Unlock Editor’s Digest for free
Roula Khalaf, editor-in-chief of the FT, selects her favorite stories in this weekly newsletter.
Anyone who has taken “Econ 101” knows that most economic models assume that decision makers have perfect information. In reality, this is rarely the case. The data may be incomplete, one party may have more details than the other, or everyone may be equally misinformed. Indeed, official national statistics – which are vital for policymakers, businesses and investors – can sometimes miss the mark. Currently, many countries experience particular difficulty in counting the number of people who are working.
The Resolution Foundation think tank recently claimed that the UK’s Office for National Statistics may have lost almost a million workers in terms of employment since 2019. It estimates that the statistical body could have significantly overestimated the scale of worker inactivity. If true, it would undermine the “vanishing workers” narrative that has been the basis of many of the UK government’s policy commitments as well as its central bank’s interest rate decisions.
There is a puzzle in America too. Job creation has stagnated over the past two years, according to its household survey. However, non-agricultural employment figures, compiled from a business survey, show continued employment growth. The data has also been volatile and subject to heavy revisions. In August The Department of Labor said the U.S. economy created more than 800,000 fewer NFP jobs than initially reported in the year ended in March.
One reason for this uncertainty could be that over the past decade, response rates to data questionnaires in the United Kingdom, the United States and the European Union have followed a downward trend, exacerbated by the pandemic. Lockdowns have also disrupted projections of population, immigration and business births and deaths, which help statisticians aggregate employment data from survey samples. This has led to bias, poor estimates and conflicting accounts from different data sources.
Worker counting is also a problem in the developing worldbut for more systemic reasons. Estimating unemployment in India has long been a challenge given that a significant portion of the population works in the informal sector. In China, opacity constitutes another limitation.
Bad jobs data leads to bad decisions. Employment figures underpin choices on taxation, spending and welfare, and are at the heart of monetary policymakers’ assessments of the health of the economy. Companies use it in hiring and salary decisions. Investors also rely on it. US NFP figures determine the pricing of interest rates in global financial markets.
What can we do? Governments should ensure that funding keeps pace with the demand for more, newer and more accurate data. In June, The U.S. Bureau of Labor Statistics said budget cuts could force it to reduce the sample size of its household survey. Data enthusiasts are also attracted to higher-paying tech companies. Authorities should also do more to hold statisticians accountable. The ONS has been particularly slow to act responses decreasingand move to online surveys first.
Even with incentives or better survey design, declining response rates can be difficult to reverse. A few studies suggest that people are fatigued by too many questionnaires. Regardless, it will be important to collect labor market data from other sources. Statistical agencies should partner more with the private sector – including job sites such as Indeed and LinkedIn – to obtain real-time statistics to support their estimates. Governments also need to share administrative figures more quickly with data agencies. In some countries, national ID cards have helped agencies better understand population and workforce data.
National statistical agencies need to better count the number of jobs. More resources, effective monitoring and wider access to other data sources will not make the numbers perfect. But at least it will give a better idea of how imperfect they really are.
