Task content of jobs and inequality

Info: 

In this research project, we focus on how new technologies affected wage inequality within-occupations. By doing so, we  contribute to the literature on skill-biased technological change in two ways: first, we incorporate a dynamic dimension to the analysis by including measures on the scope and pace of changes in task content. Second, we shift our interest from differences across occupations to inequality within occupations.

Both changes are relevant from a policy perspective. Analyzing dynamics help us to understand how occupations evovle, and how workers cope with those changes. Focusing on inequality within occupations, we can understand better who were the winners and losers from the adoption of new technologies, and possibly consider what policy instruments could ameliorate the negative consequences of technological progress.

We test two hypothesis:

Changes in the task composition towards more complex tasks were followed by increases in relative wages.

Occupations that experienced larger changes in their tasks contents present higher wage inequalities.

Budget: 

Źródło finansowania | financingNarodowe Centrum Nauki, PRELUDIUM 11

Projekt realizowany | Timeline: 06/2017 – 06/2020

Budżet łączny | Total budget: 96 454 zł

  • wynagrodzenia dla podstawowych wykonawców | compensation to researchers: 24 000 zł
  • komputery i oprogramowanie | hardware and software: 15 850 zł
  • konferencje i inne wyjazdy | conferences: 16 500 zł
  • Dane |  Data: 12 270 zł
  • książki | books: 3 000 zł
  • materiały | consumables: 2 400 zł
  • koszty pośrednie dla Wydziału Nauk Ekonomicznych | overheads for the Faculty of Economics: 6 717 zł
  • koszty pośrednie dla Uniwersytetu Warszawskiego | overheads for the University of Warsaw: 6 717 zł
Purpose: 

Objectives | Cel badania

In English We analyze the consequences of changes in the task content of jobs on wages \emph{within} occupations. In this project, we formulate two hypotheses: (1) occupations where the share of more complex (defined as nonroutine cognitive) tasks increased, experienced a larger increase in wages; (2) within occupation wage inequalities are larger in those occupations that experienced more changes in the task content.

Po polsku Zbadamy i przeanalizujemy jak zmiany w zakresie zadań wykonywanych przez pracowników oddziałują na płace wewnątrz grup zawodowych. Sformułowane zostały dwie hipotezy: (1) zawody, w których liczba skomplikowanych (zdefiniowanych jako nierutynowe poznawcze) zadań wzrosła, doświadczają większych zmian w płacach; (2) wewnątrz grup zawodowych nierówności płacowe są większe w tych zawodach, w których w większym stopniu zmienił się zakres zadań.


Method | Metodyka

In English We work with microeconomic data sources that allow to recover workers' occupations and earnings. Additionally, we will employ the DOT and O*NET databases, to recover the task content of occupations. Using these databases, we perform the following tasks.

  • Provide a descriptive analysis of changes in the demand for skills in the context of transition and developed countries
  • Identify the occupations that changed the most over the period and the direction of these changes. We will extend the existing analysis by developing a synthetic measure of the size and scope of these changes.
  • Correlate the changes in the task composition of occupations to changes in relative wages.

Po polsku Projekt ma charakter empiryczny. Wykorzystane zostaną dane mikroekonomiczne w calu pozyskania informacji o zawodach i płacach. Dodatkowo, korzystać będziemy z baz DOT and O*NET, które zawierają informacje o zadaniach wykonywanych w ramach zawodów. Na podstawie baz danych:

  • przygotowana zostanie analiza deskryptywna zmian w popycie na umiejętności w krajach transformacyjnych oraz rozwiniętych.
  • zidentyfikowane zostaną zawody, w których zmiany w zakresie zadań były największe oraz kierunki tych zmian.
  • połączone zostaną zmiany w zakresie zadań w zawodzie ze zmianami w relatywnych płacach.

Intended contribution | Planowane efekty

In English  Our project builds on the existing literature in two dimenstions. First, we focus on the dynamic aspect of technological change. The development of an index of change is an important part of this task and a valuable sub-product of our research that can be used to explore connected topics. Second, our analysis will focus on wage inequality within occupation, a phenomenon that was mostly disregarded in the existing literature.

Understanding more on the sources of wage inequality (worker or firm heterogeneity) could inform on where the main bottlenecks are and how they can be addressed. Moreover, it can inform on which workers experienced/will experience more distress, and inform on how to target them better.

Po polsku Nasz projekt przyczyni się do rozwoju literatury w dwóch aspektach. Po pierwsze, wypełniona zostanie luka dotycząca dynamicznego aspektu zmian technologicznych do tej pory pomijanego w badaniach. Stworzony zostanie wskaźnik zmian, który będzie mógł być wykorzystywany przez badaczy do dalszych prac nad poruszanym tematem. Po drugie, nasza analiza koncentruje się na nierównościach dochodowych wewnątrz grup zawodowych - kolejny temat dotychczas niepojawiający się szerzej w literaturze.

Większe zrozumienie źródeł nierówności dochodowych może pomóc wskazać, gdzie znajdują się największe problemy i jak z nimi walczyć. Co więcej, również pracownicy i pracodawcy bardziej świadomi z czego wynika ich pozycja na rynku i w rozkładzie płac, mogą wykorzystać tę wiedzę na swoją korzyść.

