One thing that has always frustrated (but also fascinated) me throughout my investing career is how dismissive economists are about signals coming from the stock market regarding economic activity. The genesis of this tendency by economists to ignore or otherwise belittle stock market signaling seems to be that equity (which for the layman is another word for "stock") market investors often overreact to apparent fluctuations in economic activity. "Stocks go up and stocks go down," people always seem to say. But while this is sort of true, it really isn't. The stock indexes that most people care about (and that most actually invest in) go up over time, and considerably so. This isn't to say they go straight up with no gyrations, but the point remains the same. Furthermore, most of the time, large stock market fluctuations are reasonably driven by perceptions of changing economic circumstances, not randomness (which has a larger influence on shorter-term price movements).
A more articulate rationale as to why economists may think they're smarter than stock market participants when it comes to forecasting the economy came from MIT Economist Paul Samuelson, who in Newsweek in September of 1966 wittily observed that the stock market has correctly predicted nine out of the last five recessions. Greater knowledge of the "wisdom of crowds" didn't exist as substantively in Samuelson's time as it does today, but this makes it easier to give Samuelson and Co. a pass than it does today's economists, who thanks to the advancements of technology have an easier time than ever incorporating stock market data into their forecasts.
But an important question arises from this: do stock markets over-react to economic fluctuations? Or do economists under-react? Or is it a little of both? Who should we pay more attention to when we're trying to figure out what the heck is going on with the economy?
The Philadelphia Fed gives us very helpful data to analyze how good economists are at predicting recessions by looking at their "Anxious Index." Each quarter since 1968, in the Survey of Professional Forecasters, the Phila Fed asks economists to estimate "the probability of a decline in real GDP in the quarter following the quarter in which the survey is taken" (so next quarter). Here is how that chart looks over time, with the shaded areas being officially designated recessions by the National Bureau of Economic Research (NBER):
Note that for each quarter, the survey is done in that quarter. So for example, the first quarter survey of 2001 was done in February of 2001, the second quarter was done in May, the third in August, etc. etc. Looking at the chart, a few things stand out:
First, a cursory glance at the chart seems to indicate that economists' anxiety about a recession does tend to go up around recessions
Second, a closer look, however, indicates that economists' anxiety usually only spikes when we're in the middle of a recession (see the 1969-70, 1973-75, 1980, and 1981-82 recessions), or, almost just as often, after it's already ended (see the 1991, 2001, 2007-09, and 2020 recessions). Economists getting nervous about a recession only after we're already in one, or worse yet, after we've already exited one, makes their insights and predictions quite a bit less helpful. It also begs the question as to whether economists are getting worse at forecasting recessions as time goes on, but that's another issue for another day (surely things like globalization and greater interconnectedness with other countries' economies may complicate matters, but this would still be particularly strange given economists have more information and more real-time data at their fingertips than ever to help them forecast).
As the chart below shows, in the 8 recessions we've had since 1968, it's quite rare for economists to forecast a recession even after the first quarter (so 3 months) we're already in one. Note that the average anxiety level for professional forecasters since 1968 is just over 19, which indicates that economists are typically forecasting a 20% chance that a recession will start next quarter. It's therefore hard to give them much credit for forecasting the 1990 recession, for example, since forecasters only believed a recession was coming with a 22.5% probability. Bottom line, even if we wanted to be generous and give economists credit for predicting the '81-82 recession (where just before that recession began forecasters thought there was a 67% chance of a recession that next quarter), that still makes them 2 for 8 since 1968. In reality though, they're 1 for 8, which is not very good.
Third, and building on point #2 above, it's actually somewhat rare that you get economists to not only agree that we're actually in a recession even when we're in the middle of one, but even worse, to agree that we've already had one even after it's over. In the 8 recessions we've had since 1968, the maximum the anxious index ever got was 75, and that was in the first quarter of 2009 after the Great Financial Crisis that previous fall and winter. Again, this means that the average economist in the survey said there was a 75% chance of negative real GDP next quarter. Given the recession ended only a few months later (in June of 2009), this was certainly a case of being "better late than never," but it's still pretty shocking that this 75% number wasn't even higher after everything the economy had been through the previous 15 months. Perhaps it's because we are not in recessions far more often than we're in them, or because economists might tend to be "glass-half-full people," but whatever the reason, it affects their ability to forecast recessions. And sure, some recessions are certainly more mild than others (thus in theory making it harder to predict them, or to notice when we're in one), but some (like the Great Recession and the COVID recession) are not mild at all. If economists can't get it right even when economic catastrophe is upon us, it does bring into question whether we should pay any attention to economic forecasts. If even in the worst recession of our lifetimes (2007-09), economists only thought there was a 75% chance we were going to have one, and that was already basically after the fact, that makes it hard to really take them very seriously. Arguably even worse was that even after the country essentially shut down the economy in March of 2020, and the unemployment rate spiked to 14.8% by April, the average economist predicted recession was no more likely than they would have any other time the next time they were asked in May of 2020. For as much flak as weather people get for being wrong, they look awfully good compared to economists.
