Financial Bubbles, Toils and Troubles

Five years after the U.S. housing bubble burst and nearly brought the world to financial ruin, regulators are creating new tools to measure bubbles.


We live in an era of bubbles and fears of bubbles — equity bubbles, housing bubbles, credit bubbles, commodity bubbles. Moody’s Investors Service even has a “covenant bubble” in deference to the recent “robust issuance of covenant-lite loans and high-yield bonds with weak protections.”

These fears are both relatively new and as old as capitalism itself. Bubbles have been occurring in markets for centuries, but until the dot-com bubble was followed by the housing bubble, the U.S. had gone through a period that since the market crash of 1929 had been mostly bubble-free.

That is no longer the case. Investors, economists, politicians, regulators and particularly central bankers are now acutely attuned to the potential presence of new bubbles. For the past five years, central bankers have poured an unprecedented flood of liquidity into markets and asset classes, at first through crisis programs designed to stop the spiral into depression and then with variations on monetary policy designed to revive, or at least support, economic growth in the presence of austere fiscal policies. The European Central Bank is frantically buying troubled euro zone bank loans. In Japan, Prime Minister Shinzo Abe has reversed two decades of tight money in an attempt to get the third-largest economy in the world growing again. And in the U.S. the Federal Reserve has continued its quantitative easing programs, despite talk of  “tapering.”

The result: a world awash in liquidity, pumping up share prices, heating up the junk bond market and sending investors reaching for yield. This has raised alarms in official circles. Institutions like the Bank for International Settlements and the International Monetary Fund have begun monitoring asset markets for signs that another 2008 is in the works and surveying current knowledge on asset price bubbles. New Fed governor and former Harvard University financial economist Jeremy Stein gave a speech in March that broached the possibility that credit markets were overheating and reviewed potential problem areas.

Stein was followed several months later by Federal Reserve chairman Ben Bernanke, who in a speech in Chicago laid out ways the central bank and the Treasury Department are monitoring asset markets, including bank stress tests and careful analyses of linkages among markets, unusual valuation patterns, trading volumes, liquidity and volatility measures. Maturity transformations — that is, short-term borrowing to support long-term lending — are now a priority, Bernanke said. He mentioned a “variety of models and methods” the Fed is trying out to pinpoint spillovers, amplification and contagion.

This is a major policy turn. For many years central bankers eschewed taking any role when it came to asset prices. Their focus was on the currency, which with the advent of Keynesianism and the decline of the gold standard allowed them to fine-tune monetary policy. Around the same time, central banks were burdened with the mission to seek full employment. But bubbles? Alan Greenspan’s see-no-evil approach was striking but not unusual among central bankers. The conventional wisdom was that bubbles could rarely be detected or defused; they were just part of the markets. Central banks had to be ready to pick up the pieces after the bust. To interfere would cause even greater damage.

That philosophy has been mothballed. In the U.S. the Dodd-Frank  Wall Street Reform and Consumer Protection Act not only set up two new bodies whose mission was financial stability — the Financial Stability Oversight Council and Treasury’s Office of Financial Research — but also laid the responsibility for monitoring bubbles with the Fed.

And this is where it gets interesting. Expectations were that bubbles are straightforward — periods when markets get irrational and prices veer from well-known intrinsic values — and that the government must act. But Congress failed to understand that bubbles are often difficult to define, not to mention detect, and perilous to time. Bubbles often produce a devil’s own brew of good and bad. They are different from one another, just as the dot-com boom was from the housing mania. They are elusive, paradoxical and strange. Economists have been studying them seriously only for about 15 years, and though lots of work has been done, they hardly always agree. There are no tried-and-true metrics to indicate a bubble in gestation; there are no readily available tools to separate dangerous bubbles from benign ones or even bountiful ones. Some scholars still question whether bubbles exist or whether they should be called bubbles at all. What does exist has never been tested in the only lab that matters — real markets in real time with real people.

Central banks and regulators now have a legislative mandate to pursue a quarry that, like Alice in  Wonderland ’s Cheshire Cat, is both here and not here. They are confronted with a two-pronged problem: First, they must understand the dynamics of how bubbles begin, grow and burst; second, they need to understand how they wreak havoc — what’s broadly known as spillover effects. These are two sides of the same coin called, a bit misleadingly, financial stability monitoring. On top of that, much of the most promising scholarly work on bubbles involves a rethinking of two pillars of modern economic thought: the Efficient Market Hypothesis and the Rational Expectations Hypothesis.

