avatar

Misunderstanding Financial History

by George Selgin July 11th, 2013 3:45 pm

I've long considered Gary Gorton one of the best economists working in the fields of banking and finance, thanks in no small part to his excellent work on 19th-century U.S. financial history. Gorton's reputation was dealt a hard blow recently owing to his role in supplying AIG with the models it relied upon in assessing the riskiness of credit default swaps it wrote on mortgage-backed securities. Gorton's part in the AIG demise hasn't itself altered my high opinion of his work. I am disappointed, however, with his apology for playing that part, as given in his 2012 Oxford University Press book, Misunderstanding Financial Crises.

Apology? Well, sort of: throughout the book Gorton refers, not to "his" mistakes but of those of "economists" generally, as if the entire profession, rather than a small (though disproportionately influential) part of it, were to blame for the fancy risk models and associated rose-colored prognostications that sank AIG and so many other financial behemoths. His is, in other words, not an outright mea culpa but a mea culpa bundled with such a large number of sua culpas as to expose him to only a miniscule risk of having to shoulder much blame. Indeed, Gorton sees himself as a victim of his profession's errant ways, chief among which was its inclination to treat fancy statistical models as substitutes for a genuine understanding of the lessons of economic history.

That inclination, Gorton says, when combined with excessive reliance upon data limited to the "Quiet Period" since the establishment of the FDIC, caused economists, himself among them, to assume that the underlying causes of financial crises had been successfully eliminated, making such crises a thing of the past. More attention to history, Gorton suggests, would have made him and his peers less sanguine. It would have warned them that the data against which they were calibrating their models did not suffice to uncover the U.S. financial system's "deep parameters." It would, in short, have shown that the root causes of crises had yet to be dealt with.

One can only applaud Gorton for rejecting the view, which is indeed all too prevalent among today's economists, that little can be learned from history because it "comes from a different structure," and for regretting the deletion of economic history courses from PhD curricula that this view has encouraged. "The relevant past," Gorton insists, is the history of market economies, not an arbitrary recent period that is largely determined by data availability" (pp. 95-96):

The past, a rich laboratory for understanding the present, lacks data richness, which is needed for some models, but we need to add economic history to the Facts. Sophisticated econometric methods come at a large cost; the data requirements narrow our field of vision. In this trade-off the loser has been economic history (p. 97).

Amen and amen. But there is right as well as wrong economic history, and wrong economic history can be just as productive of mistaken policies as the most naive formal models. Alas, Gorton's own economic history, as presented in the first part of his book, is wrong in crucial ways.

In brief, that history goes like this: Before the Civil War, banks were set-up by state governments, either through charters or (starting in 1837) by means of so-called "free banking" laws. Notes issues by those banks, though the only paper currency available, circulated, not at their face or "par" values but at varying discounts reflecting the idiosyncratic and uncertain ("secretive") content of particular banks' asset portfolios. The National Bank Acts created a uniform currency, while eliminating banknote-based runs (that is, runs to exchange banknotes for specie or legal tender) by taxing state banknotes out of existence while requiring all national banks to fully back their own notes with safe U.S. government securities. Unfortunately demand deposits, which continued to be backed by idiosyncratic and "secretive" bank assets, become increasingly important, and panics could and did still happen when bank customers lost confidence in the assets backing those deposits. Although the Fed, established in 1914, was supposed to rule-out such panics, it was thanks to the FDIC, established two decades latter, that the U.S. finally entered a "Quiet Period" during which no panics occurred. But the quiet period proved to be something of a fool's paradise, because during it, and especially during its last stages, further financial innovations made it possible for new forms of financial-institution debt, including repurchase agreements, to play a role in payments and other transactions not unlike that once performed only by banknotes and checkable deposits. Because these new types of debt were issued by "shadow" banks operating outside of the limits of the Federal safety net, they in turn became the object of a systemic run--the "Panic of 2007."

