An unconventional approach to data in unconventional times
The global economic system is undergoing an unparalleled shock. Unlike previous recessions, the economic contraction expected in many countries has, to a large extent, been purposefully imposed in order to save lives and prevent the overload of healthcare systems due to the spread of Covid-19. These necessary measures have clear consequences, not only regarding the economy itself, but also regarding the availability and usefulness of economic data on which economists typically rely. In the short term, these standard sets of data may be less informative than in normal times. At the same time, we don’t necessarily need these figures to tell us that the economy is drastically slowing or even contracting. This is not to say that data is no longer important. Standard economic data may not be able to tell us how long confinement measures will last, but they can still help us assess how well positioned an economy might be to experience a strong recovery after this shock subsides. If government measures are able to keep individuals and companies above water during this crisis, the eventual economic recovery will be far stronger.
The importance of data
Under normal circumstances, economists, analysts and especially policymakers rely on data to understand the state of the economy and how it can be expected to develop. Current circumstances, however, are anything but normal. Typically, the hope is that economic data will provide some sense of when a business cycle will peak or trough and of how strong or weak the intermittent expansions and downturns will be. Some data series can also point out risks that might derail a recovery or prolong a recession. And in the case of central bankers, data is crucial to understanding whether monetary policy is successfully transmitting to the real economy or if different approaches need to be taken. Gross Domestic Product (GDP) gives the most complete and concise view of a country’s economic state, but it is only available on a quarterly basis and with a significant lag. Labour market data, such as unemployment and payrolls, are also important and are usually available on a more frequent, but still not immediate, monthly basis. Other useful monthly data economists generally look at to gain a sense of the real economy include, inter alia, business surveys, industrial production, car registrations, and retail sales.
Problems on the horizon
Lagged data has always been a problem for economists, and forecasting techniques known as nowcasting try to solve this problem by using more frequent data to derive early estimates of indicators like GDP. The problems associated with lagged economic indicators are more glaring in today’s unprecedented crisis. The economic environment has radically changed within a matter of weeks (if not days). Most governments have introduced important health measures that roughly amount to a sudden stop of the economy. This means that the economic performances in Europe and the US throughout January and even most of February are of little importance given what we know now.
Furthermore, as monthly data that do reflect the drastic recent developments are made available, it becomes clear that the unprecedented nature of this shock will make it more difficult to pull economic signals from that data. In China, for example, the brunt of its economic shutdown happened in February, and the economy has since started to reopen. The broad business closures and stay-at-home orders during that month are clearly reflected in business surveys (PMIs), with the NBS manufacturing PMI dropping from 50.0 in January to a record low of 35.7 in February (where anything below 50 suggests a monthly contraction). The index then jumped back up to 52 in March – a fairly strong signal during normal times. However, such surveys ask how the business environment compares to the prior month, so an expansion in March compared to the sudden stop in February was almost inevitable. The March PMI doesn’t give us a sense of the strength of the Chinese recovery. It only tells us that things are not as bad as they were in February – a conclusion that could already be drawn based on the fact that businesses have started to reopen.
Another place where we may face new data-related difficulties is with inflation. Inflation, which is typically measured as the increase in a country’s consumer price index (CPI), relies on constructing the CPI each month, which in turn relies on measuring prices of a fixed basket of goods and services. That basket is generally constructed to reflect household consumption. In today’s strange new environment, that basket likely looks very different than it did two months ago. In most places, non-essential retailers have been ordered to close their businesses. There are a wide range of goods and services that suddenly no households are purchasing (think new clothes, haircuts, meals at restaurants, etc.). If it is even possible to measure these price changes given the closure of so many businesses, the average consumer’s basket of consumption will look far different than what the statistical offices usually measure. This will cloud policymakers’ ability to follow inflation developments in the coming months, which is a significant disadvantage in a world where an oil price war (disinflationary), an unprecedented drop in global demand (disinflationary), and a sudden supply shock (inflationary) are all in play at the same time.
One might be tempted to argue that in a recession caused by a non-economic shock, economic data become less important, and therefore these lags or disruptions in the data do not really matter. It is true, for example, that we didn’t need to wait for the euro area or US PMIs to tell us that business was bad in March. And scouring the data for signs that the economy is picking back up seems pretty pointless when the economy won’t restart until governments determine that lockdowns can responsibly be lifted. But economic data can still tell us a lot, and can help policymakers make decisions that will help their economies rebound once the lockdowns are lifted (for a discussion of appropriate policy responses to this particular crisis see: KBC Economic Opinion, published 24 March).
Furthermore, some statistics will help signal whether an economy is indeed poised to recover sharply in the future. The main risk, from a real economy perspective, is that the disruptions caused by the health crisis spill over and cause longer-lasting economic disruptions. A sharp rise in corporate bankruptcies, for example, could mean that some portion of the rise in unemployment will not be temporary. In economies with weaker social safety nets, temporary (or permanent) jobs losses that lead to individuals falling behind on rent or mortgage payments could lead to a rise in foreclosures and evictions that make it more difficult for people to get back on their feet once the health crisis has passed.
It is for these reasons that the extraordinary stimulus measures taken by governments and central banks of late to increase unemployment benefits, support businesses, and prop up credit and market liquidity are important. To assess whether the measures are working, or whether more support is needed, timely economic data will be crucial. Statistics on labour markets, consumer credit delinquencies, non-performing loans, credit to SMEs, and corporate bankruptcies will therefore be of particular interest to economists and policymakers in the coming weeks and months. In the US, for example, the PayNet Small Business Delinquencies Index is published on a monthly basis and can therefore give a timelier picture of the extent to which SMEs are struggling compared to quarterly, bank-reported figures (figure 1). If we see figures like this spiking sharply in the coming months, it may suggest even more policy support will be needed.