Now let's look at stock market performance around recessions (and otherwise) to compare how good the market is at predicting recessions (and how often they predict false alarms). Is it really any better? The answer is definitively yes.
Starting with the Bear Market of 1966 (which I used because it was pretty close to the 1969-70 recession), the S&P 500 has declined by 10% or more 31 times, 15% or more 16 times, and 20% or more 14 times. Since there's only been 8 recessions over this period, that means Samuelson was right: the stock market does tend to worry about recessions more than it should. But if economists have been right only 1 out of 8 times, using these indicators as a benchmark for a recession forecast from the stock market, the market has a hit rate of 25.6%, 50%, and 53% when it goes down 10, 15, and 20% respectively. That is considerably better than the track record of economists.
There's a problem with this analysis, however, and that is that the stock market doesn't always see these sizeable declines before the recession starts. Thus, we can't always just say oh, once the market declines 15% or 20% we have a 50% probability of recession. As the table below shows, history demonstrates that the market's timing on sniffing out a recession varies, and that often times much of the greater than 15% or 20% declines we've seen (more than 20% is typically referred to as a "Bear Market") occur during the middle of a recession, not before it. If forecasting the recession is what we're after, then that doesn't really help us as much either.
Consider the below table, which looks at Bear Markets' recessionary predictive power (so greater than 20% declines in the S&P 500). The data unfortunately shows that while Bear Markets usually start before recessions, they usually don't end before the recession does. Thus, things in the market usually continue to get worse as the economy worsens as well. In some cases, they last for a while longer even after the recession has already ended (see the 2000-2002 Bear Market, for example).
Some of these periods require a little bit of context, however. Take the 1980 Bear Market and the accompanying recession, which was declared on June 3rd, 1980 by the NBER to have begun that previous January. While the stock market's decline didn't begin until mid February of 1980, the extremely rapid 20.3% decline in 43 days marked the second fastest 20% drop in the market in the last 60 years, behind only the frightening COVID drop of March 2020 (where it took only 33 days for the market to fall 33.9%). So technically, yes, the market did not sniff out the recession here before it began, but it figured out one was happening very quickly. It gets credit for predicting that recession in my book.
The recession of 1990 proved similarly, where the market peaked in mid July of 1990 and promptly fell 16% over the course of the next 5 weeks. The unemployment rate jumped 0.3 to 5.5% in July (which investors learned about in early August), and fears about a war in the Middle East only compounded matters further that summer as well. The market would eventually bottom not much longer thereafter at 294.51 on October 11th (for a total peak to trough decline of 20.4%), and while a recession was deemed to have begun in July of 1990 that following April (of 1991), the market again figured out almost immediately that recession was upon us. It's hard not to give it credit for that one too.
The COVID recession was also a similar situation. The NBER declared on June 8th, 2020 that the last business cycle peaked in February of 2020 (and thus marked the beginning of a recession). So here again, the business cycle dating committee deemed the recession to begin after the very rapid stock market decline, but also here again, the market again figured out very quickly that some kind of economic decline was coming, rapidly sending down stock prices 34% in 33 days. Given economists were still saying there was less than a 20% chance of a recession in both February and May of 2020, the market deserves credit for this one too. Thus, while technically late in all three of these cases, the market figured out very quickly recession was coming. Being right a month late is far better than never being right at all.
Lastly, I'd also argue that the market deserves credit for sniffing out the 2007-09 recession as well, specifically because it took the NBER's business cycle dating committee 12 months to declare a recession had begun that previous December (of 2007). The market had peaked in early October of 2007, and had fallen almost 7% by year end. By January 22nd of 2008, the market had fallen another 10%, (for a 16.9% peak to trough decline), and by March 10th, the market had declined over 19% (this was a week before Bear Stearns was bailed out on March 16th, 2008). The market would regain some footing in April and May, rising 12% higher by May 16th (compared to the previous October's peak, the market was still down 9.6%, however). From there a rough summer ensued, followed by an even more brutal fall and winter that marked the worst of the "Great Recession." By the time the Business Cycle Dating Committee declared a recession on December 1st, 2008, the S&P had fallen 48%.
Thus, while at first glance it looks like more than half of the instances of the stock market "predicting" recession actually occurred after the recession began, some of this is both a little bit of "missing the forest for the trees," as well as circular reasoning. In four of those cases (which we discussed above), the market either sniffed out a recession immediately or well before either economists or the NBER did. The NBER Business Cycle Committee also has the hindsight of using the stock market's decline as a reference point in dating recessions. Recessions, we should be clear, are always declared after they've already happened. Thus, even if the economic data did truly begin to significantly weaken only a month before the stock market rapidly declined, we ought to again state what we did before: being right a bit late is better than being right several quarters later, or worse yet, never being right at all. Even if you want to only give the market half credit for the four recessions we discussed above, that still gives it a track record far better than that of economists.
