By David Wiczer, Economist
Microsoft paid more than $26 billion for LinkedIn, the largest acquisition in the company’s history. Though LinkedIn has obviously grown into a huge role connecting employees to recruiters and each other, this move had many tech journalists wondering why the network could command such a price, especially given that it has generally been unprofitable.
But perhaps Microsoft believes it has the wizardry to monetize a pervasive feature of the labor market. After all, there’s logic to Microsoft’s move: Many people find their jobs through referrals, and these jobs pay more. In this post, we will focus more directly on quantifying the extent and impact of job search through networks and discuss some theories behind it.
To study the empirics of job referrals, we looked to the Survey of Consumer Expectations (SCE), a detailed survey covering many aspects of an individual’s place in the labor market. The survey collected a small but representative sample of households in 2013.
Unemployed and actively searching people were asked a battery of questions about how they found jobs, including whether they directly contacted organizations about posted vacancies or received referrals through their network. For workers who were employed at the time of this survey, the SCE also asked how they found their current job. For these employed workers, the SCE asked their starting and current salaries.1 Workers searching for another job reported the salaries of their job offers.
Among currently employed workers, those who found their job through a referral from their network had an average weekly salary of $772.20, or roughly $40,000 per year. Those who did not find their job via a referral had an average weekly salary of $725.84, or nearly $38,000 per year. On average, salaries were 6 percent higher if workers found their job through their networks.
Further, their earnings are even more positively skewed. One way to interpret this is that network searchers have more “upside” risk: They can potentially draw a variety of wages, but there are more very high potential outcomes through the network, To quantify this, Kelley’s statistic is 0.6 for network-finders and 0.44 for others, meaning that 80 percent rather than 74 percent of the dispersion between 90th and 10th percentile is accounted for by the top half (from 90th percentile to 50th).
The distribution of wage offers should typically be different from the distribution of wages among employed workers. Not all offers are accepted, and workers at lower wages tend to make more over time through selective job mobility and pay increases on the job. Still, even among the distribution of wage offers, we see a premium associated with those who found jobs through their network. Workers who were searching while unemployed received offers through their networks that averaged 62 percent more than those found through direct contact. Workers searching while employed received network offers that were 12 percent higher, on average.
So what explains this pervasive difference? The premium to network referrals has also been observed in other contexts, and a common explanation is that job referrals carry additional information.2 For instance, jobs that originate from a referral may take advantage of better information about the quality of match between worker and job than can be discovered in an interview.
Alternatively, if some simply have more connections than others, this can also lead to differences. In this scenario, which I explored in a recent working paper with my co-authors Marcelo Arbex and Dennis O’Dea, the workers who tended to find jobs through their network were different than those who found jobs through direct search, and the better-connected workers had access to better jobs. Perhaps Microsoft is now best positioned to understand the underlying mechanisms.
1 Respondents may report hourly, weekly or annual earnings. We converted all salaries to weekly by dividing the annual salary by 52 weeks or by multiplying hourly wages by the usual weekly hours. We dropped workers whose implied salary is less than $80 per week.
2 A recent example is Dustmann, Christian; Glitz, Albrecht; Schönberg, Uta; and Brücker, Herbert. “Referral-Based Job Search Networks.” Review of Economic Studies, April 2016, Vol. 83, Issue 2, pp. 514-46.