“Behavioral” finance approaches are mainstream in macro

Noah Smith, a finance professor, has written a piece arguing that the finance approach to “behavioral” economics/finance has been very successful, whereas it has been largely rejected by the macroeconomics profession. The piece is confusing, however: even though Smith explains that the term “behaviorial economics” refers to a much narrower field of study than the term “behavioral finance,” he does not explore fully the implications of these two different definitions.

Smith explains the different terms:

To most economists, behavioral economics means using findings from psychology to modify models of individual behavior. But behavioral finance has come to have a much more expansive meaning, basically encompassing anything that doesn’t conform to the Efficient Markets Hypothesis (which says that you can only earn market-beating returns by taking on extra risk).

What is left out of Smith’s piece is that many of the approaches that are categorized as “behavioral” in finance, are categorized as mainstream in macroeconomics. There’s no division between saltwater and freshwater macroeconomists on whether macro needs to study frictions using the whole of the microeconomic toolbox including contract theory, mechanism design, etc. The study of the many ways in which Arrow-Debreu (and more generally the efficient markets hypothesis) fails has been fundamental to the macroeconomic research agenda for at least three decades.

By contrast, when the finance profession categorizes mainstream economic theory as “behavioral,” this has the effect of defining as mainstream finance that which is consistent with the efficient markets hypothesis. As a result finance is probably the only economics-related field that has not yet integrated the mid-20th century critiques of the competitive model into its “mainstream” body of theory. For this reason, the fact that “behavioral finance” is alive and well is hardly surprising: if this were not the case, the field of finance theory would be decades behind the theory that is used throughout the economics profession.

The real question then is why the finance profession has not dropped the “behavioral” moniker from behavioral finance, and renamed non-behavioral finance “frictionless finance.”


Note: It is clear that Paul Romer finds that macroeconomists have a lot of work to do when it comes to incorporating decades-old critiques of the competitive model into the their work. As I understand his argument, he objects to the way the critiques have been incorporated in macro, because this constitutes a pluralistic “views about the shape of the earth differ” approach, rather than an effort to  fully acknowledge and integrate the logical implications of these critiques into what constitutes mainstream theory.  I suspect that Romer would agree, however, that macro theory is several steps ahead of efficient-markets-based finance theory — if only because mainstream macro at least includes theory that is not price-taking and frictionless.


A question for our times: the role of the central bank

Peter Conti-Brown and Philip Wallach are having a debate that cuts right to the heart of what appears to me to be the most important economic question of the current era: what is the proper role of the central bank?

Conti-Brown takes what I think is a fairly mainstream view of the central bank’s role as lender of last resort: In a crisis, the central bank should intervene to rescue a troubled bank as long as, given Fed support, the bank can over time be restored to solvency. He writes:

in a systemic crisis, the problem of determining whether a specific asset class is sufficiently valuable to justify its temporary exchange for cash isn’t just “murky,” it can be impossible to determine. This is true for two reasons: first, the reason the systemic crisis exists at all is because the line between illiquidity and insolvency has become a mirage. And second, whatever line is left is endogenously determined: what the central bank does in response to the crisis has immediate consequences on both liquidity and solvency. There is essentially no way, in the depth of a crisis, to draw the line meaningfully between solvency and illiquidity. After Lehman, the Fed recognized this and extended loans through 13(3) so quickly on so many different kinds of collateral that we saw an explosion in its 13(3) lending.

The clear implication here is that if there is doubt as to whether a firm is illiquid or insolvent, the Fed should err on the side of supporting the firm.

Wallach responds that if one follows this logic to its end, there appear to be no limits to the Fed’s powers:

If I’m understanding him correctly here, Peter means to put in the Fed’s mouth some version of an infamous 2004 pronouncement of a Bush administration aide: “when we act, we create our own reality.” Amidst the chaos of crisis, it is for the Fed to decide which firms are solvent and which kinds of assets are really valuable as collateral and, whatever they decide, the markets will follow, allowing the central bank to benefit its own balance sheet and the larger financial system through self-fulfilling optimistic prophecy. As they forge this new reality, making the security on loans satisfactory to themselves will be the least of their miracles.

Teasing aside, I think that’s far from crazy, but one can get carried away. It can’t be the case that the Fed is capable of rescuing any institution through this kind of heroic thinking: if a firm is in a downward spiral, and the only collateral it has is rotten, then the Fed does not have the legal authority to funnel money into it.

