Reaching for the Stars: Who Pays for Talent in Innovative Industries?
GAICIL Insights
Innovative industries are, at heart, betting parlors for talent.1 Reaching for the Stars: Who Pays for Talent in Innovative Industries?, by Fredrik Andersson, Matthew Freedman, John Haltiwanger, Julia Lane and Kathryn Shaw, shifts attention from ideas, patents and capital to people. Innovation in the U.S. economy, the authors write, is about employing and rewarding highly talented workers to produce new products. Using unique longitudinal matched employer-employee data, they connect talent to firms in markets with risky product innovations.
A provocation up front: a country's AI strategy may be best measured by a single question. How easy is it for the world's most talented engineers to move there, and to move between firms once they arrive?
The central finding comes from software. Firms in product markets with highly skewed returns to innovation, high variance payoffs, are more likely to attract and pay for star workers. They pay more up front, in starting salaries, to attract and motivate stars, because if those workers produce home-run innovations, the firm's winnings will be huge. These firms also pay handsomely for loyalty: stars who stay earn much more in firms with high variance payoffs. In firms that have actually hit home runs, with high revenues, rewards for star talent are greater still. Earnings dispersion is higher within firms facing high variance payoffs.
The loyalty premium has a logic. Workers in such firms likely carry firm-specific human capital, and quality is identified and rewarded slowly on the job. Much of their human capital is product-specific, such as knowledge of wireless software. So firms tie workers to the firm through deferred compensation.
Here is where the paper meets the present. AI may be the most extreme high-variance product market ever assembled. Returns to innovation are wildly skewed: a few breakthroughs generate enormous value while many efforts yield little. The paper's logic predicts fierce competition for star talent, large starting salaries and big retention premiums. That is a precise description of the market for AI researchers and engineers. The eye-watering pay packages reported in AI are what the model says rational firms should offer.
Skewness explains the dispersion too. In a high-variance market, a star's expected value is enormous because one home run can transform the firm. Paying a large premium is rational when payoffs are that lopsided. Wide earnings gaps inside AI firms are the natural fingerprint of the return structure, a feature of the game.
That reframes the inequality debate. High pay in technology firms often draws criticism. This research shows star pay in innovative industries reflects the risky, skewed nature of innovation returns. Policies that compress rewards for talent in these markets could weaken the incentives drawing talented people into risky innovation. Any such policy should be weighed against that trade-off; flattening the prize changes who chooses to play.
Mobility is the other lever. Loyalty premiums and deferred compensation bind workers to firms. Where human capital is product-specific, movement across firms in the same product market can spread knowledge and accelerate innovation. Rules on non-compete agreements and compensation portability shape how talent flows in AI markets. Firms will use pay to retain stars; policy should ensure such arrangements do not unduly choke the diffusion of knowledge. The ideal market lets firms pay to keep stars without letting them lock stars in.
The research also shows what matched employer-employee data can do. Linking workers to firms and product markets revealed how market structure shapes pay. Similar data for AI firms would show how talent is allocated across the ecosystem and how compensation responds as the technology and market evolve.
The broader message is that innovation depends on matching talent to opportunity. Firms in risky markets pay for stars because stars produce the breakthroughs that generate outsized returns. In AI, with returns among the most skewed in the economy, competition for talent is a core feature of the innovation process. Policies that expand the supply of talent, through education and immigration, and keep labor markets fluid will decide how effectively the economy can reach for the stars.
Introduction
Innovation in the U.S. economy is about employing and rewarding highly talented workers to produce new products. Using unique longitudinal matched employer-employee data, this paper makes a key connection between talent and firms in markets with risky product innovations. We show that software firms that operate in product markets with highly skewed returns to innovation, or high variance payoffs, are more likely to attract and pay for star workers. Thus, firms in high variance product markets pay more up-front--in starting salaries--to attract and motivate star employees, because if these star workers produce home-run innovations, the firm's winnings will be huge. However, we also find these same firms pay highly for loyalty: star workers that stay with a firm have much higher earnings in firms with high variance product market payoffs. The large effects on earnings are robust to the inclusion of a wide range of controls for both workers and firm characteristics. One key control is that we also show that in firms that have actually hit home runs, with high revenues, the rewards for star talent are even greater. We also find that the dispersion of earnings is higher within firms with high variance product payoffs.
Takeaway
The process of innovation in the U.S. economy is fundamentally dependent on employing and rewarding highly talented workers to produce new products. This paper makes a key connection between talent and firms in markets with risky product innovations – we show that software firms that operate in product markets with highly skewed returns to innovation, or high variance payoffs, are more likely to attract and pay for star workers. Thus, firms in high variance product markets pay more up-front—in starting salaries—to attract and motivate star employees, because if these star workers produce home-run innovations, the firm’s winnings will be huge. These same firms also pay highly for loyalty: employees that stay with a firm have much higher earnings in firms with high variance product market payoffs. These striking effects are robust to the inclusion of a wide range of controls for both workers and firm characteristics. In addition, we show that in firms that have actually hit home runs, with high revenues, the rewards for star talent are even greater. There may be many factors underlying the finding that loyalty is rewarded more in firms operating in risky product markets. In these firms, workers are likely to have firm-specific human capital, and also worker quality is identified and rewarded with performance bonuses slowly on the job. However, much of their human capital is likely to be product-specific (such as a knowledge of wireless software), not firm specific.1
Thus, firms tie workers to their firms through deferred compensation – they offer options that become vested only after the employee is with the firm four years. And clearly the deferred compensation is performance pay – stock options pay off only if the firm performs. However, our model emphasizes that firms pay high wages to employees just to be in the game – to play in the market where there are potential big winners – whether the firm actually wins or not. Though the focus of this paper is on the software industry, our results should generalize to other industries that utilize knowledge workers. There were very pragmatic reasons for focusing on the software industry: the ability to examine one production function as well as use a dataset that matches data on software workers to data on software firms. In addition, firms in the software industry have several attractive features: firms are innovative, and hence have the greatest payoff potential from star talent; the activities of knowledge workers can be directly tied to the performance of the firm; and workers perform the same tasks within and across firms, so we can describe and model the production function for these workers. However, all industries that value project development that pays off should pay high wages to attract the most talented employees.
Notes
- Fredrik Andersson et al., Reaching for the Stars: Who Pays for Talent in Innovative Industries?, 119 Econ. J. F308 (2009), https://doi.org/10.1111/j.1468-0297.2009.02277.x. Takeaway source: 5. Conclusion, reproduced from a free author or repository copy (working-paper version). ↩
Cite as: Fredrik Andersson et al., Reaching for the Stars: Who Pays for Talent in Innovative Industries?, 119 Econ. J. F308 (2009), https://doi.org/10.1111/j.1468-0297.2009.02277.x
Header image: Asia Africa Library in Tokyo.jpg by Eugene Ormandy, CC0, via Wikimedia Commons.
