Bartleby the central banker

The Reserve Bank of Australia has delivered 28 years of solid growth in NGDP. Unfortunately, its recent performance has been subpar. Even more worrying is the fact that its communication has been borderline incoherent:

In its quarterly monetary policy review earlier this month, the RBA downgraded a series of economic forecasts, including growth, wages, consumption and inflation, and warned “further easing could unintentionally convey an overly negative view of the economic outlook”.

It said it was prepared to cut rates again, if required, to stimulate growth but flagged the possible use of unconventional monetary policies. Philip Lowe, RBA governor, is due to deliver a speech later this month outlining options, which are likely to include negative interest rates and large-scale buying of government bonds.

Actually, it would be more accurate to say that making the statement, “further easing could unintentionally convey an overly negative view of the economic outlook” is itself likely to unintentionally create an overly negative view of the economic outlook.  Markets will look at that sort of statement and assume the RBA doesn’t know what it is doing.

This statement is just one more indication that the problem in central banking is not the zero lower bound, it’s a much deeper failure.  Central banks seem paralyzed for some reason that I don’t fully understand.  Ben Bernanke noticed this phenomenon way back in 1999.

As an analogy, we’ve all known someone who stayed in a dysfunctional relationship that they should have left.  To an outsider, it’s hard to understand why they stay in a relationship where they are subjected to continual abuse.

Similarly, people like Bernanke and I have trouble understanding why so many central banks clearly need to do something and yet hold back for some unknown reason.  Why?  It’s one of life’s great mysteries.

As a result of their paralysis, there are now calls for fiscal stimulus in Australia:

The Liberal-National government is now under increasing pressure to abandon its election pledge to return the budget to surplus for the first time in more than a decade and instead to unleash fiscal stimulus via tax cuts and infrastructure spending.

After all, it worked great in Japan:

Instrumental beliefs, prediction and reality

Note:  Feel free to skim past the philosophy to the discussion of monetary policy at the end.

In a recent podcast, Penn Jillette said something to the effect that people don’t believe conspiracy theories because they are true, rather because they are entertaining, like a good story or a good song.

The term ‘entertaining’ has a rather frivolous connotation, so lets make the concept more general and include beliefs that add deep and profound meaning to life, in areas such as art, religion, and politics.

There is another set of beliefs about the world that are more instrumental. If I believe the two objects in front of my eye are a hammer and nail, it’s not because I enjoy this belief, rather because this belief is useful. I know (or “predict”) that if I pick up the hammer I can drive the nail through a piece of wood. Similarly, I believe that New York City exists in the sense that I predict that if I traveled to that spot I’d see tall buildings, art galleries and yellow taxis.

So you might say that our view of reality is a set of beliefs that we find either directly rewarding or at least instrumentally useful. In the rest of this post I’ll mostly focus on the latter.

The sciences contain the most famous examples of instrumental beliefs. In principle, the laws of “physics” should be able to explain the behavior of the entire physical universe, or at least the non-random portion. But in practice, people use the term ‘physics’ to refer to the subset of physics problems that are very simple and easy to model. The term ‘chemistry’ refers to physics problems that are slightly messier and more complex, whereas geology, meteorology, ecology and economics refers to highly complex areas of physics—the motion of molecules in complex and chaotic environments.

In the simpler branches of science, it is possible to do controlled experiments and reach a broad consensus about cause and effect. Two physicists will have similar predictions about the speed at which an object will fall if dropped in a vacuum at sea level. In contrast, in the more complex sciences even the experts will often disagree, as it is tough to do controlled experiments that replicate the specific empirical question you want answered. What controlled experiment would tell you the odds of an 8.0 earthquake in LA next year, or the odds of global temps rising by 2 degrees by 2100, or the odds of rhinos going extinct in the wild by 2100, or the odds of a recession in 2021?

So this raises an important question. How should “we” decide what to believe about reality in the areas where the systems are complex? If the “we” is policymakers, then Robin Hanson has argued that prediction markets are the best way to ascertain the truth. I agree.

But most people disagree and are skeptical of market forecasts; they would rather import the methods of the “hard sciences”. Let the experts decide. Let experts set monetary policy. If not all experts agree, then let a panel of 12 experts vote on the policy, majority rules. (Actually, not all FOMC members are experts.)

