Encouraging new, young businesses instead of older ones is considered a more effective path to job creation. A new study suggests that it’s not that clear.
State and local economic development policy remains, for all intents and purposes, a fact-free zone. The most popular technique—doling out incentives like business tax incentives or development zoning credits to lure business to relocate or compete for industries in the hopes that jobs will grow faster—has been shown to be a waste of money and a handout to industry by a wide body of studies.
Still, it endures. A list of 47 major cities in 32 states—not just Rust Belt cities like Buffalo and Detroit but booming cities like Dallas, Houston, and Phoenix and knowledge hubs like San Francisco and Boston—continue to fork over billions in tax revenue in the form of business incentives. The metros where these cities are located account for more than 60 percent of private sector GDP.
More recent—and supposedly more enlightened—policies are built on a different premise: Certain types of local businesses are keys to economic growth and job creation. If you encourage a specific subset of firms—small, not large; entrepreneurial, not incumbent; young, not old—then you’ll spark life into your city. Lots of places have tried to create incubators, university-industry tech transfer programs and even subsidized venture capital funds in the hope of creating the young startup firms that are thought to be the spurs to growth. But is that really the case?
A new study published in a recent edition of Economic Development Quarterly exposes the weak conceptual and empirical foundations of economic development policies that target particular types of firms. On one hand, this stems from an overemphasis on firm characteristics as opposed to other factors that truly do matter to economic development. But on the other, the study echoes Paul Romer’s concerns about “mathiness” in economics—how researchers can tweak statistical models to generate just about any finding they want.
To illustrate this, the study looks at the economic development implications of different kinds of firms in Washington State using data from the Employment Security Department and Department of Revenue. (The data cover 750,000-plus observations on nearly 200,000 firms.) The study looks at firm size, age, and job growth across the state. The average firm had 16 employees and was 11 years old, and the majority of firms had been around less than 7 years.
Looking first at the percent change in employment, company size matters—but only for the smallest firms. The growth rate for each size remains mostly relative, and the smallest firms grew less during the economic decline after 2008. However, small firms are most vulnerable to economic downturns, according to the study.
New firms (less than a year old) grow roughly 20 percent faster compared to oldest firms, at a rate of 14 percent annually from age 1 to 2. But the effect drops off rather quickly. From age 3 to 4, they grow 7 percent faster than the oldest firms, and there is statistically no difference by about age 23.
Industry matters much more than size or age to the percent change in employment. Manufacturing firms of all sizes experienced larger percentage employment declines during the recession. In health care, only the smallest firms grew less rapidly than the largest ones. In retail, food, and lodging, age did not matter. Overall, firm size did not matter to the percent change of employment.
Looking next to overall employment growth (as opposed to percent change), the opposite is true. Overall employment growth is significant for the largest firms, but in other firms the changes in employment were similar. The raw change in employment is likely to be large relative to its percent change, and the combination of the two is likely to be highly correlated with the percent change.
These findings are complicated further when the authors looked the persistence of growth year over year. Here they find no measurable persistence of growth; the effect of size is positive for the smallest firms and declines slowly as firms grow and age. But the coefficients are small and associated more with age than size. Again, the findings are murky and fraught.
If growth is sustained, there would be a positive effect from firms despite lagged growth, whereas if changes are random, there will be a negative coefficient during an economic downturn like 2008. In the latter case, size matters only for smallest firms—and only during the downturn. Otherwise, age has positive impact for young firms, but young firms are often indistinguishable with their churning, both creating and dislocating jobs.
Ultimately, the study concludes that there is little consistent and reliable evidence on one side or the other when it comes to firm size and age: No model can explain more than 15 percent of the variation in job growth. As they write: “Economic development policies that target firms of any particular size or age are unlikely to have much impact on job growth, if any, given that at least 85 percent of the variation in employment growth is explained by other factors.”
A good deal of the problem, according to the authors, lies in the way that economic development can be framed in models that can be tweaked to give the desired results, or as they put it, with models that are “very sensitive to model specification.” Or, as Arlo Guthrie put it in the song, “Alice’s Restaurant,” economic development analysis is an area where “you can get anything you want,” depending on how the issue is defined and the kinds of models that are used.
Clearly, when it comes to economic development, we need better data and better models. And most of all, given the lack of clarity around policies and the drivers of economic development, we need to resist the conventional wisdom and think local—to orient economic development away from “big game” and quick fixes, and focus instead on small-scale assets and investments.