Book review: The Money Illusion: Market Monetarism, the Great Recession, and the Future of Monetary Policy, by Scott Sumner.
This is the best book on macroeconomics that I’m aware of, with a focus on the causes of the 2008 recession.
Most of the book’s important points are based on ideas that economists respect in many contexts outside of macroeconomics, but which seem controversial in the context of macroeconomics.
It’s ironic that Sumner finished writing this book during one of the few recessions that could not have been prevented by better monetary policy.
Note that this review is primarily for people who already know something about monetary policy. It’s hard enough to do that well that I don’t want to attempt anything more ambitious.
I’ll start with the five differences from standard opinions that I consider most important.
Never Reason from a Price Change
Price changes provide potentially valuable information, but when they warn of problems, they are ambiguous as to the cause in the absence of further information.
E.g. a large drop in oil prices might mean that fracking has made more oil available (as in 2014), or it might mean that pandemic lockdowns reduced demand.
Economists know this in the abstract, but often forget it when evaluating monetary policy. They saw interest rates decline in 2008, and imagined that was due to the Fed adopting a more stimulative monetary policy. People who looked carefully could see a good deal of evidence that market forces were pushing interest rates down, and that the Fed was keeping rates higher than markets would have, in order to avoid overstimulation.
Economists have mostly conceded that the low interest rates of the 1930s coincided with tight money. There were modest similarities between 2008 and the 1930s, yet too many economists looked at low, declining, interest rates in 2008 and were unwilling to imagine that monetary policy was tight.
When respected economists are confused as to whether monetary policy is tight, it’s hard for a central bank to notice its mistakes.
Respect the Efficient Market Hypothesis
That doesn’t mean we should ignore prices. Predictions based on current financial prices can rarely be beaten.
Yet the Fed trusts its own forecasts over market forecasts (or at least it did in 2008), and few economists complain about that.
On Sept 16, 2008, the TIPS spread predicted that CPI inflation would be 1.23% over the next five years. Whereas the Fed announcement that day implied a CPI inflation forecast of around 2.3 to 2.5% (i.e. they expected to hit their target; I don’t think they specified a time period). The actual inflation rate turned out to be 1.31%.
The Fed chose not to alter course that day, when markets were saying clearly that monetary policy was going off course. So it seems the Fed accepted a weakening economy at least in part because they mistakenly believed that more stimulus would cause excessive inflation. That’s one of the clearest examples of why the Fed deserves blame for the recession.
Another example of market prices providing valuable information came on the three days in the fall of 2008 when the Fed announced interest rate increases (not the standard rate increases – this was a new interest rate called interest on reserves, which attracted less attention). Adding together the three reactions to the three rate increases, we get a total stock market decline of 15%. The Fed shouldn’t automatically treat such declines as bad, but such declines should alert it to the possibility that it was misjudging conditions.
In 1933, FDR altered the dollar to become pegged to a smaller amount of gold. That didn’t have an unusual short-term effect on interest rates or money supply. Yet it triggered a few months of high inflation, a dramatic stock market rally, and abnormally strong economic expansion (cut short by an unrelated policy blunder).
Why? Because it was a clear commitment to a sustained increase in the money supply – an increase which was incompatible with the prior policy of keeping the dollar at 1/20.67 ounces of gold.
In contrast, huge increases in the money supply can be ineffective if they’re expected to be temporary. In a 1998 paper that Sumner calls “important and underrated”, Krugman explains:
In a liquidity trap, the problem is that the markets believe that the central bank will target price stability, given the chance, and hence that any current monetary expansion is merely transitory. The traditional view that monetary policy is ineffective in a liquidity trap, and that fiscal expansion is the only way out, must therefore be qualified: monetary policy will in fact be effective if the central bank can credibly promise to be irresponsible, to seek a higher future price level.
Here’s a scenario which clarifies why increases in the money supply might be ineffective: interest rates are zero, and people mostly expect that the Fed will pursue policies which will cause steady deflation. Holding on to more cash will be a sensible investment decision, because that cash is expected to buy more in the future. Why buy a house now if it will cost less next year? (People will still buy houses to live in, but they’ll avoid investment-motivated housing purchases).
It looks like the U.S. hasn’t quite reached this kind of liquidity trap since the 1930s, when the gold standard more or less required the Fed to allow deflation.
Japan was mostly in a liquidity trap from 1990 to at least 2012, with the central bank seeming to target an inflation rate of exactly zero. It at least partially recovered in 2013 when Abe was elected with a promise of inflation.
Prior to reading this book, I was confused as to what mainstream economists think about liquidity traps. Part of my confusion came from reading commentators who imagine that the Fed can only stimulate the economy via the act of cutting interest rates. The financial
news storytellers often seem to believe that, but leading economists (e.g. Bernanke) seem wiser than that.
Now I at least understand why economists take liquidity traps seriously. Yet Japan’s ambiguous partial recovery from one suggests that I don’t fully understand the phenomenon.
A good deal of theory and evidence backs the importance of expectations, but too many people still focus on today’s money supply to the exclusion of asking what policy the Fed will use years from now to influence inflation or the money supply.
Sumner wants the Fed to compensate for recent mistakes by aiming to get the monetary target (e.g. inflation or NGDP) back onto the track that it had previously committed to. E.g. in 2020 the Fed seemed to endorse this policy, so that we ought to be able to forecast now what price levels will be in 2030.
One benefit of level targeting would be that people can make long-term plans today with less uncertainty about whether the Fed’s mistakes next year will impair our ability to predict prices in 2030.
A more important benefit is that when the Fed errs in the direction of deflation, that interacts with sticky wages to make a temporary increase in inflation more valuable at minimizing unemployment.
