by Connor O’Brien
Key Takeaways:
- Economic dynamism has been in decline across the U.S. economy for decades, but new data suggests the High Tech sector, once viewed as an engine of innovation, is no longer as dynamic as it once was.
- The High Tech startup rate, once far above the rest of the economy, has almost fully converged with other sectors.
- Young firms’ share of High Tech employment collapsed with the bursting of the dot-com bubble in the early 2000s, along with the analogous share in the non-High Tech sector. These trends only worsened in the long, sclerotic recovery from the Great Recession.
- Indicators of dynamism in the High Tech sector are increasingly concentrated geographically in a few places.
- Recent studies on the employment dynamics of U.S. inventors demonstrate that labor markets in key industries for productivity growth are also becoming more sclerotic and inefficient.
Introduction
The decline of economic dynamism has been one of the defining characteristics of the “Great Stagnation,” a period of decelerating economic and productivity growth going back to the 1970s. As EIG has previously highlighted, decreasing economic dynamism has played an integral part in the slowdown of innovation across the U.S. economy, the inability of workers to move in pursuit of opportunity or wage increases, and the shift towards markets more dominated by larger, older incumbent firms. Long thought to be an exception to broader trends making the rest of the economy more dominated by entrenched interests and less innovative, the High Tech sector, too, appears to be succumbing to these same forces, according to a new Business Dynamics Statistics dataset. Given its outsized role in research and development in the national economy, the high tech sector’s declining dynamism is a particularly worrying sign for the prospects of more robust economic growth writ-large.
Defining the “High Tech” sector and why it matters
Released in February, new Business Dynamics Statistics of U.S. High Tech Industries (BDS-HT) examines industries with high concentrations of workers in STEM occupations following a similar methodology from Nathan Goldschlag and Javier Miranda (2020). Utilizing Economic Census data between 2007 and 2017, Goldschlag and Miranda identify industries with STEM occupation intensity at least five times the national average for more than half of this window. Consequently, the following industries meet this definition:
- NAICS 3341: Computer and Peripheral Equipment Manufacturing
- NAICS 3342: Communications Equipment Manufacturing
- NAICS 3344: Semiconductor and Other Electronic Component Manufacturing
- NAICS 3345: Navigational, Measuring, Electromedical, and Control Instruments Manufacturing
- NAICS 3364: Aerospace Product and Parts Manufacturing
- NAICS 5112: Software Publishers
- NAICS 5182: Data Processing, Hosting, and Related Services
- NAICS 5191: Other Information Services
- NAICS 5413: Architectural, Engineering, and Related Services
- NAICS 5415: Computer Systems Design and Related Services
- NAICS 5417: Scientific Research and Development Services
As of March 2020, the High Tech sector accounted for 7.2 million jobs, or 6 percent of jobs nationwide covered by BDS data. The High Tech sector’s share of total employment has remained steady at between 4 and 6 percent since 1978, the first year in which data is available. Nevertheless, the composition of employment has changed dramatically over this period. Aerospace manufacturing, which once made up nearly one-fifth of High Tech employment, has gradually fallen to 6 percent, while Navigational, Measuring, Electromedical, and Control Instruments manufacturing (henceforth “NMEC”) has declined from 17 percent of High Tech employment in 1978 to 6 percent in 2020.
Perhaps unsurprisingly, Computer Systems Design has exploded from 1 percent of High Tech jobs in 1978 to 28 percent in 2020. Architecture and Engineering Services, already a major component of the sector, also increased its share to 22 percent. Notably, semiconductor manufacturing has fallen from 13 percent of High Tech jobs to just 4 percent, and has outright shed approximately 125,000 jobs since 1978 as the bulk of chipmaking moved overseas.
As a whole, the sector has shifted from one dominated by the manufacturing of physical goods in large industrial sites to smaller software and computer system design firms: in 1978, 63 percent of High Tech sector employment was in manufacturing industries, but by 2020, over 80 percent was in information or professional services.
Why does the “High Tech” sector matter in particular? As Goldschlag and Miranda argue–along with others in the related body of literature–a high concentration of scientists, mathematicians, and sophisticated technicians is a reliable indicator of innovation and research activity within an industry.
Given the indispensable role R&D and innovation play in long-run productivity and GDP growth, the health and dynamism of industries conducting research is critically important. Whether leading High Tech firms’ market positions are threatened by upstarts or whether its talent can efficiently reallocate from less productive to more productive firms to follow good ideas and spread knowledge will therefore both be outsized drivers of national economic growth and overall prosperity.
