Discussions of volatility tend to reveal eccentric views of asset price determination. ‘Uncertainty’ is deemed to increase and decrease with remarkable frequency. This is particularly odd given that what we don’t know now was presumably part of what we didn’t know before. Do perceptions about how little we actually know about the future really change – is it not more likely that correlated emotional responses periodically grip market participants?

Before considering what are the plausible determinants of asset price moves, it is worth clarifying a few areas. Implied volatility, as derived by options markets or indices such as the Vix, is predominantly determined by realised volatility. So when people say ‘markets are complacent because the Vix is low’, it is worth remembering that all they are observing is that prices have not moved very much in the last 30 days or so. There is no ‘view’ embedded in the Vix. The implied volatility market is in fact nearly always very coherent, because the statistical properties of volatility are relatively robust. Volatility clusters in the short run – i.e. yesterday’s equity volatility is the best predictor of tomorrow’s, as time passes there is a gravity towards the mode. Consequently, when monthly realised volatility is low by historical standards, one month implied volatility will be low and the implied volatility curve will slope upwards. The opposite is the case when volatility spikes. Because these statistical properties are so robust, there is almost never ‘complacency’ in the market for equity volatility. The market treats realised volatility for what it is – a fact – and usually makes very reasonable assumptions about future volatility based on that fact.

So the real question when realised and implied volatility is low is: why are prices not moving very much?

That takes us to the crux. Why do prices move, and what might cause them to move a lot? The textbook assumption is that prices respond to news. Let’s assume – as many do unconsciously – that all asset price moves are caused by news about fundamentals, like earnings and economic releases. Little movement in prices may mean that there has been little by way of significant news. But do we have a good prior sense of what is ‘significant’?

Phrases such as ‘the market believes’ are deeply unhelpful. Insightful analysis assumes that that there is a distribution of market beliefs. There is usually deep heterogeneity of both beliefs about the world – for example, the probability that rates will rise or fall – and models of price determination and economic outcomes. Thinking about the distribution of beliefs – are beliefs correlated? dispersed? – may help us better understand price moves in response to news and changing conditions.

Like everything worthwhile in finance, the most profound ideas were observed by deeper thinkers seventy plus years ago, are often forgotten, and periodically are formalised by a very good mathematician. Mordecai Kurz’s work on rational beliefs formalises how the distribution of beliefs can explain ‘excess’ volatility. It has been observed that prices more-often-than-not move by more than they should in response to news. Shiller and many others suggest a behavioural explanation, Fama and Cochrane advocate varying discount rates, but Kurz shows this is a logical consequence of certain belief distributions and differing models of how the world works.

For example, it is clearly the case that competing theories can explain the same facts. I might argue that global inflation is low because of deregulation and globalisation of labour markets. You might argue that it has been caused by increased competition in product markets due to technology. Logically, we should think differently about the pricing of implied inflation faced with higher wage growth. If you think tech-driven deflation dominates, you can be sanguine. Now the asset price response to a modest increase in CPI could then be exaggerated if it causes you to abandon your model of inflation and embrace mine. Shifts in the distribution of beliefs about how the world works should cause large price moves. To take an even more mundane example which shows how volatility can also be endogenous: many quant models have very similar models, almost be definition. A change in price behaviour which invalidates these models could itself have a price effect.

Ok. So price responses to news can be exaggerated or muted by the distribution of beliefs and the nature of competing models of how the world works. So we have two explanations of why prices move – the other being the frequency and significance of ‘news’.

Even in a efficient markets view of the world where news is (almost) everything, there is a much neglected area of asset price determination which is always overlooked – the circumstances of investors. Most people intuitively understand that news about profits is relevant to a share price, but what about changes in aggregate market PE multiples, which affect all individual shares? This should not just be determined by the beliefs about future cash flows, it should also be influenced by the correlation of investment returns with investor circumstances. To make this clear: equities are cyclical, in part because profits are cyclical, but also because unemployment is cyclical. Bonds provide insurance against unemployment (interest rates fall during recessions), equities are correlated with the aggregate probability of job loss. Logically, if the probability of unemployment diminishes the PE multiple should rise. For this reason among others – such as changes in the rights of shareholders – mean-reverting PE multiple frameworks are patently false.

So far I have considered how markets ‘should’ work. As we know, psychology is just as important – investor beliefs are biased, subject to anchoring, ideas travel slowly, and perceptions of reality and one’s own circumstances are determined by emotion and recent experience as much as by objective analyses. “The power to become habituated to his surroundings is a marked characteristic of mankind”. Probability distributions rarely change – our perceptions of them do, frequently.

