By Ana Beatriz Galvão
Economists often argue that differences in living standards across countries arise to a large extent from differences in productivity across countries. In other words, individuals living in country A will be on average richer than individuals living in country B if the amount of output produced for each hour worked in country A is larger than in country B. Hence there is significant interest in cross-country comparisons of productivity. The Office for National Statistics (ONS) used to publish such productivity comparisons.[1] These provided an estimate of how the labour productivity of other G7 countries compared to UK labour productivity. Using the published estimates, one might say that country A is X% more productive than the UK, or that country B is X% less productive than the UK etc.
There is, however, a problem with this direct comparison. The way the statistical office in country A computes the number of hours worked required for the productivity estimate may differ from the way the ONS and the statistical office in country B compile their figures. Statistical offices need to apply adjustments to estimates of the number of hours worked from labour surveys to ensure that the number of hours worked is compatible with the total output being measured using GDP. The methods that are best suited to this purpose may differ across countries.
In a research project funded by the ONS as part of the research programme of the ESCoE, I show how the ONS can effectively communicate the methodological uncertainty behind their estimates of international comparisons of productivity (see Galvão, 2022). We designed and implemented a randomised online experiment to evaluate the effect of seven different international comparisons of productivity communication tools on the UK public’s perception of productivity comparisons and their perception of the ONS. For the control group, we consider a communication tool that disregards the fact that the methods for computing hours differ across countries. For the other six groups, we show different ways of communicating the variety of estimates of the G7 countries productivity compared to the UK. They include visualisation tools such as the dot plot in Figure 1, where dots represent the variety of estimates available for how large the percentage difference is between country X productivity and the UK, and textual descriptions.
The findings reported in Galvao (2022) suggest that the specially designed communication tools can improve the UK public’s assessment of how productivity estimates compare between the UK and G7 countries. There is no evidence that the new tools conveying the uncertainty of productivity estimates affect respondents’ trust in the ONS. The communication tools are effective even for those respondents who have no knowledge of what productivity is. Still, they are likely to “be more helpful to members of the public that are familiar with the concept as they are, in general, better at making inference based on the communicated data.” (Galvao, 2022). By choosing a specific measure of productivity for international comparisons, a statistical office may provide a false sense of certainty that country A’s productivity is indeed higher (or lower) than UK productivity. My research supports the publication of the range of productivity estimates, aided by the publication of adequate supporting text and visualisation tools. Based on a variety of productivity estimates, users would then decide whether there is indeed enough evidence to support the description of country A as being more (or less) productive than the UK.
[1] See for example: https://www.ons.gov.uk/economy/economicoutputandproductivity/productivitymeasures/bulletins/internationalcomparisonsofproductivityfinalestimates/2014
Read the full ESCoE Discussion Paper here.
Ana Beatriz Galvão is Professor of Economic Modelling and Forecasting at Warwick Business School, University of Warwick, and an ESCoE Research Associate.
ESCoE blogs are published to further debate. Any views expressed are solely those of the author(s) and so cannot be taken to represent those of the ESCoE, its partner institutions or the Office for National Statistics.