From this history Gorton derives the lesson that "financial crises are inherent in the production of bank debt...and, unless the government designs intelligent regulation, crises will continue" (vii). As for what constitutes "intelligent regulation," Gorton's suggestion, informed by his understanding of the lessons of the free banking and National Banking eras, is that all forms of bank debt used to conduct transactions must be backed by collateral "produced in such a way that it is secretless," and all issuers of such transactable debt should have access to the Fed's discount window. In particular, in light of the recent crisis, so-called "shadow" banks should be converted into what Gorton calls "Narrow Funding Banks" (NFBs), which would be prevented from engaging "in any activity other than purchasing asset-backed securities, government [sic] and agency securities (p. 197). As for repos, they need to be more strictly regulated, in part by placing limits on how many repos nonbanks can engage in.

Gorton's recommendations are consistent enough with his understanding of financial developments leading to the 2007-8 crisis. However, that understanding warrants a judgement similar to the one Gorton himself offers regarding less history-conscious attempts to explain that episode, to wit, that it is a "superficial" understanding suggesting "a lack of institutional and historical knowledge" (88-9). The difference is that Gorton has the knowledge in question, as is apparent from his other writings and also from the works, with which he's evidently familiar, discussed in his "Bibiographical Notes." Nevertheless his book fails to make proper use of that knowledge.

Gorton is wrong, first of all, in claiming that financial crises are "inherent" and "pervasive" in market economies. He errs both by not allowing that different "market" economies have had very different kinds and degrees of financial regulation, and by not consistently heeding his own definition of a banking "crisis" as a "systemic" (or at least "widespread") "exit from bank debt," that is, a situation involving "en masse demands by holders of bank debt for cash" (pp. 6-7, my emphasis). According to this definition many of the "crises" listed on Gorton's Table 3.1 were not genuine financial crises at all. Canada, to take one example, did not have a genuine financial crisis in any of the years listed (1873, 1906, 1923, and 1983), though it did have to relax binding capital-based note issue regulations to avoid having a crisis in 1906.

I refer to Canada in particular because, with regard to Gorton's thesis, it is, not the only, but certainly the biggest, elephant in the room. Its record is especially revealing, because the Canadian economy of the 19th and early 20th centuries resembled the U.S. economy in many ways, though it differed in its banking structure and regulations. Unlike U.S. banks, Canadian banks could and did establish nationwide branch networks; they were also allowed to issue notes backed by their assets in general rather than by any specific collateral. It was, finally, no coincidence that the extra degrees of banking freedom that Canada enjoyed were associated with a much better record of financial stability. To put the matter differently, Canada's record suggests that the shortcomings of the U.S. banking system where not shortcomings "inherent" to all private banking and currency systems. They were shortcomings traceable to specific, misguided U.S. banking and currency laws.

Take those discounts on antebellum U.S. banknotes. Gorton attributes them to the fact that different banks, whether "free" or chartered, had different assets backing their notes, with some assets being more "suspect" than others (pp. 15-16). According to his understanding, nothing short of a rule forcing all banks to back their notes with identical, riskless assets could serve to make a uniform, par currency out of notes issues by numerous, otherwise independent banks. State "free banking" laws failed to achieve this result because different states allowed different assets to serve as note collateral, and because some of this collateral was anything but risk free. The problem was only solved when, during the Civil War, state banks were taxed out of the currency business, while new National ones had to back their notes with U.S. government bonds, which, once the war was over, were perfectly safe.

If Gorton's interpretation were correct, we should only expect to find commercial banknotes circulating at par where U.S. style backing requirements are in place. But by the 1890s Canada, despite being far less populous than the U.S., while occupying more square miles, had a uniform currency consisting mainly of private Canadian banknotes that were not subject to any special "backing" requirement. How could that be? That Canada's banking system was a "club oligopoly" may have helped. But there's another explanation, which also accounts better for other instances, such as Scotland's, of uniform currencies consisting of private banknotes backed by bank-specific assets. This is that Canadian banks, unlike their U.S. counterparts, were free to establish branch networks, and that such networks, together with note clearinghouses established in major trade centers, sufficed to eliminate note discounts, by reducing to trivial amounts the cost to banks of presenting rival banks' notes for payment. For a bank's notes to remain on the "current" list thus became a simple matter of its demonstrating a willingness to cooperate in regular (eventually daily) settlements. In the U.S. itself the Suffolk System manged to make all New England banknotes current throughout that region, even despite restrictions on branch banking, decades before the Civil War, not by telling its members what assets they could own or by otherwise monitoring their assets, but simply by insisting that they keep up their settlement accounts. These and many other examples I might cite make it clear that full backing by risk-free assets is not a necessary condition for a uniform private banknote currency.