If we needed a rough, quantitative proxy then for when to really listen to the stock market, a decline of greater than 10% is likely to be a signal that the economy is weakening. At that point, it's probably time to dig a little deeper and actually listen to what CEOs and CFOs from publicly traded companies are saying.
But this is kind of the whole point of this post, which is that we should always listen to the stock market, and that doesn't just mean following the price action. Following stock price changes is necessary, but not sufficient for understanding what's going on in the economy. Stock prices, far more often than not, are primarily responding to changes in the forecasts of companies' profits and cash flows, and thus, what the management teams of those companies actually say is very important in determining stock price moves, whether through their own "guidance" forecasts or those coming from Wall Street firms. It's therefore always been strange to me why public conference calls and the transcripts of them are not more utilized by either the Fed or economists in general. Isn't the economy ultimately made up of a collection of businesses? If so, why don't we pay more attention to what the biggest, most notable ones are saying?
After all, it's not as if the "macro" economic data that economists live and die from comes out of thin air. That too comes from businesses, only that survey data is "soft" data, which are generally based on voluntary questionnaires mailed out or otherwise given to companies. By contrast, the sales, costs and profits of publicly traded firms is "hard" data. Filling out the government surveys is not required of firms (they do not even have to answer truthfully, let alone correctly), which is at least part of the reason why survey response rates for government data are now so low (trust in government and political polarization is almost assuredly having some impact on this as well, especially amongst business owners who are Republicans). By contrast, publicly traded firms have to file information with the SEC (Securities and Exchange Commission), and it has to be right. Why would we not prioritize that data and the accompanying commentary describing that data from company management teams more than what companies say in unspecific, broad government conducted surveys? It makes little sense to me. Yet economists routinely overweight this "macro" "soft" data rather than the "hard" data and commentary coming from actual businesses on the ground feeding us intelligence through the stock market.
What's perhaps most bizarre about this is the asymmetry between stock market participants and economists in this matter: stock market investors do follow the "macro data" that economists base their livelihoods on, but the reverse is not really true. Economists generally do not closely follow individual company information. It's rare to read an economist's research note from a sell-side Wall St. firm and hear them discuss any commentary on "micro" economic results (which is another way of saying the results of "individual businesses"). Even more strange is that the Federal Reserve doesn't seem to analyze this wealth of information either, despite them actually doing some version of this exercise on their own through their Beige Book. It's good that the Fed composes the Beige Book (Chair Powell seems to specifically like it, which is even better), but again, a treasure trove of important economic information is simply being underutilized.
The biggest reason I'm writing about this now is because for some time now, and specifically right now, there has been, and still is today, seemingly large gaps in what the "macro" data is saying about the strength of the economy, and what publicly traded firms are saying about it. For at least the last several months, companies across the software, industrial and consumer sectors have all discussed emerging macro weakness. Economists continue to say that "the consumer remains strong" almost unanimously, and they often look at things like this week's government retail sales data to show evidence of that. But look at consumer stocks (especially retail stocks) and tell me if that's really the case. Look at actual hard spending data coming from the country's largest banks and credit card companies and tell me that's so (see below). Listen to what consumer companies are actually saying on their conference calls and tell me that's so. Furthermore, we recently saw a 9.7% pullback in the S&P 500, which is very close to our 10% threshold for when we should really start paying attention to what the stock market is saying. The market is sending us signals about the strength of the economy, and they are much more concerning than economists seem to appreciate.
Regardless of whether we are in a recession or not at the moment, poor response rates to government survey data (which again, are likely to be at least partially driven by lack of trust in government and political polarization), structural weaknesses in the mechanics of how that "soft" data is collected, as well as the challenge of seasonally adjusting that data on the back of all the changes and fallout from COVID, now more than ever, we ought to be soaking up as much data and commentary from the stock market as possible. Since it's literally giving us real-time information about what's happening on the ground, this hard data is likely to prove far more valuable, and accurate, than more "traditional" "soft" government data sources. The market isn't always right in predicting recessions, it's true, but it has a leg up in data quality compared to economists and policymakers, and that is likely why its track record is so much better than that of economists. Thus, economists could significantly improve their forecasting if they listened to and aggregated the data coming from the country's publicly traded companies. Most important for this change in approach is for the Fed, since they are the ones actually setting monetary policy. The longer we continue to cast aside or otherwise underutilize data from publicly traded firms, the longer we're going to unnecessarily make policy mistakes. Now could be one of those times, but it certainly doesn't have to be.
Comments