Category Archives: Banking

Institutions, Gold, and Banking

Originally, I was going to write a post on the recent volatility of gold prices and what this means for gold standard advocates. Pondering the subject, I became more interested in what this means for free banking advocates, specifically those who think that gold (or similar commodity monies) would be the primary backing asset. Since paper notes are circulated in place of gold, the relative scarcity of gold is no longer an advantage in constraining the money supply. From here, it’s not a big jump to come to the conclusion that gold loses some of its purpose. If what matters is limiting fiduciary expansion, the choice of backing assets broadens, because the bank only needs a capital reserve to make good the liabilities that its circulating banknotes represent. Many people, incorrectly, assume that it’s the “backing” that decides the sustainability of a currency. It’s not, what give sustainability are the institutions of banking.

To see my particular angle of approach, let’s quickly look at this hypothetical history of banking in hypothetical Ruritania. After some time, gold emerges as the principal medium of exchange, and people begin depositing their gold coins at specialized businesses, or money warehouses. Not only is this for safety, but these warehouses also preform the function of financial intermediaries, settling debts on net and in bulk. In place of actual gold coin, people trade with redeemable warehouse receipts. At first, any lending by part of these warehouses comes from their own retained earnings, and not the deposits of their customers. Over time, though, these firms begin loaning their customers’ deposits, as they realize that some fraction of total liabilities aren’t redeemed at any given point in time. This system slowly becomes more sophisticated, as standards are developed and new types of promissory notes are introduced Ruritanian banks eventually begin circulating the banknotes we’re most familiar with — imagine U.S. dollars redeemable for gold at your local bank branch. Gold is relegated to a reserve asset used to settle interbank clearings (when bank A goes to bank B with a large stack of B’s notes and redeems them).

Bank notes, or inside money, are circulating in place of gold coin, meaning they essentially become just as good as the commodity money they’re substituting for. To keep things simple, let’s assume that all notes trade at par (a $1 bill is worth the full amount of the assets it represents). It’s the circulation of these notes that forms part of the process of competitive price formation, so any change in the quantity of bank notes will affect prices. In this banking system, changes in the quantity of bank notes are bound to occur. The substitution of inside money for gold is essentially an act of abandoning a relatively rigid money for a much more elastic one. If banks wanted to, they could print an unlimited number of bank notes — well, until Ruritania ran out of trees (or until the saw and paper mills decided to stop working because hyperinflation had caused the pricing process to break down). This seemingly presents a dilemma.

The main argument against central banking is precisely that this state of affairs makes it too easy for a monopoly to exploit the elasticity of the currency. But, free banking doesn’t suffer from the same problem. To paraphrase Douglass North, free banking is a case of being able to conquer ourselves once we conquered nature — we gradually make the uncertain more certain. As competition in banking increases, rival bank notes (assumed to be imperfect substitutes) will circulate in competition with each other. Banks will accept rival notes to accommodate their customers, but will then take them to rival banks for redemption. This makes for a relatively fast feedback mechanism for each bank to gauge the health of its balance sheet. An increase in the number of returning notes will force a bank to constrain loans, perhaps borrowing in the short-term to temporarily shore up its assets (waiting for loans to be repaid). Even a competitive banking system of this type is bound to suffer periodic instability, but the institutions also get more complex over time. For example, specialized banks, or clearinghouses, may arise that deal solely with the inter-bank clearing process. Ultimately, these can act as central banks of sorts, loaning to banks in times of temporary distress. Banks would have to conform to certain standards to be eligible for aid.

Usually, when we think of a commodity standard we consider one of its qualities to be the physical scarcity of the commodity, like gold. But, in reality, society decided that the relative inelasticity of gold was too costly, so it adopted the more elastic system of inside money circulation. But, in doing so it abandoned one of the most important safeguards against inflation. What took their place were certain institutions, such as competitive banking. What this means to gold standard advocates is that they should stop advocating the gold standard and instead ask for competitive banking. Any gold standard that’s left unconstrained will eventually evolve into the more modern system of money substitute. The natural qualities of gold are therefore irrelevant. What matters are the institutions developed to not only take the place of these qualities, but to do the job in a superior way. Even a full reserve banking system has to develop institutions to constrain the elasticity of inside money.

