Keynesian confirmation bias
By Scott Sumner
When you read Keynesian blogs you get the impression that Keynesians think their model has been somehow confirmed by the events of the last few years. And yet figuring out what their model actually predicts can be maddeningly difficult, like nailing jello to the wall. Today I’ll use some examples from Paul Krugman, who is the leader of blogosphere Keynesians. But as far as I can see most of the others are just as hard to pin down.
1. Japan: Marcus Nunes recently documented that when Paul Krugman began writing about Japan in the late 1990s he seemed to think that the Bank of Japan needed to be much more aggressive, and that fiscal stimulus was not particularly effective. His skepticism about fiscal stimulus is not surprising given the extremely large increase in the Japanese national debt over the last two decades. Japan is famous for splurging on unneeded infrastructure projects all over the country. And yet Japanese nominal GDP is actually lower than it was back in 1993, perhaps the worst performance of aggregate demand ever by a important developed economy.
More recently, Krugman seemed to edge away from the new Keynesian view that monetary policy was the primary driver of demand. As his views shifted in an old Keynesian direction, he decided that fiscal policy was actually a very appropriate tool for boosting aggregate demand in Japan. But in order to do this it was necessary to rewrite history. Now it was a myth that Japan had engaged in a lot of fiscal stimulus, despite the huge deficits and the massive infrastructure spending. In fairness, the Keynesian model does not look at gross budget deficits, but rather “cyclically-adjusted budget deficits.” (CABDs)
The problem here is that it is often very difficult to pin down just how large the output gap actually is. Certainly Japanese output in recent decades has fallen well below its earlier trend, but how much of that is structural and how much of that is deficient demand? After all, the unemployment rate in Japan is less than 4%. Admittedly the unemployment data is of doubtful accuracy, but we don’t really have any other strong evidence either way on the size of the Japanese output gap. Nonetheless Krugman seems to have become convinced that when measured properly the budget deficits were much smaller than they appeared without adjustment for cyclical effects.
Even though I disagree with Krugman on whether Japanese fiscal policy has been highly expansionary, I’m willing to cut him some slack and assume that he became increasingly convinced of the importance of the cyclical adjustment issue in Japan, and also that the output gap is very large. Even so, this won’t explain his more recent shifts on the eurozone, Britain, and the United States.
2. The eurozone: Krugman used to hold fairly conventional New Keynesian views on the primacy of monetary stabilization policy. In 2009 Krugman suggested that once interest rates hit zero, conventional monetary policy was ineffective and fiscal policy became the key stabilization tool. When the euro zone adopted fiscal austerity and then fell back into a “double-dip” recession, Keynesians like Paul Krugman argued that this provided confirmation for the Keynesian model, and refuted the views of the austerians.
Unfortunately, the facts don’t fit this interpretation. The euro zone was not at the zero bound even before they began raising interest rates in 2011. Thus it was the monetary policy tightening of 2011 that drove the eurozone into the double dip recession, not the fiscal austerity. At one time, this market monetarist interpretation would have been viewed as standard Keynesian analysis. When interest rates are positive, and especially when the central bank is raising interest rates, we presume that the central bank is driving changes in aggregate demand and nominal GDP. But in their zeal to pin the blame for the double dip recession on fiscal austerity, Keynesian economists like Paul Krugman seem to have simply forgotten their earlier claims that monetary policy drives aggregate demand at positive interest rates and fiscal policy takes over only at the zero bound.
Even worse, they ignored the fact that the US did roughly equal amounts of austerity as the eurozone and didn’t suffer a double dip recession. Of course the only important policy difference between the US and the eurozone was the fact that the Fed adopted a more expansionary monetary policy than the ECB. The Fed did not raise interest rates in 2011, indeed it engaged in quantitative easing.
3. Britain: In 2010 a conservative government took over in Britain. At that point, Krugman became highly critical of what he saw as a foolish policy of austerity. I did some posts arguing that British fiscal policy was not particularly contractionary, as they had some of the largest budget deficits of any country in the world. Again you could argue that the policy was more contractionary than it looked, on a cyclically adjusted basis. I was skeptical of these estimates as they assumed a large output gap, whereas the British employment situation wasn’t really any worse than most other developed countries. The large output gap was mostly a productivity story.
Krugman strongly disagreed with this interpretation. In late 2012 he insisted that British fiscal policy really was very contractionary, and was completely dismissive of claims that perhaps the output gap was not all that large (and hence the cyclically adjusted budget deficit was still very large.) Mark Sadowski directed me to this October 2012 Krugman post:
So: the starting point here is the official estimate by the Office of Budget Responsibility that Britain right now has an output gap of less than 3 percent. This is a remarkable assertion, when you bear in mind that real GDP remains well below its level pre-crisis, and that we used to think that Britain’s long-run growth rate was around 2.5 percent. As the CE guys say, simple trend projection would indicate a shortfall of 14 percent; how did that become less than 3?
