Tuesday, April 20, 2010

Financial Regulation

One of our commenters (Kosta) led me to this blog piece by Krugman. I'm feeling bad that I keep giving him a hard time, so this is my opportunity to show you that I don't harbor some irrational hatred for the guy. Krugman does a nice job of laying out the issues and summarizing the key influential viewpoints, so this is a useful platform on which to build a discussion.

I especially like Krugman's analogy to health reform, though I disagree that health reform is somehow easy while financial regulation is hard. What we have just been through was a long, painful, and costly process of passing health care legislation. I think we have an improvement, but we'll see what happens when the government tries to enforce participation. The analogy to health reform draws in my running Canadian example, which I discussed here among other places. Just as there are lessons for the US in Canadian health insurance, there are lessons in Canadian banking. As one of my former colleagues at the University of Western Ontario, Ig Horstmann, was fond of saying, "Canadians love insurance." They have government-provided health insurance, relatively generous unemployment insurance, and their banks (because they are forced to do it, or in part for other reasons) are essentially self-insured. As economists know of course, there can be problems with being too well-insured, and it seems clear that the US does not want to be Canada. But that doesn't mean you can't learn something from what Canadians do.

Now, the first point of discussion has to do with a typical characterization of US banking and monetary history. The story goes like this. In pre-Civil War times there were only state-chartered banks, and these banks could issue private currency. There was a plethora of banks, and a plethora of circulating bank notes, and it seems generally agreed (with some exceptions) that this system did not work well. Bank notes circulated with discounts, counterfeiting problems were severe, and the quality of bank notes could be hard to evaluate. After the Civil War, the state bank notes were taxed out of existence, and the National Banking System was created. Given the regulatory structure, the currency issued by National banks was essentially safe, so one problem was solved. However, during the National banking era (post Civil War to 1914) there were banking panics, the last of which was in 1907. By the early twentieth century, the view was that the financial system was unstable, and that having a central bank could cure the problem. Thus, we have the Federal Reserve Act of 1913, and the Fed goes into business in 1914. Things seem to work reasonably well until the Great Depression. Then, the conventional view seems to be that the Fed was somewhat inept, and did not appropriately perform its roles as lender of last resort and supplier of liquidity to the financial system. Further, there was too much risk-taking by banks (they were somehow too closely connected to the rest of the financial industry) which, combined with the helping hand of the Fed, led to the failure of 1/3 of the banking system in 1933. Then, according to the conventional view, this was all fixed in the Banking Act of 1933 (otherwise known as the Glass-Steagall Act) which introduced deposit insurance, Fed regulation of savings deposit rates (regulation Q) and the separation of banking from other financial intermediation activity, among other things. Then we have the Quiet Period of US banking (Krugman says "quiet period" comes from Gary Gorton - I was looking for the reference but couldn't find it - anyone know?) until the savings and loan crisis (late 1980s, early 1990s). In 1980, there was a key piece of legislation, the Depository Institutions Deregulation and Monetary Control Act, an important part of which deregulated savings and loan institutions and made them look more like banks. This was followed by the Gramm–Leach–Bliley Act of 1999, which further relaxed the Glass Steagall Restrictions that prevented bank holding companies from getting into non-bank financial business. Now, one could see how deregulation could accentuate moral hazard problems in banking and lead to a need for regulators to limit bank risk-taking in new ways. However, some people (including Gary Gorton apparently, but I've heard other people say this) argue that Glass Steagall limited competition among banks (and between banks and other financial institutions) in ways that made them take less risk, and that lifting these restrictions produced the obvious result.

Part of the conventional view of this historical experience is that there is something inherently wrong with banking - in converting illiquid assets into liquid liabilities banks leave themselves open to runs and panics. A Canadian perspective on this might be entirely different. A Canadian might look at US banking history and wonder why these Americans can never get it right. They get off on the wrong foot, and it just goes from bad to worse. Canadian banking history is essentially one long Quiet Period. Before the Bank of Canada came into being in 1935, chartered banks could issue currency. This private money system worked quite well (see this paper) - notes issued by different banks traded at par, banks redeemed each others' notes, and there was an apparently efficient system for clearing the notes. Further, without a central bank, the Bank of Montreal played a quasi-central banking role, and the banks appeared to be self-policing and capable of preventing incipient panics. Private note issue was ultimately phased out in 1944. There was no deposit insurance in Canada until 1967, and yet failures of banks in the 20th and 21st centuries has been insignificant - one failure in the 1920s, the failures of two small banks in the 1980s, and no failures during the Great Depression or the recent financial crisis.

