Keynes and Friedman are always right: a reply to DeLong & Krugman

Brad DeLong has an interesting rule of thumb: if you want to understand the economy since the mid-1990s, Paul Krugman is always right. And if you ever think he’s wrong … Don’t.

I disagree, and a detour into how much Hicks, Tobin, Hansen or Wicksell is involved is a distraction. Krugman goes awry – and here is a profound irony – because he is overly influenced by Lucas. Yes, Lucas.

One of the strangest developments in the history of economic thought is that the predominantly Keynesian economics profession completely capitulated to the ideas of Robert Lucas, when the institutional structure on which these ideas depended was being dismantled – precisely because Lucas’s diagnosis was right for the time.

First let’s be clear that Krugman has been more right than most about post-crisis macroeconomics. These are the issues that mattered: 1) QE never threatened a major inflation problem; 2) there was never a risk of a bond panic outside of the Eurozone – in fact the opposite was more likely; 3) when interest rates and inflation are extremely low (near zero), fiscal policy is more powerful than conventional monetary policy (including QE); 4) Because 1-3 are true, tightening fiscal policy is a form of self-harm.

I agree with Krugman that 1-4 should be reasonably clear to someone familiar with a high school or undergraduate economics textbook. (The exception is the Eurozone – understanding that a government doesn’t have a credit problem if your inflation-targetting central bank is buying more debt than your government is issuing, requires basic deductive reasoning – knowledge of economics seems to impose additional cognitive burden.)

Where I disagree with Krugman is that the diagnosis of 1-4 is clearer than he makes out – and this difference is substantial because it leads to different policy prescriptions. It’s true that a simple (Lucas-free) IS-LM framework gives mostly the right answer. But you don’t even need that, and I’m with Simon Wren-Lewis on ditching the LM curve.

I use a different rule of thumb to Brad: Keynes and Friedman are always right.

The reasons why QE has negligible sustained effect on inflation is explained most clearly by Friedman (1968), when discussing wealth effects and changes in the money supply:

“This revival [of belief in the potency of monetary policy] was strongly fostered among economists by the theoretical developments … that pointed out a channel – namely, changes in wealth – whereby changes in the real quantity of money can affect aggregate demand even if they do not alter interest rates. These theoretical developments did not undermine Keynes’ argument against the potency of orthodox monetary measures when liquidity preference is absolute since under such circumstances the usual monetary operations involve simply substituting money for other assets without changing total wealth [i.e. QE]. [italics added]”

Friedman even repeats this point in a footnote, contrasting QE with helicopter drops:

“Open market operations are different, because they result merely in a substitution of one type of asset for another.’”

So here is Krugman’s first key error. He is absolutely right to make the empirical observation that QE in Japan had no impact on inflation or growth, but he was completely wrong about why. He – likely many other New Keynesians – is a child of Lucas, obsessed with expectations. Now he won’t like me saying this. He doesn’t like Lucas. And Keynes believed in expectations. But there’s the rub. Keynes never said that a central bank could change an economy by doing something that has no discernible effect simply by “committing to being irresponsible”. I hate to say it, but that’s the New Keynesian version of the Austerians’ “confidence fairy”.

To be clear: Krugman did not argue that QE has no effect because net wealth is unchanged – which is Friedman’s argument. Krugman argues that the BoJ didn’t commit aggressively enough to raising inflation expectations – with all the emphasis on whether or not the increase in base money is “permanent”.

I totally disagree with that as a policy prescription, and also as a viable policy. “Permanence” has little meaning in the context of expectations about QE, given the uncertainty. And I think Keynes would view this policy prescription as confused. Keynes made it clear how to get out of a recession when interest rates have run their course (if you don’t have infrastructure, do public works; if you have enough infrastructure, bury notes and get people to dig them up – or just post a check). Friedman in 1968 also thought this self-evident: at zero rates and in deflation he thought monetary policy did work – if it involved increasing household net wealth through money-financed transfer payments.

The expectations error: under-estimating Pigou, over-estimating Lucas
Now it’s probably worth saying a little more about how Bob Lucas led most of the Keynesian economics fraternity, the great Krugman included, up a blind alley. Of course Lucas alone is not responsible for this, so too is Muth, Friedman, Phelps and others. But I’m blaming Lucas because he had the ‘best’ models – and that’s the real reason why Keynesian economists became obsessed with expectations.

Now some will counter that expectations are central to the General Theory. Yes they are, and here I would deploy my rule of thumb: Keynes is always right. When Keynes writes eloquently about profit expectations, or “vague panic” and “fears”, he is not talking about beliefs which policy makers can change with “announcements”. In Keynes’s world view, if you want to raise profit expectations – do something that has a high probability of raising profits. Keynes has uncertainty at the heart of beliefs and expectations formation. This is a totally different take on expectations, which new Keynesians think central banks’ control. For example, there are new Keynesian economists who actually think that the central bank can raise the rate of inflation simply by announcing that it wants a higher inflation rate, even though it is currently completely failing to meet its inflation target! Tony Yates discusses growing support for this idea. Keynes would view that as beyond silly – we know this because he never advocated it.

