I’m increasing inclined to believe that the Keynesian way of thinking about the world (that spending shocks drive the economy) is so intuitively appealing that it becomes hard to visualize any other worldview. As a result, Keynesians start to think that any piece of empirical evidence, or logical argument, somehow supports the Keynesian model.
1. One obvious example from a while back is the expansionary effect of tax cuts. Yes, but how do we know whether they are expansionary for demand or supply-side reasons?
2. A much more interesting example occurred recently, when a series of studies showed positive “regional multipliers.” More federal spending in a given area tends to increase regional GDP. I would have thought that rather obvious, as all the anti-Keynesian models that I am aware of would predict the same result. But people like Paul Krugman and Brad DeLong latched on to this as evidence in favor of the Keynesian model. How can it be evidence in favor, if it would also expect positive regional multipliers if the national multiplier was zero? Take the famous Alaskan bridge to nowhere. Does anyone serious doubt that the GDP of that tiny Alaskan hamlet would have increased had the $100 million bridge been built? Even Chicago economists would say yes.
So it seems to me that Keynesians sometimes make the mistake of simply assuming the Keynesian model is true, and then seeing any and all real world events as somehow confirming that belief. I’m not claiming they are stupid; indeed I’m sure if they thought about the alternative explanations for a few minutes they see that those are also logically possible.
Recently we saw another example, when Wren-Lewis claimed that consumption smoothing somehow refuted Cochrane’s claim that the balanced budget multiplier is zero. The problem, of course, is that it doesn’t refute Cochrane, as consumption smoothing is also expected to occur in anti-Keynesian models. Milton Friedman’s Permanent Income Hypothesis is a great example of a consumption smoothing model, and by the mid-1990s Milton Friedman had reached the conclusion that fiscal stimulus is ineffective.
There are of course many reasons why the balanced budget multiplier might be zero. Many of them I find weak, although one (the monetary offset argument) seems pretty persuasive to me. But the important point is not whether you agree or disagree with these non-Keynesian models, what’s important is whether you can understand them on their own terms. Let’s take three quick examples:
1. The supply-side argument: Supply-siders claim that higher tax rates discourage work, and thus lower GDP. They also tend to be skeptical of the efficacy of public sector output. If tax increases do discourage private sector output (C+I), then there is no particular reason why you won’t expect consumption smoothing to occur, with consumption falling by only a modest fraction of the total decline in private sector output.
2. The crowding out argument: In this view the decline in after-tax income will reduce saving, and hence crowd out private investment. Again, this is completely consistent with consumption smoothing, indeed in a sense it assumes consumption smoothing. Investment is assumed to be a big part of the decline in private sector output.
3. My favorite argument is the monetary policy offset argument. Under modern monetary regimes such as inflation targeting, any government attempts to boost AD are offset by less monetary stimulus, and hence AD is unaffected. Again, in that case it’s easy to envision consumption smoothing occurring, as consumer theory predicts it will occur any time after-tax income falls.
So why do Wren-Lewis and Krugman think that consumption smoothing shows Cochrane was wrong? What could they have possibly meant by this argument?
Both make the same simple error. If you spend X at time t to build a bridge, aggregate demand increases by X at time t. If you raise taxes by X at time t, consumers will smooth this effect over time, so their spending at time t will fall by much less than X. Put the two together and aggregate demand rises.
Read these over several times, especially the “put these two together” and the “so their spending at time t will fall by much less.” I’m sorry, but I simply can’t accept the convoluted explanations being offered by their defenders. Let’s do this with numbers to make it easy to see. Wren-Lewis is basically saying that if you raise taxes by $100, then C will fall by much less than $100 (consumption smoothing), so their spending will fall by much less than $100. Assume, for example, that C falls by only $20. Then he says “put these two together.” Correct me if I’m wrong, but isn’t the straightforward interpretation that he was putting together the $100 increase in G and the $20 fall in C, and noting that the fall in C is smaller than the increase in G? Isn’t he implicitly assuming that a fall in C is equivalent to a fall in “spending?” I can’t read it any other way. Otherwise why use the term “so” in explaining the impact on “spending?” I’ll bet 99.9% of readers read it this way.
The best counterargument I’ve seen is from Andy Harless, who suggests that consumption smoothing somehow underlies the Keynesian result. Even if he’s right I’d still claim they made a bogus criticism of Cochrane, is it doesn’t weaken his (non-Keynesian) argument. But is he right? Isn’t the balanced budget multiplier one in the Keynesian model regardless of the MPC? And isn’t the regular multiplier actually higher with a bigger MPC, i.e. a smaller amount of consumption smoothing? Andy understands the Keynesian model better than I do, so I’ll keep an open mind on that issue. But it’s going to be pretty hard to convince me that their “consumption smoothing” criticism of Cochrane makes any sense.
Of course both Wren-Lewis and Krugman could say; “we simply assume the Keynesian model is right, and if you assume it is right, then it is right.” And Cochrane could simply say “I simply assume the Keynesian model is wrong, and if you assume it’s wrong, then it’s wrong.” I hope we don’t fall into that sort of debate in the comment section. Let’s actually evaluate the quality of the arguments, not whether they happen to be true or not.[Update: I obviously meant whether the conclusions happened to be true.] Are they convincing? I say both sides are completely unconvincing. I say both sides simply assume income changes or doesn’t change according to their preconception. And I think many find the Keynesian model so intuitive appealing that they conflate “Keynesian reasoning” with “logical reasoning.”