Are the markets ignoring reality?

Descriptive transcript

Our first presenter today is Professor Thalis Putnins, a professor of finance at the UTS Business School. He will present for roughly 15 to 20 minutes, followed by five minutes of questions. Afterwards, we continue with Professor Warren Hogan, who will similarly present for just under 20 minutes, followed by five minutes of questions for him, and we finish up with a Q&A.

Thank you, Gerhard, and a warm welcome to everyone. I'm going to be talking about stock markets and how they've behaved during the COVID pandemic, and what we can make of those stock market reactions—what we can read from the stock market movements.

The first part I'll cover is about market dynamics: just having a look at what's happening in markets around the world before we start digging into the interpretations of those dynamics.

Here I have a plot of global stock markets since the beginning of the year, since 1 January 2020. The striking feature across these examples of global stock markets is that they're all moving in tandem.

Despite substantial differences across countries—the US, Japan, Germany, Australia, the UK—in how governments have responded to the pandemic, the economic policies introduced, and the differences in infection and death rates, the markets seem to be moving as one. This is a global problem, and the markets are treating it as if we're all in this together. Analysing any one of these markets is fairly representative of what's happening globally across many markets.

I'm now going to break it up into three phases: the build-up—what markets looked like in the lead up to the pandemic; the crash; and what we've been seeing in the most recent two months up until today.

First, the build-up. The starting point of where we entered the year was that we were at the end of the longest bull market on record. Here I've got a diagram showing stock market movements going back 120 years. The striking feature is that this last run, this bull market run, has basically taken us from the GFC through to the end of last year.

As a result of this extremely strong bull market run, stock market valuations were also at record highs leading into the pandemic. The price-to-earnings ratios, with a few adjustments, were at levels we've only seen on two other occasions historically: during the dot-com bubble of 2000, and immediately preceding the Great Depression. So, equity markets were extremely hot, with extremely high valuations going into this crisis.

We also saw that the yield curve in the US, and in some other currencies as well, was inverting, which is typically a sign of trouble up ahead. Yet the markets hadn't responded to that yield curve inversion. So from many regards, equities markets looked like a ticking time bomb simply waiting for a catalyst.

Then the catalyst arrives—and boy, did it arrive. It was quick, bad information. The market peaked around late February. If we take the US market as an example, we saw a 35% decline in the market in the space of a month—an extremely rapid fall with a huge amount of volatility. The market was hitting market-wide circuit breakers on a number of occasions, which is quite rare for a market to have such extreme price movements on a daily basis.

If you look at a plot of the volatility, what we see is that volatility spiked tremendously. In fact, it hit levels that were equal to or even slightly higher than those seen during the global financial crisis of 2009—so, extreme volatility.

Then, from mid to late March, comes a very strong rebound in markets. If you measure the price movements in the US from 23 March to today, US markets have rebounded by about 32%. It's a similar number in Australia, a little less, but still a very strong rebound. This has happened in the space of two months. What this does is take the stock markets back to valuations last seen in October 2019.

Historically, that means markets are back to where they were immediately prior to COVID breaking out in China—back where we came from. It's as if this whole thing never happened.

What's really puzzling about this strong rebound, or bounce, is that it came at a time when there was a flow of really bad economic news. For example, the IMF announced we were heading into the worst recession since the Great Depression, far worse than the GFC. That's the sort of news coming out while the market is rising 32%.

A number of market commentators have looked at this and concluded that the market simply seems disconnected from reality—the stock market seems to be ignoring the economy. So, puzzling dynamics in stock markets, to put it lightly.

The question is: how do we make sense of this? Does this strong rebound in the stock market mean we're headed for a strong economic rebound? Can we expect a V-shaped recovery? Or is the market response somehow fooling us about the economic prognosis going forward? Were stocks overvalued to begin with, going into this crisis after this 12-year bull market? Why didn't the market correct earlier? Why did we have to wait for a pandemic like this to see the correction? More generally, what drives market movements? Are they efficient? Are they linked to economic outcomes? Can we read information about the future of the economy from what the stock markets are doing?