One final point. In our hypothetical world of free banking, gold is relegated to the task of paying net debts through the inter-bank clearing process. Why couldn’t alternative assets take the place of gold? A bank could transfer ownership of another asset, such as a securitized loan portfolio. One thing that decides the quality of an asset, though, is its information sensitivity. An asset that varies in value is going to be more information sensitive, because the counterparty is going to have to track changes in price. For example, a $1,000 securitized loan portfolio with a 90 percent repayment probability may, at first, be valued at $900 by the counterparty (.9 × $1,000). As it turns out, this asset is volatile, with default probability fluctuating between 10 and 30 percent. If a bank is is using volatile assets as a capital reserve, then there will be a relatively greater likelihood that this bank is going to be unable to make good of all short-term liabilities (if the value of its “backing assets” falls). The banknotes of these banks are going to be circulating at less than par. Competitive pressures are going to force them out of business, and customers are going to choose more stable currencies. Another consideration is liquidity. Counterparties, or rival banks, are hardly going to want to be repaid in relatively illiquid assets. The advantage of gold as a backing asset is that it’s relatively information insensitive (if the money supply is constrained enough) and it’s relatively liquid. But, there are a broad range of assets, and new assets are likely to be developed over time, so we can’t discount the possibility of other assets being mixed with, or replacing, gold.

As I write above, it seems to me that what I argue here is true whether you advocate for a free fractional reserve or full reserve banking system. If inside money is circulating in place of outside money, what matters is constraining the supply of inside money. Banks can still choose amongst different backing assets, but even full reservists are interested in constraining money supply volatility. Ultimately, the end goals of the two approaches aren’t too different. Fractional reservists look to competition as a means of developing means of restricting excess money creation. Full reservists look to legal constraints, or some simply claim that a competitive banking system would be forced to full reserves because of instabilities allegedly inherent in fractional reserve banking. Whatever the case, what matters are the institutions, not gold.

Gorton with a Twist

1 Reply

In yesterday’s post, I talked a little about Gorton’s idea of “information insensitive” debt. Well, this framework might be a good one to interpret some of Bernanke’s “unconventional” monetary policies, especially those concerning the purchase of agency mortgage backed securities and long-term U.S. treasury bonds. Of course, QE1′s objective was to restore some liquidity to wholesale short-term credit markets, such as repo and commercial paper, by taking off weak banks’ balance sheets a large amount of mortgage backed securities. A couple of years later, the Fed began “Operation Twist,” which was about buying long-term treasury bonds by selling the Fed’s holdings of short-term bonds, with the purpose of reducing the average maturity rate of treasury bonds held by private agents.

Why? Well, I have no definitive answer, but one possibility relates to these short-term credit markets. Specifically, traditionally, banks operate by borrowing short and lending long. Similarly, banks accumulated large numbers of treasuries as a means of collecting collateral to back debts borrowing on the wholesale credit market (a role that MBS’ and other related assets played prior to late 2007). But, long-term treasuries mature over the long term, and so Operation Twist can be interpreted as a means of forcing the private market to accumulate short-term assets, which means that banks would be borrowing short and lending short. In other words, it’s a method of reducing the chance of maturity mismatching, and the consequent drying up of wholesale credit markets, by making private assets better suited for what banks intend them for.

What is interesting about Operation Twist is that, usually, the opposite is considered preferable. That is, governments prefer to lengthen the maturity of their debt, to avoid having to roll over debt on terms that may become increasingly unfavorable. For the United States, of course, this may not currently be a problem.

Also, this explanation should take away from a common narrative that the Fed is purposefully pushing down interest rates on U.S. treasuries. I’ve taken some flak for not supporting this anti-Fed position, but I simply don’t think it’s very credible. The fact is that after the 2007–09 crash there was a spike in private demand for treasuries, because there was a dearth of safe assets after the collapse of the mortgage market. Even currently, there are really no private assets that are as safe as treasuries. The only caveat is that investment banks haven’t really recovered, in the sense that they haven’t recovered their status, so I’m not sure what wholesale credit markets look like currently. Without a doubt, though, a major concern of the Fed has been to revive these markets (as opposed to supporting fiscal policy by monetizing debt).