Much of this research is technical, theoretical and sketchy. Bubbles turn on human factors — psychology, incentives, perception — and on economic fundamentals; there are no easy answers. Bubbles give up their secrets grudgingly. They are a second-guesser’s delight: You’ll never know if the bubble you deflate was dangerous at all. And that makes regulatory choices difficult. As Princeton University economist Wei Xiong says: “We understand some things: leverage, increasingly supply, volatility. These offer hints. Although how it all works has to be refined.” He adds, chuckling, “Maybe we shouldn’t be talking prediction here but red flags.”

It’s a week before graduation, and Princeton is downshifting to summer. In his office — soon to be vacated when he moves to Columbia University — Brazilian economist José Scheinkman is talking bubbles. In March he offered up a distillation of work that he and his colleagues have pursued over the past decade or so in the Kenneth J. Arrow Lecture at Columbia, titled “Speculation, Trading and Bubbles.”  The group makes up one of the largest “schools” in bubble theory. It operates out of the 15-year-old Bendheim Center for Finance, an interdepartmental institution set up by Bernanke when he chaired the economics department at Princeton. Scheinkman was one of the first hires for what the Wall Street Journal once called “Bernanke’s bubble laboratory.”

Scheinkman’s lecture is a typical combination of narrative, historical and empirical fact and forbidding math. In a world of relatively efficient markets, how do bubbles grow? The traditional answer, reflected in everything from popular views of tulip mania to  Yale University economics professor Robert Shiller’s Irrational Exuberance, has been that a sort of mass delusion occurred: Market participants fooled themselves. That’s not enough for Scheinkman and his colleagues.

“Bubbles grow in times of innovation,” he says. This innovation can be disruptive change — American colonization, the railroad, the Internet — or a financial technology like securitization or derivatives. Bubbles tend to be associated with high trading volumes, he adds, and are linked to a surge in the supply of an asset, particularly near their tops: the flood of dot-com shares released after venture investors’ lockup periods ended or real-estate-related assets created by securitization. This growth in supply leads to a bubble’s implosion.

These are empirical observations from past bubble episodes — what economists call stylized facts. But the heart of Scheinkman’s story is a model powered by “heterogeneous beliefs and overconfidence.” Although the Rational Expectations Hypothesis suggests that participants share a common mental model of the market, he believes that’s rarely the case. Market participants develop different views of the future or of fundamental valuations, even if they have much the same information. “Most economists don’t like to think about that,” he says.

Scheinkman is not discounting the presence of ignorance, fads, misaligned incentives and greed: The engine of his model is a clash of beliefs about where the market is heading. Most bubbles spawn in periods of calm — the so-called volatility paradox. Prices rise for “rational” reasons. Even deep into the process, not everyone has to be convinced that asset prices will increase. Speculators may believe a market is overpriced, but they know they cannot arbitrage that “excess” away by themselves. They play British economist John Maynard Keynes’s beauty pageant game: guessing what others believe. This is particularly true when bearish techniques like shorting are constrained by regulation, cost or complexity; it wasn’t easy at first to short real estate. As long as there are buyers who accept the logic of an asset’s rise — the Internet changes everything; real estate prices never fall nationally — and act as marginal buyers, the bubble grows.

Scheinkman’s model features two kinds of investors: those who use asset price signals “rationally” — that is, those who believe prices have swung from intrinsic values — and less discerning “noise traders” who react to so-called useless information, including technical trading. The dynamic between the two fuels price rises, even if the number of “rational investors” — who in the Efficient Market Theory clean up excesses — swells. Scheinkman defines a bubble “as the value that a buyer pays for the option to resell. Thus a bubble occurs when a buyer pays in excess of  her valuation for future dividends, because she values the opportunity to resell to a more optimistic buyer in the future.”

The Brazilian economist and his colleagues believe that overconfidence, roughly equivalent (if less arresting rhetorically) to Keynes’s “animal spirits,” is a likelier explanation of what occurred in 2008 than the popular belief that excessive pay, bad incentives and greed were primary culprits. Princeton’s Xiong and two co-authors recently published a paper, “Wall Street and the Housing Bubble,” that analyzed the private real estate actions of midlevel Wall Streeters involved in subprime securitizations. What they discovered is that despite specialized knowledge of real estate securitizations, these individuals in their private lives acquired second homes at high prices, as if the bubble were “real.”

Scheinkman’s model, with its behavioral engine, captures some empirical realities. Frothy markets often seem to feature heterogeneous beliefs and great draughts of optimism. They do appear to be triggered by innovation. They do feature speculation, liquidity and leverage — and they are pumped up by feedback loops.