These are the issues I want to tackle. I'll do this by pointing out what I think are three important considerations for trying to read stock market reactions, based on three recent research studies with co-authors.

First, market efficiency. We know markets are good at forecasting in many contexts—they give very strong, often very accurate, forward-looking predictions of what's going to happen in the future. But it's important to understand the subtleties of different types of market efficiency, and when markets can be efficient or show signs of inefficiency.

In our study, we split market efficiency into two types. One is micro-level efficiency, which is about how efficiently one stock is priced relative to another—relative prices within the stock market. For example, is ANZ correctly priced relative to BHP? Is one too cheap and the other too expensive? If you construct a long-short position—long one, short the other—can you predict future returns? If you can, the market isn't particularly efficient in a micro sense. The related concept is micro-level informativeness of prices: do stocks with higher valuations, like a high P/E ratio, predict stronger future earnings?

Contrast that with macro efficiency, which is about the efficiency of absolute stock prices—market-wide valuations. Is the market as a whole overvalued, undervalued, or correctly valued? Can you take the current market-wide valuation ratios, compare them to historic averages, and use that to predict future market returns? If valuations are high, does that predict low future returns, and vice versa? If so, there's a degree of inefficiency at the market-wide level. The related concept is market-wide informativeness: does the current valuation level of the stock market, or the stock market return, give you a signal about future market-wide earnings or economic activity? That would be the case if you had a highly informative market at the macro level.

We've got a number of measures of these concepts, but let me get into the evidence. What we find is that markets are far more efficient at the micro level—in relative pricing—than at the macro level, that is, absolute market-wide valuation levels. You can see that in the plot here: the top curve is the degree of micro-efficiency for US stock markets, and the darker line is macro-efficiency. Micro-efficiency is much higher than macro-efficiency, consistent with Samuelson's dictum. Paul Samuelson predicted this effect long ago, but it's only recently been tested.

Another observation is that the wedge between micro- and macro-efficiency is getting larger. Stock markets are getting better at pricing stocks relative to one another, but worse at setting market-wide valuations at levels that correlate with future economic activity. In other words, markets are becoming decoupled from the economy. We find several drivers of this: the rise of passive investing, the rise of delegated funds management, and rigid asset allocations like 60/40 portfolios (60% equities, 40% fixed income). For example, when markets crash and the equity value in such a portfolio declines, the fund is mechanically forced to buy the stock market, irrespective of the future economic outlook.

You get bounces in markets that are unrelated to fundamentals as a result of this type of investment management. A couple of implications: it explains why markets are very efficient in one sense—relative pricing is really accurate, so fund managers find it hard to beat the market (generate alpha)—but at the same time, market-wide valuations often seem uncoupled from the actual economy. You get this low degree of macro-efficiency. From an investment perspective, it implies it's difficult to gain a lot of value from stock picking, but there is potentially value to be added from strategic dynamic asset allocation that considers market-wide valuations.

The second consideration is what I'll call post-traumatic stress disorder, or the PTSD of markets. Consider the effect when you walk from outside where it's sunny into a poorly lit room—the room seems really dark at first, but after a while, once you adapt, it seems reasonable. Or, if you stare at a red screen for a while and then look at white paper, the paper appears to have a greenish tinge, which fades as you adapt. These are examples of perceptual biases that happen when someone shifts from a strong stimulus environment to a more normal one. The bias lasts until perceptions adapt and catch up with reality. This is known as habituation or neuronal adaptation.

What does this mean for markets? We tested these effects in financial markets in two settings. First, lab experiments: we took people through trading simulations where we could control risk and volatility, and observed their perceptions of risk. Second, we looked at the pricing of S&P 500 options and inferred how implied volatility relates to future volatility to look for distortions. The key result is that people tend to underestimate risk following periods of very high volatility. When you drop from an extreme to a normal level of volatility, people feel the environment is safer than it actually is.

