Questioning popular narratives
Open up the CNBC website and you will find statements such as:
- “Stocks could take another leg higher into year end, after the four major market indexes all closed at record highs — a bullish sign.”
- “More than two-thirds of Netflix’s business will be international in 15 years, and that could be a good thing, one analyst said.”
- “A global rollout of [Amazon’s] streaming video service could mean further upside for the stock following weakness postelection.
The statements include observations and opinions pushed by analysts or journalists. How should a rational person react to this type of statements?
The inner writer and the inner fact-checker
In the book Being Wrong, journalist Kathryn Schultz offers a compelling metaphor for how the mind works. Humans tend to think in terms of stories – narratives – that explain real-world observations. We behave as if there is an “inner writer” and an “inner fact-checker”. When we direct our attention somewhere, we try to make up a story that explains that particular observation. The inner fact-checker makes sure that our stories correspond to the world around us.
For example, if we see snow on the ground, we may assume that it has been snowing sometime in the last couple of days. If we observe a negative real GDP growth number for Turkey, we may fit it into a story that Turkey is facing a financial crisis.
The stories we create are influenced by our existing set of beliefs. Our beliefs shape where we decide to direct our attention and also shape the stories that we create – and which stories we deem to be plausible.
As it takes a lot of imagination and effort to create our own interpretation of reality, most of the time we simply borrow narratives from others: “Housing prices are going up because of low interest rates” or “Gold is going up because of money printing on a massive scale”. If the narratives seem reasonable, fit with our own observations and we hear them over and over again, then chances are high that we will eventually buy into them.
A key point however is that it takes considerable effort to analyse and question a given narrative. It is not easy. And challenging your existing set of beliefs is stressful, especially when there is money and ego is on the line. So instead of actively questioning the narratives thrown at us, most of us end up exposing ourselves to those that fit our world view. Zerohedge people keep on reading Zerohedge. Liberals keep on reading The New York Times.
Ripples in calm water
Word-of-mouth propagation of ideas spread like ripples in calm water. As one person often tells his view to several of his friends, and those friends tell pass on the story to their respective friends – ideas tend to spread in an exponential fashion. How far the ripple ultimately goes depends on how convincing the narrative is to the majority of people hearing it – how likely it is to “infect” people hearing it.
The infection rate is a function of how well it corresponds to “common sense” and common beliefs. But it also depends on the narrative’s ability to evoke emotions such as hope, fear of loss, dominance and connectedness. In the investment world, the propagation of narratives is exacerbated by the need to produce short-term performance. If Greece becomes the topic du jour, then we feel that we need to be involved. Not only to demonstrate our superior understanding of the situation, but also to make a quick buck on what we perceive to be a near-term event.
The crux of the problem is the following: the likelihood of catching a ripple early in the process is very low. Imagine that a story-teller tells his stock recommendation to person 1 and 2, who then spread the tip to person 3, 4, 5 and 6, who in turn spread it to another two people. If you are a random person, the likelihood of hearing the tip in stage 1 is 1/7. The likelihood of it in stage 3 is close to 60%. Meanwhile, most of the profits comes from investing in stage 1 when the price hasn’t yet responded to the positive story. So the expected profit from acting on a tip is very low, because you will most likely be one of the last to invest. And if the recommendation proves to be faulty in some respect, you run the risk of persons 1 to 14 liquidating their positions before you get a chance to react.
By continuously trying to assess narratives in mainstream news reports, you are likely to be one of the last to read and act on them. You end up chasing events that have already been priced-in.
Sometimes narratives reinforce each other. The story that Amazon Web Services is growing fast and the story that Jeff Bezos is a genius may reinforce a broader narrative that AMZN stock is a screaming buy.
Narratives can also affect prices negatively. The Deepwater horizon oil spill caused BP’s stock price to fall more than 50% in a matter of weeks as headlines such as “Oil Spill Puts a Target on BP’s Back” spread like wildfire. The stories exaggerated the impact on BP’s business and the stock quickly recovered half of its initial drop.
Strategies to cope with information overflow
What is the trigger that causes us to start researching a particular stock? What is the trigger that causes us to make the final investment decision? These days, social or traditional media typically provide those triggers.
But given the nature of how narratives spread like ripples, we should be very careful about exposing ourselves to popular narratives – because most of the stories we hear are likely to be late-stage ripples. Meanwhile, we are human – we don’t have the knowledge or the mental stamina to question everything we read. It’s too much to ask of anyone.
- Stage 1: Ignore all types of opinion and look forward instead
Seek solitude. Try to imagine how the future will be different 18-24 months from today. Use first principles to predict the future decisions of customers, competitors, business leaders or central bankers. Then figure out how investors will interpret and react to those events.
- Stage 2: Listen to non-mainstream experts and then enter the trade quickly
If you want to copy ideas from others, make sure they are unique, expert in their subject areas and that you are able to trade fast before everybody else catches on to the idea. Otherwise you will run the risk of getting into a crowded trade. The narrative doesn’t need to be flawless – all that matters is that it is new and that it appears convincing enough for others to buy into it at a later stage.
