Wrong in a very confusing way

There are lots of macro models out there: old monetarism, market monetarism, old Keynesianism, new Keynesianism, supply-side economics, Fiscal Theory of the Price Level, NeoFisherism, Austrian, Real Business Cycle, etc., etc. People who believe in one tend to view the others as being at least partly wrong. But where they disagree, it’s usually possible to pin down some specific points of disagreement.

MMT is not like that.

In his Twitter account, Paul Krugman again tries to show what’s wrong with MMT, by pointing to a specific example of how very little of the national debt is financed by printing money.

I’ve made similar arguments on a number of occasions, as has Nick Rowe and many other people. But I’m increasingly coming to the view that none of this will work. MMT has constructed such a bizarre, illogical, convoluted way of thinking about macro that it’s almost impervious to attack. Krugman’s right that a reasonable person would view his evidence as demolishing their claims about fiscal policy—but it won’t be seen that way.

As far as I can tell, MMT created this monster by combining the following:

  1. Bizarre (and unconventional) definitions of terms.
  2. The tendency to confuse accounting relationships with causal relationships.
  3. Being wrong about basic questions of causality.

If it was just one problem, it would be easy to figure out how to attack the model. Thus Krugman is trying to present evidence that they are wrong about certain causal relationships. But because they define terms differently than the rest of us, this evidence will have no effect on their views.

For instance, normal economists would think about government spending being financed by a mixture of taxes, debt and money creation. AFAIK, MMTers think spending is paid for with money creation. When they describe their views in detail, however, it looks like they believe that spending is paid for with taxes, debt and money creation, not just money creation. They simply characterize that fact differently. So how to attack this view?

If you say, “You’re wrong, spending is paid for with some mix of money creation, debt and taxes,” they’ll respond, “no, it’s just paid for by money creation.”

If you bring to bear all sorts of evidence that implies that spending is paid for by all three, they’ll respond, “We know all that, it’s in our model. Taxes are used to drain money from the system to prevent inflation.”

I’m struggling to think of an analogy from everyday life—perhaps someone can help me. But try this:

I ask my friend, “Did you pay for that new car out of your savings, or did you have to borrow the money? And your friend responds, “Neither, I paid for it with a check”. You say, “I get that, but where did the money for the check come from?” And the conversation keeps going around in circles.

So one problem is their weird definitions, insisting that government spending creates money that pays for the spending, which is based (AFAIK) on a misinterpretation of the implications of an accounting relationship involving the Treasury account on the Fed’s balance sheet.

But there are substantive problems too. They seem to not understand that when nominal rates are positive, high-powered money is several orders of magnitude more inflationary than T-securities. (High-powered money is zero interest base money in a positive interest rate environment). They don’t seem to understand the Fisher effect, and instead assume that flooding the economy with base money drives interest rates to zero. While it’s true that you can flood the economy with high-powered money when the equilibrium nominal rate is zero, if it isn’t zero then you’ll create lots of inflation and thus much higher interest rates—as the UK discovered in the 1970s. Thus here is Stephanie Kelton, ignoring the Fisher effect:

It helps to break the argument into a two-part thought experiment. First, think about what happens if the government is running huge budget deficits. As I explained, these deficits would result in a massive injection of reserves into the banking system. Unless something is done to prevent it, banks will scramble to offload the excess funds in the overnight market. But with massive supply and no demand for these balances, the overnight bid heads toward zero.

If you foolishly follow the “As I explained” link, you will find no explanation at all, merely a repetition of the bizarre claim that deficits lead to a massive injection of new reserves. (I think she is assuming a money financed deficit, but who knows?) And here is Kelton misrepresenting the views of Krugman:

He called our nation’s finances “a fiscal train wreck” and confessed, “I’m terrified about what will happen to interest rates once financial markets wake up to the implications of skyrocketing budget deficits.”

He insisted that the U.S. faces, “a looming fiscal crisis,” adding, “the only question now is when foreign investors, who have financed our deficits so far, will decide to pull the plug.”

He mused about the potential for accelerating inflation under quantitative easing, writing that the Fed is, “printing $1 trillion of money, and using those funds to buy bonds. Is this inflationary? We hope so!”

