Goldman's "Early Warning Signals"

Tyler Durden's picture

Fed officials have repeatedly emphasized the importance of financial stability for monetary policy. But, as Goldman Sachs points out, knowing which financial and macroeconomic imbalances to monitor is challenging, not least because of the limited number of past crisis episodes in the US. To help The Fed, Goldman surveys a large economic literature that studies the effectiveness of "Early Warning Systems" (EWS) in detecting banking crises, costly asset price busts, and currency crises across a broad range of countries. While they suggest subtlely that the Fed is clueless with regard what to look for, they note that credit markets and asset-price run-ups (especially when they occur together) provide a statistically clear warning signal... and as we know, both are flashing red currently.

 

The results, Credit markets are the canary most often cited...

 

And as we explained here in detail (and numerous times in the last few weeks), Credit is flashing redder than ever.

 

 

As Goldman notes, we draw three key lessons from our survey:

First, several indicators turn up repeatedly as valuable warning signs. In particular, credit growth, run-ups in asset prices (especially house prices), the current account, the real exchange rate, and GDP growth are frequently cited across studies. Other indicators appear more frequently in studies of particular types of crises, such as weak governance and regulation in the case of banking crises, or reserve holdings in the case of currency crises.

 

Second, many studies note that the joint occurrence of multiple signals provides a statistically clearer message. For example, rapid credit growth in the context of weak bank regulation or an asset price run-up fueled by large current account deficits might send a stronger warning sign than the indicators do independently.

 

Third, EWS nonetheless tend to have a modest signal-to-noise ratio. The models fail to predict some crises, and in other cases struggle to assess the likely severity of a bad outcome. For example, they might have been unable to distinguish between the consequences of US asset price collapses in 2000 and 2007. The models also sometimes send false positives, calling to mind Paul Samuelson's remark that "Wall Street indexes predicted 9 of the last 5 recessions." In addition, some indicators are as coincident as leading, and studies that are more ambitious in testing for an advance warning sign tend to have less success. Finally, the lessons drawn from one set of crises do not always apply to the next. Indeed, research by San Francisco Fed economist Mark Spiegel reports little success in predicting which countries would have been hit by the global financial crisis.

*  *  *

The bottom line is that despite Fed officials proclaiming omniscience, Goldman believes the Fed cannot accurately identify bubbles and therefore the costs of tightening in response to a potential bubble are too high no matter what...

To The Moon Alice (unless a 'free' market decides otherwise)