Saturday, May 13, 2017

Maybe macroeconomic forecasts aren't so bad

Macroeconomists... are asked to routinely produce forecasts to guide fiscal and monetary policy, and are perhaps too eager to comply. ...

...the supposedly most embarrassing forecast errors come with regards to large crises. Yet, these crises are rare events that happen once every many decades. Since typical economic time series only extend over a little more than one hundred years, statistically forecasting the eruption of a crisis will always come with large imprecision.

Compare how economics does relative to the medical sciences. ...

Imagine going to your doctor and asking her to forecast whether you will be alive 2 years from now. That would sound like a preposterous request to the physician, but perhaps having some actuarial mortality tables in her head, she would tell you the probability of death for someone of your age. For all but the older readers of this article, this will be well below 50%. Yet, one year later, you have a heart attack and die. Should there be outrage at the state of medicine for missing the forecast, with such deadly consequences?

One defense by the medical profession would be to say that their job is not to predict time of death. They are driven to understand what causes diseases, how to prevent them, how to treat them, and altogether how to lower the chances of mortality while trading this off against life quality and satisfaction. ... This argument applies, word for word, to economics once the word disease is replaced by the words financial crisis. ...

A more sophisticated defense would note that medical sciences are about making conditional forecasts: if you make some lifestyle choices, then your odds of dying change by this or that much. These forecasts are at best probabilistic. ...

Economics is not so different, even in 2007-08. Within days or weeks of the failure of Bear Sterns or Lehman Brothers, economists provided diagnoses of the crisis, and central banks and finance ministries implemented aggressive measures to minimize the damage, all of which were heavily influenced by economic theory. ... The economy did not die, and a Great Depression was avoided, in no small part due to the advances on economics over many decades.

Too many people all over the world are today being unexpectedly diagnosed with cancer, undergo enormously painful treatment, and recover to live for many more years. This is rightly hailed as a triumph of modern oncology, even if so much more remains to be done. After suffering the worst shock in many decades, the global economy’s problems were diagnosed by economists, who designed policies to respond to them, and in the end we had a painful recession but no melt down. Some, somehow, conclude that economics is at fault. ...

Macroeconomists are... asked to predict what will happen to the changes in the CPI or GDP over the next 1-5 years. The comparison of forecast quality must be made for the same time horizon and for a similar level of aggregation. The fairer comparison would be to ask doctors to predict what will be the percentage change in the annual number of patients that eventually die after being admitted to an emergency room due to a stroke. For these similar units, my guess is that medical forecasts will look almost as bad as macroeconomic forecasts.
--Ricardo Reis on comparing apples to apples