People often define fault lines in economics in terms of left/right, Keynesian/classical, Austrian/Marxist, etc. But the fault line that really matters is methodological. How do we decide what we know?

The standard view is that reality is best understood in terms of what the experts believe to be true. I see reality as what the markets believe to be true. Expert opinion is an input, but only one of many inputs, into market forecasts.

In the past, the Fed has tended to rely on the experts. (Albeit not exclusively, they have always paid some attention to markets.)

You might wonder why I spend so much time fighting against the asset price bubble view of markets. If bubbles exist, if they are a part of reality, then they are useful for making forecasts. (I’m ignoring the fact that people might get utility merely from believing in bubbles.) If they are useful, then market forecasts are not reliable, and that makes expert opinion relatively more valuable.  The fight over bubbles is a fight over the future of macroeconomics.

I’m actually not ideologically opposed to rule by experts—after all, I’m an expert on monetary policy. I’d like to be a ruler, to have others ask me where the Fed should set interest rates. But my reading of the evidence suggests that market forecasts are superior.  Thus I try to infer the market prediction of the interest rate most likely to achieve the Fed’s policy goal.

Robert Shiller is one of the most famous proponents of the view that asset price bubbles are important. Thus you’d also expect him to be skeptical of the view that markets can guide monetary policy. And that seems to be the case.

Consider the past 12 months, a good example of the difference between expertise and markets. Monetary policy experts tend to rely on Phillips curve type models, which suggest that very low unemployment is a sign the economy is in danger of overheating. Here’s a discussion of Robert Shiller’s ideas from July of this year:

Nobel-prize winning economist Robert Shiller sees justification for a quarter-point interest rate hike.

That’s right: A hike — not the cut Wall Street is expecting Wednesday from the Federal Reserve.

“We still have a very low unemployment rate. The economy is hot,” the Yale University professor told CNBC’s “Trading Nation” on Monday. “One could easily make a case for staying the course and doing another interest rate increase at this meeting to cool this economy.”

That’s an almost perfect example of the methodological split that I described earlier.  “Nobel-prize winning” vs. “Wall Street”.  The financial markets were suggesting that inflation would stay low even if the Fed cut interest rates; whereas Shiller worried that the economy would overheat, even without a rate cut.  Olivier Blanchard recently expressed similar concerns, although he later backed off a bit.

This year, the Fed decided to follow the markets and ignore the models constructed by experts.  That’s partly because even the experts are losing a bit of faith in Phillips curve models as a policy tool.  Even some of the experts are beginning to follow the markets.

The view that markets should guide monetary policy is just one part of a much broader agenda—markets should determine what is true, what is reality.  

Consider the following two cases:

Los Angeles policymakers decide to spend $300 million in a new high school, believing it will make LA better off.

A small town in New Hampshire holds a town meeting, and decides to spend $2 million remodeling an elementary school, believing if will make the small town better off.

I would argue that while neither decision is, strictly speaking, a market outcome, the New Hampshire town more closely mimics a market.  That’s because the decision-makers in LA have almost no personal stake in what happens.  They are engaged in “expressive voting”.  It makes them feel good to build a shiny new high school for mostly low-income students.  Sort of like when Penn Jillette’s acquaintances believe in conspiracy theories.  In the New Hampshire town, the residents who vote at the town meeting have a real stakes in the decision.  It will affect their property taxes and their children’s education.

[Yes, even LA policymakers pay property taxes, but the gains they personally derive from “big government” far outweigh the cost of their taxes going up.]

This also explains why Switzerland is more successful than most other countries; its policymaking apparatus more closely resembles a market outcome.

At the Fed, some people feel good when they vote in a “dovish” or “hawkish” direction.  They have a lot invested in their ideology.  Contrast that with Wall Street.  In 1932, New York financiers voted for Herbert Hoover.  But in 1933, the financial markets “voted” that FDR’s policies were likely to boost prices and output.  Markets are unsentimental, and hence more likely to produce useful predictions, useful maps of “reality”.

Unlike markets, Fed officials are reluctant to reverse course soon after a major decision, as it makes them look bad—to most people, not to me.  I have a higher opinion of Powell after he reversed course on interest rates.  I believe the Fed was right to raise rates in 2017-18, and right to cut them this year.  Why? Because the outcome was good.

