What is the effect of the mobile Internet on the economy? This question is an important one, and one that has drawn significant attention by researchers, policymakers, and even the President. What makes answers to this question difficult to come by is that while the Internet may influence things like income, education, depression, and so forth, Internet use may in turn be influenced by income, education, depression, and so forth. Establishing the causal direction of the relationship, and its magnitude, can be challenging.
Due to the present economic woes, the effect of the Internet on job creation is an empirical question of substantial importance. The Phoenix Center has published a number of studies on the relationship between Internet use and job search. These studies use advanced statistical and econometric methods to quantify the impact of Internet use, and in all of these studies the Internet is found to have sizeable and beneficial effects. The most recent of these papers, Mobile Broadband and Job Search: An Empirical Test, quantifies the effect of mobile broadband use on job search. We also consider how the effect of mobile use compares quantitatively to the effects of other connection modalities. To summarize, the study found, “mobile Internet use has a large and statistically significant effect on sustaining active job search, cutting in half the probability an unemployed person abandons efforts to find new employment due to discouragement about labor market prospects.” That’s a very large effect. And, it’s not just that mobile use has a large effect; we found that “mobile use reduces labor market discouragement even more than broadband use at home.”
While these studies of job search address an important issue, they do not answer the question of whether Internet connectivity, or mobile connectivity in particular, has any effect on employment itself. Last week, a new study was released—The Employment Effects of Advances in Internet and Wireless Technology: Evaluating the Transitions from 2G to 3G and from 3G to 4G, by Robert Shapiro (Sonecon) and Kevin Hassett (AEI)—which attempts to answer that question.
This study reports that the investment in mobile network upgrades, and the resulting adoption of smarter devices and the apps that ride on them, have stimulated significant job creation in the US. Indeed, the authors claim, the “shift from 2G to 3G Internet and wireless network technologies led to the creation of nearly 1.6 million new jobs across the United States, between April 2007 and June 2011—even as total private sector employment fell by nearly 5.3 million positions.” Shapiro and Hassett approach the problem with an econometric tool designed for the quantification of causal effects. The relevant question for a study like theirs is whether the advancement of wireless network causes employment growth. With time series data, a widely-used econometric technique for assessing causality (at least, to the extent possible) is Granger Causality. Granger Causality is a technique that tests whether past changes in the value of some variable affect future changes in the values of another. There are no perfect econometric techniques and there is never perfect data. But, Granger Causality is widely-used among professionals and academics, has good theoretical properties, and is in most cases easy to implement. Shapiro and Hassett apply Granger Causality to a unique dataset constructed by the authors for their own research (using data from the Nielsen Mobile Insights Survey).
Since some of the data analyzed in the Shapiro-Hassett study are not publicly-available, I can’t analyze the data myself. With that caveat in mind, the results from the study look good. The null hypothesis of the Granger test is “no causal relationship,” so a linkage between mobile technology and employment is established if this hypothesis is rejected. As shown in their study, the null hypothesis is easily rejected—the diffusion of advanced mobile services appears to affect employment, and in a favorable direction. (The joint test—the F-test—is not reported in the study, but Dr. Hassett told me that the joint-test permitted a rejection of the null hypothesis of “no effect,” which is not surprising given the t-statistics on the individual coefficients.)
Based on computations using their estimated relationship between employment and wireless technology diffusion, the authors conclude that the advancement of wireless technology created about 400,000 jobs annually (1.585 million jobs over about four years). This is a big number (which is a good thing given current economic conditions). My own rough calculations suggest that is not much different from what our own econometrically-derived employment multipliers would produce using investments in mobile networks over the same period.
Based on the Shapiro-Hassett study and our own work, it appears that mobile broadband communications technologies are assisting not only in job search but in job creation as well. In light of this evidence, and the spectrum shortage that threatens to limit ongoing expansion of next generation wireless broadband networks, pending legislation aimed at getting more spectrum in the hands of as many commercial mobile wireless companies as are interested and able to deploy the resource is economically important and should be quickly passed and signed into law.