The irony is that a policy-driven view of expectations was a helpful diagnosis in the 1960s and 1970s. The New Keynesian error was a failure to recognise the institutional dependence of this model. In a unionised world, with capital controls, with large, often state-owned monopolies operating in many industries – i.e. the 1970s – a simple view of forward-looking expectations determining short-run growth and inflation dynamics is the right way to think about the economy.

To put it bluntly: if you have economy-wide wages being set by collective wage bargaining and national incomes and wages policies, and large shares of industry are protected or state-owned monopolies – government attempts (even covert attempts) to raise inflation will feed-through quite quickly. Partly because of this correct diagnosis, governments – and societies – dismantled these institutions.

This institutional structure does hold still in some countries – Argentina for example. New Keynesians are right in Argentina: policy makers determine inflation expectations and inflation.

But they’re wrong in most of the developed world. What would happen, for example, if the UK economy slows down and inflation starts to fall further away from the Bank’s target, and the Bank of England (on advice from Olivier Blanchard) says that it is raising its inflation-target from 2% to 3%? Most people don’t know what inflation is (they certainly don’t know the current number, and have little grasp of how its determined – including economists) and they don’t know what the Bank of England does, other than changing their mortgage rates – which it stopped doing years ago. Trade union leaders know about inflation, but they don’t coordinate wages anymore. And firms don’t set prices based on what the Bank of England says (sorry Mervyn).

A debate would start about their chances of success. There would be a spectrum of opinion. If all the Bank had at its disposal was more of the same – more QE, maybe another 75bps of rate cuts – people, would say, “Well if they can’t hit 2%, why should I expect them to hit 3%”. Hard to argue with that. (Which is not to say, of course, that you might not get a demand boost from a 1.25% cut in base rates.)

It is for very similar reasons that most central banks (ok, all central banks) got the 2007/8 oil price shock wrong. Oil price shocks mattered in the 1970s – to inflation – they don’t in deregulated, competitive, globalised economies. Because the economic structure has changed. We don’t live in an indexed, price-setting world. We live in a deregulated, price-taking world with profound and rapid competitive technological threat.

Why does all this matter? Particularly, if Krugman was essentially right about fiscal and monetary policy post crisis? Mainly because policy was sidetracked, and still is, in three areas: an obsession with irrelevant “inflation expectations”; with QE – it provides liquidity in a panic, then it signals rates and influences risk premia, end of; and the prevailing Keynesian policy prescription narrowly focuses on infrastructure – we all agree it’s a good idea, but also that it takes years, often decades, to implement and there are huge vested interests who block it.

With the correct diagnosis, the economics profession could have rallied around Friedman’s Keynesian solution: transfer payments to all households financed with base money. Forget Hansen, Hicks, Tobin, and Wicksell. It’s simpler than that: money-financed transfer payments raise demand. Central banks can do it quickly. There are no vested interests who lose: this is a “failure of the mind”.

Not only would the recessions have been shorter, the recoveries stronger, but the intellectual debate might have focused on under-researched areas, such as the relative merits of monetary policy centred around payments to households versus one centred around interest rates and QE.

In conclusion, Brad’s rule is good – very good. But maybe there’s a better one: Keynes is always right, and (1968) Friedman provides clarification.

Postscript
Re-reading Krugman’s seminal paper on Japan. I am reminded how insightful and in many ways prescient it was. Consider, for example, this remark on the possibility of a negative equilibrium real interest rate:

“An immediate question is how this can happen in an economy in which … productive investment can take place and the marginal product of capital, while it can be low, can hardly be negative.

One answer that may be extremely important in practice is the exis- tence of an equity premium. If the equity premium is as high as the historic U.S. average, the economy could find itself in a liquidity trap even if the rate of return on physical capital is as high as 5 or 6 percent.”

This fact is lost on many commentators in the US today.

I also find ample support of my current thesis, however. Consider this:

“The ineffectuality of monetary policy in a liquidity trap is really the result of a looking-glass version of the standard credibility problem: monetary policy does not work because the public expects that whatever the central bank may do now, given the chance, it will revert to type and stabilize prices near their current level. If the central bank can credibly promise to be irresponsible – that is, convince the market that it will in fact allow prices to rise sufficiently – it can bootstrap the economy out of the trap.”

It is also worth bearing in mind that Krugman assumes that the liquidity trap in Japan has emerged because the equilibrium real interest rate is negative. It follows that the only way out is positive inflation. That is true. I would still argue, however, that the best way of generating the inflation – if it is needed – is through money financed transfer payments. One caveat is warranted. One point repeatedly ignored in analysis of Japan is that because JGBs are held by the private sector, the huge public sector debt/GDP ratio is also a huge private sector wealth/GDP ratio. So it is fair to ask: why if you give people cash will they spend it, but not if you give them bonds. Well, bonds and cash are not perfect substitutes, and different people hold them. I would argue that because base money is not a liability, Ricardian equivalence does not apply to money financed cash transfers – it is an increase in consolidated net wealth. Also, the ownership of bonds is intermediated, which is not trivial.