What's more, it isn't a sufficient condition. For although Gorton claims that the National Currency and Banking legislation of 1863 and 1864 succeeded in finally eliminating banknote discounts by requiring full (or more than full) backing of all national banknotes by U.S. government bonds (p. 18), the truth is otherwise. National banks were no more willing than their state predecessors had been to bear the cost of sorting and shipping rivals' notes to thousands of other (unit) banks, many of them located long distances away, for payment. Consequently national banks did at first occasionally refuse to accept other national banks' notes at par. That changed in 1864, not because national banks suddenly realized that all their notes were equally good, but because a provision of the 1864 Act (sec. 32) required that every national bank receive every other national bank's notes at par.* Similar legislation, had it been imposed on antebellum banks, might also have made their notes current, though not without causing other, perhaps more serious mischief.

Remarkably, Gorton makes hardly any mention in his book of the role of unit banking laws either in preventing the emergence of a unified U.S. currency market or in contributing to the likelihood of bank failures and crises by creating a system consisting of many thousands of mostly tiny and under-diversified banks. In listing the provisions common to the so-called "free banking" laws, for example, he omits the one disallowing branching (pp. 12-13). (Neither "Unit banking" nor "Branch banking" appear among the terms listed in the book's index.) To say that telling the history of U.S. financial instability without mentioning the part played by unit banking is like staging a performance of Hamlet without the Prince of Denmark is to resort to a very tired cliche. But in reading Gorton's book I could not help having the cliche insistently come to mind.

The difference between Gorton's conclusion regarding the "inherent" vulnerability to crises of any economy having lots of bank debt and that of one of his occasional co-authors, Charles Calomiris, in his own recent study, is striking:

[E]mpirical research on banking distress clearly shows that panics are neither random events nor inherent to the function of banks or the structure of bank balance sheets....The uniquely panic-ridden experience of the U.S., particularly during the pre-World War I era, reflected the unit banking structure of the U.S. system. Panics were generally avoided by other countries in the pre-World War I era because their banking systems were composed of a much smaller number of banks operated on a national basis, who [sic] consequently enjoyed greater portfolio diversification ex ante, and a greater ability to coordinate their actions to stem panics ex post.**

Despite the debilitating effects of barriers to branch banking, U.S. panics prior to the passage of the Federal Reserve Act generally did not involve outbreaks of distrust of most, let alone "all" (p. 32) bank deposits. Instead, distrust tended to be confined to banks that had suffered from prior shocks, or to banks that were associated with others that had suffered from such shocks. Bank runs appear, in other words, to have been informed, if only imperfectly, by bank-specific information. According to George Kaufman, a general flight to currency appears to have occurred during one pre-Fed National Banking era panic only--that of 1893. And even in that case, as Gorton himself recognizes (p. 77), the general flight was a response to prior, exceptionally widespread industrial and mercantile failures which, given banks' limited opportunities for portfolio diversification, gave depositors good reason for anticipating similarly widespread bank insolvencies.

That most panics didn't involve general flights to currency doesn't mean that the public's desired currency ratio didn't increase on other occasions, or that those increases were not a cause of financial distress. Of such occasions the most important was the harvest and subsequent "crop moving" season, roughly from August through November, when currency was needed to pay migrant farm workers. Depositors' attempts to convert deposits into currency for the sake of making such payments, for which checks were unsuitable, had nothing to do with them fearing that their banks might be insolvent. However, thanks to binding national banknote collateral requirements such attempts could leave banks with no choice but to draw upon their legal reserves. Those reserves might consist--again thanks to unit banking laws, and also to national banking laws sanctioning the practice--of country-bank deposit credits at so-called "reserve city" banks, whose own reserves might in turn consist of deposits at New York ("central reserve city") banks. To pay out a single dollar of currency a country bank lacking any surplus bonds might, in short, find itself triggering a three-dollar decline in total banking system reserves, with the brunt of the burden being felt in New York. Consequently a sharp-enough rise in the public's desired currency ratio, though itself based on routine transactions motives, might nevertheless lead to a credit crunch and even, in extremis, to a currency famine. That credit tended to tighten every autumn under the pre-Fed national banking system, resulting in a marked seasonal pattern in interest rates, is well established, as is the fact that several panics took place during the harvest and crop-moving months, suggesting, not necessarily that harvest-related currency demands triggered the panics, but that such demands may have contributed to their severity.