P.S. Just to get an idea of what wholesale credit markets currently look like, check out the status of asset backed commercial paper (ABCP) outstanding,

Some Quibbles on Gorton (2012)

1 Reply

The title is referring to Gary Gorton, Misunderstanding Financial Crises (Oxford: Oxford University Press, 2012). This post is a collection of “complaints,” as I read through the book. More than “complaints,” these are points of contention. This is largely meant to keep track, but also for the sake of stirring up conversation. Also, these are likely to reappear in any review I write of the book, and it’s worth hashing out these ideas (for example, in case I’m missing something). Finally, these quibbles shouldn’t imply that the book is “bad” or not worth reading. Whatever my issues are, this is an incredibly interesting and informative book (like most of Gorton’s work).

1. Private information insensitive debt

Gorton holds that private banks cannot create information insensitive debt. He doesn’t offer any evidence of this, except for the fact that state banks during the “free banking” era, in the United States, could not create information insensitive collateral. Gorton explains information sensitivity as the degree to which it pays to know “secrets.” I don’t like this definition; more than about knowing, it’s about uncertainty. Collateral is information sensitive when you’re not sure of the true value — when market “signals” suggest that the collateral backing debt is unsafe. Since the National Banking era, that collateral has been Federal public debt, which in the United States has historically been “information insensitive” (Gorton writes as if this were a sure thing, but Federal public debt would be “information sensitive” if there were a risk of sovereign default, for instance).

Prior to the U.S. Civil War, some states passed “free banking” laws. Gorton explains that these laws required banks to back their money substitute with state bonds; but, these were not always information insensitive (e.g. if the state was suffering from fiscal problems) and so most of the time notes would trade under par (also, given the era, they’d also have to factor in distance from the issuing bank, et cetera). But, then, he goes on to blame this on the inability of private banks to create information insensitive debt. I don’t understand, because the “private market” does have information insensitive collateral: outside money (e.g. gold, silver, et cetera). No less, the ability to produce alternatives is severely handicapped when one of the conditions of being granted a charter is to back liabilities with a specified bond, in this case state bonds.

This being said, something Gorton doesn’t directly point out is that information sensitive collateral may be useful. Let’s say that a bank has reached its limit on fiduciary extension backed by outside money, so it decides to use different forms of assets (e.g. asset backed securities). Information sensitivity is a form of market discipline, where notes trading below par signal over-issue by the banks: it limits credit expansion.

But, my main point is that outside money is information insensitive (unless there are wild fluctuations in its supply, but “hard money” evolved largely to avoid these transaction costs). Why doesn’t Gorton consider it?

2. Moral hazard

I’m sympathetic to the notion that moral hazard, in banking, is not as relevant as some people assume. But, one of the arguments used by Gorton, I think, is ridiculous. He invokes an economist by the name of T. Bruce Robb, arguing that moral hazard is irrelevant because depositors don’t know enough to discipline banks anyways. Again, I’m somewhat sympathetic, and I can see it as relevant to an extent. What bewilders me is that Gorton doesn’t realize that the notion of information sensitivity undermines this argument. Notes trading below par are examples of depositors disciplining their banks: banks with notes trading below par are likely to get less business, and in fact customers will probably opt to withdraw their outside money and re-deposit it in a safer bank (one with higher value notes). This, of course, is another advantage of competitive note issue (that Gorton doesn’t consider).

3. Liabilities v. assets

I have to find the exact page number, although it’s a subtle theme throughout the book, but Gorton emphasizes the liabilities side of banks’ balance sheets more than the assets side. His reasoning is that financial crises are mostly about banks being unable to meet their short-term liabilities — that is, banks become illiquid. While he casually mentions this during his description of certain historical cases, the fact is that banks have a hard time meeting short-term obligations only because their assets have deteriorated in value. Specifically, short-term credit markets dry up because the collateral banks offer in return is information sensitive, which really means that their value is suspect and their liquidity declines. The assets side is just as important as the liabilities side.

