Scott Sumner  

Coase and Krugman

Henderson at Troy University... Uber Scam?...

There were lots of good answers after my previous post. Commenters dlr (first) and then Rajat provided my preferred answer, and there were some other good options as well (Friedman, Lucas, etc.) Here I'll explain the special connection between Coase and Krugman.

In 1960, Coase developed a radically new way of thinking about externalities. At the time, Pigou's interwar theory of externalities was very well established, almost unquestioned. When a person or company does something that imposes external costs on others, there is a market failure. The optimal public policy is a remedial tax, equal to the size of the external cost.

Coase's alternative view was the sort of shocking "bolt from the blue" that almost never occurs in a mature science like economics. There was a famous seminar at the University of Chicago, where almost everyone went in convinced Coase was wrong, and he convinced them all, one by one. Coase's basic insight is that external costs, by themselves, are not market failures. The victim would have an incentive to bribe the entity imposing external costs. That bribe has a similar impact to an optimal tax. Thus before Coase, economists thought there was an economic rationale for government regulation of indoor smoke. After Coase, economists recognized that the owner of the property, not the government, should regulate indoor smoke.

But Coase did not stop there. He also showed that when many people are harmed by externalities, there may be "transactions costs" in privately negotiating an agreement. In that case, Pigou's suggestion that a remedial tax is needed might be correct. But the real problem is not externalities, it's transactions costs.

Screen Shot 2018-03-17 at 12.52.37 PM.png
In 1998, Krugman came up with a radically different way of thinking about liquidity traps. During the interwar period, Keynes had argued that monetary policy may become largely ineffective at zero interest rates, as money and bonds become very close substitutes. He recommended government actions such as fiscal stimulus.

Krugman showed that even at zero interest rates, monetary injections should be effective. That's because the liquidity trap is presumably not expected to last forever (an assumption that Krugman himself later questioned) and thus an increase in the money supply should raise prices once the liquidity trap had ended. Krugman showed that in a rational expectations model, the mere expectation of a higher future price level would tend to raise the expected long-term rate of inflation, and reduce real interest rates on long-term bonds.
Thus monetary policy would continue to be effective at zero interest rates. And if real interest rates did not decline, then nominal rates would rise, which would end the liquidity trap---also making monetary policy effective. No need for activist governments engaging in fiscal stimulus

But Krugman didn't stop there. He noted that (conservative) central banks might not be able to convince the public that currency injections are permanent. In that case, future expected inflation would not rise, and the monetary injections would be ineffective. Krugman argued that the real problem was not that cash and bonds are perfect substitutes at zero rates, producing a "liquidity trap", but rather that central banks might not be able to convince the public that they will allow higher inflation in the future, creating what Krugman called an "expectations trap."

If there is an expectations trap, then the original Keynesian policy of fiscal stimulus might make sense, but not for the reason assumed by Keynes. This is similar to Coase's argument that corrective taxes might be called for, but not for the reason originally assumed by Pigou. But Krugman also indicated that something like a higher inflation target ("promising to be irresponsible") was a first best policy, and indeed Krugman recently cited Abe's decision to raise Japan's inflation target to 2% as an example of what he had in mind (although Krugman would probably prefer an even more aggressive target.)

I'll add my own wrinkle here. In a 1999 Economic Inquiry piece I argued that the constraints of the gold standard were always lurking in the background of Keynes's thinking. When I first wrote that paper, I was not aware of Krugman's 1998 paper. But in retrospect, I was claiming that the gold standard created a sort of expectations trap, which prevented central banks from raising the expected rate of inflation.

Screen Shot 2018-03-17 at 12.53.41 PM.png
My previous post was not just intended to be a diverting puzzle. I believe that seeing these sorts of underlying similarities allows us to better understand each theory separately. Indeed being a good economist is largely a matter of seeing common underlying factors behind a lot of seemingly disparate phenomena. This sort of intuition is what puts Coase and Krugman ahead of most economists. Coase had written a paper back in 1937, pointing out how "transactions costs" help to explain why firms are big. If transactions costs did not exist, the principle/agent problem would cause films to contract out almost every single specific task they do to other smaller and more specialized firms--even to individuals. Big firms would be almost completely hollowed out.

Then, 23 years later, Coase recognized that these same transactions costs explain why the private sector may have trouble negotiating solutions to complex externality problems.