Incidentally, the Fed’s behavior this year seems to be an example of why some economists have claimed there’s a long and variable delay between when the Fed adopts a policy and when the economy responds. A better description of this phenomenon is that there’s a long and variable delay between policy announcements and when, if ever, the Fed clearly decides to follow the new policy.
NGDP versus Inflation
Sumner wants the Fed to target NGDP growth rather than inflation.
Mainstream macroeconomic theory advises governments to influence aggregate demand. Sumner comments:
a high inflation rate is viewed as an indicator of strong aggregate demand, and low inflation implies weak demand. My view is that, rather than using inflation as an indicator of aggregate demand, we should use aggregate demand as an indicator of aggregate demand. And nominal GDP, which measures total spending, is basically what people mean by aggregate demand.
He provides a few examples of when targeting NGDP would have produced better results, such as mid-2008, when NGDP forecasts seemed to start falling a bit earlier than inflation forecasts. Targeting NGDP will cause inflation to be higher when economic activity slows, which seems likely to stabilize the economy a bit more than constant inflation would.
Targeting NGDP can’t be a complete answer here, because a semi-arbitrary growth rate such as 4% per year could cause problems if we had very different productivity growth or population growth. Sumner seems satisfied with improving on existing policies, rather than aiming for perfection.
Sumner provides both theory and evidence that the main alternatives to monetary policy seem to be weak explanations for recessions. The evidence is hardly conclusive, but in combination with theory it’s good enough to be taken seriously.
Did fiscal policy cause the recession? Sumner provides examples that seem inconsistent with standard claims about how budget deficits stimulate the economy:
- the 1968 tax increase, which eliminated the budget deficit, without keeping inflation from accelerating (the economy slowed a bit, but slowed more when the tax expired).
- tax increases in 2013 caused the deficit to shrink nearly in half; many leading economists predicted a double-dip recession. Instead, the economy grew a bit faster than it did in 2012.
- Japan’s 2013 Abenomics example of fiscal austerity and monetary stimulus improved the Japanese economy.
Did the stock market crash cause the recession? The 1987 crash is a clear piece of evidence that such crashes don’t necessarily cause recessions.
Did the banking crisis cause the recession? Sumner’s best evidence here involves comparing Iceland to Ireland.
Iceland had a banking crisis, about which Wikipedia says:
Relative to the size of its economy, Iceland’s systemic banking collapse was the largest experienced by any country in economic history.
Whereas Ireland had a an ordinary enough crisis that it was able to bail out its banks.
Does the size of the banking crisis tell us which economy was hurt worse? The Washington Post compared them in a 2015 story titled The miraculous story of Iceland, saying “No, the biggest difference between the two is that Iceland has its own currency and Ireland has the euro.”
The ECB adopted a tighter monetary policy than the U.S. around 2008-2015, which coincided with slower growth in the eurozone than the anemic U.S. economy. Whereas Iceland’s monetary policy created rather high inflation for a few years.
I suppose that we can’t rule out the possibility that Iceland’s success was due to not bailing out its banks. I doubt that bailouts have big economic effects, but we should keep this example in mind the next time banks ask to be bailed out.
Did the housing bubble cause the recession? No. Or at least not directly. The slump in housing construction was more than halfway to its bottom in April 2008, yet nearly all the increase in unemployment came later. Also, the eurozone had a worse recession without having had a notable housing bubble.
Erdmann’s book Shut Out did a better job than Sumner at showing that the housing bubble had mostly stopped causing disruptions by the start of the recession, so it’s pretty hard to see the bubble as a direct cause of the recession. It still seems quite plausible that the bubble generated conditions under which central banks were likely to make mistakes.
Sumner goes further, denying that bubbles are even a useful concept, as bubbles are inconsistent with the EMH. That requires an implausibly strong version of the EMH.
He’s right that bubbles are a less useful concept than the EMH. At least 30% of investors would do better if they respected the EMH more, whereas maybe 3% of investors benefit from worrying about bubbles.
But 3% is not zero. It’s probably not much different from the number of investors whose investing would benefit from the improved understanding of this book.
If we had to choose between the EMH and the bubble hypothesis, then Sumner would be correct to choose the EMH. But there’s no need for a strong enough version of the EMH to rule out all bubbles.
As an investor, I found it useful to say in 2005 that the U.S. had a housing bubble. It would have been useful for me to identify a bubble involving irrational confidence in the Fed’s competence.
But I’m not typical. It’s unclear whether it’s useful for central banks or regulators to believe that bubbles are possible. Without strong incentives to distinguish bubbles from healthy markets, most people who think they can identify bubbles will be correct less than half the time. This is especially true for institutions that are influenced by the average voter’s opinions – almost by definition of a bubble, the average voter will be mistaken about whether a market is experiencing a bubble. Elite opinion isn’t much better, unless the elites are selected for their skill at beating markets.
Erdmann provided some evidence that regulators, and maybe the Fed, caused significant harm via mistaken attempts at deflating an already-deflated housing bubble in 2008-11. I see signs that China is currently making similar attempts to deflate a housing market that seems not to have a bubble.
Sumner has a well thought out procedure for a good monetary rule. I’m reluctant to summarize it – the one paragraph version would sound weird to most people, and I don’t have enough time to explain it in a way that does it justice. I prefer to focus most readers’ attention on how the worst mistakes can be avoided. That will likely produce more value than arguing about which policy is best. The only politically feasible path toward his favored monetary rule is through incremental changes to avoid specific mistakes.
We can’t be sure that Sumner’s policies would have avoided a recession, but it’s pretty clear they would have at least made it substantially milder. I’m pretty sure they would have avoided more than a half trillion dollars of harm from monetary conditions.
This book won’t be the final word on monetary policy, but it represents an important step in the right direction.