I think that there are actually three question raised by this exchange: First, what are the Fed’s potential powers; that is, what is it feasible for the Fed to do? Second, what were the Fed’s powers in 2008; or alternatively, what was both legal and feasible for the Fed to do? And, third, what should the Fed have legal authority to do? Conti-Brown and Wallach are debating the second question, but I think it’s important to explore the first question regarding what the Fed can do, before moving on to the second and third questions regarding what the Fed is legally authorized to do.

A little history on the concept of the lender of last resort is useful in exploring the first question. A previous post makes the point that the term lender of “last resort” was initially coined, because the central made the self-fulfilling determination of whether or not a bank was solvent and worthy of support. The fact that the central bank has the alternative of saving a bank, but chooses not to is what defines the power of a lender of “last resort.” From the earlier post:

The term “lender of last resort” has its origins in Francis Baring’s Observations on the Establishment of the Bank of England and on the Paper Circulation of the Country published in 1797. He referred to the Bank of England as the “dernier resort” or court of last appeal. The analogy is clear: just as a convicted man has no recourse after the court of last appeal has made its decision, so a bank has no recourse if the central bank decides that it is not worthy of credit. In short, the very concept of a “lender of last resort” embodies the idea that it is the central bank’s job to determine which banks are sound and which banks are not — because liquidity is offered only to sound banks. And the central bank’s determination that a financial institution is insolvent has the same finality as a last court of appeal’s upholding of a lower court’s death sentence.

In short, for a partial reserve bank “solvency” is a state of affairs that exists only as long as the bank has access to central bank support. Solvency in the banking system does not exist separate and apart from the central bank – and this concept was fundamental to the 18th and 19th century understanding of banking in Britain [where the concept of lender of last resort was developed].

There is a long list of banks that were deliberately allowed by the Bank of England to fail in Britain, including the Ayr Bank in 1772, and Overend, Gurney, & Co. in 1866. The latter was, second to the Bank of England, the largest bank-like intermediary in England at the time, and its failure triggered a Lehman-like financial crisis — that was, however, followed by only a short, sharp recession of unexceptional depth. Bagehot made it very clear in Lombard Street that he did not believe that the Bank of England had mishandled Overend Gurney. He argued, on the contrary, that it was always a mistake to support a “bad bank.”

In short, just as it is in some cases the job of a court of last appeal to uphold the law in the form of a death sentence, so it is in some cases the job of a central bank to pronounce a death sentence on a bank in order to promote healthy incentives in financial markets. The fact that the bank would still be alive in the absence of the death sentence is as obvious and irrelevant in the case of the lender of last resort as it is in the case of the court of last appeal.

So let’s go back to the original question: What are the potential powers of the Fed? Can it in fact determine “which firms are solvent and which kinds of assets are really valuable as collateral” and expect markets to follow that determination? We have a partial answer to this question: from past experience we know that a central bank can choose not to support a bank in a crisis in which case it is almost certain the bank will fail, or that a central bank can choose to support a bank and with equal certainty carry it through a crisis of limited duration. We also know that sometimes a bank that was saved fails a few years or a decade after it was saved (e.g. City of Glasgow Bank). The British history also indicates that it is possible for a central bank to have a similar effect on assets (see here).

Thus, the fact that ex post the Fed did not lose money on any of the Maiden Lane conduits — or more generally on the bailout — is not evidence that the Fed exercised its lender of last resort role effectively. Instead this fact is simply testimony to powers of a central bank that have been recognized from the earliest days of central banking.

What we don’t know are the limits of a central bank’s ability to “create it own reality.” Can a central bank continue to support banks and assets for a prolonged period of time and still be successful in leading markets? At what point, if ever, does the central bank’s intervention stop being a brilliant act of successful alchemy, and end up looking like fraud?

What makes a lot of people in the financial industry nervous about the current state of central bank intervention (see for example here, here or here) is that they are not sure that the central banks will be able to exit their current policies without causing a crash in financial markets of the sort that none of us has ever seen before. Of course, we are sailing uncharted waters and literally nobody knows the answer. Let’s just hope that Janet Yellen and Mario Draghi are brilliant and creative helmsmen. (Should that be helmspeople?)

