Unreal GDP

“The only function of economic forecasting is to make astrology look respectable,” John Kenneth Galbraith.

A recent article in The Economist talks about the difficulties even the world’s best economists have forecasting economic growth. Approaches based on theory or empirical data or both yield poor results. Economies are just too complex to compress into accurate models, and forecasters give too much weight to the recent past when they look ahead into the future.

Take the International Monetary Fund as an example. The IMF publishes forecasts for 189 countries for the current year and the next year every April and October, and even with that short a time horizon it’s regularly wrong. From 1999 to 2014, the IMF’s April forecast missed every recession in every country it covered; and its October forecast missed half of them.

The IMF is not alone. The Federal Reserve Bank of Philadelphia publishes a survey of forecasts from economists at banks, consulting firms, businesses, and universities (as does the ECB). They are shockingly inaccurate.

Chart 1

As the chart shows, one year in advance forecasters in the United States have consistently failed to predict when the economy goes into recession and how deep the fall in GDP is. This is at least understandable to the extent economists are confused by complexity. It’s at best regrettable to the extent they are reluctant to be the bearers of bad news.

Chart 2

This chart shows that they do better 90 days ahead: they get the timing right more often, but they’re still wrong about the severity. The danger signs get clearer as the downturn gets closer, so it’s harder not to see a recession coming. But forecasters still seem to hesitate to think – or say – the worst.

Most credit risk takers make loans, buy bonds, or deal with counter-parties. They’re not economists, and they’re not expected to forecast GDP. So why does all this matter?

It matters because for many of our credit exposures default probability is linked closely to the course of the economy. Making a loan to a cyclical company like an auto parts manufacturer in a growing economy is one thing. Holding that loan through a recession is a much worse thing altogether.

The more advance warning we have, the better we can protect ourselves from the credit impacts of a recession. We can avoid losses by cutting or hedging our cyclical exposures. We can reduce the losses we have to take by improving security or strengthening our covenants.

It also matters because financial modeling is so important in risk analysis and debt structuring. We often use downside case projections to help evaluate credit risk and estimate funding needs, and those projections often include a recession. Having a realistic sense of when the next one will occur and how bad it will be makes scenario analysis much more useful.

Taking credit exposures involves a lot of guesswork about the future. That includes future recessions. We can’t escape relying on professional forecasters.  We just can’t count on them too much.

Submit a Comment

Your email address will not be published. Required fields are marked *