The Annual Congress of the European Economic Association took place in Geneva in August. One of the many papers to attract interest was by Oliver Falck, Alexandra Heimisch, and Simon Wiederhold1 on the effect of digital skills on earnings. This is a shortened version of that paper.
Objective measures of ICT skills
For our analyses, we use data from the Programme for the International Assessment of Adult Competencies (PIAAC), providing information on ICT skills in 19 developed countries. PIAAC is the product of collaboration between participating countries through the Organization for Economic Co-operation and Development (OECD), and uses leading international expertise to develop valid comparisons of skills across countries and cultures. The survey was conducted between August 2011 and March 2012. PIAAC was designed to provide representative measures of the cognitive skills possessed by adults aged 16 to 65 years, and had at least 5,000 participants in each country.
Using novel data on ICT skills from PIAAC, the OECD's "PISA for Adults", we show that ICT skills are substantially rewarded in modern labor markets. These positive wage effects of ICT skills can partly be attributed to the fact that individuals with higher ICT skills work more often in jobs that involve abstract tasks (requiring problem solving, adaptability, and creativity), which pay substantial wage premia. At the same time, the higher an individual's level of ICT skills, the less often she works in routine, easily automatable jobs. Regarding the determinants of ICT skills, show that ICT skills are developed by performing ICT-related tasks, while having access to (high-speed) Internet is a precondition for this learning-by-doing.
The PIAAC data are unique as they provide objective measures of numeracy, literacy, and ICT skills (each measured on a 500 point scale), which can be compared within and across countries. To assess ICT skills, participants were given a series of problem scenarios and asked to find solutions to them using ICT-based applications such as an Internet browser and web pages, email, word processing, and spreadsheet tools. Often, solving the tasks required a combination of several applications, for example, managing requests to reserve a meeting room using a web-based reservation system and sending out e-mails to decline reservation requests that could not be accommodated. Previous studies instead had to rely on measures of ICT skills reported by survey participants themselves, which are prone to substantial mismeasurement.
Estimating causal wage effects of ICT skills
The major empirical challenge was to assess whether any estimated association between ICT skills and wages indeed depicts a causal effect. A prime concern is that more able people may be more likely to accumulate ICT skills and may independently receive higher earnings because of their higher ability. Also, better jobs may more likely use and reinforce skills or they may provide the resources to invest in adult education, training, or computer courses. These are only two examples of the potential problems in a simple estimation of the wage returns to ICT skills. To address these problems, we exploit technological peculiarities in the broadband Internet infrastructure, because of which some persons got access to high-speed Internet earlier than others simply by chance. We show that having (early) Internet access kickstarts learning-by-doing in ICT skills.
In our first empirical strategy (cross-country strategy), we exploit the fact that existing voice-telephony networks were upgraded in most countries to provide broadband access. In consequence, countries with better-developed voice-telephony networks before the introduction of broadband in the late 1990s or early 2000s could roll out broadband faster than countries lagging behind in voice-telephony infrastructure. Our second empirical strategy (within-country strategy) makes use of the fact that some German municipalities, by chance, are too far away from the necessary infrastructure to have early access to broadband. We observe that people who were lucky enough to have early broadband access, because they live either in a country with a well-developed traditional voice-telephony network or close enough to the appropriate infrastructure in Germany, have higher ICT skills than their unlucky counterparts, but do not differ from them in their numeracy or literacy skills, which were also tested in PIAAC.
To address the potential concern in the cross-country strategy that richer and more pro-ductive countries have better-developed voice-telephony networks as well as higher wages and more skilled workers, we exploit that different age cohorts were differently affected by broadband. In particular, Figure 1 shows that the age pattern in the impact of broadband availability (determined by the initial voice-telephony networks) on ICT skills is inverted-U-shaped. The youngest cohorts in PIAAC (aged 16-34 years) were toddlers or still at school when broadband emerged; the oldest cohorts (45 years and above) were already reluctant to use the new technology when it emerged. The middle cohorts (aged 35-44 years), however, entered the labour market or started university in the early years of the Internet era. Thus, these persons had a lot of reason to use the Internet, so learning-by-doing should be strongest for them. Survey evidence shows a similar age pattern for time spent on the Internet, suggesting that age is indeed a suitable proxy for the take-up intensity of the Internet and therefore for the strength of the learning-by-doing channel. To use this age pattern, we instrument ICT skills with an interaction between the extent of a country's initial voice-telephony networks (determining the timing of introduction and diffusion of broadband) and age cohorts (determining the intensity of use of broadband). Doing so, we combine instrumental-variable (IV) estimation with a difference-in-differences strategy, effectively identifying returns to ICT skills based on differences in ICT skills and wages between age cohorts within countries.
Notes: Coefficient estimates on fixed-line voice-telephony diffusion (in 1996) for indicated five-year age groups in a regression of ICT skills (standardized to standard deviation 1 across countries) on fixed-line diffusion. Regression weighted by sampling weights (giving same weight to each country). Sample:employees, no first-generation immigrants. Solid lines show average effect of fixed-line diffusion on ICT skills by age groups (separately for ages 16-19, 20-34, 35-44, 45-54, and 55-65). Data sources: ITU, PIAAC.
