A few weeks back, Harvard’s David Laibson gave a fascinating keynote lecture at the Geary Institute’s Economics and Psychology Conference. A key theme was the way people form expectations when macroeconomic time series have what he calls “hump-shaped dynamics”. These dynamics and their implications for expectations are described in a recent paper for the Journal of Economic Perspectives:
Many macroeconomic time series have long‐horizon hump‐shaped dynamics – processes that show momentum in the short run and some degree of mean reversion in the long run. Such dynamics will generally not be captured by simple growth‐regressions. Hence, agents with natural expectations will make approximately accurate forecasts at short horizons, but poor forecasts at long horizons, because the economy has more long‐run mean reversion than the agents impute from their intuitive models. In other words, agents with natural expectations will overestimate the long‐term persistence of good news or bad news.
David explained how even a skilled econometrician facing relatively short time series will tend to miss the longer-term mean reversion. The difficulty of seeing the mean reversion can mislead us into believing that a string of good draws on the fundamentals reflects a permanent improvement – with Ireland’s property bubble a good candidate. But equally a string of bad news can lead us to excessive pessimism – Wolfgang Munchau’s expectation that Ireland’s nominal growth will not exceed 1 percent for a decade comes to mind as a possible example.
We have certainly experienced a string of bad news on both economic growth and fiscal cost of the banking losses. Just as during the boom, extrapolation has led to extreme expectations about the economy and solvency. Of course, this pessimism could turn out to be justified. But it is no harm to remember that mean reversion works both ways.
Today’s Q3 growth numbers can be considered mildly good news. It is still too early to tell if we will be “bumping along the bottom” for some time or have “turned the corner”. (See here for graphs of real and nominal GDP/GNP based on today’s release.) For a mild antidote to the competition for who can come up with the biggest number for the banking losses, it is worth taking a look at Ronan Lyons’ analysis of potential mortgage-related losses.