Part of the answer is the assertion that the UK economy was operating well above sustainable levels in 2007, even though there were none of the usual signs of overheating. Beyond that, however, is the claim that the financial crisis somehow reduced potential output by a huge amount. As CE says, there is no plausible story about how that might have happened.
Then in 2013 the British economy began growing much faster, and now has what the Economist magazine calls a recovery that is the envy of the developed world. Now the Keynesians began insisting that British fiscal policy in 2012 actually wasn’t all that contractionary. Here’s Britmouse expressing frustration with the Keynesian flip-flop:
I find it astonishing that Krugman and Wren-Lewis, having done post after post in 2012 describing how the UK does have real fiscal austerity in 2012, are suddenly happy to now argue that a relaxation of fiscal austerity in 2012 is the “reason” for GDP recovery in… erm, 2013.
In this comment, Britmouse provides links documenting the shift in Wren-Lewis’s position.
Sadowski also directed me to a more recent Krugman post, where he argued that Britain was doing better because fiscal policy in 2012 was actually not contractionary:
But the basic picture is surely right: Cameron/Osborne imposed a lot of austerity in their first two years [2010 and 2011], then let up substantially. In effect, they spent a while banging Britain’s head against the wall, and are now claiming vindication, because it feels good when they stop.
Read the entire post for context.
Even if you put this flip-flop in the most favorable possible light, it doesn’t help the Keynesian position. Suppose it were true that it was impossible to know in 2012 whether fiscal policy was contractionary or not. We wouldn’t know until much later when better estimates of the output gap came in. In that case fiscal policymakers would be flying blind. How can you use fiscal policy as a stabilization tool if some of the most knowledgeable Keynesian economists in the world cannot even identify whether fiscal policy is contractionary or or not during the year in which it’s being implemented?
4. The USA: In early 2013 Mike Konczal argued that 2013 would provide a test of the market monetarist proposition that changes in fiscal policy would typically be offset by shifts in monetary policy. At the beginning of 2013 numerous taxes were raised and in April the sequester kicked in. Krugman agreed that 2013 would provide a test of market monetarism.
I’ve always been a bit skeptical of this sort of test, arguing that there is no “wait and see” for policy initiatives. The markets immediately provide us with the best estimate of the expected impact of the policy, and any results the deviate from market forecasts are presumably due to random shocks. But I also understood that it would inevitably be seen as a test of market monetarism. So I gave my prediction that GDP growth in 2013 would be about the same as in 2012, as monetary stimulus offset fiscal austerity. Keynesian models predicted a sharp slowdown due to the fiscal austerity.
We now know that I was a little bit off, both real and nominal GDP growth in 2013 actually exceeded the pace of 2012. At that point the response of Keynesians (including Krugman) seemed to be something like; “what makes those crazy market monetarists think that 2013 provides a test of their theory?” That takes some gall given that the test was their idea. Again, history is rewritten, now the market monetarists who were bit skeptical of this test are seen by pundits like Brad DeLong as the people who naïvely believe a single data point somehow confirms their theory. In fact this “one data point approach” has been the Keynesian modus operandi from the beginning. Austerity in the eurozone, then double dip recession in the euro zone, ergo austerity must be to blame, even though the US did just as much austerity and only avoided the double dip recession due to a less contractionary monetary policy. Ceteris paribus only comes into play when needed to rescue the Keynesian model, otherwise it’s ignored.
So we have a record of Keynesians redefining the stance of fiscal stimulus whenever the results don’t seem to confirm their theory, as in Japan and Britain. Or they redefine the basic theory of AD determination so that fiscal stimulus can somehow still be effective even when central banks are raising interest rates, as in the euro zone. Or they propose tests, and then when the test results don’t come in as expected they blame the other side for naïvely believing that fiscal austerity followed by increasingly rapid GDP growth was any sort of test all. I could go on and on, but I don’t want to abuse my privileges over here at Econlog. This post at TheMoneyIllusion is another important example of events contradicting the Keynesian model.
PS. Some might argue the euro zone austerity thesis is not a single data point. Krugman sometimes cites cross-sectional studies of austerity and growth. These studies ignore the effect of monetary offset, and hence are essentially worthless. Mark Sadowski shows that the results go away in a regression that does account for monetary offset.
PPS. The figure Keynesians generally use, RGDP growth, rose from 1.95% in 2012 to 2.74% in 2013. The figure their models imply the Keynesians should use, NGDP growth, increased from 3.8% in 2012 to 4.15% in 2013. Some Keynesians dismissed the strong third quarter as an inventory story. Not so, they just announced Q4 RGDP growth at 3.2%.
HT: Mark Sadowski, Britmouse, Vaidas.