PS.  The question of whether reality is actually “out there” or is merely a mental construct is not important for the purposes of this post, or indeed for any other purpose.

Recessions in a post-inflation world

The Financial Times has an article pointing out that inverted yield curves are not a foolproof predictor of recessions, a point I’ve frequently made. (It’s actually a pretty good forecasting tool, just not perfect.)

In the article, Gillian Tett cites BIS research:

But as this inversion-watching game intensifies, it is worth reading a recent paper from the Bank for International Settlements, the central banks’ bank in Basel. It suggests that term spreads — what the shape of the yield curve measures — are not as good at predicting downturns as widely assumed and that there are other, better indices that economists and central bankers could (and should) use.

The authors start from the belief that the nature of business cycles has subtly shifted recently. In decades past, downturns were often sparked by rising inflation. But today, consumer price inflation seems increasingly benign, if not downright boring. They write that “there has been a shift from inflation-induced to financial cycle-induced recessions”. For this argument, the BIS staff define financial cycle as “the self-reinforcing interactions between perceptions of value and risk, risk-taking, and financing constraints”. The 2008 financial crisis is a case in point: a boom-to-bust financial cycle sparked a recession.

Obviously I don’t agree with that.  But there is a real change in the nature of recessions now that inflation is no longer a major problem.

In the past, some recessions were at least partly intentional. When inflation rose to unacceptable levels, the Fed tightened monetary policy to slow NGDP growth. A recession occurred. Even in 2008, inflation played a role, as the Fed was reluctant to cut rates during the late spring and summer months because of inflation fears.

Nonetheless, I do believe that financial cycles now play a bigger relative role, but not in the way the BIS assumes. (Recall that this institution was consistently wrong about monetary policy during the decade after 2007.)

Financial cycles do not directly cause recessions, but they may indirectly do so if they lead interest rate-targeting central bankers astray. When a financial cycle enters a downturn, the natural rate of interest falls sharply. If the central bank doesn’t respond in a timely fashion (by keeping its eye of forward looking market indicators), then money will get too tight and a recession will occur.

If the central bankers of the 1950s were in charge of the Fed during 2019 then we would now be in recession. Because they did not place enough weight on market indicators, we had 4 recessions between 1949 and 1960. We also had 4 recessions between 1970 and 1982. That’s way too many.

In another post I pointed out that central bankers following Phillips curve-type Keynesian models would have pushed the US into recession in 2019, as the very low unemployment rate suggests (in those models) that the economy was in danger of overheating.

Instead, the Fed looked at market indicators and did an abrupt shift from raising rates to lowering rates. There was no recession in 2019, and most forecasts now call for no recession in 2020. The longest expansion in US history is likely to go on for at least a few more years.

Because most developed countries have inflation under control, recessions caused by tight money aimed at restraining inflation will become very infrequent. By itself, this suggests that we will have fewer recessions than in the past. If the Fed continues to pay increasing attention to market signals, we will also have fewer recessions caused by the Fed not responding quickly enough to financial cycle-induced changes in the natural rate of interest.

I was taught that the average business cycle in the US lasts about 4 years. If I’m right (and I am pretty sure that I am right), then in the 21st century the average business cycle will last much more than 4 years, at least 15 to 20 years. Unfortunately, I won’t live long enough to know whether I am right.

PS. This made me laugh:

Four months ago, the yield on long-term US Treasury bonds fell below that for short-term ones, creating what is known as an “inverted yield curve”.

This sparked jitters, given that yield curve inversions preceded “seven of the last seven recessions”, with a lag of “8-60 months”, according to a recent Bank of America Merrill Lynch client note.

60 months? Why not 120 months, then the prediction would be even more reliable.

UK endorsements

Liberal Democrats > Conservatives >>>>>>> Labour

In England and Wales vote for Liberal Dems where they have a chance, otherwise vote Conservative. In Scotland, replace Liberal Democrats with Scottish nationalists.

I presume the Northern Irish will vote for their tribe. Wales? I know nothing about Wales.

I expect the Conservatives to win. Will Brexit be delayed long enough for Bryan Caplan to keep his perfect record intact?

PS. Tony Blair is a sad reminder of what happens when you are in office at the wrong time. Imagine if Bill Clinton had been in office during 9/11 and also the collapse of the housing boom.

