Recent Posts

Sunday, February 18, 2018

Why Lender-Of-Last-Resort Operations Are Inevitable

Lender-of-last-resort operations by central banks (or "bailouts of the financial system") are deservedly unpopular across the political spectrum. Malcontents have argued that risky lending ought to be handled by markets, and that deposits should be fully backed by reserves or Treasury bills. Unfortunately, the believers in these theories never bother to look at the economics of short-term lending. The money markets have the structure that they have for a reason, and they only will function if there is a lender-of-last-resort that is able to step in and prop the system up. Any attempts to make the system bailout proof would have far-reaching consequences into the structure of the economy.

The logic of my argument can be summarised as follows.
  • Money market participants lack the analytical capacity to fully analyse the credit risk of short-term debt.
  • Therefore, commercial paper ends up having to be backed by credit lines issued by banks. In turn, those credit lines are backed by discount window borrowing by the central bank.
  • Attempting to sever those credit lines will either force the economy into industrial groups built around a financial firm that acts the group's financing arm, or will just go back in time and rebuild the merchant bank rediscounting system. These alternatives have drawbacks.
The reality that the implications of the reforms are not in the interest of any powerful groups makes their implementation extremely unlikely. Hence, the potential need for lender-of-last-resort operations will not plausibly go away.

Note that I am using "money market fund" in this article to refer to any fund that invests in money market instruments. This is distinct from a technical definition of money market funds, which operate under special regulations. Although this distinction may matter for retail investors, for institutional investors, what matters is the underlying instruments the fund invests in, not the ornamental wrapping around the fund.

Lack of Analytical Capacity in Money Markets

Many of those who call for the reform of the financial system put their faith in "market-based" approaches to lending. The belief that this will function is largely based on faith on free markets, without consideration of how the money markets actually function.

It is no accident that the serious financial crises in the post-1960s world generally started in money markets. (Hyman Minsky's works detailed the rise of the money market financing, and the associated crises.) As the business cycle rolls on, unwanted credit risk accumulates in the hands of those who are least able to analyse that risk. Many money market funds unfortunately fall into that role.

One important initial problem is that public money market investors are operating at a huge information disadvantage versus banks, and even specialist lenders like receivable factoring firms. They have to make lending decisions on low frequency public company data. This is enough for analysing long-term solvency risks, but not liquidity risk. This means that this works for corporate bonds, but not money market instruments, which are almost only exposed to borrower liquidity risk. The available information on a corporation's liquidity profile can be affected by end-of-period window dressing. Conversely, banks have a privileged view on at least some of their client's real-time liquidity situation. (A factoring firm will have a good view of transactions across the industry.)

The economics for money market fund analytic capability is terrible. There are two drivers. Firstly, fees are too low to support a large research staff. Secondly, the payoff for research is minimal. I discuss these in turn.

Imagine that we set up a money market fund that has attracted $1 billion in assets. That is smaller than some of the giant money funds, but is larger than the money market funds in smaller investment firm. Let is then make a very aggressive assumption that we could charge 10 basis points a year to hire analysis and trading personnel. (I apologise to any investment managers that sprayed coffee out of their nose when they read that.)

Ten basis points on $1 billion gives us $1 million to pay compensation. In Montreal, you used to be able to hire a junior credit analyst for $100,000/year salary, plus 50% to gives us $150,000 total compensation. (I have no idea what the current state of the market is.) A $1 million budget gives us a maximum team of six analysts and an intern (stagiaire). However, the numbers would be less than that, as we would need to hire portfolio managers/traders to execute trades, and they will not work as cheaply. So we are looking at a small team of analysts, at most.

Of course, the above numbers were wildly optimistic. A money market fund is lucky to get 10 basis points in fees in total, and so we also need to cover running expenses, risk management, back office, and marketing. (Plus, the firm may want to make a profit.)

The implication is that unless the money market fund is a behemoth, there are not going to be hundreds of credit analysts running around scouring for opportunities in the money markets.

The situation is even worse when you consider the risk profile of the money markets. A credit-investing firm normally tracks a pool of issuers. However, in order for the money market to function like an idealised market, investors have to be able to bid on a wider range of securities outside that pool.

Imagine that a investment broker calls you up and offers you $100 million of a security that is trading 20 basis points cheap to fundamentals. You have yet to analyse the company, but the initial pitch sounds interesting.

If the security is a 10-year bond, the 20 basis point cheapness translates (roughly) into $2 million excess return over the life of the bond. If the spread closes to fair value quickly, that profit will be front-loaded.That is enough excess return to justify digging further.

Imagine instead that it is a one-week maturity. If we round off the period to 1/50 of a year, the excess return over the life of the security is $4000. Do you really want to put $100 million of capital at risk to make $4,000? There is almost no incentive to dig into the issuer. (If you are already comfortable with the credit risk of the issuer, you might as well pick up the pennies off the road.) Money market credit analysis is generally going to piggy-back on the firm's corporate bond credit risk analysis. However, the low upside of these securities implies that the analysis will always have to be skewed towards the analysis of long-term bonds.

