Why recessions are hard to predict
By Scott Sumner
On occasion, a pundit will accurately predict a recession. As far as I know, however, no one is able to consistently predict recessions. Why is that?
I thought of this question recently while at a conference in Nashville, where there was a discussion between Paul Krugman and Tyler Cowen. As you’d expect, there were lots of interesting comments. Unfortunately, I lack a transcript, so the following might not be entirely accurate.
Both Krugman and Cowen had noticed some excesses during the housing bubble period, and both seemed to regret that they had not done better at predicting the problems ahead. I also failed to predict the Great Recession; indeed I probably did even worse than they did, not even seeing the excesses in housing. But I don’t worry much about my failure to predict the recession, as I don’t see that as a realistic goal for economists. On the other hand, I’m even more optimistic than Tyler about our ability to prevent recessions with better monetary policy.
Last week, I happened to notice a MRU video on the Chinese economy, done in 2015. In the video, Tyler mentions that China was sliding into a “Great Recession”. He also discusses 5 structural problems with the Chinese economy. In retrospect, China did not enter into a Great Recession, although in fairness their economy did stumble a bit in 2015, and many believe that growth slowed considerably more than what you see in the official data. Nonetheless, China avoided the worst predictions (we’ll have to see how they come out of the current trade war.)
Perhaps Tyler’s pessimism in 2015 related in some way to his chagrin at not anticipating the extent of problems after the US housing bubble. After all, China has recently had some of the same debt-fueled excesses as the US. Indeed if you agree with Kevin Erdmann’s claim that the US in 2006 was not nearly the housing bubble that most people assume (and I do), then the Chinese situation is arguably much worse. So why no Great Recession?
I’m not going to say that the US housing bubble had nothing to do with the Great Recession. Rather I’d claim that it did not cause the Great Recession. The necessary and sufficient condition for a major demand-side recession is bad monetary policy. It’s true that problems in housing and banking helped to throw Fed policy off course, but with a better monetary policy that need not have happened. Instead, the problem could have continued being managed as it was in 2006 and 2007, when housing construction fell very sharply.
This is why it’s so hard to predict demand-side recessions:
1. The Fed basically sets monetary policy at a position where it expects adequate growth in AD.
2. Fed policy generally reflects roughly the consensus of the economics profession.
3. Thus Fed policy is usually not far from what a consensus of the profession expects will lead to appropriate growth AD.
4. This implies that the consensus of the economics profession will generally not forecast a sharp drop in AD, except possibly when at the zero bound—if you believe that monetary policy is ineffective in that situation (I don’t.)
Of course a few individual heterodox economists will occasionally predict recessions. But that’s not really very helpful, as the public has no idea which alternative views to rely on, especially as success in one prediction generally won’t carry over to the next business cycle.
If Tyler did correctly diagnose structural problems with the Chinese economy, why didn’t that lead to a correct business cycle prediction? Because recessions are very different from structural problems. In the early 1930s, the US had perhaps the most structurally sound economy in all of human history, up to that moment in time. China’s economy was a big mess. Yet it was America that had a Great Depression in the early 1930s, not China. Demand-side Great Depressions aren’t caused by structural problems; they are caused by monetary policy mistakes, by excessively tight money. In contrast, severe structural problems cause countries to be very poor, as with China in the 1930s. It’s about levels vs. rates of change.
If you could predict recessions, you could also predict the asset price changes that come with recessions, such as sharp declines in interest rates and equity prices. That would allow you to become rich. It’s simply not realistic to expect the economics profession as a whole to ever be able to reliably predict in a way that would allow economists to market time much better than Wall Street experts. As a result, we’ll never be able to develop macro models capable of predicting demand-side recessions.
And we shouldn’t even try.
Instead, we should think about policy regimes that will tend to stabilize expected NGDP growth at roughly 4%, even when there are shocks to industries such as housing and banking. What would that regime look like?
PS. This was the first time I had met Paul Krugman in person. I just had a few minutes to speak with him, and mentioned that I thought his 1998 Brookings paper would end up being the most important macro paper of the past 25 years. (I thought I ought to say something positive, after frequently criticizing him in my blogging. But I really do think it’s an extremely important and underrated paper, for reasons explained in this post.)
PS. Just to be clear, I agree with those who say China has some major debt-fueled distortions, and I also believe that this may make it more difficult for Chinese monetary policymakers to avoid a demand-side recession. I don’t know when their next recession will occur.