One issue is that by overemphasizing the liabilities side, Gorton seems to suggest that crises are, in a sense, forced on banks. For example, he explains the Livingston Doctrine, which holds that banks during crises are really illiquid, not insolvent, and so should be bailed out (including temporary suspension of redemption and bank holidays). I’m not questioning the value of these measures, rather what I’m drawing attention to is that financial crises are as much about assets as they are about liabilities. In fact (actually, as Gorton’s book shows), crises are usually caused by a deterioration of banks’ assets, since the values of these are directly related to “market fundamentals.” Similarly, Gorton writes that banks have a hard time turning illiquid assets liquid, because “fire sales” invariably lead to drastic declines in assets’ prices. But, these declines are “efficient,” in the sense that they force banks to price assets reflecting the true nature of their risks (although, admittedly, fire sales can push assets’ prices down below their clearing price). We have plenty of competing theories that all agree that much of this deterioration is caused by the banks themselves: either through bad loans, or through price distortions.

I don’t think this quibble is particularly important (although, I have to re-read some sections of the book to better remember some of the implications that Gorton drew from his emphasis on the liabilities side). Nevertheless, it’s as strange an error on Gorton’s part as the previous two points. This being said, near the end of the book, Gorton does make a good point: many assets unrelated to subprime mortgages, and, in any case, those which actually had a heightened propensity for failure, also fell in value. He makes the case that changes in expectations also undermined the value of assets which weren’t at risk, such as those securitizing school loans, credit card loans, car loans, et cetera. I’m not at home, so I can’t look at some sources I have in mind, but I have some skepticism because I thought that most investment banks, what Gorton calls “dealer banks,” mostly held mortgage-backed securities — I’ll have to confirm this when I get back home (or someone can do it for me!).

Credit Injection, Empirically

5 Replies

I haven’t been too involved with the debate on Cantillon effects, largely because I’ve been busy with other things. But, I’d like to make a quick comment on a recent post by Nick Rowe (H/T Robert Murphy) on the topic, just to clarify what Hayek and other Austrians actually have in mind. Near the beginning, Rowe writes,

Hayek assumed that newly-printed money is injected via the credit market, where it pushes down the market rate of interest relative to the natural rate of interest, and this causes people’s plans and expectations to be mutually inconsistent, and this causes an unsustainable change in the time-structure of production.

But Hayek’s assumption is obviously usually wrong.

My first reaction was, “No, Rowe is obviously wrong.” My second reaction was, “In part he’s right, but there’s something about this argument that is ignoring key aspects of the Mises–Hayek theory.” Because I haven’t been involved with the debate, it’s difficult for me to frame this within the context of said debate, but I think excessive focus on the Federal Reserve (which could be the fault of any [or both] side[s]) is drawing attention away from what makes Hayek’s assumption “obviously usually right.”

When the Fed injects money, Rowe is right that it can do so through various avenues. It can buy U.S. Treasuries through open market operations, both from banks holding these assets or from the government proper (less likely?), or it can control base money through short-term interest rates, et cetera. But, the Fed, in the role of credit expansion specifically, is really a “minor player,” in that most of the credit expansion comes as a result of private bank lending — this being said, the Fed is a “major player” in that it shapes the banking industry and decides the extent of credit expansion (i.e. if Selgin’s banking model in the Theory of Free Banking is correct, credit would be much more restricted in a competitive banking system). Hayek, in “Monetary Theory and the Trade Cycle” (e.g. p. 77 [in the Mises Institute's hardback edition of Prices and Production and Other Works]), explicitly recognizes the role of private banks (the “existing credit organization”) in perpetuating credit expansion.

So, when Hayek is talking about the “inconsistency of plans and expectations,” in the context of the business cycle, the credit injections specific to Fed action are quite limited relative to the participation of private market financial firms. Although Rowe told me he doesn’t get Hayek’s theory, the last few paragraphs of his post say otherwise — what matters is the effect of base money on the extent of private market credit expansion.

To make my point clearer, let’s reformulate my argument to directly respond to the following claim by Rowe,

If Austrian economists are right to insist that it really really matters where the new money is injected into the economy, then Hayek was making a very special assumption, one that is nearly always empirically false, and false in a way that matters a lot, and so Hayek’s analysis is mostly irrelevant.