Opublikowane | Published

  • A cautionary note on the reliability of the online survey data – the case of Wage Indicator | Sociological Methods and Research

    We investigate the reliability of data from the Wage Indicator (WI), the largest online survey on earnings and working conditions. Comparing WI to nationally representative data sources for 17 countries reveals that participants of WI are not likely to have been representatively drawn from the respective populations. Previous literature has proposed to utilize weights based on inverse propensity scores, but this procedure was shown to leave reweighted WI samples different from the benchmark nationally representative data. We propose a novel procedure, building on covariate balancing propensity score, which achieves complete reweighting of the WI data, making it able to replicate the structure of nationally representative samples on observable characteristics. While rebalancing assures the match between WI and representative benchmark data sources, we show that the wage schedules remain different for a large group of countries. Using the example of a Mincerian wage regression, we find that in more than a third of the cases, our proposed novel reweighting assures that estimates obtained on WI data are not biased relative to nationally representative data. However, in the remaining 60% of the analyzed 95 datasets systematic differences in the estimated coefficients of the Mincerian wage regression between WI and nationally representative data persists even after reweighting. We provide some intuition about the reasons behind these biases. Notably, objective factors such as access to the Internet or richness appear to matter, but self-selection (on unobservable characteristics) among WI participants appears to constitute an important source of bias.

    We provide weights and full documentation here.

    Magdalena
    Smyk-Szymańska
  • Within occupation wage dispersion and the task content of jobs | Oxford Bulletin of Economics and Statistics

    Most analyses linking task content of jobs to income inequality focus on the effects between occupations, e.g. the growing dispersion between lousy and lovely jobs. Theory, meanwhile, provides insights on links between task content of jobs and inequality also within occupations: models predict compression of wages in more routine jobs, that is those where capital is a direct substitute for labor, and an increase in dispersion in jobs where capital and labor are complements. I document that within occupations dispersion of wages is empirically relevant, as it represents around half of total wage inequality across Europe. I then link wage inequality to the task content of jobs. Using matched employee-employer data from Europe for the period 2002-2014, I show that occupations where tasks complement newer technologies exhibit higher wage dispersion. This relationship is robust to adjusting for a variety of confounding and mitigating channels.

     

    Replication files

    Lucas
    van der Velde

W toku | Work in progress

  • Phasing out: routine tasks and retirement

    Population ageing poses new challenges to the sustainability of the pension system and possibly to economic growth in advanced economies. In such context, calls are made to increase participation of workers close to their retirement age. Ageing occurs in a period where technological progress has changed the patterns of labor demand, away from physically demanding tasks (opportunity) and into more cognitive-interpersonal type of tasks (challenge). To understand the net effect, we analyze the relation between automation and labor supply of older workers. We explore whether exposure to technological change, measured by the task content of jobs, was connected to labor supply of older workers in Germany and Great Britain. Using panel data, we show that the adjustment in the number of hours of workers in occupations exposed to automation was small, and only negative for a subset of workers. The exposure to automation is related to somehow earlier retirement, but the size of the relation is small.

    Lucas
    van der Velde

The first stage of the research project involved the creation of comparable measures of inequality for different occupations. To obtain these measures, I employed matched employee-employer data from the European Union, i.e. the Structure of Earnings Survey (EU-SES). While to a large extent the data are comparable across countries, some differences persist. The files presented here help to obtain consistent definitions of the main variables of interest across databases. They are the first step for the computation of different inequality measures. 

 

The files below contain the next step of the analysis: measures of inequality within occupations. These measures are obtained using the databases harmonized with the codes presented earlier. The .zip file contains both the code used to obtain those measures of inequality, a database with the measures themselves and a read me file.

 

Please contact by email if you find any mistakes / errors when using these files.

Did you ever find in need to measure how far apart two objects are? Regardless of whether one considers task content or other variables, measures of distance help to illustrate both how distant two or more observations are and how they evolve over time. Of particular interest is the index proposed by Walesiak (1999 - and similar). The author extends measures of distance to cover variables that are at least ordinal in nature, as it is the case with most task content variables.

I developed a STATA package that estimates this index. The package grants users flexibility in setting up their distance measures. Users can choose to weight different variables, to compare observations against a pre-specified pattern or to obtain an average measure of distance to all other observations. The program files, help and an example database are available for download here.

  • distord -  a package to estimate distance between observations

One central issue of the task content literature is how workers decide which tasks they perform at their workplace. The main vision is that workers specialize in the tasks where they have a comparative advantage. Workers who are more productive in non-routine interpersonal tasks would spend more time in doing those tasks than other workers.

This approach assumes that there is an agent that combines these tasks in a way that maximizes output. What would happen if instead one assumes that the process is decentralized and occupations require a task profile? In that case, workers might choose to either provide tasks themselves or specialize, with the catch that specialization requires finding an adequate partner and leads to coordination costs. Workers need to balance the efficiency gains from specialization against these costs. 

In the slides below, I present a toy model and initial empirical findings based on PIAAC and Eurostat data.