The reliance of broadband rollout on traditional voice-telephony networks also led to an uneven distribution of broadband Internet access within countries in the early years of the Internet era, which we exploit in the within-country strategy. Specifically, in West Germany, the general structure of the voice-telephony network dates back to the 1960s when the provision of telephone service was a state monopoly having the declared goal of providing universal telephone service to all German households. While all households connected to a main distribution frame (MDF) enjoyed voice-telephony services in the same quality, only those households closer than 4,200 meters (2.6 miles) to their assigned MDF could gain access to broadband Internet. Past this threshold, broadband access was no longer feasible due to signal attenuation, and parts of the copper wire had to be replaced with fiber wire to provide broadband access. Since this replacement involved costly earthworks that increased with the length of the bypass, certain West German municipalities were effectively excluded from early broadband Internet access.
Modern labour markets substantially reward ICT skills
In the cross-country analysis, one additional point in PIAAC's ICT skills assessment leads to roughly a 0.6 percent wage increase. This implies that if an average worker in the United States — a country with ICT skills below the OECD-average (see Figure 2) — increased her ICT skills to the level of an average worker in Japan — the top-performing country — her wages would increase by about 8 percent. This effect size is similar to the wage effect of an additional year of schooling in developed countries. Strikingly, estimated returns to ICT skills are very similar in our within-country strategy, exploiting variation in ICT skills at a fine geographic scale between German municipalities. Comparing our results for the returns to ICT skills to those on PIAAC’s numeracy and literacy skills in Hanushek et al. (2015), we find at least suggestive evidence that ICT skills are somewhat more valued in the labour market than other, more traditional cognitive skills.
Testing the validity of the identification strategies
A series of validity checks and robustness tests bolster confidence in our IV strategies. First, we show that our instruments do not predict the ICT skills of first-generation immi-grants, who are unlikely to have acquired ICT skills in the PIAAC test country. Nor are the instruments associated with any appreciable changes in numeracy or literacy skills, which we consider strong evidence that our identification strategies isolate the effect of ICT skills (vis-à-vis generic skills or general ability) on wages. With the latter placebo test, we address DiNardo and Pischke’s (1997) concern that computer users have unobserved skills that might have little to do with computers per se but that increase their productivity.
We also provide a careful assessment of the exclusion restriction of our IV approach that exogenous broadband availability affects today’s wages only through individuals’ ICT skills, and not directly in any other way. In fact, we account for the two most prominent channels of direct productivity effects of broadband: the introduction of online job search channels, which increases the quality of job matching, and that broadband is directly productive for firms, affecting wages of workers irrespective of their own usage of ICT skills. We also show that households in Germany without broadband Internet access do not selectively relocate to regions where broadband is available.
What explains the positive wage effects of ICT skills? A task-based approach
A unique feature of the PIAAC survey is that it combines individual-level information on ICT skills, wages, and detailed occupation in a single dataset. This allows us to find a potential mechanism behind the positive returns to ICT skills, namely, that the proliferation of personal computers caused a shift away from routine tasks — that is, those more amenable to automatization — toward problem-solving and complex communication tasks (typically called ‘nonroutine abstract tasks’). We expect that the complementarity of computers (requiring ICT skills) and abstract tasks allows workers with high ICT skills to select into abstract jobs and to benefit from the wage premia these jobs pay. Indeed, we find that higher ICT skills increase the abstract task content of jobs and decrease their routine task content. Back-of-the envelope calculations suggest that occupational selection explains a significant portion of the wage increase caused by higher ICT skills.
By showing that ICT skills are rewarded quite substantially in the labour market, our results support Neelie Kroes’s notion of ICT skills as ‘the new literacy’.2 Since ICT skills are likely to become increasingly important in our increasingly technology-rich world, their acquisition (or absence) has serious implications for individual labour market success and overall inequality. Thus, the rather lackluster performance of some European countries in the PIAAC assessment of ICT skills is all the more worrisome — in particular, the post-communist countries and Ireland constitute the bottom of the international league tables (see Figure 2). However, we show that ensuring access to the Internet so as to kick-start learning-by-doing is an effective way of reducing the wage gap between ‘digital natives’ (those who are capable of using modern ICT tools to get along in a digital world) and digital illiterates. This is especially important for the discussion about social inequality in access to the Internet, also known as the ‘digital divide’. The fundamental insight of this paper — that ICT skills can be promoted by providing access to ICT infrastructure — suggests that the efforts by policy-makers worldwide to expand broadband coverage (e.g., the EU’s ‘Digital Agenda’) may prevent a drifting apart in employment opportunities when advances in ICT change job demands.
1. Falck, Oliver, Alexandra Heimisch, and Simon Wiederhold (2016): ‘Returns to ICT Skills’, CESifo Working Paper No. 5720.
Available for download at: https://www.cesifo-group.de/ifoHome/publications/working-papers/CESifoWP/CESifoWPdetails?wp_id=19185466
2. http://getonlineweek.eu/vice-president-neelie-kroes-says-digital-literacy-and-e-skills-are-the-new-literacy/ (assessed September 19, 2016).
Di Nardo, John E and Joern-Steffen Pischke (1997): ‘The Returns to Computer Use Revisited: Have Pencils Changed the Wage Structure Too’ Quarterly Journal of Economics, 112(1): 291-303.
Hanushek, Eric A, Guido Schwerdt, Simon Wiederhold, and Ludger Woessmann (2015): ‘Returns to Skills Around the World: Evidence from PIAAC.’ European Economic Review, 73(C): 103-130.