I can't disagree with people who pointed to Friedman and Phelp's insight that it's not high inflation that matters, but rather higher than expected inflation. That was a really important breakthrough. But the way Coase and Krugman reframed the longstanding dogma on externalities and liquidity traps seems like more of a radical intellectual shift--not just adding one derivative. And in both cases the original policy suggestion became a sort of special case, a policy that is called for when other options are not available.

COMMENTS (11 to date)
Steve F writes:

Even if transaction costs of externalities created by the private sector suggest government regulation as a fix, government regulation creates externalities.

My interocular trauma test shows that the costs of externalities caused by government regulation surpass the costs of the externalities that government regulation attempts to fix.

R.L. Styne writes:

In any discussion of "the Coase theorem" I think it's important to note that Coase himself said "the Coase theorem" isn't his theorem, that many people misunderstand Coase's point, and that everyone should read his essay "Notes on the Problem of Social Cost in his collection "The Firm, The Market, and the Law."

I'm not saying that Scott's point on Coase here is wrong, but we should be careful that we aren't attributing to Coase what actually came from Stigler. Coase's point is much more interesting from a policy perspective, anyway.

Thomas Sewell writes:
In that case, Pigou's suggestion that a remedial tax is needed might be correct. But the real problem is not externalities, it's transactions costs.
An interesting complication is that in some cases, transaction costs are significantly increased by government laws/regulation around the externality. For example, if someone is subject to a fine for some behavior which has been made minorly illegal, than without a bunch of lawyers and some lobbying the "criminal" can't just pay off their "victims" in a Coasian bargain, even if that's what the victims would prefer. An anti-pollution, anti-noise or whatever regulation might in that way actually work more harm while trying to "correct" an externality via a one-size-fits-all process.
If transactions costs did not exist, the principle/agent problem would cause films to contract out almost every single specific task they do to other smaller and more specialized firms--even to individuals.
Implied by this theory is that as transaction costs are reduced, say perhaps by some sort of technology where people can transact asynchronously using pre-built systems which have standardized rules around common transactions (we could call it something like application software using the Internet), then firms will contract out more and more to specialized firms and even individuals. That's an easy prediction to make, as it's what we're seeing in the "gig" economy working around the fringes.

It also makes me think we're just at the tip of the iceberg for that process, as beyond the Ubers of the world, I'm seeing the same thing with markets in everything from trucking to book cover design taking advantage of the reduced transaction costs to become smaller and more efficient.

Cloud writes:

I like this comparison, it is interesting.

I just hope that one day the concept of Transaction cost can be applied to macro and make some breakthrough in the field.

Pollard writes:

I hate to quibble about details, but Coase did not present his theorem at a seminar, it was at a dinner party at Friedman's residence. (See Coase obit. in NYT). This doesn't change the substance of his theorem but it tells you what the social life of Chicago faculty was like.

Scott Sumner writes:

Pollard, Thanks for that correction.

Antischiff writes:

Dr. Sumner,

I mentioned Krugman, but I thought you might be referring to him because of new trade theory. I thought the liquidity trap paper was probably only news to various shades of Keynesians, as RBC and ABC types probably don't care about papers like that, and I thought at least some monetarists(market), like yourself for example, were already there.

I disagree with the statement that economics is a mature science. There isn't enough data for it to be a mature science. If not for this fact, there probably wouldn't be different schools of macro.

Matthew Waters writes:

There are some comments here that externalities are minor, or that government action is innately worse than any externality. 1970's LA is a big counterpoint. Refiners also did not love the elimination of leaded gasoline and lead has had a big effect on childhood development.

The Coase theorem is an interesting thought experiment, but libertarians go astray when they think the zero-transaction-cost model approximates the real world.

Miguel Madeira writes:

I suspect that the big problem with solving externalities with transactions is that many have a public-good element - if you (in the apartment B) make much noise, you can't make an agreement with your neighbour in the apartment A (where he pays you for not making noise) but continuing to making noise for your neighbour of apartment C.

Miguel Madeira writes:

Cloud: "I just hope that one day the concept of Transaction cost can be applied to macro and make some breakthrough in the field."

The whole "sticky prices" thing can be conceptualized as a version of the transaction costs - more, in some way, in ceertain way the whole point of having firms is to make prices "sticky", instead of negotiating the terms for every single transaction.

robc writes:


The brilliance of Coase is that it doesn't only apply to zero-transaction-cost situations.

With transaction costs, it helps determine where the property rights should lie, by minimizing transaction costs.

Comments for this entry have been closed
Return to top