It’s become a truism that corporations must invest in leading-edge digital technologies to move ahead and outpace competitors. Digital laggards pay the consequences in lost revenue and customers. Even the strongest traditional firms have collapsed under pressure from disruptive, tech-savvy upstarts.
But can you actually measure the value of new technologies like AI? And what about the value of the engineers and much-sought-after software talent driving these new technologies — can their value be measured, as well? Daniel Rock, a PhD Candidate at MIT Sloan, and a researcher at the Initiative on the Digital Economy (IDE), studied these issues and how firms make and earn returns to investments in technology. He is particularly interested in the economics of Artificial Intelligence. Some of his recent projects have focused on predicting the economic effects of machine learning, measuring the market value of human capital and intangible assets, and understanding investment returns to technological labor.
Rock discussed his recent research, “Engineering Value: The Returns to Technological Talent and Investments in Artificial Intelligence,” at a recent IDE seminar. He examines the relationship between technical talent and corporate employers, both in general and in the specific case of the open-source launch of TensorFlow, a deep learning software package, by Google. He found that an additional engineer at a U.S., publicly traded firm is correlated with approximately $854,000 more market value for the firm. Rock recently spoke with IDE Editorial Content Manager, Paula Klein, about his work. The following highlights key points in the conversation.
Q: It seems like you’re charting new territory with this research: quantifying the ROI and value of engineers and AI to a corporation. What led to this research, and why focus on engineers?
Daniel Rock: There’s been lots of discussion lately on these topics. First is recent work on monopsony — a market structure in which buyers have market power that can be an issue when the supply- side is competitive — and on labor-market concentration. A monopsony employer — who has market power in hiring workers — can take advantage of the fact, for instance, that sometimes workers can’t go elsewhere and earn more. These companies might also make their employees specialize in narrow domains. Engineers are a natural context for that since they solve problems for their employers that are often specific to the firm.
I wanted to examine firm-specificity as a source of monopsony power and its impact. If there is only one main employer of labor, then it has market power in setting wages and choosing how many workers to employ. In a more competitive labor market, on the other hand, workers would go to other firms willing to pay a higher wage. Even if a firm is not a pure monopsony, it may have a degree of this type of power because it’s difficult for workers to switch jobs and find alternative employment in some markets. Further, the employer can decide how to set tasks for their employees and that can introduce another means of getting value from the workforce.
The second reason to study the topic is to measure intangible assets and capital. What proportion of firm value lies in what employees know? Engineers — and also managers and sales people, to a lesser extent — get to see technology changes and tools before others. If we can track what engineers know, it helps explain why a firm gains more value from them than other labor. Discontinuities exist for technology in a way that don’t for management style or sales techniques. My preliminary findings, in fact, are consistent with the idea that as early implementers of new technology, engineers are highly complementary to the intangible knowledge assets that firms accumulate.
Q: Can you give an example of how talent bottlenecks can be eased?
DR: When new tools like AI are first introduced, talent is a bottleneck. A company may have all the assets to implement the advances, but not the talent. Labor supply is inelastic, and it’s hard to find enough skills, even with high wages. In addition, there’s friction and costs associated with finding these workers and training them — which adds to their high value. The same is true for in-demand programmers. Skilled developers at Google, Facebook, and other high-tech companies, were rare, and that drove up wages. Because it wasn’t in the corporation’s best economic interest to keep that high wages and low supply, they turned to open source, and tools like TensorFlow that made it easy for anyone to learn their programs, easing the bottleneck and also creating more market equilibrium.
Q: Describe your methodology. How accurately can you measure the value of labor?
DR: Broadly speaking there are four ways to measure the value of labor: value to workers; value to the company; value to consumers, and value to society. I deal with the company side where the value of a firm is equal to the value of its assets. Loosely, I treat labor as an asset — similar to physical buildings or tractors — to measure its value, but the ways in which labor adds to market value can be a little different than ordinary buildings or equipment. In the past, it was harder to find data to improve measurement, but it’s more available now and we can decompose the labor force. I used over 180 million position records and over 52 million skill records from LinkedIn, as a foundation for our data analysis.
A key finding is that an additional engineer at a U.S., publicly traded firm is correlated with approximately $854,000 more market value for the firm. In other words, the value of engineering talent partially goes to the company; complementary investment is required to realize returns. I try to examine the causal effect of adding an additional engineer on market value, but find it statistically indistinguishable from zero. As far as I can tell, there isn’t a lot of free money out there on the aggregate.
Q: How can managers use this information and what are the broader policy implications? Is it a good time to choose a career in engineering?
DR: As a manager, if you think carefully about your competitive advantage, you might realize it’s in talent that no one else has. Then, you have to decide how you use it. Labor has its own objectives that don’t always match the corporation, of course, so if talent is really scarce, you may want to remove the gates and make it more democratic for all. Open source becomes a strategy to add complements to what you have. It’s one lever to take some control of the pipeline.
From a policy perspective, we need to compete more vigorously. Entrepreneurship has declined in the last 20–30 years among smaller U.S. firms and we may need to import more top talent to offset scarce domestic talent. There are a few ways to alleviate talent bottlenecks where they exist. We also need to encourage academia that focuses on future work skills that complement technology skills and promote lifelong learning. Antitrust consideration of factor markets, like labor, is worth considering. And building a competitive and productive economy requires a commitment to inclusivity, equity, and diversity. Of course, all of these are big challenges. It’s definitely a good time to choose a career in engineering.