In summary, the term lender of “last resort” itself makes it clear that a fundamental aspect of a central bank’s duties is to refuse to support firms such as Lehman. Thus, in my view Conti-Brown, even though he gives a description of a lender of last resort that many modern scholars would agree with, envisions a lender of last resort that is very different from that of Bagehot and 19th century bankers. Whereas Conti-Brown appears to argue that, because the line between solvency and insolvency is so murky in a crisis, if a bank can be saved, it should be saved, Bagehot clearly understood that even though Overend Gurney could have been saved (ch X, ¶ 11), it was correct for the Bank of England to choose not to save it.

This very traditional view of the central bank, as the entity that determines which banks are managed in such a way that they have the right to continue operating, indicates that the Fed’s error in 2008 was not the decision to let Lehman fail, but the failure to prepare the market for that decision beforehand. The Bank of England announced its policy of not supporting bill-brokers such as Overend, Gurney & Co. in 1858, fully eight years before it allowed Overend to fail. This failure was followed by a full century of financial stability. The Federal Reserve, by contrast, never clearly stated what the limits of its lender of last resort policy would be in the decades preceding the 2007-08 crisis. Indeed, the Fed was busy through those decades expanding the expectations that financial institutions had of support from the Fed. Thus, the post-Lehman crisis was decades in the making, and was further aggravated by the inadequate warning signs provided to markets subsequent to the Bear Stearns bailout.

The definition of the proper role of the central bank is probably the most important economic question of our times. We are learning through real-time experimentation what are the consequences of extensive central bank support of the financial system — and whether financial stability is better promoted by a 19th century lender of last resort that very deliberately allows mismanaged banks to fail or by a 21st century lender of last resort that provides much more extensive support to the financial system.

Discount Markets, Liquidity, and Structural Reform

Bengt Holmstrom has a paper explaining the “diametrically opposite” foundations of money markets and capital markets.* This dichotomy is also a foundation of traditional banking theory, and of the traditional functional separation that was maintained in the U.S. and Britain between money and capital markets.

Holmstrom explains that “the purpose of money markets is to provide liquidity,” whereas price discovery is an important function of capital markets. In a paper I extend this view a step further: money markets don’t just provide liquidity but a special form of price stable liquidity that is founded on trade in safe short-term assets; by contrast capital markets provide market liquidity which promotes price discovery, not price stability.

A century ago in Britain privately issued money market assets were, like capital market assets, actively traded on secondary markets. The two types of assets traded, however, on completely different markets with completely different structures that reflected the fact that money market assets needed to be “safe.”

To understand why the markets had different structures consider this question: how does one ensure that the safety of the money market is not undermined by asymmetric information or more specifically by the possibility that when the owners of money market assets have information that the assets are likely to default they do not use the market to offload the assets, adversely affecting the safety of the market itself, and therefore its efficacy as a source of price stable liquidity? The answer is to structure the market as a discount market.

In a discount market, every seller offers a guarantee that the asset sold will pay in full. (You do this yourself when you endorse a check, signing its value over to a bank — while at the same time indemnifying the bank against the possibility that the check is returned unpaid.) This structure was one of the foundations upon which the safety of the London money market was built. The structure ensures that the owner of a dubious asset has no incentive to attempt to sell it, and in fact is very unlikely to sell it in order to hide from the public the fact that it is exposed to such assets.

From their earliest days it was well-understood that discount markets were designed to align the incentives of banks originating money market assets and to promote the safety of the assets on the money market. (See van der Wee in Cambridge Economic History of Europe 1977.) Any bank that originates or owns a money market asset can never eliminate its exposure to that asset until it is paid in full. For this reason a discount market is specifically designed to address problems of liquidity only. That is, a bank that is illiquid can get relief by selling its money market assets, but if it has originated so many bad assets that it is insolvent, the money market will do nothing to help.

Contrast the structure of a discount market with that of an open market. On an open market, the seller is able to eliminate its exposure to the risks of the asset. This has the effect of attracting sellers (and buyers) with asymmetric information and as a result both increasing the riskiness of the market and creating the incentives that make the prices of the risky assets that trade on open market informative. Thus, it is because price discovery is important to capital markets, that they are structured as open markets. Capital markets can only offer market liquidity — or liquidity with price discovery — rather than the price stable liquidity of the money market. On the other hand, an entity with asymmetric information about the assets that it holds can use the open market structure of capital markets to improve its solvency as well as its liquidity position.