- Stage 3: Identify popular narratives (“late-stage ripples”) and dig deep to find potential flaws
A number of “mental models” that can be used to dissect a narrative:
- Base rate: Find the base rate value/correlation/expected outcome/etc.
- Premises: What would have to hold for the statement to be true?
- Disconfirming evidence: What made those situations special?
- Invert the statement: Is the opposite statement true – if not, then why?
- Reductio ad absurdum: Take the argument to its logical extreme
- Find a conflicting narrative: Which version of the story is the most plausible?
- Cause-effect: What can be attributed to as the cause of a particular outcome?
In practice, just finding disconfirming evidence may be enough to be able to tell that a narrative isn’t water-tight. The next step would be to examine correlations and think through what might be the cause of the phenomenon we are observing.
Sometimes the reason or the story behind buying a particular stock isn’t necessarily flawed, but active investors are so over-exposed to a particular asset that it’s a terrible bet to join them. There was nothing wrong with the Nifty Fifty in the early 1970s – most of the companies kept on growing for decades. The stocks just happened to be overpriced and over-owned among a large group of active investors. Extreme positioning and leverage is enough to cause a fragile state where the probability of a panic is larger than the probability of further buying.
In George Soros’s boom-bust model, the three stages can be illustrated in the following fashion.
- Stage 1 is when you are able to foresee a future event before any other investors has caught on to the story
- Stage 2 is when you hear the story from an insider or expert investor
- Stage 3 is when the investor hear the story from amateur investor
In stage 1 and 2, the investor will want to be long the security. In stage 3 he will want to bet against the common wisdom. The length of the cycle depends on fundamentals and reflexive feedback loops. The strength of the price reaction then depends on how convincing the narrative is – how far it is likely to spread.
Example: the Fed Model
Consider the following statement by Warren Buffett on 18 November 2016:
“Stocks are cheap if long term rates are at 4%, four to five years from now.”
Buffett is espousing the so-called “Fed Model”, arguing that government bond yields and equity earnings yields have a direct relationship.
|Dissecting the narrative:
Base rate correlation: Let’s start with calculating the correlation between bond yields and earnings yields in a simple scatter plot for the S&P 500:
The R2 of 0.02 is practically zero. On average, there has been no relationship between bond yields and earnings yields.
Premises: What would have to hold for the statement to be true?
Dissecting the above premises exposes flaws in Buffett’s statement. Banks tend to buy government bonds when there is no demand for borrowing, and they are unlikely to see equities as a good substitute. Investors tend to buy long bonds as bets on the projected path of future Fed Funds rates. Meanwhile, if the yield curve steepens or shifts upwards to achieve a long term rate of 4% – would corporate earnings be unaffected? Probably not. The third premise also exposes a flaw in the theoretical validity of the Fed Model: investors buy a security because the probability-weighted upside is higher than the probability-weighted downside – and the growth rate is clearly a key input to where the price will end up.
Disconfirming evidence: If you plot the S&P 500 earnings yield vs the yield for the 10-year UST you will find that the Fed Model worked well between 1970 and the mid-2000s. But there were also long periods of time when it did not work – from 1931 to 1960 for example. Japan since 1990 is another example when low government bond yields have not been associated with very low earnings yields.
Invert: Are stocks expensive if long term rates are at 10%, four or five years from now? Long-term rates at 10% can be justified by a very high nominal GDP growth, and it is unclear to me why a high nominal GDP should necessarily mean that stock prices should be low.
Reductio ad absurdum: If long-term rates reach zero, four to five years from now, should earnings yields also reach zero, giving stocks average PE ratios in excess of 100x? The absurdity exposes flaws in Buffett’s way of thinking. The potential loss of buying stocks at 100x is massive while the upside isn’t great. It doesn’t matter whether bond yields are currently low.
Conflicting narrative: Another narrative that describes the behaviour of earnings yields and bond yields is that “Stocks are likely to go up if liquidity is flowing into the stock market”. This narrative explains why stock prices were low in the private sector deleveraging of 1931 to 1960, while the Fed Model does not.
Cause-and-effect: If long-term interest rates go up to 4%, what is going to cause investors to buy and not sell stocks? What is the actual trigger that will cause them to buy or hold on to their stocks? I can’t think of psychological or institutional triggers. If anything, the negative wealth effect from lower bond prices should cause investors to sell stocks.
Accepting that you are fallible makes it easier to challenge your existing set of beliefs. We all fall prey to convincing narratives from time to time. But until beliefs have been supported by plenty of real-world observations – they are nothing else but initial assumptions.
A way to deal with narratives is to front-run their propagation, either by foreseeing a future event or by trading fast on a seemingly convincing tip from an insider or an obscure expert investor. Another way to profit is to bet against popular narratives that are flawed in some key respects – especially when you are able to pinpoint when those flaws are likely to be exposed.