One interpretation of what is going on in Canadian banking is consistent with Gary Gorton's ideas, and one commenter (Kosta) suggested something similar. We could say that Canadian banks have a lot to lose from bad behavior. They enjoy significant monopoly rents, and they risk losing them. If they take on too much risk and there is a large chartered bank failure, the Canadian government will step in and break them up like ATT. In equilibrium the government never has to impose the punishment, but the banks understand that the government is committed to this off-equilibrium behavior. The more I think about this story, the less I like it. One lesson from the US financial crisis is that some of the players who contribute to the riskiness of large financial institutions really have nothing to lose. The professionals who designed financial instruments and marketed them for Bear Sterns and Lehman Brothers, for example, have valuable skills, and some of them are now employed at the Fed, Fannie Mae, Freddie Mac, and in other financial firms. They were handsomely compensated and are doing just fine now, thank you. It seems to me that we can explain the behavior of Canadian banks adequately by looking at how they are regulated - it doesn't have much to do with the implicit threat of punishment.

Now, particularly since Krugman brings up the Diamond-Dybvig (DD) model, I should talk about that. This paper by Gary Gorton is also essentially a DD story. I have found DD useful in modeling (e.g. here and here). The model captures in a nice way the insurance role played by banks - essentially a story about how liquidity is shared through the banking system in an efficient way. But that is about as far as it goes - people just give DD too much credit (sorry Doug and Phil). In the original paper, it can be easy to write a bank contract that prevents a run, and when it's not DD don't solve the problem properly. Ultimately, the model has nothing to say about deposit insurance, moral hazard, or central banking - all the things we really care about. For the bottom line on the DD model, see work by Huberto Ennis (Richmond Fed) and Todd Keister (NY Fed).

I have no doubt that the possibility of runs and panics is a real problem in some banking systems, but DD is not instructive. One thing that seems clear is that deposit insurance is one fix put into the US banking system that appears to be working. Runs (other than the phenomenon Gorton discusses in the paper I linked to in the previous paragraph) were essentially absent in the US banking system during the financial crisis, and the the FDIC seemed capable of resolving small bank failures (and even larger ones like Washington Mutual) quickly.

If we get away from a Diamond-Dybvig view of banking, and think in general equilibrium terms, in monetary economies with central banking (like this one, we can come away with a very different view, both of history, and of the reasons for the financial crisis. Panics, of the type discussed by Gorton here could just be viewed as liquidity shortages, as could the National Banking era panics (see this paper). It was a problem for some financial institutions during the financial crisis which were funding long-maturity assets with short-term repurchase agreements that they could not borrow from the Fed (though of course the Fed was lending to pretty much anybody who had decent collateral and maybe even some who did not), just as it was a problem in the National banking era that the currency was not "elastic." Canadian banks in 1907 facing a line of depositors wanting to withdraw their deposits could print notes and simply swap the deposit liabilities for notes in circulation - they created the liquidity privately. In banking systems without private note issue we rely on the central bank to supply liquidity appropriately.

Here's a question. Once there is a central bank lending facility in place, why isn't deposit insurance redundant? Surely the central bank can stand ready to lend on good collateral, and if the collateral is no good and the bank is deemed to be insolvent, the central bank can resolve it in the same way the FDIC does. Why not?

That's all for today. Comments are welcome. I'm all ears.


  1. Stephen,
    I agree that a Central Bank can resolve it the same way FDIC does in case of bank insolvency. However, it is important to make the distinction between a country with its own floating currency (therefore, which has its own Central Bank) and a country that is on a peg or in some kind of monetary-only union (eg Greece!).