Historically it appears that in order for a money market to have active secondary markets, it must be structured as a discount market. (Does anyone have counterexamples?) That is, it appears that when the only option for secondary trading of money market instruments is an open market, then secondary markets in such instruments will be moribund. This implies not only that the absence of incentives to exploit asymmetric information plays an important role in the liquidity available on money markets (cf. Holmstrom) — but also that price stable liquidity is an important benefit of the discount market structure.

Both discount markets and open markets can be adversely affected by extraordinary liquidity events. But only one of the two markets is premised on safe assets and price stable liquidity. Thus, the lender of last resort role of the central bank developed in Britain to support the money (discount) market only. (In fact, I would argue that the recognized need for a provider of liquidity support to the discount market explains why the Bank of England was structured as it was when it was founded, but that goes beyond the scope of this post. See Bowen, Bank of England during the Long 18th c.) One consequence of the fact that the central bank supported only assets that traded on a discount market is that it was able to support the liquidity of the banks, without also supporting their solvency.

Given the common claim that one hears today that it is unreasonable to ask a central bank to distinguish illiquidity from insolvency in a crisis, perhaps it is time to revisit the discount market as a useful market structure, since acting through such a market makes it easier for a central bank to provide liquidity support without providing solvency support.


*His focus is actually money markets and stock markets, but in my view he draws a distinction between debt and equity that is far less clear in practice than in theory. In a modern financial system unsecured long-term bonds are not meaningful claims on the assets of a firm, because as the firm approaches bankruptcy it is likely to take on more and more secured debt leaving a remnant of assets that is literally unknowable at the time that one buys an unsecured long-term bond.

Yes, the Fed can be constrained by a law granting it discretion

Peter Conti-Brown argues that the Fed could have bailed out Lehman Bros. but chose not to and is using the law as “ex post rationalization” for political action. Philip Wallach argues that the fact that the Fed wasn’t satisfied with Lehman’s collateral means that it could not bail out Lehman.

The underlying question is this: What does the criterion in section 13(3) of the Federal Reserve Act, “secured to the satisfaction of the Federal Reserve Bank” mean? In my view the text itself makes clear that the interpretation of the meaning of the term “secured” has been delegated by Congress to the Federal Reserve Bank in question, subject in theory (although arguably not in practice) to a “reasonableness” standard.

For this reason, I am having difficulty following Conti-Brown’s argument. He writes in a follow up post:

The point is that “satisfaction,” in the midst of a financial crisis, is an entirely discretionary concept. . . . Instead, my argument—and critique—is that Bernanke, Paulson, Geithner, and others made a decision. They exercised the discretion they were entitled to make. They made these decisions knowing that there would have been enormous political fallout if they had bailed out another Wall Street “bank.” And they made it knowing that the legal authority to go in another direction was broad, robust, and entirely left to the discretion of two bodies of decision-makers within the Federal Reserve System. The Fed wasn’t legally obliged to do anything. But nor was it legally prevented from doing something.

I think there is confusion here. The Fed was granted discretion to interpret the meaning of the term “secured” within reasonable bounds.  It was not, however, granted discretion to do whatever it thought was necessary in a financial crisis.

Thus, I don’t think it makes sense to call the Fed’s claim that the law prevented it from lending to Lehman an “ex post rationalization.” Conti-Brown appears to be arguing that he knows that the Fed first decided not to bail out Lehman and then later determined that it did not have adequate collateral to “secure” a loan. But it seems much more likely to me that these two determinations were so closely intertwined that they were more or less determined at the same time: that is, it is at least equally likely that the Fed was unwilling to bail out Lehman, because in its discretion it found Lehman’s collateral to be inadequate. Given that the Fed is by law the entity that interprets the meaning of the term “secured” in section 13(3) of the FRA, such a finding by the Fed is close to a final determination on the issue. Thus, far from being an “ex post rationalization,” the Fed’s explanation that a bailout of Lehman was not permitted seems to me simply a description of its decision-making process.

In my view, the fact that the Fed could potentially have exercised its discretion differently is irrelevant. What makes this complicated is, of course, the fact that the Fed turned around found that collateral that was deemed inadequate on September 15 had become adequate a few days later. In short, Conti-Brown appears to be arguing that, if the Fed had authority to bail out AIG et al., then it must have had authority to bail out Lehman.