What Rowe may not realize is that the “special assumption” is actually his: the Fed is not the only organization/firm that injects new money into the economy. Banks also create money through loan over-extension, or credit expansion, “in a way that matters a lot,” making Hayek’s analysis mostly relevant. If you want some empirical evidence,

Treasuries and the GSEs

Freddie and the Crisis

7 Replies

The Eugene Fama interview I linked to yesterday created some Twitter discussion that I managed to catch. There was a hint of a suggestion that government sponsored enterprises (GSEs), specifically those responsible for assisting mortgage markets, were, at least partially, responsible for the 2007–09 crisis. The reasoning is that by buying mortgage backed securities and related financial instruments the risk of these assets was transferred to the public. Since GSEs were willing to buy these assets, paying very little scrutiny, banks were willing to accept and package riskier loans that otherwise wouldn’t have been made.

The evidence, however, suggests that the role played by GSEs was relatively minor. It’s true that GSEs purchased large amounts of mortgage-backed assets and securitized private assets. This latter action, in particular, distorted risk assessment, since mortgage-backed securities were categorized under a lesser risk basket under the recourse rule. But, these distortions were relatively minor, knowing that the recourse rule already benefited investment into the mortgage market — not to stimulate mortgage creation, but because these were considered low-risk assets, at the time.

Investment banks were also loading up on mortgage-backed financial instruments, driven in large part because of the United States’ macroprudential regulations (Friedman and Kraus [2011]). These banks were preforming a role similar to that of the GSEs, and this becomes clearer when we separate the conventional banking industry into two parts: commercial and investment banks. Commercial banks were the “loan originators,” or the institutions which directly financed commercial and residential mortgages. These were then purchased by investment banks, transferring the risk of the loan away from the originator. Investment banks, and GSEs, then packaged securities to hedge risks between individual loans, with loans of different quality organized into tranches. In theory, this reduces risk, therefore reducing interest rates and benefiting borrowers. The point is, if it was risk distribution that caused the financial crisis, then investment banks were just as, or more, responsible than the GSEs.

Why did investment banks begin to concentrate their portfolios on mortgage-backed assets? As aforementioned, existing capital regulations rewarded “low-risk” investments by requiring lower reserves. Mortgage-backed assets were highly rated; individual loans that were poorly rated were simply pooled with higher quality debt, and the security received a high rating. This has led many scholars to point their fingers at rating agencies, with some authors suggesting that the system was “gamed” by investment banks. An uncompetitive rating market — since only three rating agencies are considered “nationally recognized” — may be partially at fault (White [2009]), but I don’t think it’s the full story. Rating agencies were operating under some basic assumptions, including,

Housing prices were likely to continue to rise;
Declines in housing prices would be local and isolated, rather than nation-wide (justifying the pooling of subprime tranches).

The crisis comes down to changes in housing prices. Why did they rise, why did they stagnate, and why did they fall? The Austrian explanation for the first part of the question is: credit expansion. Following the 01–03 recession Alan Greenspan’s Fed lowered the Federal Funds rate, stimulating loan extension, especially for investments particularly sensitive to changes in central bank policy — i.e. mortgages. Housing prices rose, especially where purchases were concentrated, and bubble began to form. The common explanation for the stagnation that occurred between 2005–06 is usually attributed to the increase in the Federal Funds rate, inciting loan originators to issue higher volumes of adjustable rate mortgages (ARMs). These had teaser rates to draw borrowers, but allowed for changes in the rate of interest in accordance with changes in Fed policy. Banks weren’t particularly worried about the risk that some borrowers wouldn’t be able to re-pay loans at higher rates, because it was assumed that the borrower would simply refinance.

In conjunction with a higher volume of ARM loans, we also see a steady rise in the rates of default. What this suggests is that housing was becoming less affordable; credit expansion slowed and the rate of price change fell and soon stagnated. Without rising prices investment in housing soon became even more unaffordable, ultimately targeting a reversal in price changes (i.e. a decline in prices). The banking crisis followed, as mortgage-backed assets plummeted in value and secondary markets dried up. The fact that multiple local markets collapsed simultaneously is a strong indication that there was a common factor, and I think the evidence favors the Austrian story of price distortion.