But the model is just that, a simplified version of complex market dynamics. The psychology is still rudimentary: irrationality and rationality, optimism and pessimism, those who are capital-constrained and those who are not. Is this just one class of bubble or a sketch for a metabubble? Scheinkman assumes rational investors have some knowledge of the ultimate value of the asset and that irrational investors are captives of  TV stock touts and chartists. And although his model offers a mechanism for rising prices, it doesn’t say much about how “negative” bubbles — falling asset prices — work. Scheinkman himself cautions about its predictive power. He worries about false positives from using trading volumes or leverage measures. His recommendations in the end are modest, even traditional: Limit leverage and facilitate shorting.

Roman Frydman and Michael Goldberg reject the use of the word “bubbles” as a way to describe the habitual tendency of asset prices to swing, sometimes dramatically. The term frustrates them because it suggests that swings are not related to fluctuations of fundamental economic variables. Their stance on bubbles captures just how far their views are from most economists’, not to say the public’s. The long-collaborating pair — Goldberg from the University of New Hampshire; Polish-born Frydman, who left his native country for the U.S. in 1968, from New  York University — don’t believe assets are fated to seek an intrinsic value, reject the idea of rational versus irrational participants and ridicule the notion of mass delusion.

The common conception of bubbles, says Goldberg, is based on the belief that a fundamental value for every asset exists. However, value is not a Platonic essence, he and Frydman argue; it’s the product of a mass of variables in flux. They call this “nonroutine change”; knowledge of the future is subsequently “imperfect.” Frydman and Goldberg claim Friedrich Hayek, Frank Knight, Hyman Minsky, Karl Popper, George Soros and especially Keynes as advocates of nonroutine change and imperfect knowledge.

The reality of nonroutine change knocks the stuffing from attempts to calculate fundamental value. There is no common map — no distribution of future probabilities, in economic jargon. Frydman and Goldberg view this approach as “predetermined” and mechanical, as if economists have some magic algorithm of future prices, whether it comes from rational expectations or New Keynesian models — the “formalization,” or mathematization, of Keynes — or behavioralists, who assume a correct “rational” value.

Their 2010 book outlining these ideas is titled Beyond Mechanical Markets: Asset Price Swings, Risk and the Role of the State. Notice: no mention of bubbles.

Does it matter? Well, what we call these things shapes our thinking about them. We see bubbles as balloons to be punctured. We see bubbles as irrational divergences from a rational intrinsic value. Yet notions of “rational” and “irrational,” say Frydman and Goldberg, are often predetermined; the way ahead is never obvious.

The pair take a fundamentally different approach from most other economists. Their larger argument is something they call imperfect knowledge economics, or IKE. (The Soros-backed Institute for New Economic Thinking has begun to fund IKE research centers.) The distinction between rational and irrational loses its force once you realize there are many reasonable ways to think about the future. Some investors follow fads, biases or practices like momentum trading that the behavioral school has tapped as drivers of bubbles; others do not. The market contains many views, some more expert than others, and investors with an array of aims: short-term, long-term, speculative, investment-driven. Investors can be “rational” and hold diverse views about the future that conventional and behavioral theories would deny. In fact, given the uncertainty, the most important drivers of asset swings are short-term economic fundamentals.

Do participants sometimes miss patterns that might enable better forecasting? Certainly. But the pair contend that that’s rarely enough to sustain asset swings over many years. “I just find it naive to believe that the same people will make the same mistakes over and over again,” Goldberg says.

Frydman and Goldberg do not believe markets are perfect. Because the future is uncertain and beset by nonroutine change, markets will fluctuate, producing sudden and wrenching shifts. The damage can be real. The pair take a middle position between the polarized beliefs that markets are always right or that they are casinos full of fools. “The markets never get it totally right,” admits Goldberg. “But they’re better than anything else in allocating capital. They’re better than people or governments, but they still produce asset swings.” Market participants have to use emotion and intuition because they know the future is uncertain. “Once you realize you don’t really know the truth, you need intuition rather than just some calculation,” Goldberg adds.

Frydman and Goldberg believe the government should not try to stamp out asset swings in their infancies but instead should move to “dampen” excesses when prices drift out of a range of historical benchmarks. That range, they say, should be fairly wide. When action needs to be taken, they suggest using monetary policy to alter interest rates or dampen equity or housing markets, or differentiating margin requirements for bulls and bears. Goldberg talks about IKE metrics that are contingent and qualitative.

