The Great Moderation and Leverage

In response to my earlier post, The Great Moderation Never Ended,  the perceptive Kevin Drum noted that the moderation seems to have been asymmetric–the booms have moderated more than the busts. That’s correct but it’s more than lower economic growth–expansions also last longer. It’s as if the booms have been smoothed over a longer period of time but not the busts.

Søren Hove Ravn points me to a paper of his with co-authors, Leverage and Deepening: Business Cycle Skewness that documents this fact and also proposes a theory.

The authors argue that financial innovation made credit more easily accessible and easier credit led to more leverage. Leverage, however, has an asymmetric feature. When asset prices are up everything is golden, wealth is high and credit is easy because lenders are happy to lend to the rich. When asset prices decline, however, the economy takes a double hit, wealth is low and credit is tight. The net result is that booms are smoothed but busts become, if anything, even more violent.

The theory is promising because it explains both the negative skewness and the great moderation. It’s also important because higher leverage, longer expansions and greater negative skew are new features of business cycles that appear across many developed economies as shown by Jorda, Schularick and Taylor in Macrofinancial History and the New Business Cycle Facts. In this paper Jorda et al. create new data series using over 150 years of data from 17 economies and conclude:

…leverage is associated with dampened business cycle volatility, but more spectacular crashes.

and more generally:

We find that rates of growth, volatility, skewness, and
tail events all seem to depend on the ratio of private credit to income. Moreover, key
correlations and international cross-correlations appear to also depend quite importantly
on this leverage measure. Business cycle properties have changed with the financialization
of economies, especially in the postwar upswing of the financial hockey stick. The manner
in which macroeconomic aggregates correlate with each other has evolved as leverage
has risen. Credit plays a critical role in understanding aggregate economic dynamics.


So it's true: reliance on rising asset prices for economic growth produces a moderation in growth, greater instability, and deeper busts. Who knew? Two questions: why rely on rising asset prices for economic growth and what are the alternatives. The answer to the first question is that, for mostly political reasons, the Fed is all that stands between economic stagnation (or worse) and the illusion of growth. The answer to the second question must come from trained economists since they are paid the big bucks to answer the big questions.

Sir, I think the visceral problem is relying on asset price inflation as a substitute for slow GDP and wage growth. In the run-up to the recent crisis, the US experienced asset (housing) prices rapidly rising (solely due to liberal credit underwriting) while GDP growth and median household income rises were much lower. Not only did everybody miss that, Federal agencies that were established to prevent recessions actively promoted it. The Federal agencies whose purposes were to finance housing went way overboard in adding liquidity/loans to the real estate market.

Regarding leverage. Truth: "leverage is associated with dampened business cycle volatility, but more spectacular crashes." In a 100% equity (no debt) situation - say you paid $1,000,000 in your own cash for a home that five years before was sold for $400,000. When the market busted, you lost $400 or $500,000. The only loss incurred is only incurred by the owner. In a heavily-leveraged situation, both the owners and creditors incur large losses. The owner loses the down payment and the bank loses several hundred thousand on its loan. when that happens six or eight million times, we have a catastrophe.

The "visceral problem"? I'm glad you cleared that up. "The owner loses the down payment and the bank loses several hundred thousand on its loan." You must be from one of the few states that prohibit deficiency judgments. Do you know what a deficiency judgment is? Do you know that most states allow deficiency judgments?

I don't make the big bucks (actually occasionally I do), but the solution is here:

Of course, autonomous vehicles might be an answer to the big question. Here is something I find both fascinating and frightening: developers of autonomous vehicles are now using a combination of algorithms and simulators to "train" autonomous vehicles. Algorithms? Are they training autonomous vehicles to think, and to make choices (i.e., react) based on what they "see" and what they have "learned" from having spent so much time in the simulator, much like a pilot of an aircraft learns how to react to situations by spending many hours in a simulator. I'm not sure which is more dangerous: an autonomous vehicle making choices based on simulations or economists making choices based on simulations.

For those who don't quite get the analogy, the autonomous vehicles are being trained to predict the future, i.e., the most likely events based on what the vehicle "sees" and what the vehicle "learned" will follow from what it "sees" after spending many hours in the simulator. It's not unlike what those quants do in predicting the future for a market. Soon enough everyone will be a soothsayer, even autonomous vehicles.

Vehicles and humans both making decisions based on prior experience and training. Simulations are a great training tool. The military uses them. Sports has scrimmages based on the expected opposition. Sparring partners are brought in to simulate the opposing boxer.

What's the problem exactly? Isn't most learning based of experience?

It's certainly true that human drivers of vehicles take into account past experience in making current choices, including past experience with the way other drivers respond to a particular situation. Human drivers expect other drivers to stop at a stop sign, but sometimes they don't. In particular, human drivers sometimes drive right through a stop sign because they don't see it or aren't paying attention. Hence, I anticipate that possibility when I approach an intersection and look to see if another car is approaching the intersection at the same time. If so, I try to make a prediction whether the other driver sees the stop sign and will stop, the clues including whether she is slowing as she approaches the intersection. I am assuming that the autonomous vehicle will "see" the other car approaching the intersection and will have "learned" through many simulations the clues whether the other vehicle approaching the intersection will likely stop. Really? The safe alternative would be for the autonomous vehicle to stop at every intersection to avoid the possibility that the other vehicle will not stop. Safe, perhaps, but highly inefficient. Of course, I have presented one simple situation, but the reality of driving in the city is that the driver is presented many situations some simultaneously and is constantly relying on past experience in determining the appropriate response to the many clues she sees.