Canada avoided both panics and any seasonal tightening of credit thanks again to its banks' ability to branch and also to their ability to give customers all the notes they wanted in exchange for their deposit credits, without having to make costly (let alone impossible) adjustments to their asset portfolios. Although entry into Canadian banking was very strictly regulated, established Canadian banks were genuinely (and not just nominally) "free." That many contemporary experts favored granting national banks Canadian-style freedoms, by allowing them to branch and by repealing the bond-deposit requirement of the National Bank Act, as the most straightforward way to put an end to U.S. currency shortages and panics, is yet more evidence contradicting Gorton's account. If Bordo, Redish, and Rockoff are right, even Canada's dodging of the recent financial crisis is attributable to a significant degree to the freedom it awarded its banks back in the 19th century:

Because of the fragmented US banking system, and because of various restrictions placed on the assets the banks could own, securities markets emerged to finance most economic growth,unlike Canada which developed a bank-based system. Mortgage markets and housing finance also developed differently in the two countries. Investment banks, which participated in the creation and marketing of securities, became an important part of the system. Thus the United States always had something like the ‘Shadow Banking system’ that has been the subject of so much recent discussion.

Unsurprisingly, the lesson taught by this different understanding of financial history itself differs dramatically from the one Gorton offers. It is that there are better ways to avoid financial crises than by trying to regulate risky bank debt out of existence. They are better both because they can actually succeed (whereas the war on debt Gorton proposes would probably prove as futile as the war on drugs) and because they get rid of the financial crisis bathwater without sacrificing the financial intermediation baby. For that reason I'm convinced that, should Gorton's version of history prove persuasive, it could end up proving no less misleading, and far more costly to society, than the models he concocted for AIG.

______________________________________
*The same law provided facilities--though very inadequate ones--for the centralized redemption of national banknotes. In 1874 a new and and better, though still far from adequate, redemption agency was established.
**In correspondence Professor Calomiris has alerted me to his forthcoming Princeton University Press book, with Stephen Haber, Fragile By Design: The Political Origins of Banking Crises and Scarce Credit, which "provides much more evidence that banking crises are the outcomes of political choices, not inherent fragility." The book is, as, Tyler Cowen might say, "Self Recommending." (Added 7-12-2013).

23 Responses to “Misunderstanding Financial History”

  1. avatar JonCatalan says:

    In the book, he mentions that prior to using Federal debt as collateral, many state-banking laws required banks to back their currencies with state-issued debt. Gorton writes that this debt was often information sensitive, which implies that state finances were not always in order. Then he writes that the market failed to provide an information insensitive backing asset. He doesn't reconcile this latter conclusion with the fact that banks, for the most part, weren't allowed to use private collateral to back their banknotes.

    • avatar George Selgin says:

      Jon,

      Gorton wants bank debt to be backed only by essentially risk-free and homogeneous assets, which basically means U.S. Treasury and agency bonds, and MBSs backed by Federal authorities. Left to their own devices, state banks would have backed their notes largely with loans, which were of course neither risk-free nor homogeneous.

  2. avatar hollander says:

    Bravo! Exellent piece.

  3. avatar zlatanprst says:

    Is Gorton repeating what George Soros said in his book "The Crash of 2008?" in his theory of reflexivity? Why does Soros discuss only the banking crisis since 1980 to support his view that financial markets are best interpreted as an historical process? Perhaps Soros also misunderstands financial history.

  4. avatar JP Koning says:

    Great article.

    "Take those discounts on antebellum U.S. banknotes. Gorton attributes them to the fact that different banks, whether "free" or chartered, had different assets backing their notes, with some assets being more "suspect" than others (pp. 15-16). According to his understanding, nothing short of a rule forcing all banks to back their notes with identical, riskless assets could serve to make a uniform, par currency out of notes issues by numerous, otherwise independent banks."

    I've read that the discounts on US banknotes during the Free banking period were due to the incredible amount of counterfeits being passed in the US, not differing credit quality. (Smith, 1942) If so, this problem wouldn't be solved by homogenizing banks' assets but by adopting new technologies and methods to police against forgeries. It seems to me that unit banking would only add to the problem of counterfeiting since the multiplicity of notes created would make identification harder and the passing of fakes easier. It would be interesting to see if Canada had the same problems with counterfeiting.