Asking the wrong question

David Beckworth recently directed me to a paper by Gauti Eggertsson and Kevin Prouix, discussing how much QE a central bank might have to do when in a liquidity trap:

The required intervention in real assets needed to generate this outcome in Eggertsson (2003) corresponds to about 4 times annual GDP. Moreover, the intervention is conducted under the ideal circumstances under which the assets bought are in unlimited supply, their relative returns are not affected by the intervention (but instead equal to the market interest rate in equilibrium), and assuming that the world is deterministic so there are no risks associated with using real asset purchases as a commitment device.

More generally, however, if the government buys real assets corresponding to something like 400 percent of GDP it seems exceedingly likely that all of these assumptions will be violated in one way or the other. First, an operation of this kind is likely to have a substantial distortionary effect on pricing – which is not modeled. Second, it is likely that the government may run into physical constraints such as running out of assets to buy. Third, as the scale of the operations increases and uncertainty is taken into account, the risk to the government’s balance sheet may be deemed unacceptable, thus lessening the power of this commitment device. Finally, with an intervention of this scale it is very likely that the central bank will hit some political constraints, either due to public concerns, or concerns from trading partners in the case the assets in question are foreign. Indeed, all the considerations mentioned above have proved to be relevant constraints for banks conducting large asset purchases since 2008.

I don’t wish to contest the specific technical findings of this paper, or their political analysis. All you need to do is look at the central bank balance sheets of Japan and Switzerland to see that QE is not a panacea. Rather, I warn against misinterpreting these findings. Indeed the authors conclude their paper with a similar warning:

We do not wish to interpret this as suggesting that monetary policy is impotent at the zero bound, however.

They advocate monetary/fiscal coordination, but I don’t believe the fiscal aspect adds much. Instead, central banks need a new policy regime, such as level targeting at a growth rate high enough to generate positive nominal interest rates, combined with a “whatever it takes” approach.

Eggertsson’s paper is pushing back against the thought experiment that argues, “Of course sufficient QE must work, otherwise a central bank could buy up the entire world without creating inflation.”

Here I’d like to reframe the debate. Asking how much QE is needed is no more useful than asking how far interest rates need to be cut. If the policy is truly effective, then you don’t need to cut interest rates at all, nor do you need to do any QE.

I’ll illustrate this with an alternative thought experiment. Suppose the BOJ promises to depreciate the yen by 5%/year against the US dollar. Because of interest parity, nominal short-term interest rates in Japan will immediately rise to about 6.5% (5% plus the US short-term rate.) Also assume the BOJ pays zero interest on bank reserves.

Obviously, this policy would be highly inflationary over time. (If you don’t believe me, replace 5% with 50%). But would the BOJ actually be able to do this? One counterargument is that they’d have to do a lot of QE to depreciate the (normally strong) yen so sharply, exactly the problem discussed by Eggertsson and Prouix.

But that can’t be right, because the demand for yen base money at 6.5% nominal interest rates is likely to be quite low, say less than 10% of GDP. In fact, the BOJ would probably have to reduce the monetary base by roughly 90% after this policy was established and made credible.

The US would complain about this sort of (exchange rate-oriented) policy regime, but almost the same results would occur if the BOJ targeted CPI futures contracts at a price rising at 6.5%/year, combined with a “whatever it takes” commitment to keep CPI futures prices growing along the target path.

The willingness to do “whatever it takes” creates an equilibrium where you don’t have to do anything, indeed you do less than nothing, you actually reduce the monetary base sharply.

This is why these estimates of QE at levels of 400% of GDP can be misleading if not interpreted in the proper context. They describe what might be done in a dysfunctional monetary regime, not what would be required in a sensible monetary regime.

One final point. The central bank balance sheet will depend on the trend rate of growth in NGDP (and inflation). There’s no point is whining about the need for a large central bank balance sheet. Ultimately, central banks must always accommodate the demand for base money if they wish to prevent severe deflation. (Indeed even if they don’t want to, as the ECB discovered in the mid-2010s.) If you don’t want a big balance sheet, then set the NGDP growth rate (or inflation) path high enough to so people don’t want to hoard lots of your zero interest base money.

PS. Some people worry about the “credibility” issue. I don’t. If the central bank plans to actually do it, they will be believed. Past examples of credibility problems occurred where markets were rightfully skeptical of the central bank’s commitment. If you build it, they will believe you.