I give it even chances that the AI is still overall safer, even it it fails to correctly anticipate bad drivers simply by being a better than average driver.

The impediment to a safer future with AI driving is human drivers; the solution is to stay with human drivers.

This is what Taleb predicted in Black Swan. Leverage makes you fragile; it allows you to pick up pennies before a bulldozer.

Isn't this always true and for a much simpler reason, not having anything to do with credit? It takes time to build confidence on the bullish side, but panic can induce rapid crashes to the bearish side. It's normal human psychology.

Right, but how the underlying psychology manifests in the markets/economy (and how that feeds back to psychological state) probably has something to do w/ the relative amount of credit, leverage, and counter-party risk in the system. The nature of the transmission mechanism matters. More speculative excess and bad debt that needs to be unwound -> more pressure on asset prices and credit markers -> more panic -> more pressure on prices -> more panic ... more severe crisis.

Yeah, but there's no such thing as panic to the upside. What would that be? Fear you'll miss out on a rise of the market? On the other hand, panic to the downside occurs often. Fear that you'll lose the assets you've already got. And yes, panic builds on itself. But it always builds to the downside. During the last market contraction, my sister was saying we needed to move my mother's assets out of Vanguard because they had declined so much over several months. I told her that it was the worst possible time to get out of Vanguard. Fortunately, she listened, and that was right before the run-up which is continuing today.

There is often a ton of "panic" on the it "manic" instead and you'll see it.

Yes, real estate buyers in manic phases will hurry to close deals because the price will go up. Sellers may hold out for higher prices.

Correct, although "mania" is probably a better word. Kindleberger's _Panics, Manias, and Crashes_ was and probably still is an oft-cited classic.

There's also bubbles, I don't know if there's an agreed upon dividing line but I think of manias as extreme bubbles (and crashes as extreme bear markets).

Very true. Many financial disasters, back to the great tulip crash, were marked by new financial instruments that either increased leverage or created illusion that they had decreased risk. Often both. Savings and loan was about deregulation of risk on upside with bailouts on downside. The housing collapse was in part the result of a belief that you could hedge the risk of default to very low levels with derivatives and other measures. In many cases the lowering of risk is actually a transfer of risk to government agencies

Risk weighted capital requirements allowing much larger leverages for what is perceived, decreed or concocted as safe put economic cycles on steroids

So... Minsky rehashed? Am I missing something?

This seems obvious. If central bank's can take volatility out of the business cycle, the system will add leverage to maintain the desired volatility of wealth. This further increases the pressure on central banks to successfully achieve their mandate. At the same time, central banks are at increasing risk of being hamstrung by political change. A very dangerous set-up.

In 1972 we went off the gold standard, defaulted on our paper gold promises and tripled the wealth of gold horders. Until 1981 or so, gold bugs managed the investments and it was pricing chaos.

By 1983 we went to a monetary standard based on government treasury rates, interest rate targeting. That put us into disinflation but it also made Congress the bailout agency. Since then, all the volatility is centered at the regime change point, every eight years. One other major volatility component is the monetary cycle, every 60 years or so government defaults because we enter long term deflation.

One chart says it all:,#0

This chart looks like something a very accurate price neutral, accounting system would create. First 50 years is building up debt to cover inflation, then from 2018 to 2080, we pay off all that inflation with deflation. But the taxpayers won't do it, they won't pay for all that inflation and we default.

Re: In 1972 we went off the gold standard

Actually we quit the gold standard back in 1933.

Sounds like Alex has discovered the financial cycle. But I'm not seeing any explanation here for why higher leverage smooths or slows growth.

Could I nominate this for most worthwhile Alex post? Insightful, informative, and worth reading. This is definitely an area worth exploring and it is worth noting that this bust - bust cycle is not specifically an American phenomenon. Though Americans do have a particular way of doing things. My general observation is that Americans like recessions and so this bust - bust business cycle is an American social phenomenon.

1. Two of the three recessions since the mid-1980s were very mild (1990-91, 2001). It's hard to say that recessions are worse when we have only three examples in the Great Moderation period.
2. Financial innovation was a key factor in the early 1990s. The savings and loan industry went bust, and it had been making most of the home mortgages. Securitization arose, which was critical to continued home construction and resales. If we had another credit blockage, I don't think Dodd-Frank would allow any innovation to get around the blockage.

As a Sumnerite this seems nearly obvious to me. Modern central banks are hyperalert to moderate inflation, and less on the ball when it comes to preventing huge sharp drops in NGDP. What more needs to be said?

Yep - its all about incentives. People are constantly hyperventilating about runaway inflation, but no-one is too upset by a bit of deflation. So the policy makers follow the incentives. The key innovation was the invention of inflation targeting, prior to that CBs pretty argued that inflation was nothing to do with them, and monetary growth was something they observed but didn't try to control.

True Dat. But the Great Moderating Influence should be a Job Guaranty as part of a Full Employment Fiscal Policy. Why should we wait for the tea leaves of private sector testosterone deficiency/surplus. The dog nailed it here:

Yes, Minsky rehashed, who, after all, was largely right.

Re: leverage/debt

What are you referring to?

- Sovereign?
- Corporate?
- Household?

Your choice matters.

But the Great Moderating Influence should be a Job Guaranty as part of a Full Employment Fiscal Policy. Why should we wait for the tea leaves of private sector testosterone deficiency/surplus. The dog nailed it here:

Comments for this post are closed