    ECB will lend to Greek banks in exchange of Greek bonds, but these are full recourse loans. If Greece default, Greek banks are on the hook. So the ECB has the power (and don't underestimate its willingness... Germany is watching) to bankrupt Greek banks. And guess who is on the hook if Greek banks go broke? The Greek government through deposit insurance. Of course, I'm not sure what these deposit insurance are worth if the Greek government default on its debt in the first place! If I would live in Greece, you bet I would take out my money from a Greek bank and stack it under my mattress (once Greek default, EUROs will still have value given that Greece represent a small portion of the Euro zone). This illustrates how crazy the Euro arrangement is: monetary policy is centralised but everything about fiscal policy (including deposit insurance) rest with member countries!!

    Now, you can quickly see that for a country with its own floating currency like the U.S., this type of scenario is absolutely impossible. The FED could never say to private banks that since the US government defaulted on its debt, it wants its US dollars back! The US dollars is not decoupled from U.S. bonds like the Greek bonds are from the Euro for one thing. And for starter, a country with its own floating currency can not default on its debt unless it makes a political decision to do so.

    Keep an eye on the Greek situation, things could get ugly real quick. Portugal may not be far behind. And then a domino effect. I don't worry the least about the UK, it was clever enough to keep its own floating currency, everything will be just fine there. If I would live in the UK, I would have full faith in deposit insurance just like I do in Canada or the U.S..


  2. One thing to note that unlike many other countries the Central Bank does not guarantee interbank payments in Canada. Every morning the clearing banks in LTVS set the level of credit exposure they are willing to take from each other just like the old system of exchanging each others notes at par. In addition the LTVS clearing banks are an elite bunch composed of the Big Six, Laurentian, Desjardins, ATB(guaranteed by the Alberta govt), HSBC Canada and the Credit Union Central. Thus anyone who wants to be a new clearing member has to convince the all of the existing members of its financial standing a cosy little club indeed. Thus most of the bank failures up to now such as the two in the 1980s where because a particular instition was cut off by its clearing bank minimizing any direct risk to the payment system. In many post 1935 the real job of the Bank of Canada has been to support government securities through open market operations and repos. In fact the govt still to this day can and does use chartered banks as direct fiscal agents ala its relationship with BMO pre 1935.

    On the subject of deposit insurance in Canada I would argue that CDIC has only served a purpose in protecting depositors of the smaller deposit taking trust cos which don't have the liquidity of the chartered banks and anyways are increasingy few and far between in terms of a real presence in Canada's financial system. In fact the the two small banks in 1980s which failed where supported by the Big Six for several weeks in order for basically all retail depositors to take their money ou leaving very little for the CDIC to deal with at the end.

  3. Tim Smyth:
    It is my understanding that interbank payment through the LVTS is fully collateralized. So the BoC does not need to guarantee payment. If a given Bank runs out of acceptable collateral, of course it will be in deep deep trouble and will not be able to borrow from its peers.

    See http://www.bankofcanada.ca/en/review/summer09/dsouza.pdf

  4. Anon:

    LVTS is only partially collateralized. This is an advantage of the system since it saves on collateral costs compared to say Fedwire or CHAPS. The trade-off is the Bank of Canada assumes the residual risk when it guarantees finality.

    Tim brings up a good point that DI helps unconnected banks (i.e. ones without a direct relationship to the CB). For example, in the recent crisis Lehman Brothers spent a huge amount of time right before its failure to borrow from the Fed in the hope this would stop the run on it by other market participants. True there was no DP for it but you can see that this is probably how a small bank without direct access to the CB would need DI to restore confidence.

    Also CB charters usually limit the type and riskiness of collateral they can accept in normal times. So if depositors think their bank is holding assets that are ineligible as CB collateral they may be tempted to run.

  5. Anonymous,
    Good point. I can see DI being usefull for banks without direct access to the CB.

    Regarding your comment on LVTS, are you positive it is only partially collateralized? See the following extract taken from http://www.cdnpay.ca/faqs/lvts.asp :

    How are the certainty of settlement and irrevocability of LVTS payments achieved?

