What do the following two graphs have in common?
Points for the best answer: they both have a blue line, they both demonstrate limited finesse on Excel, they both don’t have labels on their axes? You’re all correct, but the true problem hidden beneath these graphs is far bigger. In fact, these two graphs represent the exact same economic data. In the last few days I have seen both of these graphs used in their varying colours and camouflage to depict the status of US employment. The chart on the left is a misleading depiction of the US jobs market, the right a more realistic interpretation of the impact of the virus on the US economy. I’ve picked up on this sort of reporting before, resetting the balance of journalism one daily brief at a time (or not!). And, as usual, the culprits shall remain anonymous. All I’m saying is that the chart on the left Can Not Be Considered correct.
In the United States there is a popular statistic published monthly that demonstrates the number of jobs added of lost in the United States in that given month as a result of payrolls added or lost. It offers an accurate snapshot of the economy with the headline not concerning itself with the level of unemployment (which encourages estimations of the size of the underlying labour force). Farming payrolls are not considered due to the highly seasonal nature of this type of employment which would, if included, have a tendency to distort the statistics based upon weather and agricultural trends. If interpreted properly therefore, the non-farm payroll statistic can be a powerful tool.
With so much talk of V- U- L- shaped recoveries, a chart like the one on the left showing the marginal change in non-farm payrolls in a given month would lead the observer to conclude that the downturn in employment as a result of the coronavirus has concluded and the economy is back (or even higher) than where it left off. When we aggregate that data and compare the change in the total number of jobs in the exact same period we end up with the graph on the right – one that reliably informs us that the US economy still has 20 million less jobs that it did 3 months’ ago.
There have seldom been more important times for the public to have a clear and transparent picture of health and economic statistics. Perhaps even the stock market is believing this kind of data. Following a 1.2% rally yesterday in the US stock index the S&P 500 we can now claim stock valuations have clawed back their 2020 losses to now trade flat on the year. These gains in no small part have been from Friday’s non-farm payroll data. The lion’s share of the move I concede has been central bank action but the continued exaggeration between fundamentals and prices within stocks this week is alarming.
A final note on the data. The reason why the increase in jobs reported in the US on Friday had an important impact was because the market had been anticipating a decline of around 8 million, adding to the 20.6 million loss registered for the month of April. If the decline in US jobs has abated, as the data appears to show, then quite rightly the market might scale up its expectations of economic recovery. That doesn’t change the misleading nature of the graph. Within the report released last week was also a note signalling a potential misclassification and error in the collection of the data. The caveat has spurred discussions of a potential political influence in the statistic released and at best undermines the accuracy of this months’ data. All the more reason to be cautious in our depiction of the data.
Discussion and Analysis by Charles Porter