The High Tech sector’s worrying convergence with the national economy on measures of economic dynamism is essential context for understanding the large investments Congress committed to making in the CHIPS and Science Act. Beyond making a major down payment on semiconductor supply chain resiliency, the law aims to seed new clusters of High Tech growth throughout the country, smooth the pathway for innovative startups with new ideas to bring them to market, and scale up to challenge incumbent firms.
Dynamism is declining across the economy, and the High Tech sector is no exception
Although the High Tech sector still remains more dynamic than the non-High Tech sector of the economy, the gap between the two has gradually faded. Beginning with firm entry rate, we see a long-running decline in both the High Tech sector and the broader economy. The national firm entry rate, hitting a series high of 13.5 percent in 1984, fell below 10 percent during the Great Recession and has yet to recover. The High Tech sector saw much higher startup rates during the 1980s, peaking at 17.9 percent, also in 1984. High Tech industries saw another huge surge in new startups in the 1990s, coinciding with the dot-com boom, but their startup rates have yet to recover.
In 2018, the gap between the national and High Tech startup rates fell below one percentage point to an all-time low, down from 5 to 6 percentage points in the 1990s. The rate of new firm creation in the High Tech sector, once far ahead of other sectors, now looks similar to that of the increasingly-sclerotic national economy.
Breaking down firm creation within the High Tech sector by component industry, we also see broad declines in startup rates. Software publishing is one notable exception; after plummeting in the early 2000s, the startup rate in software has subsequently doubled, though it still makes up only a small share of new firms within the sector.
The ability of new firms to both start and grow is a key feature of economic dynamism, enabling upstarts to challenge large, older incumbents. The startup rate is one measure of this phenomenon, but it fails to capture the ability of new entrants to scale and challenge existing firms for market share. An accompanying measure, the share of industry employment in young (less than 5 year-old) firms, captures this second quality. In this data, we see a clear, steady decline in the non-High Tech sector going back to the early 1980s. In the High Tech sector, young, upstart firms grew their employment share dramatically during the dot-com boom of the late 1990s in an expansion that was geographically and economically broad-based, until the subsequent wave of failures, bankruptcies, and consolidations sent this share plunging. It has since continued to gradually decline, a fall perhaps accelerated by a new era of strong network effects and platform economics in software and other High Tech industries.
Perhaps relatedly, the share of High Tech sector employees working in large companies (at least 500 employees), long declining before the dot-com bubble’s burst, stopped falling and held steady around 60 percent in its aftermath. In the non-High Tech sector, the share of employees working at such firms has increased steadily, from 42 percent in 1985 to a slim majority (50.2 percent) in 2020.
Workers moving between firms is another indicator of economic dynamism, as talent migrates from low-productivity to high-productivity firms. Job-switching is a key channel of knowledge diffusion, allowing hard-to-codify know-how and process knowledge to spread through the rest of the economy. Year-to-year labor market churn showed no discernible trend throughout the 1980s and 1990s, but has since declined in both the High Tech and non-High Tech sectors. In 2001, the reallocation rate in the High Tech sector was over 31 percent, but subsequently fell to 21 percent in 2020, the lowest level since 1980. Notably, job switchers see substantially faster wage growth than those who remain with their employer.
High Tech industries are geographically concentrated in large metros
High Tech Industries are more geographically concentrated in large metro area clusters.Two-thirds of High Tech firms are located in 25 metro areas ranked by firm count, while five metro areas alone account for over 27 percent of all High Tech firms. High Tech firms are modestly more concentrated than other kinds of firms, as the top 25 metros are home to about 68 percent of such firms compared with 62 percent of all firms.
High Tech industry employment is even more concentrated in top metros. The top 25 metros ranked by High Tech employment account for 56 percent of all High Tech employment nationally, while the top five host 28 percent of High Tech jobs. In contrast, the 25 largest metros by employment across all industries account for less than half (46 percent) of all jobs nationwide.
Putting both pieces together, the top 25 High Tech metro areas contain 6 percent more High Tech firms than their size would predict but 10 percent more High Tech jobs.
