services a worker produces per hour of work – is a lingering economic concern, as it remains a critical source of prosperity.  Since 2004, productivity has been expanding at its slowest clip in the post-war era. While there is broad acknowledgement that the slowdown is occurring, its causes—from declining breakthroughs in major technologies, to inefficiencies in major sectors like education, healthcare, and housing, to simply a mismeasurement issue—remain the subject of much debate.

What is missing from this debate, however, is a serious exploration of productivity growth at the local level.

Why might examining productivity at the regional level matter? To start, the nation’s 382 metropolitan areas provide a potentially useful source of spatial variation to help understand productivity trends in a very large, economically diverse country. Economic regions, after all, differ significantly in key determinants of productivity: natural resources, industrial structures, the quality of local workforces, infrastructure, and technological capabilities. Firms tend to cluster together to draw on these shared resources, many of which are actually shaped by investments and policies at the sub-national level. Of course, national policies around competition policy, taxes, trade and research and development matter greatly, as do firm-level factors like size, management processes, and the use of capital and technology matter.  Even so, what happens locally remains influential.

Yet, while its drivers may be partly locally-controlled, productivity is typically overshadowed as a factor for local economic success given the emphasis typically afforded to traditional economic development metrics like job creation or new capital investment. This is a mistake, since rising productivity is a prerequisite for long-term real wage growth and increased living standards. Improving workers’ productivity increases their value in the labor market. It is the main mechanism by which they are able to command higher wages and improve their well-being. And, while in a world of heightened income inequality, the link between productivity and wages has weakened, it is generally acknowledged that it will be very hard for incomes to rise without increases in productivity.

Lack of measurement also stems from the challenges of measurement. The only statistic available at the regional scale—the amount of economic output per worker—is admittedly crude.  In light of that, this short report provides metro-level productivity statistics by generating metro level output figures and relevant metro employment data.  Metro-level economic output is estimated by summing county-level GDP data from Moody’s Analytics (which draws from the Bureau of Economic Analysis GDP data). Employment data also comes from Moody’s and draws on two U.S. Bureau of Labor Statistics series: the Current Employment Statistics (CES) and the Quarterly Census of Employment and Wages (QCEW).

Using these data, we observe massive variation across the U.S. economy. We estimate that U.S. labor productivity averaged $113,000 per worker in 2015. Yet, among the nation’s 382 metropolitan areas, that figure ranged from $299,000 per worker in Midland, TX to $38,000 per worker in Jacksonville, NC.

Overall, the nation’s largest cities and regions tend to be the nation’s most productive areas.

Final Figure 1

The largest 100 metro areas have higher average labor productivity ($119,000 per worker) than smaller metro areas ($99,000 per worker) and the country as a whole. This is in line with a large literature showing that large cities achieve higher productivity through the learning, knowledge sharing, and specialization benefits of agglomeration. Therefore, the nation’s most productive industries outside of natural resources cluster in its largest cities.

Final Figure 2

Yet, even within the nation’s 100 largest metropolitan areas there are striking differences in labor productivity. The 10 most productive metros, on average, are 66 percent more productive than the bottom 10 metros (Table 1). At the extreme ends, San Jose, CA, the nation’s most productive large metro area, is about twice as productive as McAllen, TX, its least.


Map 1 reveals some clear regional patterns. Large metro areas with the highest productivity levels tend to be located on the coasts in California (San Jose, San Francisco, Los Angeles, San Diego, and Oxnard) and the Northeast (Bridgeport, CT; Hartford, CT; and New York). Houston and Seattle round out the top 10. At the low end of the productivity spectrum are smaller metro areas in the South (McAllen, TX; Greenville, SC; Jackson, MS; Charleston, SC; Augusta, GA; Columbia, SC; and Deltona-Daytona Beach, FL) and West (Provo, UT; Boise City, ID; and Wichita, KS).

Final Map 1

Many small metro economies are highly productive as well, especially those that specialize in oil, gas, and mining. The energy belt—spanning from Louisiana from Oklahoma—concentrates very high productivity metro areas. The least productive small metros are much more widely distributed.

As noted above, the significant differences in productivity levels between U.S. regions derives from many factors. But it is important to note that the very existence of such wide interregional gaps in labor productivity runs counter to prevailing economic theory, which predicts that regions within the same country should converge to similar productivity levels over time. This “catching up” phenomenon—known as convergence—occurs as lower productivity regions import or copy the skills, business processes, and new technologies generated in the high productivity regions.

Gaps clearly remain, but there is evidence of at least partial convergence over the past several decades. To examine this trend, Figure 3 divides the nation’s 382 metro areas into ten equal categories based on their 1978 productivity levels. Even as the nation’s productivity growth varied during the 1980s, 1990s and first half of the 2000s, the trend towards metropolitan convergence remained consistent: lower productivity metro areas tended to see their productivity grow faster than their more productive counterparts.

Final Figure 3

The pattern then changes between 2004 and 2015. Most metro economies have been unable to escape the nationwide productivity slowdown—a trend that parallels other recent work, such as by Elisa Giannone, that has shown a slowing over time of regional convergence. Consider the two maps below. Between 1978 and 2004, just under half of U.S. metro areas, representing a diversity of sizes and geographic areas, expanded productivity at a one percent per year clip. While one percent per year is by no means a blistering pace, the geographic spread second map shows just how few metro areas were able to attain that same clip in the 2004-2015 period.