In my view, this gets the reality of the situation precisely backwards. I think that since the Fed didn’t have the authority to bail out Lehman, it probably didn’t have the authority to bail out AIG. Indeed, the AIG trial has made it clear that regulators believed that an AIG bailout was necessary and that they pushed legal interpretations of Fed authority to their absolute limits. In fact, I suspect you could even get some of the attorneys involved to admit the latter — though they would almost certainly also assure us that no lines were actually crossed. (For an example of this, see the Sept. 21 email from Fed General Counsel Scott Alvarez indicating that a term sheet produced five days after the AIG loan-for-stock bailout was announced had to be changed because the Fed couldn’t own AIG stock.) Whenever you are that close to a boundary, however, it seems very likely that there are lawyers and regulators close to the action and possibly even at a decision making level who believe that lines were in fact crossed — but that they were crossed in order to do what was necessary and in that sense in good faith. On the other hand, I would not expect any such privately held views to come to light until the passage of time has rendered them statements of only historical interest. (And even then only the non-lawyers are likely to speak out.)

In short, it seems to me that a position equally valid to that taken by Conti-Brown is that the Fed didn’t have the authority to bail out Lehman and didn’t have the authority to bail out AIG et al. But the more important point is that the law as drafted deliberately renders this whole discussion moot. The Fed was granted the authority by Congress to make two decisions within a week of each other that would appear to be contradictory. This, in fact, is the essence of what “to the satisfaction of the Fed” means.

How does a central bank stabilize the economy?

Nick Rowe has a post up from a week ago on macroeconomic policy. He points out that as long as we have a central bank, that central bank will be affecting the macroeconomy, so it doesn’t really matter whether we need a central bank to stabilize the economy. The only relevant question is what the central bank should do. In Nick’s view the old question was whether the central bank should follow rules or act on a discretionary basis, but now there’s a fairly strong consensus in favor of rules and the debate centers on whether central bank monetary policy can be effective in contemporary circumstances (i.e. at the zero lower bound).

Nick argues that monetary policy can be effective — the central bank just needs to keep buying something in order to put more money into circulation. He rejects the view buying private sector assets ends up being fiscal policy and argues that it can be kept neutral (as opposed to the picking of winners and losers that characterizes fiscal policy) by, for example, having the central bank buy the index of all marketable securities.

My model of a central bank is based on the role of the Bank of England in the early 20th century and thus it is very different from Nick’s (and from that of most macroeconomists). I don’t understand how it could ever be possible for monetary policy — or for a one-dimensional policy tool that adjusts either “M” or “i” — to stabilize the economy. To stabilize the economy the central bank needs to monitor and discipline the flow of credit through the banking system. It needs a window on how credit is flowing to the different sectors of the economy. And when the banks are allowing so much credit to flow to a certain sector that prices are becoming (or are likely to become) misaligned in that sector, the central bank needs to force the banks to restrict the flow to that sector.

In short, I think the focus on one-dimensional monetary policy is the reason we had the mother of all financial crises — and let’s be clear, from the British perspective this crisis was indisputably worse than any crisis experienced over the past two centuries with the possible exception of the outbreak of World War I.

Macroeconomists need to start thinking outside their general equilibrium models where there are no relative price distortions, and understand that the banking system has the capacity not just to inflate prices, but also to distort prices. And that when these prices get so out of line that readjustment is forced, the result is often disruptive. Thus, the important tasks of the central bank include keeping an eye on the flow of bank credit and taking quick action when this flow is clearly out of kilter.

Let me give a specific example. From mid-2004 to mid-2006 the Federal Reserve was raising interest rates. At the same time the origination of adjustable rate mortgages which have interest rates that are very sensitive to the short-term rate did not fall, but actually rose and remained elevated throughout the period of rising rates. (Compare Figure 14 with Figure 17 here.) When significant increases in interest rates have no effect on the quantity of credit, alarm bells need to start going off at the Fed. This is a clear indicator of Ponzi finance.

And we all know what was happening to housing prices as the same time that mortgage credit markets were clearly out of whack. The Federal Reserve should have observed the clear indicators of dysfunction in mortgage markets, and it should have taken action to adjust the flow of credit. If the Fed recognized this duty, the housing bubble would have been much less severe.

Unfortunately the Fed was very much focused on one-dimensional policy. If the Fed wants to stabilize the economy, it will have monitor the flow of credit through the economy and be ready to take action to address any markets where the data makes it clear that something is going very wrong.