    • avatar George Selgin says:

      "I've read that the discounts on US banknotes during the Free banking period were due to the incredible amount of counterfeits being passed in the US, not differing credit quality." I don't think that's correct, JP. Counterfeit notes were a big problem, to be sure, and I agree that unit banking contributed to it. (In branch-banking systems with plural note issue, including Canada's, counterfeiting was in fact less of a problem than it is in many modern central-bank based systems, including ours.) But the discounts in question are ones that were given by note "brokers," who offered them only for notes that they determined to be genuine.

  5. avatar edeast says:

    I think after 1873 or maybe the previous crisis in Canada, bank note holders were given priority over bank shareholders in event of bankruptcy.

    • avatar George Selgin says:

      There can't have been any such change, edeast, for the simple reason that banknote holders are creditors, and bank creditors necessarily have priority over bank shareholders, who are residual claimants. I believe that the statute to which you refer is one that gave Canadian banknote holders a first lien on any insolvent bank's assets, in effect giving them priority over the bank's other creditors.

      • avatar edeast says:

        I think you are correct, I just can't find the source I was going from memory, of an old book on history of Canada's banking. There was a debated change in the banking act, that put bank note holders first in line. Ok, this isn't the source I was going off of, but after the 1880 recession, they did change the act to let noteholders have first lein. http://faculty.marianopolis.edu/c.belanger/quebechistory/encyclopedia/BankinginCanada-CanadianBanks-CanadianHistory.htm

        • avatar edeast says:

          Ok, non specialist, but I did like Gorton's paper on the information content of money/near monies when the crisis came out. So I'd like to get this all sorted out. From the article I linked to just some interesting things, it says that banks were called into being in Canada, due to the desire for money, after the war bills went out of print 1816:1820. Also the charters closely mirrored Alexander Hamilton's. I might try to find the relevant debates, 1867:1870 on the bond backed currency by Sir John Rose. Apparently the legislature relied on the newspapers to transcribe the parliamentary session and the official Hansard didn't appear for a couple years later. Also the author highlights good data records were kept from the Banks reporting to parliament semiannually, I'm going to try to get my buddy to get me a copy of CC Arthur's "Statistical Contributions to Canadian Economic History", which is in a local long term storage library. But does anyone know of where to get the data on Canadian banks statements from the 1830s or 1870's on?

          I suppose my only relevant question, would be during the harvest season, the bank notes in circulation in Canada, increased massively, how did they maintain the one to one value of the bank notes between banks? Couldn't some banks/branches massively overloan/print?

          • avatar edeast says:

            Sorry for assaulting you with ignorance. (another line out of the Walker essay, which is awesome.) the paper you linked by Bordo, Redish, and Rockoff is excellent. I will no longer be pursuing the financial data of canadian banks. It would be interesting to see what happened in 1907, through to the installation of the central bank. That's all from me though, very good essay, and I think it is important.

  6. avatar edeast says:

    I've got a better understanding of how it works. Due to the double liability of shareholders, and the first lien status of note holders banks really didn't want to overissue notes. However issuing notes was profitable. Also because banks preferred to have their own notes in circulation they would send back other banks' notes daily for redemption. The author argues Canadian currency was elastic to follow trade cycles, but because of the daily redemption, there was no inflation.

    They were also subject to fines of 1000 or 100000 dollars for overprinting.

    In an paper read before the world Congress of Bankers and Financiers at Chicago, B.E. Walker. Used in the introduction to Banks and Banking by J.J. Maclaren. 1896

    • avatar George Selgin says:

      You've done good work, edeast--and I'm sorry I hadn't time to reply until now. Concerning the role of regular interbank note redemption in regulating the currency stock of competitive note issue systems, see also the paper by Larry White and I referred to in the first footnote above, and (for the theory) my Theory of Free Banking. I agree that the Bordo, Redish, and Rockoff article is very good; presumably the published version will be even better, Hugh Rockoff will be among the participants in a November 1 conference Larry White and I are organizing called "Instead of the Fed: Past and Present Alternatives to the Federal Reserve System."

