An enticing looking book has arrived in the post. It’s, by Walter Friedman.
[amazon_image id=”0691159114″ link=”true” target=”_blank” size=”medium” ]Fortune Tellers: The Story of America’s First Economic Forecasters[/amazon_image]
It’s easy to make fun of economic forecasts, which are always wrong. Nate Silver’shas a chapter explaining with care why this is so, without bothering to score the cheap shots many critics resort to. Essentially, he points out that the macroeconomy is a large complex system with many feedbacks, about which we have very little data. Economic forecasting lags well behind weather forecasting in its gathering and use of statistics. David Hendry and Mike Clements have written, for my money, the best book on how to do time series forecasting given our current data and knowledge, .
There has been progress. W Stanley Jevons famously correlated economic activity with sunspots. The theoretical basis for this might in fact have grown stronger now there is so much electronic communication for solar storms to disrupt. Fans of Kondratiev cycle-type analysis sometimes stretch the insight that applying disruptive technology can require generational change to trying to forecast the cycles.
I’m sympathetic to macro forecasters as I used to be one myself for a couple of years. It was an eyeopener to me, a relatively freshly minted, idealistic PhD, to realise how much fiddling there is to make any forecast look even plausible – they all require it – and therefore how strong the herding instinct among forecasters. It took me another giant stride on my journey from macro to micro. I was working for Data Resources Inc, founded by the eminent US macroeconomist Otto Eckstein in 1969 (now part of Global insight).
stops before the Second World War, however – that is, before modern macro models. It looks great fun.