What do computers do?
In an influential paper, Autor et al. (2003) address the question of ‘what computers do’ by suggesting that computers complement work that involves the execution of complex, cognitive-intensive workplace tasks, while to an equal or greater extent substituting for work that is highly routine, or lacking in complexity. Since then, substantial long-run evidence consistent with this thesis has been presented. Yet, an important question remains, namely what is the direct, short-run impact of investment in computers, or other Information and Communication Technology (ICT), on the demand for different types of workers within the firm.
In addressing this question, researchers must overcome a significant obstacle, namely that over long-time horizons the relative supply of different types of workers responds to the level of ICT adoption by firms, and vice versa. In other words, the direction of causation can be difficult to assess. This point has been emphasised by Acemoglu (1998, 2002, 2007) as well as by Goldin and Katz (2008), who document the long-run ‘race’ between education and technology. As a result, in order to isolate the direct, causal relationship between ICT adoption and the demand for different types of workers, it is necessary to hold both the economy-wide supply of skill and the overall level of technology fixed.
Firms’ response to an ICT tax incentive
One way to do this is to focus on a large, unanticipated, short-run shift in the incentives faced by firms to invest in ICT. We find exactly this scenario in the form of a tax incentive that targeted small UK businesses—those with 50 or fewer employees—over the period from 2000 to 2004. Specifically, the tax break allowed firms to write off 100% of their investments in ICT from their taxable profits. In a recent paper (Gaggl and Wright 2015), we compare these small firms with otherwise similar firms and, using confidential data from Her Majesty’s Revenue and Customs (HMRC), find that thousands of small UK firms claimed this tax credit. Figure 1 plots average tax claims for firm investments in plant and machinery, of which ICT investments are a component, against firm size for the period during which the tax credit was available. The clear discontinuity at 50 employees (and the lack of a discontinuity after the tax credit expired) is quite strong evidence that the tax credit was taken up, particularly in light of the fact that there were no other tax credits specifically targeting these firms during this period.
Figure 1. Average plant & machinery tax deductions claimed
Notes: The figures plot average first-year tax allowances claimed by firm employment in British Pounds (GBP). Panel A pools the treatment period (2000-2004) while panel B pools the three years after the ICT tax incentive expired (2005-2007). The solid lines are locally weighted regressions based on the Stata command lowess.
Using data on UK firm capital investments, we find that these tax claims did in fact correspond to new ICT investment for the targeted firms.
Figure 2 illustrates this graphically, as we find a discontinuity at 50 employees in both software and hardware investment, consistent with the increased tax claims (Figure 1). This discontinuity reflects a £1,684 of additional software and hardware investment per worker each year, approximately equivalent to an extra PC and associated business software per worker at the prices prevailing in 2000.
Figure 2. ICT investment by firm-level employment
Notes: Panel A illustrates software investment along the firm size distribution during the treatment period (2001-2004) whlie panel B shows an analogous figure for hardware investment. The graphs are based on the UK Quarterly Capital Expenditure Survey (QCES).
Effect of the ICT investments on workers
Given that both the economy-wide supply of skills and the overall level of technology are likely to be fixed over the short horizon during which the tax incentive was available, we were able to exploit this policy in order to identify the causal effect of the ICT investment on the demand for different types of workers within the investing firms.
We find that workers who were engaged in complex, cognitive-intensive production tasks – such as architects and engineers – saw immediate gains from the ICT investment.
Panels (A.1) and (B.1) in Figure 3 depict a sharp positive discontinuity in wages and hours for these workers. In other words, the average complex, cognitive worker in a 50-person firm saw wage gains relative to similar workers in firms with 51 people. Importantly, we find no evidence of such a discontinuity beyond the period during which the tax credit was available. On the other hand, workers engaged in relatively routine, cognitive-intensive tasks—for instance, office assistants and bookkeepers—were displaced and also suffered a loss in earnings and working hours, as depicted in panels (A.2) and (B.2) in Figure 3, yet these effects were small compared with the gains for complex, cognitive workers.
Figure 3. Weekly earnings and hours: Routinea, & non-routine cognitive jobs
Notes: The figures plot average weekly earnings (panel A) and hours (panel B) against firms size for both non-routine cognitive (panels A.1 and B.1) and routine cognitive (panels A.2 and B.2) occupations. These occupation groups are based on work by Autor, Levy, and Murnane (2003).
The benefits that accrued to complex, cognitive workers were large; within investing firms, an additional £1000 of ICT led to an additional half hour of work per week in these jobs and a wage increase of about £13 per week. One possibility is that simply handing a computer to these workers raises their productivity, and therefore their wage and hours worked—what we can think of as a direct effect of the investments. A more subtle possibility is that, in addition to the direct effect, firms may reorganise their workplace in order to give computer-oriented tasks a more prominent role, further increasing the value of these workers to the firm. To explore this possibility, we exploit data from the UK Innovation Survey, which records several types of organisational changes made by UK firms over the period from 2002 to 2004. We find that indeed the tax incentive was, on average, associated with significant changes in organisational structure for the affected firms.
Given both the scale of the ICT revolution and recent concerns about wage inequality, it is important to have careful evidence on the relationship between these two phenomena. As many researchers and market observers have conjectured, our evidence indicates that, in the short run, ICT has a significant and positive impact on workers engaged in complex, cognitive tasks. At the same time, we find that the losses are quite mild over the time horizon we explore, suggesting that the large displacements observed in long-run analyses occur with a lag. Beyond these basic findings, we show that firm reorganisation is a key part of the ICT adoption process, with important implications for the distribution of earnings within the firm.
Acemoglu D (1998), “Why Do New Technologies Complement Skills? Directed Technical Change And Wage Inequality”, The Quarterly Journal of Economics 113: 1055–1089
Acemoglu D (2002), “Technical Change, Inequality, and the Labor Market”, Journal of Economic Literature 40: 7–72.
Acemoglu D (2007), “Equilibrium Bias of Technology”, Econometrica 75: 1371–1409.
Autor D H, F Levy, and R J Murnane (2003), “The skill content of recent technological change: An empirical exploration”, The Quarterly Journal of Economics 118: 1279–1333.
Gaggl P and G C Wright (2015), “A Short-Run View of What Computers Do: Evidence from a U.K. Tax Incentive”, SKOPE Working Paper No. 123.
Goldin C, and L Katz (2008), The race between education and technology, Harvard University Press, ISBN 9780674028678.