He asked, “couldn’t America still end up like Greece?” answering, “Yes, of course. If investors decide we’re a banana republic whose politicians can’t or won’t come to grips with long-term problems, they will indeed stop buying our debt.”

And he puzzled over the different interest rate environments in Japan and Italy, asking, “Why are the interest rates on Italian and Japanese debt so different?” confessing, “I actually don’t have a firm view. But it seems to be an important puzzle to solve.”

No economist is going to get everything right. But the odds of getting things right improve dramatically when you’re working with a macro framework that doesn’t lead you astray. 

I have often criticized Krugman, but I don’t think I have ever misrepresented his views so egregiously. Look how the QE quotation is taken out of context. Implying that Krugman was a QE/inflation scaremonger doesn’t even pass the laugh test. He’s probably the world’s most famous QE skeptic. He was merely pointing out that inflation was the goal of QE. Perhaps he thought it would produce a tiny amount of inflation, but it did!

So don’t tell me she’s just quoting Krugman. Her snarky “No economist is going to get everything right” tips her hand. She’s accusing Krugman of making lots of false predictions, in areas where MMT is “correct”.

MMTers would say that Krugman doesn’t understand the distinction between Italy using the euro and Japan having its own currency (with zero default risk.) In fact, it’s the MMTers that don’t understand that having your own currency doesn’t guarantee low rates if investors believe that the only way you’ll be able to handle your debt is via inflation.

FWIW, I suspect one difference is that Japan has far lower government spending than Italy, and thus more room (fiscal space) to raise taxes before relying on money creation. And Japan has less to worry about in terms of their best people moving to better run eurozone countries. But having fiat money is no cure-all. The UK discovered this in the late 1970s, when the IMF bailed them out. So Krugman’s puzzlement was not completely unjustified.

The quotes provided by Kelton don’t provide specific dates, and thus one can’t really say that Krugman was “wrong”. In fairness to Kelton, Krugman probably did think the day of reckoning was coming sooner than it has, but then during the 1990s very few people predicted the super low interest rates of the post-2008 period. That sort of mistake is hardly indicative of a flawed model.

PS. In the US, I expect the day of reckoning to take the form of higher taxes, not high interest rates/high inflation.

We are in an NGDP factory

Arnold Kling has a new post discussing the issue of wages and productivity. He argues that productivity is increasingly difficult to measure:

We are not in a GDP factory. As the share of GDP devoted to health care and education goes up and the share devoted to manufacturing goes down, we are giving more weight to a sector where real output and the quality of labor input are extremely difficult to measure.

This is also my view, if GDP means RGDP (which, in context, it clearly does.) Indeed Kling and I are probably out on the extreme, in terms of being especially skeptical (relative to other economists) of the usefulness of measures of RGDP, real productivity, the price level, etc.

Over at Econlog, I argue that while the problem of measuring real productivity is very real, it’s largely unrelated to the so-called “wage decoupling issue—which is mostly about the gap between nominal wages and nominal productivity. Fortunately, errors in measuring real productivity have no impact on measures of nominal productivity.

So while I entirely agree with the Kling quotation above, I also strongly believe the following to be true (please read carefully):

We are in an NGDP factory. As the share of NGDP devoted to health care and education goes up and the share devoted to manufacturing goes down, we are giving more weight to a sector where nominal output and the quantity of labor input are relatively easy to measure.

So why is this second claim so different from the Kling quotation? Because RGDP and NGDP are radically different concepts, almost unrelated. Thus we can say with some confidence that the nominal health care industry has expanded from (say) $36 billion to $3.6 trillion since the mid-1960s, but I have absolutely no idea how much real health care output has grown, as I don’t even know what the output of the health care industry is. What are they trying to produce? (Recall that Robin Hanson says that health care is not about health.)

And even if health care is about health, how much of the improvement in health is due to health care, and how much is due to less smoking, better nutrition, better sanitation, etc.

In many ways, NGDP is not a very interesting variable (think Zimbabwe), except when it shows short run volatility. In that case, it destabilizes labor and financial markets, because there are lots of nominal wage and debt contracts. In those situations the “NGDP factory” is a useful concept, even though NGDP is also measured imperfectly. But NGDP is at least an order of magnitude more clearly defined and more easily measured than RGDP.

