How can employers attract talented engineers in tight labour markets? This question has challenged managers for at least a generation, and interest in it shows no sign of abating as investment in AI and data science technologies continue to create demand for workers with specialised technical skills that can support development in these areas. The question of how the available pool of technical talent affects employers’ decisions also poses a challenge to policymakers, who are tasked with ensuring that their cities, states, and countries remain competitive in an increasingly digital age. For example, for these public officials, encouraging businesses to invest and create new jobs in the regions they represent often requires convincing them that they will have access to the skilled workers they need to produce goods and services that, more often than not, have a digital component.
Indeed, technology-oriented employers devote a great deal of attention and energy to the question of how best to attract and retain the technologists who can build and support cutting-edge digital systems. Indeed, one might argue that an employer’s ability to attract these workers is a key differentiator in the modern, digital economy. How well firms can attract this talent, and what this means for performance has also attracted attention from social scientists. Recent work has shown that it has implications for the speed of technology adoption and for the productive gains that can be realised from a new technology (Branstetter et al. 2019, Tambe 2014).
Although wages for technology workers have certainly seen growth in recent years, a challenge for employers is that technology workers, perhaps more so than other high-skill professionals, often care about many factors other than wages. Work culture, mission, co-workers, perks, and access to interesting data and problems matter a great deal to being able to attract talented technical workers. For engineering hiring, therefore, employers often compete on a number of dimensions, not just on salary alone.
One of the most interesting and important of these dimensions is knowledge, skills, and abilities, especially when the market is moving towards new technical paradigms, such as data science or machine learning. During these periods, workers ‘pay’ for the opportunity to acquire training in new technology by taking lower wages. The notion that workers choose employers depending on where they can learn new skills has a number of interesting implications, both for where they choose to work and how much they earn.
Dating back at least to the work of Gary Becker (1964), there has been a long tradition of work on the question of when employers pay for training workers and when workers pay for their own training (Acemoglu and Pischke 1999, Autor 2001, Cappelli 2004). Our recent work in this area tests the hypothesis that for engineers, employers compete, in part, on the technologies that they use (Tambe et al. 2019), and that high-tech workers ‘pay’ for the opportunity to acquire training in a new technology. Such an exchange implies that for a given wage, technical workers are more likely to be attracted to employers who use cutting-edge technologies. For a given wage, employers that use cutting-edge technologies can attract better-quality workers. We also show that younger workers – i.e., those who more likely have long careers ahead of them – are more likely to be attracted to those employers who can offer them the chance to work with these types of technologies.
Directly testing the notion that workers place value on working with more interesting technologies is a challenge, because workers generally do not publicise – or even know – the different components of the value they gain from working in a particular job. One way to elicit the value they associate with working with a particular technology is to just ask workers how much they value it, but they may not know, or depending on the context, may choose not to answer truthfully. A more reliable approach might be to offer them a schedule of wages and different technologies to use, and observe how they make choices, but this would not be realistic in a labour market setting. Our approach was to essentially ask workers, in a labour market context, how much would it take for you to leave your current job? Under certain assumptions, this amount can be broken down into its component values, and can tell us something about the amount of value that technology workers place on working on a particular technology in their current job. Workers who are happier in their current jobs should require more to leave it.
The data we used were from a job board that explicitly asks participants (i.e., workers applying for jobs) for these data. On this job board, job seekers post the target wage they would require to move to a new employer. Because potential employers see this number when considering applicants, there are real consequences to misreporting it. When combined with data on the elements of the job they are currently working in, statistical techniques can be used to reveal how much workers value individual elements of their current job. We measured the kinds of technologies they use while on the job from information on their work experience, and to measure other aspects of employer brand, we studied what workers say about their employers in online reviews. We also worked with the company Glassdoor.com to analyse the textual content of reviews posted by tech workers on their websites, and identified key features of employers that their workers particularly value. For tech workers, some of these key features include levels of compensation, benefits, quality of co-workers, and the emphasis they place on skills development and learning.
Some of our key findings about these relationships are summarised in Figures 1 and 2. Figure 1 shows that tech workers require more to leave their current employers when they are working with more interesting technologies. The data are from 2007, so it shows that for technologies that were ‘hot’ – at that time, such as technologies related to web frameworks – workers placed value on working with them and therefore were more difficult to poach from their employers. For older and more established technologies, this premium disappears, and workers using these technologies would require less to move to a new employer.
Figure 1 Technology age and predicted wages difference
These findings are not the same for all workers. Figure 2 shows that these findings are concentrated in younger workers, which makes sense, because workers are more likely to value a chance to learn new skills if they have a long time over which they can use those skills and improve their future earnings. Workers who are close to retirement should not be expected to place as much value on learning new skills.
Figure 2 Younger workers value new skills more
Of course, there is no ‘free lunch’, and these findings do not suggest that employers who want to attract and retain tech workers should all go out and invest in the newest technologies. First, a better technology infrastructure is costly, both in terms of the technology itself and the labour required to support it. Even though these workers accept less than they might otherwise earn, they are not necessarily less costly to hire than workers with older technology skills. Second, allowing workers to use ‘hot’ technologies means they will be attractive to outside employers as well. In fact, we found that while employers find it easier to attract top-flight workers when they use better technologies, these workers also churn faster, perhaps because they develop skills that are in high demand on the external market.
For managers, our analysis adds to the growing interest in the concept of ‘employer brand’ as a critical differentiator in a highly competitive labour market. It makes explicit the tradeoff between providing an environment in which workers can use new skills and the ability to attract talented engineers. Tech workers require higher pay to move to an employment context where they are using less interesting technologies than they would otherwise.
This type of relationship also has important implications for policymakers. For example, different economic regions vary markedly in terms of the freedom with which workers can move between employers, due to non-compete restrictions, the portability of benefits such as health insurance, unemployment insurance, and other differences. Our results suggest that some of the most important benefits to using new technologies depend on workers’ abilities to transfer their skills to future employers.
Acemoglu, D and J S Pischke (1999), “Beyond Becker: Training in imperfect labor markets”, Economic Journal 109(453): 112-142.
Autor, D H (2001), “Why do temporary help firms provide free general skills training?”, Quarterly Journal of Economics 116(4): 1409-1448.
Becker, G S (1964), Human capital: A theoretical and empirical analysis, with special reference to education, New York: Columbia University Press.
Branstetter, L G, M Drev and N Kwon (2018), “Get with the program: Software-driven innovation in traditional manufacturing”, Management Science 65(2): 541-558.
Cappelli, P (2004), “Why do employers pay for college?”, Journal of Econometrics 121(1-2): 213-241.
O'Mahony, S and B A Bechky (2006), “Stretchwork: Managing the career progression paradox in external labour markets”, Academy of Management Journal 49(5): 918-941.
Stern, S (2004), “Do scientists pay to be scientists?”, Management Science 50(6): 835-853.
Tambe, P, X Ye and P Cappelli (2019), “Paying to program? Engineering brand and high-tech wages”, NBER working paper 25552.
Tambe, P (2014), “Big data investment, skills, and firm value”, Management Science 60(6): 1452-1469.