How that's relevant today is that we've just dropped off from a period of extreme volatility. Remember, the VIX was at all-time highs, and we've just backed down a bit into a period still highly elevated, but no longer as extreme as a couple of months ago. Traders are saying, "thank God we're back to normal", which is an example of the bias, because we're actually not back to normal—we're at a level much higher in terms of risk and uncertainty than historical averages.

The implication is that following extreme volatility, as people underestimate risk, it appears as excessive optimism. This effect is driven by a deep-rooted neurobiological adaptation to the high-risk environment. When you underestimate risk, you underestimate the discount rates required, which means you inflate valuations. Markets will bounce following extreme volatility until perceptions catch up with reality. Once perceptions catch up, discount rates adjust and valuations come back towards fundamentals. We found evidence for this effect in the lab and in field studies, and it's likely to explain some of the bounce in markets lately.

The third major consideration is shifting away from free markets to what I call Fed markets. The US Federal Reserve has said, "we'll do everything it takes to stop this." The RBA has echoed that sentiment, saying they'll transact in whatever quantities necessary to achieve their objective. These days, the biggest market participant is not a hedge fund—it's the central bank. The Fed, ECB, RBA, Bank of Japan, and Bank of England are all intervening in markets.

Here's a plot of the US Fed's balance sheet since they started quantitative easing in response to the GFC. Since then, they've expanded the balance sheet, even with some periods of contraction. Overlaid on this plot is the S&P 500 index, and visually you can see a link between the two.

Zooming into the most recent period, just as markets crashed, the Fed expanded its balance sheet enormously and markets recovered in response. The Fed's asset purchases have been at an unprecedented rate—the balance sheet went from $4 trillion to about $7 trillion in about one month, or 33% of GDP. Compared to other quantitative easing programs, the speed and magnitude of this expansion makes previous programs look negligible.

What does this do for markets? Early analysis of lead-lag correlations between Fed balance sheet expansion and stock market movements suggests that when markets decline, that's followed by strong balance sheet expansion by the Fed, typically within two to five weeks. Subsequently, the Fed's asset purchase activity tends to precede market rebounds by about zero to three weeks. These correlations are very strong.

If we model this more formally in a time series model that captures these lead-lag causations, we can construct a counterfactual: what would market prices look like had the Fed not intervened? The model tells us that a shock to the Fed's balance sheet takes about eight weeks to play out in the stock market, with a substantial magnitude—a decent fraction of the total balance sheet expansion is reflected in stock prices. If we take the Fed's asset purchases in March and translate them according to this model, the estimated impact on the S&P 500 is about a 13% increase. Contrast that with the actual movement of the S&P 500 of 32% since its low—this tells us the Fed's actions alone explain over a third of this bounce in US stock markets.

If you overlay the Australian and US market reactions and look at the point where the Fed really stepped in, two things happen: the markets turn and spring back, and from that point onwards, there's a disconnect between the US and Australian markets—consistent with the notion that the US Federal Reserve's actions had a major impact on the stock market.

Bringing this all together: what can we read from the current market reactions? Number one, stock markets are increasingly decoupled from future economic conditions, driven by factors like passive investing, delegated funds management, and static asset allocations. In general, there's not much to read from stock market movement if you're trying to predict future economic outcomes.

Second, following extreme volatility, risk tends to be underestimated, which looks like excessive optimism. This leads to inflated valuations and market bounces until perceptions have time to adjust and catch up with reality. So, a strong market recovery doesn't imply a strong economic recovery.

Third, central banks are now playing a major role in setting overall market price levels. Central bank actions contribute to rebounds in stock markets precisely when the economic outlook is gloomy, but balance sheet expansion cannot continue indefinitely.

The bottom line: I do not interpret these market rebounds as an indication that the economy is back on track for a quick recovery.