I do occasionally refer to RGDP, as despite its flaws it tells us something about business cycles and international comparisons. A sudden drop in RGDP usually indicates a recession, as all those imponderables associated with measuring “real heath care” don’t change much from one year to the next. Think “law of large numbers in BEA errors”. The BEA bureaucrats are making similar errors, one year after another. Thus it’s a problem if measured RGDP suddenly falls by 5%, despite the many flaws in RGDP data.

As far as international comparisons, a country with a $50,000 per capita GDP is not necessarily richer than one with a $45,000 per capita GDP, but it is almost certainly richer than a country with a $5000 per capita GDP. So RGDP has some value, if used with care.

Most importantly, don’t ask any statistic to do more than it can.

neutral inflation/real inflation

Over at Econlog, I have a new post pushing back against the view that inflation is caused by “global factors” such as economic slack. I point out that the UK and Switzerland have faced the same global factors since 1971, but have had vastly different inflation rates, due to vastly different monetary policies.

This confusion is actually a symptom of a deeper problem, the tendency of many economists and other pundits to view inflation as a sort of “real phenomenon.” At a fundamental level, inflation is nothing more than a change in the value of the medium of account, sort of like switching from using feet to using meters. It doesn’t change the size of objects being measured. People in Turkey and Argentina have adjusted to inflation that is higher than in India, and Indians have adjusted to inflation that is higher than in the UK, and the Brits have adjusted to inflation that is higher than in Switzerland.

In the 1980s, Americans easily adjusted to 4% inflation, as their wages rose faster than today. The economy was not “overheating”.

In the short run, unexpected changes in inflation can have real effects on output due to wage/price stickiness, nominal debt contracts, money illusion, etc. (It’s as if the size of the objects being measured changed as a result of switching from feet to meters.) The mistake many people make is to work backwards from these short run real effects and assume they actually cause the inflation (or deflation)!

No, no, no, a thousand times no! An overheating economy does not cause inflation. Economic growth is deflationary. Rather, higher than expected inflation can cause an economy to overheat. But you can also have high inflation without any overheating at all, indeed even with lots of “slack”.

Inflation is caused by monetary policy, and that monetary inflation might or might not have a wide range of real effects associated with it, depending on all sorts of contingencies. If I read that Argentina has 20% inflation, I don’t immediately think “overheating”, I think “printing money to pay the bills.”

Many New Keynesians rely on models where inflation is associated with some sort of overheating. In fairness, those models generally have some sort of natural rate component, and include expectations of inflation. They are still “reasoning from a price change”, and thus highly flawed, but at least they understand that money is neutral in the long run and that economic slack doesn’t explain long run changes in inflation.

Old Keynesian models are even worse (and AFAIK) this also applies to MMT. They simply assume that inflation is not a problem as along as there is economic slack. Milton Friedman once said that in 200 years we’d merely gone one derivative beyond Hume. Unlike the monetarists and New Keynesians, lots of pundits and economists still haven’t even incorporated the first derivative into their models.

And while looking at changes in inflation is better than looking at the absolute level of inflation, it’s still not good enough. It’s still reasoning from a price change. We need to focus on changes in NGDP growth, which may or may not have real effects, depending on all sorts of factors. But at least with a model of:

Monetary policy —> NGDP —> inflation/unemployment/RGDP

we are on the right track.

If you begin with:

Economic slack —> inflation/deflation

then you aren’t even on the right track.

PS. Consider the following scenario. During an economic recovery, RGDP grows as 3% as unemployment falls rapidly. Once full employment is reached, growth slows to 2% and unemployment stabilizes. A soft landing. Now assume NGDP grows at 5% throughout this entire period. What happens to inflation?

Inflation would rise from 2% to 3% once we had reach “full employment.” But this is NOT at case of strong growth pushing inflation higher, indeed in my example the rise in inflation was caused by a slowdown in economic growth (given the stable path of NGDP.)

As I predicted

This morning I read a post by noted MMTer Stephanie Kelton, which responded to some questions by Paul Krugman. I predicted that he’d be pulling out his hair, as her responses were a complete mess—as if she didn’t understand the simple questions he was asking. Here’s an excerpt from my Econlog post:

I can’t even imagine Krugman’s reaction to all this.  First of all, although she says “no” in answer to question #1, her explanation makes it quite clear that she is actually answering “yes”, especially when you combine the answers to questions #1 and #2.  Krugman is asking whether, assuming a given macroeconomic situation (including a given level of private spending), there is only one budget deficit consistent with full employment.  She clearly thinks the answer is yes.  So why does she answer “no”.  I suspect she doesn’t understand Krugman’s question (which is pretty straightforward.)