    LVTS payments are backed by collateral pledged to the Bank of Canada by the participants. Before an LVTS payment is approved, the system checks it against risk controls to verify there is sufficient collateral available to support it. The level of credit available to participants is adjusted in real time to reflect incoming and outgoing payments. The risk controls ensure there is enough collateral available in the system to permit settlement of any participant’s obligations on the books of the Bank of Canada, even if a participant is declared insolvent


  6. Excellent! These points are all extremely useful. Generally, we need to pay more attention to payments arrangements in thinking about monetary policy and regulation. The more I hear about overnight lending in the fed funds market, for example, the more bizarre it seems - Canadian arrangements seem much saner.

  7. In response:

    I'm positive its only partially collateralized.

    See http://www.bankofcanada.ca/en/financial/lvts_neville.pdf last paragraph of page 23. The amount it's short is if more then one participant defaults (even then it probably still has enough collateral in most cases).

  8. But the risk associated with any LVTS member must be extremeley low. Then, whether this is collateralized or not doesn't matter much, except that you have to go to the trouble of posting the collateral in one case, and not in the other.

  9. I have a question as well. In normal times, typically reserves are zero in the LVTS overnight, unless someone gets stuck with a large payment in their reserve account that they don't have time to park somewhere else at the end of the trading day. Then, during the day, the LVTS members have fluctuating reserve balances. Is there interest on positive and/or negative intraday balances, or a limit on how negative a member's balance can go during the day?

  10. What is interesting and a little big scarier from the standpoint of the Bank of Canada is looking at its balance how little on the liabilties side is actually LTVS deposits. Before the financial crisis 95%+ of its liabilities where notes in circulation which as the demand for paper currency declines in the future it is going to start to be hard for the bank to control monetary policy. The real interesting current figure is the deposit the Government of Canada has placed at the Bank($13 Billion) which is almost a third of the current federal deficit and is essentially makes part of the current deficit a wash as the bank holds a much greater amount of govt bonds($50 Billion). Is part of the current deficit in Canada simply to create govt bonds to use collateral? Is this why Flaherty talks about eliminating the deficit so quickly when the economy recovers? Also I do have to mention there is $22 Billion in repos to the primary dealers outstanding by the Bank although this is normal part of Bank of Canada operations unlike the US Fed.

  11. Stephen,
    Right now targetted end-of-day LVTS balance in aggregate is 3 billion, but it is usually 25 million. It was increased during the crisis by the BoC to make sure that the overnight rate trade at the lower bound (0.25%). The Bank of Canada will of course not say that this 3 billion is quantitative easing, but -excuse my french- this is horse shit. Of course, the 3 billion increase in LVTS represents unsterilised amount and is therefore quantitative easing. Any substantial increase like this in end-of-day aggregate LVTS balance can either result from 1)unsterilised purchase of assets by the Bank of Canada or 2) deficit by the federal government for which no T-Bills were issued (so unsterilised deficit so to speak). The Bank of Canada has the capacity to do the latter option because it acts as the Government fiscal agent on a day to day basis. These are the only two ways I know where the BoC can increase LVTS end of day balance in any given day. The increase in LVTS end-of-day balance (and consequently in deposits from financial institutions staying overnight at the BoC at 0.25%) happen precisely on the week ending April 22, 2009 (check BoC's balance sheet ).

    The Bank of Canada is less than transparent on this unfortunately.


  12. Krugman has been referring to Gary Gorton's new book lately, so the reference might be from that. SLAPPED BY THE INVISIBLE HAND http://www.amazon.com/Slapped-Invisible-Hand-Management-Association/dp/0199734151/ref=ntt_at_ep_dpi_1

  13. Steve,

    There is no interest payments in LVTS balances during the day. The limit that an FI's account can go negative is equal (actually a function of but equal is close enough) to the sum of the credit limits that the FI's counter parties have granted it.

