Tuesday, September 17th, 2013

A Blameless Activity

This week many otherwise excellent economists will hang their heads in shame, while others will have bragging rights for the rest of the year. This week we find out whether the much discussed taper will begin in September or whether it will be delayed. The people who have got the answer right will be able to say, “We forecast this back in March” (or whenever), while those who didn’t will argue that there were very good reasons why they were wrong. On occasions, the whole process can get quite gladiatorial. For reasons which will become obvious as the article goes on, we have no intention of jumping into the debate.

At Harlyn, we have a well-publicised aversion to forecasts. At a very basic level we think that too many are wrong too often for them to be the basis of a successful investment process. At another level the information contained in accurate forecasts is often already incorporated into the price of the security under analysis. By the time we know that the forecast was right, investors are worrying about something else. It would be nice to think that the economists who have correctly forecast the outcome for this week’s FOMC will have equal success in the future. Unfortunately no such evidence exists. You don’t have to be religious to believe that people (even economists) learn from their mistakes and can be misled by success.

Forecasting is a natural and (mostly) blameless human activity. We are curious about the future. Forecasts are a way of systematizing that curiosity. There are obvious intellectual and psychological benefits and there is a clearly a commercial market for them. But does that make it a helpful discipline for investors? We think the benefits are overstated, and here are the reasons why.

The main point of earnings forecasts is to predict the evolution of share prices, and there are good examples of when it works. Most quant teams will tell you that there is a strong relationship between the change in consensus earnings forecasts and changes in share prices. What they may not tell you is that there is a significant variation in the time lag between the publication of the forecast and the impact on the price. (At the bottom of a bear market it can be longer than three months, and the height of a bull market it can be less than a week in a hot sector.) But what really matters is the behaviour at inflection points. At both peaks and troughs share prices change direction before earnings forecasts. Just when investors need it most, the forecasting process lets them down.

Investors tend to validate forecasts by the impact they have on prices. A useful or valuable forecast is not necessarily one which is correct, but one which has the capacity to change a share price, hopefully in a direction which suits their position. Every analyst has had the conversation with a fund manager along the lines of, “if the market really believed that, the share price would be different.” A forecast without a price attached to it is mostly pointless. In the end most investors think that price performance is more important than earnings forecasts.

If that is the case, why not study prices directly? What matters is not the ability to predict the future, but to understand the present, and respond to the recent past. Forecasts reduce flexibility. Organisations which have lots of forecasts take longer to change their mind than ones which don’t. When the unpredictable happens, you should change your position before you change your forecast.

Last week I had a long conversation with an economist and close friend who was originally in the December taper camp. He began to step away from his published forecast, so I asked him which data release had caused him change his mind. His answer was, “None really: the data are not decisive either way. It’s the price action over the last month which tells us what to expect.”

Leave a Reply

Full Blog Archive:


Other Recent Posts