Economic dynamism within the High Tech sector appears increasingly concentrated in fewer metro areas, too. The share of metro areas with High Tech establishment entry rates above the national High Tech rate has been in steady decline over the past three decades. Concurrently, only one-quarter of metro areas have reallocation rates in the High Tech sector above the national rate, suggesting a subset of high-churn regional High Tech economies are driving national rates.
Declining dynamism and innovation
Two recent working papers by Goldschlag and fellow economist Ufuk Akcigit linking U.S. patent data with employment records shed further light on the decline in dynamism within the High Tech sector. While not limited to High Tech industries by the Census dataset, both papers aim to examine the health and dynamism of markets employing and deploying scientists, technicians, and inventors. Just as data on firm dynamics reveal a High Tech sector increasingly dominated by older and larger incumbents, this line of research on patents finds older and larger firms employing an increasing share of inventors, with potentially negative effects on aggregate invention.
The first paper analyzes demographic and employment trends among inventors, finding that inventors increasingly concentrated in larger, older firms between 2001 and 2016. They find that the share of inventors at young firms roughly halved over this period, falling from 15 percent to 8 percent. Goldschlag and Akcigit also find that many so-called “super star” inventors–the most productive according to patent data–migrated to larger firms in the aftermath of the Great Recession and were significantly less likely to found a new business.
The authors also find inventors are increasingly concentrating in a handful of metro areas. In 2000, the top 20 counties by inventor count were home to 39 percent of inventors in their sample, rising to 47 percent by 2016. Just as the indicators of dynamism in the High Tech sector have become increasingly concentrated geographically, so has talent contributing to research and innovation ecosystems.
The second paper more explicitly demonstrates the cost of declining dynamism on innovation. Again examining inventors’ employment dynamics over time, the authors find inventors increasingly concentrated in larger incumbent firms, where their compensation rises but their productivity falls, relative to those at younger companies. Rather than demonstrating the healthy migration of inventors and technical talent to firms that can more effectively leverage their skills, this finding suggests a new degree of talent hoarding and risk-aversion among large firms.
The challenge of dynamism in the High Tech sector
Reviving economic dynamism in the High Tech sector requires reckoning with the sector’s continued transition towards software and information technology and away from hard manufacturing in large, centralized industrial facilities. Legacy manufacturing industries were rife with high barriers to entry as well, given economies of scale and the importance of tacit knowledge in complicated production processes. Yet in the 1980s, these industries remained dynamic within a context of large federal investments in scientific and technological research, Cold War-era procurement requiring near-constant innovation, and demographic and macroeconomic factors otherwise reinforcing pro-dynamism trends throughout the economy.
Even the initial information technology boom in the mid-1990s proved to be a highly dynamic time for the High Tech sector, as it picked off the low-hanging process improvements in one industry after another. Thanks in part to low barriers to entry, this period of rapid growth was broadly-shared, both across industries and geographies.
Yet in the aftermath of the dot-com bust, a different set of trends have set in even as High Tech has become even more service-intensive and software-heavy. The sharp rise of what economists call “intangible assets” may now be further entrenching incumbents and slowing the diffusion of closely-guarded know-how from leading firms to the rest. Evidence of talent hoarding among top firms and the build-up of large patent portfolios further demonstrate this phenomenon.
Further, the rise of platforms characterized by strong network effects also create a degree of winner-take-all incentives, at least in the short run. Dislodging market leaders in these areas is by no means impossible, but it looks increasingly challenging than in prior eras.
Many of the industries we classify as within the broader High Tech sector have long histories of dynamism and disruption. Over the last 40 years, they have produced some of the world’s most impressive new technologies, widely-used services and platforms, and have built local economies characterized by very high churn and knowledge diffusion. That this sector is now rapidly converging with the rest of the economy on measures of economic dynamism presents a fascinating paradox, but also should call renewed attention to the health of one of the country’s most important sources of innovation and growth.
Conclusion
The problem of declining economic dynamism has become endemic, both geographically and across sectors. Important markets are increasingly static, characterized by larger and older incumbents. Labor markets, too, are less dynamic, as workers move between companies at slower rates. The High Tech sector was formerly shielded from such phenomenon, serving as an enclave where disruption and constant change persisted. Yet now it appears that this sector–which employs much of the economy’s most prized scientific and technical talent–looks increasingly similar to the rest of the economy on these measures. This new development should serve as yet another set of indicators of the need for a comprehensive agenda to revive economic dynamism and new investments in STEM research ecosystems across the country.