There is a second way of considering the role of GSEs. Remember, the idea is that GSE involvement distorted risk assessment by transferring risk to essentially public balance sheets. This implies that bankers were knowingly taking greater risks in allocating capital. While not directly commenting on the role of GSEs, there is a literature that deals with financial risk-taking before the crisis. It revolves around the question of whether CEO compensation promoted risk-taking, increasing management compensation through short-term profits at the expense of long-term health. There is strong evidence that there was no significant relationship between compensation and risk taking (see: Fahlenbrach & Stulz [2011]; Acrey, McCumber, & Nguyen [2011]; Murphy [2012]). Further, most banks were over-capitalized and management took huge hits during the crisis, usually because much of their income was in the form of claims on equity (in a paper written as a reply to Fahlenbrach & Stulz [2011], which was originally published as a working paper, Bebchuk, et. al., concede that the losses of 07–09 were greater than whatever gains CEOs earned between 2000–06). The evidence simply doesn’t support the thesis that banks purposefully, or knowingly, chose higher-risk investments. This applies to the GSE thesis as well as it does to the compensation thesis.

(Also, the evidence against the purposeful risk-taking thesis is slight support for Austrian business cycle theory, in the sense that it hurts the case of alternative explanations of unstable price bubble, viz. Minsky’s financial instability hypothesis. This doesn’t mean that some loan originators weren’t extending loans to people they knew wouldn’t be able to re-pay, but ultimately these originators are restricted or financed by investment banks.)

I certainly don’t think that GSE involvement in mortgage markets are a good thing, but history simply doesn’t support the argument that it was this involvement which caused the financial crisis. Overall, GSE involvement only accounts for a small part of the crisis, and it definitely had little to do with the root cause: price distortions caused by fiduciary over-expansion.

Modeling and Ignorance

19 Replies

I recently read part of a two-year old study by the IMF, “Sovereigns, Funding, and Systemic Liquidity.” In chapter two, amongst other things, it talks about modeling techniques used to rate counterparty risk, which basically means accurately gauging the value of the collateral. The financial crisis — referring to the systemic bank failures of 2008–09 — was caused by the mispricing of risk. To some people, especially those who don’t blame the market per sé, the point I’m about to make might seem obvious. On the other hand, it really is lost on the majority.

Everyone is beset by radical ignorance, or radical uncertainty: the unknown unknown. There are factors that models can’t control for, simply because their creators can’t fathom their existence. It’s simply unrealistic to argue that we need to develop better models. More than unrealistic, it’s unhelpful.

Actually, radical uncertainty speaks favorably for the market process. Jeffrey Friedman and Wladimir Kraus, in Engineering the Financial Crisis (pp. 126–128), make a good case as to why this is so. Weighing counterparty risk, to a large extent, fell on the rating agencies. The rating agency industry is heavily regulated, and is dominated by the “big three” which are given their status by the government. In a more competitive market, it’s certainly possible that better models would have been developed. This is the benefit of heterogeneity in decision-making.

At the risk of detracting from the principal point being made, it’s also worth considering the role of prices in propagating mispricing. Why were mortgage backed instruments so highly rated? Because, (i) prices were rising, and (ii) nobody expected a broad decline in these prices. Few, if any, studies have really answered the question of what caused housing prices to continue to rise. There’s a good case to be made that it was intervention, either through the Federal Reserve’s policy of lowering the Federal Funds Rate in the early 2000s, as a result of the organization of the banking industry (around a monopolized currency, as opposed to competitive currencies), or both. In any case, the evidence clearly suggests that most people were misled by prices, and this was out of their direct control.

That few economists are considering the role of price distortions, or the more general concept of radical ignorance/uncertainty, is unfortunate. It really hurts the quality of post-recession debate, and ultimately it inhibits progress towards any truly effective solution (in my opinion, the market, but you don’t have to accept this conclusion to acknowledge the importance of ignorance).

Economic Thought

A heterodox perspective on theory & history