Still, details remain sketchy. The Fed has resisted using monetary policy to affect asset prices. It’s a tricky technical and political task for officials to wade into markets to dampen bullishness — and using historical prices is always vulnerable to the argument that “this time it’s different.” Frydman and Goldberg admit that more analysis needs to be done to provide better tools for pinpointing appropriate times to act. IKE, they say, suggests the need to reconsider the question of rules versus discretion in policymaking and the framework that governs the interaction among regulators, policymakers and markets.

Their notion of an uncertain future makes sense. As Goldberg notes, that’s why markets continually reset prices. But its implications create vertigo in economics, a discipline built on quantitation and prediction. We have had 40 years of efficient markets and rational expectations. And although these ideas have been battered, not least from asset swings, the reflexive belief that markets offer a right and a wrong answer persists. Fundamentally changing minds can take a long time.

Markus Brunnermeier has spent a lot of time thinking about spillovers. Brunnermeier, a German native, is a Bendheim colleague of Scheinkman’s and has written a number of surveys, papers and chapters summarizing current bubble research, including the encyclopedic “Bubbles, Financial Crises, and Systemic Risk,” with Columbia Business School’s Martin Oehmke, which links the growth of a bubble with its aftereffects. In 2003 he coauthored an early paper on behavioral bubbles with fellow Princeton economics professor Dilip Abreu: “Bubbles and Crashes” described how a bubble could persist despite the presence of “rational” arbitrageurs. The reason: Arbitrageurs adopt different strategies to ride the bubble and fail to take a common view, thus prolonging excesses. Since then Brunnermeier has coauthored papers on maturity mismatches, liquidity, systemic risk and limits to arbitrage — many of his ideas sharing a familial relationship with Scheinkman’s model.

Brunnermeier is now exploring the fallout from a bursting bubble. “CoVaR,” a paper written with Federal Reserve Bank of New York researcher (and former Princeton colleague) Tobias Adrian — one of the co-authors of a Fed “Financial Stability Monitoring” report cited by Bernanke in his Chicago speech — sets out a methodology that tackles the systemic risk of so-called spillover effects, the bubble aftershocks that transmit amplifications and contagion, liquidity spirals, network externalities, frictions, flights to safety and feedback mechanisms.

CoVaR, which is one of Bernanke’s new approaches, takes the ubiquitous (and much criticized) measure of risk for a single firm, value at risk, and tries to apply it to the system as a whole. The “co” stands for conditional, contagion or co-movement — that is, measuring the effect of a single firm in distress on the system. Brunnermeier describes two forms of spillover: direct and indirect. Direct spillovers are the immediate damage produced by firm failure or acute decline, like the losses visited upon shareholders, creditors and counterparties following the bankruptcy of Lehman Brothers Holdings. Indirect spillovers, he explains, are more important and harder to get at and spread through interconnections, often in the area of credit. Indirect spillovers produce the post-Lehman long-term bank woes and seizing up of markets.

Yet another Bendheim economist, South Korean game theorist Hyun Song Shin, tackled some of those spillover issues in Oxford University’s 2008 Clarendon lectures, published as “Risk and Liquidity” in 2010. The Oxford-trained Shin, an adviser to the South Korean government, delved into the power of market prices to not only signal information but induce action. Much of this action in a mark-to-market world, he asserts, is “pro-cyclical,” not unlike leverage, creating herding, illiquidity, feedback and amplification, transmitted through interconnections.

How does someone capture those often opaque interconnections? “What we try to measure is how much your firm may be affected if my share price tanks,” says Brunnermeier. “I don’t have to know all of the underlying links. It’s a top-down approach.” He and the Fed’s Adrian take historical data from bubble periods — they only have weekly data back to 1986 — and try to get some sense of how firms with similar underlying characteristics are moving in a coordinated fashion. The key is to develop a useful set of attributes.

“What we’re saying is, what is the danger to the financial system if Lehman Brothers goes down?” Brunnermeier explains. “We can also turn it around: If the system is in trouble, how will Lehman fare?”

“We are still fine-tuning,” he adds in his clipped, brisk English. “We’re searching for the right characteristics — maturity mismatches, size, leverage, liquidity risk. If you see leverage building up, that’s a sign.”

Although “CoVaR” has attracted attention — a flock of new papers has tackled aspects of it — relevant data is patchy, the metrics themselves are indirect, and lots of work still needs to be done. Regulators are working with other approaches that seek the same macroprudential goal. “It’s actually good to look at [the problem] from different angles,” says Brunnermeier. One tool, the “distress insurance premium,” developed by Xin Huang, Hao Zhou and Haibin Zhu from, respectively, the University of Oklahoma, the Fed and the Bank for International Settlements, uses credit default swaps to indicate firm vulnerability. But that method is limited because CDS data is available for only one crisis — 2008. A third measure, the “systemic expected shortfall,” advanced by a group of NYU economists, examines how firm capital will be affected by a decline in equity prices.

