My last post (here) quantified a mean historical growth line for the S&P Composite Index (currently represented by the S&P 500) and how to measure the market level relative to that line. As you can see in Figure 1, the market level varies significantly from the mean growth line most of the time. The market’s level at the end of 2019 was 21% above the historical mean, about as high as just prior to the Great Recession, but not nearly as high as at other times. The ride has been wild since then: getting even higher at 24% in January; falling to -0.4% in March; and rising to 3.5% in April. The fall of course was due to Covid-19 infection fears and government lockdowns, while the rise appears to be predicated on the assumption of a V-shaped economic recovery. Is this likely?
Figure 1 April 2020 Real Total Value About The Historical Mean Growth Line
Referencing daily closing prices, the S&P 500 index dropped to 2,237 in March from the February high of 3,386. This is a drawdown of 34% as seen in Figure 2. The recovery line in the figure measures how far the price has “recovered” to the prior peak. 100% means the price is all the way back. Recovery back to the February high shows some failed attempts in March, but as of early May appears to be on a steady climb out. The unemployment rate was 3.5% in February and trending flat. It rose to 14.7% in April, and the Congressional Budget Office (CBO) predicts quarterly average unemployment will peak at 16% in the 3rd quarter of this year. Historically, unemployment peaks about 6-1/2 months after the S&P Composite Index price buckets, so a 3rd quarter peak is consistent with a March bottom in S&P 500 Composite Index. Add-in that earnings estimates from Capital IQ and Factset project 2021 trailing twelve month (TTM) earnings will exceed 2019’s, and the V-shaped recovery looks plausible.
Figure 2 Covid-19 Daily Price Drawdown & Recovery To Date
What the V-shaped recovery narrative does not recognize is that the very high current level of unemployment is the forcing function leading to reductions in revenue and earnings in the future, and reduced revenue and earnings have historically led to higher unemployment rates. As the lockdown’s are lifted, some people will return to their jobs, if their jobs still exist. Other people will resist returning to work out of fear of infection. Earnings will be negatively impacted by revenue reductions, e.g., for quite a while at least, airlines and restaurants won’t be the revenue machines they were just a few months ago. Earnings will be further reduced by the debt loads corporations and small businesses have been forced to take on. Some businesses won’t survive. We won’t see for a while what the level of unemployment becomes as a result, but we can use history as a guide to see how long this process takes to play out and what the impact could be on market price.
Figure 3 is a graph similar to Figure 2, but for the Great Recession. The shock to the system then was the collapse of the housing market and its connections within the financial system. S&P 500 price drew down 55%, and it took 18 months to get there. The market made multiple attempts at recovery, some very large (76%) late in 2007 and early in 2008, but the market didn’t bottom until March of 2009. The unemployment rate peaked at 10% seven months later, and S&P 500 price finally re-achieved its prior’s peak in 2012. The price recovery took 4.8 years.
Figure 3 Great Recession Daily Price Drawdown & Recovery
Taking one more step back in time, Figure 4 is the same type of graph for the early 2000’s recession. There were two shocks to the system then, the popping of the internet-driven price bubble and 9/11. Hopefully, we won’t have a second shock this time, but Covid-19 is a continuing depressant in lieu of herd immunity or a vaccine. S&P 500 price drew down 48% in October of 2002, and it took 30 months to get there. As above, the market made multiple significant (54%) attempts at recovery early on. The unemployment rate peaked at 6.3% eight months after peak drawdown, and S&P 500 price finally re-achieved its prior peak in 2006. Price recovery took 6 years.
Figure 4 Internet Bubble Burst & 9/11 Price Drawdown & Recovery
A common thread in these graphs is the failed market recoveries soon after the shock that triggered the recession. It appears to take a while to reset market participant’s expectations to the emerging reality. Another common thread is that both prior recoveries took double-digit number of months to reach the worst S&P 500 price drawdowns, and then multiple years for the price to achieve its prior peak. Are these a defining characteristics of periods with significant increases in unemployment?
To answer this question, I took the monthly average price data documented by Professor Shiller at Yale University and unemployment rate data from the Bureau of Labor Statistics and created Figure 5. This figure documents the change in average monthly price during the twelve periods of significant increase in unemployment over the last 91 years. The dashed lines on the right of the graph show the Congressional Budget Office’s (CBO) unemployment rate projections for the rest of this year and 2021. CBO projects an average of 10.1% in 2021 after peaking at 16% this year.
Figure 5 Historical Unemployment, Price Drawdown & Peak-to-Peak Price Duration
Figure 6 is a scatterplot of the change in average monthly price from the peak to trough versus the change in unemployment in each of the twelve high unemployment periods. There is a clear trend of larger drop in price as the change in unemployment rate gets larger. In fact, the regression line is nearly the same even if you exclude the rightmost point for the Great Depression. Assuming it takes the rest of this year for the economic damage caused by Covid-19 fear and forced unemployment to manifest itself in reduced revenues, earnings and the resulting unemployment, the CBO’s 10.1% projection for 2021 represents a 6.6% change from early this year. This leads to a 20% to 50% reduction from the peak average monthly price early this year of 3,260, or 1,630 to 2,608. The regression average is a 36% reduction yielding 2,086, probably towards the end of this year since unemployment rate lags price by about 6-1/2 months. The S&P 500 opened at 2,827 on May 15th, so the historical record suggests a significant reduction is likely.
Figure 6 Average Monthly Price versus Unemployment Rate Scatterplot & Regression
Figure 7 plots the number of months for recovery to the prior monthly average price peak versus peak to trough change in monthly average price. A quadratic regression fits the data well and highlights the unsurprising result that deep price drawdowns lead to nonlinear price recovery durations. A 36% average monthly price reduction corresponds to a 60 month recovery duration. This is at odds with a V-Shaped recovery occurring next year.
Figure 7 Duration of Average Monthly Price Drawdown versus Maximum Price Drawdown
The levels of unemployment that we are seeing now and that are projected by the CBO rival the highest levels on record. Lessons-learned since the 1930’s were employed in the last two recessions to lessen the impact on the economy, yet these two recessions led to the second and fourth worst price drawdowns. The current administration and the Federal Reserve have taken significant fiscal and monetary steps to mitigate the impact of Covid-19 on the economy and unemployment. These steps go well beyond those taken during the last two recessions, but damage to the economy has already occurred, and the threat of Covid-19 infections will continue to be a drag on the economy until we get to herd immunity or a vaccine. We should expect a market level drawdown and market churn for the next few years.
References:
Earnings projections from Capital IQ and Factset
Projections from Congressional Budget Office (CBO)
Historical price data
Historical unemployment rate data here and here