Now Krugman has responded and he’s every bit as frustrated as I expected. Here is his response to the first point I raised:

Her response to my first question totally misses the point; I was asking if *given private behavior* there’s a unique level of the deficit needed for full employment, and argued that there wasn’t. She just assumes that there is

And here’s how Krugman summarizes his twitter thread:

Sorry, but this is just a mess. Kelton’s response misrepresents standard macroeconomics, my own views, the effects of interest rates, and the process of money creation. Otherwise I guess it’s all fine. See what I mean about Calvinball? 6/

MMT desperately needs a spokesperson capable of conversing with economists like Paul Krugman, if they want to be taken seriously. When they give bizarre answers to serious questions from a highly respected economist on their side of the ideological spectrum, it’s a problem.

Not surprisingly, I think it’s perfectly fine to be heterodox, but first you need to understand orthodox.

Are “normal” yield curves actually normal?

The US yield curve usually slopes upward. Hence positively sloped yield curves are termed ‘normal’ and negatively sloped yield curves are termed ‘inverted’. But are “normal” yield curves actually normal?

The US business cycle has an unusual feature, an absence of soft landings. A soft landing is when unemployment recovers to the natural rate, and then stays low for at least four years. The only plausible example is 1966-69, but that wasn’t really a soft landing because it came at the expense of rapidly accelerating inflation. The plane soared off into the stratosphere before the wheels even touched the runway. Otherwise, we have only a year or two of stable and low unemployment, before the next recession.

He not busy being born is busy dying — B. Dylan

So 90% of the time we are either recovering (mostly with a normal yield curve) or in recession (mostly with a normal yield curve.) Inverted curves tend to occur right before a recession, when unemployment is low. But what if we did have a true soft landing—persistent low unemployment without accelerating inflation? Would the yield curve be flat? After all, when output is already near the natural rate, you don’t expect further recovery. You don’t expect the future economy to be better (stronger) than the present.

So I decided to seek out a country that did have a soft landing—Britain from 2001-08:

2001-08: Not busy being born or busy dying

Notice that in 1991, the UK entered recession almost immediately after the unemployment rate stopped falling—the US pattern. But in 2001, unemployment reached the low 5s and stayed near there until mid-2008. So how about the UK yield curve? Below I have data from 2000-08, using June data in each case. The first number is the nominal yield on 6-month UK T-bonds, and the second number is the yield on 10-year bonds. I’m not used to using UK data, so someone tell me if I made a mistake:

2000: 6.0%, 4.13%

2001: 5.33%, 4.92%

2002: 4.38%, 4.80%

2003: 3.19%, 4.87%

2004: 4.89%, 4.93%

2005: 4.16%, 4.28%

2006: 4.70%, 4.47%

2007: 5.85%, 4.94%

2008: 5.26%, 5.12%

During these 9 years, the UK yield curve was slightly inverted, on average. But pretty close to flat. This suggests that a flat yield curve might be expected when the economy is still expanding, but the unemployment rate is not expected to fall much further.

Note that the US yield curve was pretty flat in May 2006, when unemployment had fallen to 4.6% and was expected to stay around that level. And unemployment did stay around that level for another 18 months. The yield curve was also fairly flat in 1966. The yield curve today slopes gradually upward, but it’s flat between 2 years and 5 years, which suggests to me that markets may be expecting the US to achieve a soft landing.

The unemployment rate today (4.0%) is not much different from 15 months back (when it was 4.1%.) I expect it to fall a few more ticks (the trend is actually in the high 3s), but then stop falling. If I’m right that the US will achieve its first ever non-inflationary soft landing, then I’d expect a pretty flat yield curve to be the new normal. There’d still be a slight tendency for an upward slope, but not much.

Of course this happy scenario assumes that the Fed keeps NGDP growing at a reasonably steady rate (say a range of 3% to 5%). There are no guarantees, but they seem to be trying to do something like that in recent years, even if they don’t admit it.