In fact, there are dozens of so-called risk indexes under development. NYU’s Leonard N. Stern School of Business has started something called the V-Lab (for Volatility Laboratory), run by Nobel laureate Robert Engle, which provides online a variety of daily measures of various risks, including systemic risk, listed by both country and financial institution.

Will any of these methods sound the alarm? Will anyone respond? We’ll only find out in a live crisis. So far, there are limitations on these predictive forays that are not just theoretical and empirical but data-driven. Brunnermeier hopes better data will be generated, particularly with the involvement of the Office of Financial Research. He has worked on a liquidity mismatch index that indicates how easy — or difficult — it is to sell assets off in a crisis. But the index relies on flow-of-funds data, a measure that has grown more complex and opaque with the diffusion of derivatives.

Meanwhile, there’s another aspect that spillover analysis helps illuminate: Different kinds of bubbles vary in their costs and benefits. Princeton’s Harrison Hong and David Sraer, for instance, have developed a model that tries to distinguish “loud” equity bubbles — characterized by high prices, volatility and trading volume — from “quiet” bubbles, such as credit bubbles, with more-subdued pricing, volatility and volume. Noise has nothing to do with threat. Generally, equity bubbles tend to cause direct losses. Stealthy credit bubbles, with their thicket of interconnections, spawn more havoc.

In his recent book, Doing Capitalism in the Innovation Economy, longtime venture capitalist and  Warburg Pincus senior adviser William Janeway argues for the need to differentiate bubbles along two dimensions: “the object of speculation” — whether it’s a fundamental technology or a financial instrument — and “the locus of speculation,” whether confined to capital markets or spilling over into credit markets. Stamping out every suspected bubble may erase risk that produces reward, Janeway contends.

Where does that leave stewards of financial stability around the world? With broad generalizations, some traditional measures — leverage, liquidity — and a flock of systemic risk measures, like CoVaR, that may or may not prove effective. We know generally that equity bubbles in relative isolation are less systemically dangerous than credit bubbles, which drag down interconnected banks. But you still have to be aware of an equity crash that takes down banks, like 1929. We know, as Janeway lays out, that bubbles that seem to focus on a technology may be worth letting ride. We also know that some bubbles may be fueled by disagreement and that leverage and unusually high volume and volatility are dangers. Last, we know that the key evidence of excessive asset swings comes from a comparison with historical benchmarks.

What don’t we know? A lot, starting with timing. If we don’t know when a bubble will burst or when the most reasonable time is to dampen it, we need to act carefully. In that sense, as one Fed official notes, stability monitoring resembles monetary policy. It’s less a science than a data-driven art.

Eventually, we will understand more, we will have more data, and we will do more testing — though public trial and error spawns its own damaging spillover. But all the data in the world cannot ensure certainty. What we can’t calculate is the judgment and will of decision makers. Will they be prepared to yank away the punch bowl, and do they possess the wisdom to discriminate among assets, even with Wall Street or a politically polarized public claiming that “this time it’s different”? Will they be willing to act even if they’re coping with the effects of their own policies? Data, metrics, models and analytics can help, but they may be ambiguous and can’t make the tough decisions. As Frydman and Goldberg might say, there’s no algorithm. And what lurks if we get it right a few times? Moral hazard.

Bubbles resemble yet another metaphor, the black hole: still poorly understood but such an integral part of the market universe that Janeway titled one chapter of his recent book “The Banality of Bubbles,” followed by one called “Explaining Bubbles” and a third dubbed “The Necessity of Bubbles.”

Bubble fears are now part of the public consciousness. Like it or not, our monitors of financial stability operate in a system shaped by popular sentiments that begin with the belief that markets naturally rise — at best, a frothy notion. Bubbles, or asset swings, challenge that thinking. But it’s a foreboding sign when we can’t agree on what “rational” means. As Janeway writes, “In this term rational there is a nexus of confusion that infects both academic and popular discussions of how economic and financial agents think and act.”  The idea that bubbles represent a decline of virtue and that everything that occurred during bubble years is a lie continues. The concept of the “good” bubble remains counterintuitive. The belief that the future is predictable lingers despite contrary evidence.

Yes, it’s a new era. • •