At Sokrates advisors, we think very hard about what makes a good forecast, or prediction, and what produces a bad one. And, thankfully, we are not alone. In 2006, the US government formed the Intelligence Advanced Research Projects Activity (IARPA), as part of the Office of the Director of National Intelligence. Inside IARPA, there are several programs dedicated to forecasting and human judgment. IARPA’s mission is no secret: “At IARPA, we take real risks, solve hard problems, and invest in high-risk/high-payoff research that has the potential to provide our nation with an overwhelming intelligence advantage.”
And plainly, for us in the private equity world, making good forecasts is exactly that: it’s solving hard problems, high-risk/high-reward analysis that should provide an overwhelming advantage. So, of course, we follow everything that IARPA publishes in the area of human judgment, including all kinds of challenges, tournaments and competitions designed to reveal what makes good forecasting and good forecasters.
IARPA ran such a forecasting tournament from 2011 to 2015. Over this period IARPA asked approximately 500 questions, geared toward yes or no answers, and a timeframe. For instance, a possible question would be whether Ali Bongo would remain president of Gabon in the next 12 months. Thousands of forecasters participated, grouped in several teams.
One of the outcomes of this tournament was that experts, or insiders, are not very good forecasters. They tend to be overconfident; they take into account new and relevant information less often that they should. They favor continuity over change a little too often.
When we conduct a buy-side due diligence at Sokrates, we call this the continuity bias. Let’s see how the continuity bias works. To start with, let’s reduce all due diligence questioning to one single yes/no question and a time horizon: Will the sales (topline) of the target be higher in 5 years?
Let’s now take the example of a business that delivers a product to a finite installed base, for instance:
- anti oil-spill equipment for ports
- medical mattresses for hospital beds
- equipment (joints, flow-meters, etc.) for oil platforms.
In all these cases, the market is determined by an installed base: ports, medical beds and oil platforms. Let’s assume now that the installed base remains stable over our time horizon, that is 5 years. This is not a far-fetched idea:
- in many parts of the word, no new ports are built
- in many developed countries the number of hospital beds does not grow any larger
- and surely, there are periods of time when no new oil platforms are built (and actually old ones are closed).
So, let’s assume a stable installed base, nothing more complex. In this quite simple example, market sales come from two sources:
- Sales from penetration are sales of products that go on previously unoccupied installed base. For example, anti oil-spill equipment is installed at a port for the first time. The Company is increasing the penetration of its product within the stable installed base. A higher percentage of beds are equipped with medical mattresses, more ports are equipped with anti oil-spill equipment, etc.
- Sales from replacement are sales of product that replace older ones – products that were operating on the already occupied installed base. Anti oil-spill equipment is replacing an older installation.
We have spoken to many managers, certainly in the high hundreds, and all kinds of market insiders active in such markets. Of course, everybody is aware that the installed base is stable. We always ask the following question: what will happen when the market reaches the saturation point? Overwhelmingly, to the point of no exception, the answer is: the market growth will slow down, of course. The dominant, the controlling narrative is: there still will be growth, but at a lower rate.
But is this really what will happen? Here is our alarmingly simple example.
|Year 1||Year 2||Year 3||Year 4||Year 5|
|Sales from penetration||20||25||20||15||10|
|Sales from replacement||80||81||82||83||84|
The fact is that the market will contract after reaching saturation point. Yes, the market will get smaller, because these markets contract. Market insiders see the tipping point (at 25 in Year 2) but they assume that the market will continue to grow after that, albeit at a slower rate. This is the continuity bias at work, a propensity to presume that the future will look like the past.
The incentives of all market insiders to see continuity are so strong that the dominant narrative is locked on a simple slow-down. Market participants do not want to be part of contracting markets. The sellers, of course, even less. The potential buyers, engaged in the acquisition process, will easily accept this narrative because they do want the acquisition opportunity to be a good one. They have their own confirmation bias to confront, a whole other fight.
For us, it is sometime really difficult to stand on the position that the market will contract. The difficult part here is that the market will contract sua sponte, from within, due to internal constraints – and without any major external events, even less any extreme events. The management would often say: what could happen in the worst-case scenario? The installed base will always be there. The answer is hard to sustain precisely because nothing big will happen, the market will simply dwindle. In reality, there is continuity in the market, but this continuity implies contraction, not slower growth.
In practice, it is hard to place a tipping point on a timeline. It is even harder to predict what happens after the tipping point. The example above is purposefully simple. At Sokrates, we use data science to forecast much more complex and realistic markets: S-curve, Bayesian modeling, Power-rank laws. When direct data is not available, we use proxies. On top of this, we think in probabilistic terms. Paradoxically, the harder the analytics, the more difficult it is to defend the idea of market contraction in the face of continuity bias.
Every now and then, a perspicacious market insider would say: well, of course, trees don’t grow to the sky. But even this analogy does not do justice to market contraction. Trees stop growing, but do not shrink. Markets do.