    As for posting collateral, you could make two arguments for it. Either it's their to reduce the risk that the Bank of Canada has to step in (this could be related to one of the acts that Bank of Canada operated under). Or two to reduce the chance of a strategic default. The latter of which is (essentially) argued by the work of (subsets of) Koeppl, Monnet, and Temzelidies.

  14. Stephen,

    Your characterization of the conventional view on the U.S. banking system is spot on. It is a shame that those who advocate this view get so much wrong. For example, you reference Canada as a counter-example. In fact, when you look at the so-called free banking era, the losses tended to be concentrated in rural areas. Most bank notes actually traded at -- or close to -- par (in New York a $1 note never traded for less than $0.99). Rockoff (1974) finds that the total loss from bank failures during this era was $1.9 million, which was roughly 0.01% of national income. When you consider that $1 million of these losses were concentrated in Michigan that substantially changes the picture as well.

    The conventional view is thus particularly troublesome given how much it tends to shape opinions on banking and regulation.

    In any event, I recently discovered the blog and I am enjoying it thus far.

  15. Thanks for your comments, and please keep reading.

  16. Thank you for the excellent post exploring whether Gary Gorton's hypothesis that the lack of competition may have promoted banking stability (and thank you for the reference). I like your counter-argument; if I understand it correctly, you're basically saying that the mechanism of guaranteeing profits for the industry (by having an uncompetitive sector) and then threatening those profits if the industry takes undue risk can not explain why individual players choose to refrain from excessive risk taking, given that those players will still be rewarded whatever the outcome. Rather, you argue that its regulation that explains why individual players don't take excessive risk in Canada.

    It's a good counter-argument, but I think there are elements of Gorton's story which are still compelling. A key issue is why aren't there individual players in the Canadian system pursuing risky behaviours? Your suggestion is that Canadian regulation is sufficient to curtail this behaviour. I'm skeptical of the effectiveness of any regulatory system to curtail risky behaviour on its own (just look at ECB's success with Greece or the rest of the PIIGS). Rather, I suggest that in the Canadian banking system, risky behaviour has been largely curtailed because Canadian banks have chosen to self-regulate, and given that the banks have the manpower to watch their own players, they have largely succeeded.

    As you've noted in your great review, Canadian banks have always tended towards self-regulation. Why have they done so? Has it been a unqiuely Canadian trait, like their tendency to over-insure? Or has there been some incentive?

    This is where Gorton's argument is very compelling. Canadian banks enjoy significant monopoly rents and risk losing them if they undertake risky behaviour or fail to prevent their own players from doing so (i.e., they fail to self-regulate). The self-regulation is key to Canadian banking stability.

    This does not mean that the system can not remain stable without prudent regulation from the gov't. Rather, the monopoly rents encourages Canadian banks to self-regulate within the bounds of those regulations. Removal of the rents will lead to a system like Europe. Prudent regulations largely exist (after all Germans designed them), but banking is competitive, and there is no incentive to not push the bounds (no monopoly rents to lose).

    Removal of the regulations, on the other hand, could very well lead to a situation similar to the US's, where the incentive to self-regulate disappears, and individual players are allowed to pursue risky behaviours which are ultimately destabilizing.

  17. You may have a point about self-regulation in Canada. I need to learn more about exactly how OSFI operates in Canada. If you think about some of the hurdles for regulators, the problem of bank regulation appears quite formidable. How do you keep up with financial innovation, overseas operations, etc. ?

  18. I agree that it's hard to imagine how effective bank regulation can be implemented, given how extensive financial activities can be, and how there's an obvious incentive to push the envelope. Encouraging self-regulation might be one way to promote stability.

    It's quite interesting that the Canadian banking culture has such a strong tendency to self-regulate. The Gary Gorton story might provide one piece of the puzzle why self-regulation is maintained, but I'm not convinced that's the whole story. In particular, it requires that the big banks in Canada are far-sighted enough to realize that certain kinds of financial innovation will lead them into trouble. It's hard to imagine that the bankers (or anyone for that matter) won't, at times, make short-sighted decisions to gain extra profit. But the Canadian banks, for the most part, have chosen not to make these decisions? Is it just self-regulation incentivzed by monopoly rents? Or is there something else going on?