If we think about where markets are headed, I'm not in the business of predicting market movements and can't tell you with certainty where the market will be at the end of the year. But if you think about these three effects—three different factors that distort stock markets from fundamentals—all three are currently pushing in one direction: to keep the market propped up. If something is fundamentally being propped up, and you consider the downside risk on the table compared to upside potential, it looks to me that, on the balance of probabilities, there's more downside risk than upside potential from this point onwards. It only takes one of these propping devices to fail or attenuate, and the market could come back down towards fundamentals.

Thank you for your attention.

Thank you very much, Thalis. And boy, that horse doesn't look really good. But we have a couple of questions for you.

Question from attendee Nancy: Five major companies—Amazon, Apple, Facebook, etc.—in the US control 20% of US stocks. They also saw increases in revenue. For example, Amazon recorded a 17% increase in revenue in the last quarter. What role do these companies play in determining the stock price and affecting the market? Also, world debt in the last quarter is $17 trillion, whereas last year it was $8 trillion. How will this debt affect the market post-COVID?

Very good question. It's a very good observation that markets are highly concentrated, particularly US markets, driven by a small number of extremely large companies. That is yet another factor—one I didn't discuss—that drives the disconnect between stock market movements and the economy. If you think about which companies have been hardest hit by the COVID pandemic, it's not the largest companies—it's not necessarily the tech companies driving the US stock markets—it's the small and medium enterprises. So that's another reason why stock market movements aren't really reflecting the broader economy.

Regarding the huge amounts of debt, particularly on government balance sheets, I view them as similar to Fed balance sheet expansion. You can't keep expanding government debt, nor can you keep expanding Fed balance sheets indefinitely as a proportion of GDP. The fact that these levels are very high at the moment gives us less room to expand fiscal or monetary spending in the future. While it's unlikely that governments or central banks will rapidly unwind these positions and cause a sharp decline in markets, if they have to unwind over a longer period, that simply means a dampening force on economic activity and on stock markets, which doesn't look good for long-term returns. But then again, long-run returns on fixed income are also not looking that good at the moment.

Another question: What are potential factors driving the underestimation of risk in such circumstances?

The underestimation of risk, from the evidence we have in lab and empirical studies, is due to deep, ingrained neurobiological biases. Once you become immersed in an environment with very high stimulus, you become desensitised to it. For example, traders in a market where, on a given day, the market can move 7%—unprecedented movements—after a while, operating in such a high-risk environment with millions of dollars on the line every day, you become somewhat desensitised to that risk. Once the risk jumps back down towards an elevated but closer-to-normal level, that seems like calm to a trader who's been operating in an extremely high-risk environment. That's what triggers the underestimation of risk: even though there's a high degree of uncertainty about earnings and economic conditions going forward, it just feels so much calmer compared to where we were in March. That's one of the drivers of the underestimation of risk post-extreme volatility.

Another question: Are your findings consistent with a correlation between implied volatility and historical volatility? One of our guests, Alan, says he recalls earlier analysis showing realised option payouts are less than those implied by market prices—for example, future actual volatility is less than implied volatility. Could you comment on that?

Sure. I think the question relates to the variance risk premium: there tends to be a gap between implied volatility and realised volatility, which is explained by the variance risk premium—people willing to pay to hedge the possibility of spikes in volatility. Implied volatility, from the empirical analysis we've done, closely tracks realised volatility. It's a forward-looking measure, and realised volatility is backward-looking, but future expectations of volatility are heavily driven by current levels of volatility. Volatility is a highly persistent variable—the current level is a good predictor of the future. This is one of the empirical facts we exploit to identify distortions in risk perceptions in US markets in the field. In those types of tests, we can see systematic distortions in implied volatility compared to future realised volatility at particular times—typically when you come back down from a very high volatility level. Even highly sophisticated and liquid markets like the US S&P 500 options market show these distortions in implied volatility at those points.