So far, so philosophical. Can we use these observations to help us? Let’s stay philosophical a little longer. Some will argue that this outline of the various causes of asset price moves is not complete. What about flows and positioning, for example? These are really variants of what we have already discussed. Investor ‘positioning’ is an attempt to identify the distribution of beliefs. Speculative positions in FX may reveal the distribution of beliefs about the model of exchange rate determination. Of course the complete market cannot be net long or short. But market professionals look at the positions of different investor types. For example, there are currently very large ‘speculative’ short positions in sterling versus the US dollar. Is there information in this? It must reveal a view of price determination. The prevailing model in FX is trend-following. Large short speculative positions in GBP may simply reflect the fact that sterling has been falling and exchange rates have a statistical tendency to trend. Some will say that “investors are bearish on the UK economy due to Brexit”. But that is incomplete, ‘speculative’ investors have to be more bearish than those currently transacting at current prices. It is more likely that the scale of speculative shorts increases the probability of GBP strengthening sharply in response to what appear appear to be minorly positive noise – as we’ve just seen.

Analysis of flows usually masks the really interesting question – what is causing the flows? The answer to which typically reverts to discussions of the behavioural models of asset price determination outlined above. The same is true of the recent ‘technical’ discussion of trends in volatilty, which Jon Sindreu and his colleagues describe.

How might these reflections bear on current market behaviour? The realised volatility of most major assets was low in the first quarter of the year. And despite the hoo-ha around geo-politics, the global asset price response remains relatively muted. My prior would have been to anticipate more volatility in Korean assets than we have seen. You would barely notice anything significant had occurred from a chart of the Korean stock market, let alone the suggestion that we are on the brink of nuclear war. Complicated rationales for a lack of price response based on positioning in the derivatives market don’t convince. Markets which are immune to derivatives activity – such as the sovereign CDS market – have reacted in a very measured way.

There are many interesting features of the current environment. Low volatility does have a feedback. The prevalence of targeting constant levels of volatility and VAR in portfolios means that formal risk management processes reinforce the behavioural bias that Keynes observed – we get used to recent volatility. Now risk managers tell investors that risk has declined, merely because trailing volatility has fallen – without any changes to portfolio positions occurring. That is a hugely correlating feature of market behaviour. And a great potential source of opportunity to the active manager.

It is also widely observed, but rarely explained, that volatility usually rises when prices fall, and declines when prices rise. Keynes also observed that recoveries were gradual, recession abrupt. One reason may also be the role of investor circumstances. The PE multiple rises as investors gradually believe that cyclical conditions are firmer. The memory of crisis fades and fears of recession around every corner recede. Many beliefs and perceptions of risks travel slowly, less like contagion and more like osmosis, as the great Jack Treynor observed. Confidence has a different rhythm to fear.

Observably there is also a lot of talk of equity bubbles. These are largely superficial observations on the PE multiple of the S&P500. Yes, the PE multiple is higher than its mean. It’s expected return is lower than its historic average. So what? It spends little time at its mean, ands the earnings yield is closer to its long run mean than any competing asset. Bad statistics is no substitute for thought. Most equity markets, by contrast, have in fact done little for over two years. But earnings are growing and the global economy is the strongest it has been for almost a decade, maybe more, given that a decade ago it wasn’t as strong as it appeared. All the talk of ‘complacency’ seems analytically weak, or worse … complacent.

About The Author

Eric Lonergan is a macro fund manager, economist, and writer. His most recent book is Money (2nd ed) published by Routledge. He is also a supporter of Big Issue Invest (BII), the investment arm of The Big Issue, and is one of the initial limited partners in BII’s Social Enterprise Investment Fund LP. In a personal capacity, he makes direct investments in social enterprises. He also supports and advises The Empathy Museum.

6 Responses

  1. Valter Buffo

    “It has been observed that prices more-often-than-not move by more than they should in response to news. Shiller and many others suggest a behavioural explanation, Fama and Cochrane advocate varying discount rates, but Kurz shows this is a logical consequence of certain belief distributions and differing models of how the world works.”. It would be useful to define HOW they “should”. We published two papers on this subject il The Long Rooom, Financial Times, February the 17th and March the 24th, 2017. Comments more than welcome.

    Reply
  2. John

    Eric
    Good article, but I don’t understand the comment ‘mean reverting PE multiple frameworks are patently false’?
    If you say equities earnings are cyclical due to the business cycle, itself due to changes in employment, then why wouldn’t PE multiples also be cyclical?
    I get that they are just an output (i.e. not cause and effect), but your statements about the employment would imply you would endorse a mean reversion model of employment, and the business cycle …. so aren’t PE mean reversion models just a short hand (and crude) summary of that?

    Reply
    • Eric Lonergan

      Hi John – apologies for not replying sooner. I think P/E multiples are likely to be cyclical around trend earnings because stock prices are cyclical. But that is not the same as suggesting some kind of stable mean. Precisely because one of the risks of owning stocks is that their prices tend to fall in recessions, if the probablity of recession declines, the PE multiple should rise.

      Reply

Leave a Reply

Your email address will not be published.