  7. avatar edeast says:

    I will maybe post the essay tomorrow, I can't find it online, for he also contrasts with the bond backed currency of the National Banking System. Where banks would only supply currency when it was profitable, so overshoot, undershoot, depending on bonds interest rates. However B. E. Walker, advised the U.S. on the federal reserve legislation; It would be interesting to see his statements to congress. https://en.wikipedia.org/wiki/Byron_Edmund_Walker#Banking_career
    So I wish the conference the best, but I think convincing gov's or electorates to give up their money printing is going to be impossible? But I will try in Canada, I am more intrigued by the possibility of building a banking institution on top of bitcoin, but I digress.

    The one interesting thing is the localization of American banks, which spawn the creative financial markets, vs the pan canadian branches which efficiently move value from eastern depositors to western borrowers. The panic retractions described by savings and loans, and the cascading failure as banks withdraw support, this is interesting from a distributed system perspective( on which I have nothing to say). Local banking is information, I have a friend who operates in a branch bank, in Canada, she is required to be in the community at events etc, representing the bank and I believe understanding the local conditions. So maybe Gorton has a point about the lack of information within products which are exchanged on markets, and their being some type of internal information benefit to large banking organizations, perhaps the monolithic bank is able to more finely tune the failures, this branch manager, this branch etc, vs the loosely coupled unit banks, which must fail as a unit but this is speculation.

    Your theory of freebanking does cover 1907, canada so that was helpful. I skimmed over the Ruritania story and it seems coherent. And the final paragraph sort of puts a stake in my bitcoin bank, enlarging my banks' interest earning assets in an extrajudicial way would probably require criminal enforcement. Unless there is a similarly clever way to enforce contracts, not being allowed into the block chain, or something. Maybe now I'll read the story again, with the commodity money being bitcoins. damn i had only read the intro. I think the Walker essay is a good tl:dr.

    So what now.. is the future, internet giants, and big data, providing the branch banking work of information gathering. Maybe the u.s. won't be reformed, but products will be tagged, with increasing information. So mortgage products, will be linked to data profiles. I guess this is conceding the point to Gorton that it is information that is the problem.

    This may be better than NGDP, but I need to get a handle on credit creation, and information content.

    • avatar edeast says:

      argh looks like I'm a little late on the bitcoin bank. today this was announced. https://news.ycombinator.com/item?id=6041464 bitcoinlendingclub.com

      For an institutional challenge in Canada, it looks like federal credit unions, are now covered under the Bank Act, so I may try and create a credit union, similar to how Dwolla.com funcitons out of a credit union in iowa. However to make my canadian dwolla substitute function as bank notes, would require the ability for note holders to pass them to each other without going through my institution as an intermediary. Uncounterfeitable digital notes, that can be passed around. Sounds dificult.

      Also I was too negative on the prospect of change, Sumner had an enormous effect on central bankers, and it is possible that a true good banking system can be implemented, if it is convincing.

      I have now read the previous blog posts. But I will need to put some effort into credit creation. predicting the future etc, similar to expectations and ngdp, but its at the local level.

    • avatar edeast says:

      Found the B.E. Walker essay. Readers will need to scroll down to the introduction, ( First batch of text) the essay forms the preface to the book.

      https://archive.org/stream/banksbankingbank00macluoft/banksbankingbank00macluoft_djvu.txt

  8. avatar Phil Rothman says:

    Thanks for this interesting piece.

  9. avatar McKinney says:

    I really learned a lot from Gorton’s writing about “shadow” banking and the crisis, but he seems to have missed a lot in his history of banking.
    Have you read the Bank of England paper on reform of the international monetary system? It’s at http://www.bankofengland.co.uk/publications/Documents/fsr/fs_paper13.pdf
    Here’s are interesting quotes: “…the incidence rate of banking and currency crises in the Gold Standard was much lower than in today’s system…The direction of net capital flows during the Gold Standard seemed broadly consistent with the efficient allocation of capital across countries.”

  10. avatar edeast says:

    Canadian banks in the early 1900s weren't allowed to issue mortgages, so my argument of information content of branch banking may not apply.

    • avatar George Selgin says:

      Yes. But bear in mind that U.S. national banks were subject to the same restriction until after the passage of the Federal Reserve Act.

Leave a Reply

You must be logged in to post a comment. Login now.