Final Map 2

Unless a metro economy participated in the energy boom (see metros in North Dakota, Oklahoma, and Texas) or the tech boom (see Austin, TX; Pittsburgh, PA; Portland, OR; San Jose; CA; and Seattle, WA), it was very hard for it to grow at more than one percent per year. Beyond the impact of these two sectors, many of the small metro areas that were able to grow at this clip housed major research universities (Ames, IA; Blacksburg, VA; Durham-Chapel Hill, NC; Madison, WI; and State College, PA).

A second reality of the 2004-2015 era is stalled interregional convergence in productivity levels. In fact, the most productive decile of metro areas actually increased productivity at the fastest clip during the 2004-2015 period. This trend partly stems from the energy boom that occurred during this period, as many metro areas in the top productivity decile specialize in commodities, but the pattern holds even when the mining, oil, and gas sectors are removed. The strong have been getting stronger.

Convergence between high productivity and low productivity regions is not an end to itself, but stagnant productivity growth and entrenched gaps in productivity levels between metro areas signals broader challenges. For starters, productivity is associated with three key metrics of rising living standards. Metro areas with higher labor productivity tend to also have higher incomes—as measured by household median income and average real personal income per capita (the latter of which controls for regional differences in cost of living)—and lower poverty rates, and this relationship is statistically significant across nearly every period going back to 1980 (Figure 4), even when controlling for a region’s education levels.

Final Figure 4

Productivity growth alone may not be enough to end poverty or enduringly raise middle class incomes. But its relationship to both suggest that it is an important mechanism to improve these challenges in our nation’s cities and regions, and thus the country as a whole. It is worrying that only small pockets of the country—tech hubs, energy boomtowns, or college towns—have registered productivity gains in the recent decade.

So what do these findings suggest for policy?  For starters, policies should address artificial barriers that prevent U.S. workers from accessing high-productivity regions experiencing rapid growth. As Peter Ganong and Daniel Shoag have shown, high housing costs in these regions, exacerbated by supply restrictions, has prevented Americans from accessing them. This dynamic partly explains declining intra-state migration and declining regional income convergence. Zoning and land use reforms that encourage more housing development in these fast-growing regions is necessary so that more workers can affordably access the economic opportunities they generate. Allowing more people to access dynamic regions would boost growth for the entire country according to a recent study by Chang-Tai Hsieh and Enrico Moretti, which estimates that housing regulations have lowered the overall size of the U.S. economy by about 14 percent because many productive metro areas are smaller than their economic dynamism would predict.

However, not everyone can move, so boosting the productivity of a broader swath of American regions is also paramount. Some responses are obvious but admittedly difficult.

Most notably, the nation and especially its states and cities need to address the stark education and especially technical skills disparities that are now contributing to what Enrico Moretti calls the “great divergence.”

To the extent that Giannone is correct — that “skill-biased” technology change now especially favors technically educated workers — skills lags are now a major driver of productivity gaps that regions and the nation must respond to. To the extent that digital technologies are increasingly central to productivity gains among firms, industries, and places (as forthcoming Metro Program research will show), imbuing local education and training efforts with strong digital content may maximize the gains from investment.

Similarly, the nation would do well to maintain and expand its innovation and technology development initiatives, with a renewed focus on distributed delivery formats.

On this front, the nation should continue its recent embrace of both new and familiar forms of regionally distributed innovation and translation investment.  For example, just as federal R&D flows to universities have long provided a crucial support of regional “catch up,” so do newer experiments in the creation of networks of far-flung local energy hubs and manufacturing innovation institutes. Likewise, recent efforts to better leverage the nation’s energy laboratories for local economic development represents another approach to catalyzing local innovation and productivity growth.

Finally, states and the nation need to do more to facilitate the diffusion of productivity from leading firms, industries, and regions to their lagging counterparts.

The concentration of productivity growth in particular regions stems from the fact that global integration and technological advances have created winner-take-all dynamics in many industries. In this fashion, those “frontier firms”—and by extension their regions—that possess the technological capabilities, sophistication of management, and global networks to navigate the complexities of the modern economy are pulling away from the pack. New evidence from the OECD, moreover, finds that the technological capabilities of firms at the frontier are not diffusing across the rest of the economy.

So, what should be done about this?  Restarting the diffusion engine requires a concerted effort on the part of government and industry to invest in technologies that help upgrade legacy firms and industries. An expansion of federal programs like the Manufacturing Extension Partnership to other sectors of the economy could help small and medium-sized businesses upgrade their management and technology processes. Regulators should foster competition within industries to ensure that incumbents are not stifling productivity growth by maintaining profitability through rents. And here again, supporting the diffusion of digital technology and the managerial and worker capabilities to deploy it in secondary and tertiary regions remains imperative as well.

In sum, addressing shortfalls of regional productivity needs to become a priority.  The 2016 election laid bare the political ramifications of inter-regional economic disparities. Going forward, expanding the productive capabilities of all U.S. communities will be critical to ensuring shared prosperity nationally.

View an appendix on labor productivity growth across 382 metropolitan statistical areas >>

The authors would like to thank Zhongyi Tang for her analysis and contributions to this post.