Time to stop forecasting, and to start thinking

By: Wolfgang Münchau

This is slightly - but only slightly - off topic. The degree to which pundits - in newspapers, TV etc - got the US Democratic primaries persistently wrong is quite astonishing - and has some interesting lessons for economic forecasters as well. The pundits had written Hillary Clinton off in New Hampshire, where she bounced back to a large victory. And they have made exactly the same mistake again ahead of the Texas and Ohio primaries. Being a columnist myself, one of the things I have learned is never to make predictions that are likely to be falsifiable at a time when the memory of the column is still fresh - about a week or so. That means concretely: Do not forecast next week's inflation figures, and certainly do not forecast next week's elections. This is why I said last Monday that the Spanish elections are too close to call, even though I am aware that the Socialists have had a persistent lead in the polls. But the margin of error in all of these polls is so large that they do not tell us anything at all.

 

The polls ahead of the Texas and Ohio primaries were of similar quality. The political punters were factually mistaken when they concluded uniformly that only a miracle could save Clinton. How that? The polls had her in the lead in Ohio, and the polls were too close to call in Texas, given the statistical margins of error. From those polls, it was absolutely not possible to deduce that her campaign was likely to end on Tuesday. It was possible, of course, but not likely in any way. There was no miracle at work when she won. The problem was that numerically illiterate reporters misjudged the numbers, and that other numerically illiterate reporters multiplied the nonsense. The morale is: look at the numbers, use your own judgement, and try not to listen to pundits - who in fact do not know more than you do.

 

These observations are true of political forecasts of a Democratic primary just as they are of economic forecasts. Over the last year, we have heard for the umpteenth time that inflation is about to fall sharply, based on some model prediction. Yet month after month that predicted fall in inflation has "surprisingly" failed to materalise. That should tell us something about our ability to forecast inflation. The financial markets, just like political commentators, not only got the oil price wrong once, but they kept on making the wrong forecasts for years, using the same models which had given them the previously wrong forecasts.

 

We at Eurointelligence are fortuitously not in the forecasting business. The simple truth is that forecasting models are not built to handle uncertainties of such scale. You can shock them with a small fall in export demand, or a modest decline in share prices, but when you shock them with the biggest slump in the US property market since the 1930s, an unprecedented meltdown of securities credit markets with spillovers, a rebalancing of global capital flows, a sharp fall in the dollar, and a slump in the equity markets, you may as well throw those models in the bin. In such circumstances, they are about as good as astrology.

 

So use your brain, not your model. My brain tells me that inflation has been persistently higher than forecast, which suggests that a dynamic must be at work which is not understood by those models. That dynamic is almost certainly a monetary dynamic. Inflation is a monetary phenomenon - just as economic theory always told us. All we are seeing is that something is happening that is supposed to happen. So I wonder, why should this be a surprise. In the euro area, where both monetary aggregates and credit continue to expand at levels that are clearly not consistent price stability, it is clear now that monetary policy has been persistently too loose, and that price pressures will remain for the forseeable future.

 

Based on what we see, I have some difficulty comprehending how the ECB could start to cut interest rates in Q2, which is what the majority of market participants seem to believe. If the ECB was serious about its inflation target, it should have had to raise interest rates in the autumn of last year. In fact, even now, the present level of rates is not consistent with medium term price stability. But to cut interest rates would be the clearest signal yet that monetary policy has abandoned the goal of price stability altogether. While the Fed could not care less about price stability - with the exceptions of Messrs Fisher and Poole - I believe that the ECB is unwilling to take that risk.

---