The Flaw At The Heart Of Keynesian Economics

by: Scott Sumner

The term "Keynesian economics" means different things to different people. But one thing almost everyone can agree on is that Keynesians believe that the 2009 fiscal stimulus boosted aggregate demand, at least compared to the no-stimulus alternative. Alternatively, Keynesians believe the fiscal multiplier is positive. In my view that's a sort of sine qua non of modern Keynesianism. Yes, it has many other things to say, but without the fiscal multiplier there’s nothing particularly Keynesian about the rest of the theoretical apparatus.

Even if you don't agree with me, surely you've read dozens of pundits, reporters, economists and politicians make the argument for a positive multiplier, and call that argument Keynesian. Indeed many go further and suggest that only unenlightened conservative ideologues could question something so obvious, so well established by both theory and empirical studies.

I'm here to tell you that it's all a fraud. There is no empirical study that shows the 2009 stimulus was effective. It's not even clear new Keynesian theory implies it was effective. It might, but it also might not. There is nothing scientific about "the multiplier." And many of the arguments made by pundits (with a few notable exceptions like Avent and Yglesias) are deeply misleading to readers.

Let's start with the easy part. New Keynesian theory predicts that the fiscal multiplier will be zero if the central bank is targeting inflation in a forward-looking fashion. That is, increased deficit spending will not increase expected future growth in aggregate demand. The smarter Keynesians know this, but the smarter Keynesians are a very, very small group. If you polled PhD trained economists in America, I'd guess less that 10% know this, maybe less than 2%.

Even worse, many of the Keynesians who do know this fail to mention it in most of their "pro-stimulus" screeds, thus giving average readers the impression that a positive multiplier is the default assumption, and that it's up to those who disagree to explain why. No, it's up to those who believe fiscal stimulus would cause the Fed to stop targeting inflation (or stop Taylor Rule-type policies) to explain why they believe this.

Long-time readers of mine know that the monetary offset problem quickly degenerates into debate over whether the Fed would sabotage fiscal stimulus. I don’t like that framing, because it implies monetary policy forces the Fed to "do something" to offset fiscal stimulus, or more precisely to "do something that looks like it’s doing something" to offset fiscal stimulus.

In a recent comment section, Statsguy asked if I thought the Fed would have offset a lack of fiscal stimulus in 2009. I do think so, indeed I believe that the current unemployment rate would probably be lower than 7.7% if there had been no fiscal stimulus in 2009. I will explain why, using what I hope everyone will see as eminently reasonable assumptions. I'm not claiming that I know exactly what would have happened, but rather that quite plausible counterfactuals can leave us with a negative multiplier, contrary to the claims of Keynesians. Here are my assumptions:

1. In early 2009 there were fears of depression. Stock prices had collapsed and unemployment was soaring. People were frightened.

2. If the fiscal policymakers provided no stimulus, Bernanke and the Fed would have felt an incredible burden to save the economy.

3. However, I would not expect the Fed to do anything that looked radical or dangerous.

4. I would have expected them to take off the shelf the most highly regarded models of how to save an economy with monetary policy when stuck at the zero bound. That would be the Woodford model, which calls for level targeting of prices.

5. I would have expected them to use at least some of the ideas that Bernanke suggested the BOJ employ. One of those was level targeting of the price level.

6. They would have been reluctant to abandon the 2% inflation target, with level targeting of the price level they would not have had to.

7. Do you see where I'm going? Level targeting of the price level along a 2% path is the overwhelmingly most likely "nuclear option" to be employed by the Fed if the economy is falling off the cliff and the fiscal policy makers are doing nothing.

8. The level targeting theory suggests you want to make up ground lost in the crisis period where monetary stimulus is (supposedly) ineffective. I think they would have started the trend line for the price level in September 2008, which is both the peak month for the price level (it started falling in October) and also the month the financial crisis blew up.

9. The Fed uses the PCE price index, so we'll assume they targeted that index (although the CPI or the core PCE would give you similar results.) The PCE on September 2008 was 110.275. That means the PCE for January 2013 should have been 120.165.

10. The actual PCE in January 2013 was 116.342, well below the Fed's likely target under PLT regime. In other words, under the most likely alternative (no fiscal stimulus) policy, prices would have likely risen much faster than the actual path of prices, assuming the Fed was able to hit its target.

11. Because both fiscal and monetary stimulus impact the demand-side of the economy, the depressed price level of January 2013 means, ipso facto, that AD and NGDP have also risen more slowly than under the no-fiscal stimulus PLT alternative. We are worse off after having wasted hundreds of billions of dollars in fruitless deficit spending.

I can anticipate the objections: "What makes you think the Fed would have hit its price level target?" Initially I think they would have failed, but recall that even in the NK model a higher inflation target is far more effective than QE. As prices fell in 2009, the expected inflation rate would tend to rise under PLT. So even using the assumptions of the NK model, the Fed would have been doing a policy that is far more effective than the policy they actually conducted. Surely that counts for something! Indeed, with a higher inflation target you don't even need to do as much QE, other things equal. If you add in the lack of fiscal stimulus, then it's unclear whether the PLT regime I propose would have required more or less QE than what we actually saw.

To be clear, I am not claiming that I "know" that this counterfactual would have occurred. But I think any fair-minded person would admit that it's a plausible counterfactual. Here's what I am claiming:

1. My counterfactual is plausible.

2. Almost all Keynesian discourse on fiscal stimulus, multipliers, neanderthal conservatives, etc., implicitly implies that my counterfactual is not plausible.

3. Hence Keynesianism is nothing but a sandcastle built on a pile of flawed assumptions.

Maybe the 2009 fiscal stimulus helped a lot. That's possible. But don’t pretend that your belief in the 2009 fiscal stimulus is any different from blind faith in some sort of religious dogma. There's no science to back it up, and none of the empirical studies I've seen have any bearing on the counterfactual I just presented. None.

P.S. Remember those Keynesians telling us that higher payroll taxes would slow retail sales in Q1? It looks as if they might want to revise their models.