The reason why the money markets can function is that there is a reliance on the credit lines backing commercial paper programmes. If the issuer has a hard time issuing commercial paper to roll over the facility, it is expected to draw down the credit line (and probably enter into restructuring talks with the credit line issuer). The role of the money markets is to provide funding; the serious credit analysis is done when the commercial paper programme is being set up. (The corporate bond market has the capacity to take on more of the credit analysis tasks.)

All Credit Lines Lead to the Central Bank

In order for a credit line to be credible, it needs some form of backstop. Banks are in the liquidity management business, and at worst, they can use the discount window. The historical experience of the Bank of England resulted in an evolution towards this model.

Non-banks struggle to offer credit lines. Unless they themselves engage in the credit line equivalent of fractional reserve lending, they would need to hold an offsetting position in Treasury bills to match the amount of the credit line. This structure is economically equivalent to lending the entire amount of the credit line to the borrower. However, that is a large non-diversified exposure, and there are not a lot of investors that are interested in doing that.

In other words, if the authorities want to eliminate the special privileges of banks, they are effectively axing the credit line business. In turn, that wipes out the existing model for the money markets.

Economic forces would result in one of three outcomes, or a hybrid mixture of the three.

  1. Short-term lending is cut off for most firms.
  2. There will be oligopoly controlling short-term lending, and firms will fall into the orbit of finance providers.
  3. Specialist lenders will rebuild a rediscounting system.

Short-Term Lending Cut Off

It is much easier to build a boutique firm that specialises in buying long-term corporate debt, particularly low quality. There is enough potential for excess returns to pay for analyst salaries.

If we reform banks so that they have no inherent advantages in short-term lending and cannot provide credit lines, we could just cut smaller firms off from short-term credit. They issue either long-term debt or equity. An economy without short-term debt is going to be relatively safe, we just need to look at the lack of crises in the aftermath of World War II. But as Hyman Minsky argued, once the trauma of the Great Depression wore off, borrowers and lenders became increasingly aggressive. 

There is a large desire in the private sector to hold cash instruments, and there is a large desire by the business sector to borrow short term. One of the economic roles of the financial system is to intermediate those desires. Attempting to abolish short-term financing by firms is most likely going to end up with financial innovations to work around the regulations. Meanwhile, it would be very easy for a "no bailout" policy to go horribly wrong in the near-term, as firms would need to restructure their balance sheets to match the new reality. The end result is that such reforms would be unpopular across the business community, dangerous, and face immediate pressure of being nullified by financial innovation.

Keiretsu System

The only way a money market fund could survive in world without credit lines is to be very large. It could then have the resources to employ a large staff of analysts. By definition, we cannot have a lot of very large funds; we would end up with an oligopoly.

Imagine you are a borrower that in some way competes with the owner of the money market fund. Do you think you would get funding? Of course not.

Economic forces would force firms into the orbit of one of the large financial funds. There would be pressure for the businesses in these groups to align their interests. The end result would resemble the Japanese keiretsu system, with an industrial group built around a financial firm, although it would likely be more informal than the Japanese structure.

Such a model can be successful; it worked for Japan, and the South Korean chaebols. If these industrial groups are lean and hungry, they can be an extremely dynamic force. The risk is that devolve into protected, senile, giants -- like the American automotive industry of the 1970s.

Erecting huge barriers to entry into finance or industry thus poses a risk to the long-term dynamism of the economy.

It may be that the previous discussion was alarmist; perhaps concentration rules might force some diversity of funds. For example, if a fund is limited to holding 10% of an issuers' paper, it would imply that there would have to be at least 10 funds holding the paper. The problem is that it would be very difficult for a small or medium firm to attract credit coverage from 10 or more large funds. Realistically speaking, funds aimed at institutional investors would arise that could take more concentrated lending positions. Since the funds' clients are institutions, they would be able to bypass concentration limits that would apply to public funds.


The alternative approach would be for merchant banks to lend to industrial firms, and they themselves borrow in the money markets. For example, they could buy a receivable at a discount, and then borrow against that receivable by discounting it again (rediscounting) with an end investor. This is the old merchant bank money market model. Such specialist lenders still exist, such as receivables factoring.

End investors have a greatly reduced analysis burden. They only need to track the fortunes of a handful of merchant banks, plus some extremely large non-financial issuers, including governments.

This will work well, until it doesn't. The money markets using this structure were prone to runs when the credit of any of the merchant banks was called into question. This is why the credit line model evolved in the first place.

Concluding Remarks

Attempting to eliminate the possibility of bailouts by clamping down on banks is not an easy task. Either market-based intermediaries will step in -- and require eventual bailouts -- or else it would require a restructuring of the economic structure. Short-term finance would be expensive, and might require moving into the orbit of a finance firm.

(c) Brian Romanchuk 2018


  1. There is a lot here to comment on. First, lender of last resort for liquidity purposes is not really the same thing as bailing out banks that made bad loans. And I don't think it is as politically unpopular. They are different. Maybe I am wrong to be, but I was way more upset with the Fed purchasing mortgage backed securities than I was with them loaning money to banks or buying US debt from them.

    Second, there obviously was a financial system in the US before a lender of last resort was set up (1913 creation of Federal Reserve I believe). That system just had a lot of problems that eventually led to the creation of the Fed. And even after that, until 1933 or so with the start of FDIC, bank failures cost depositors their money. So really for most purposes, the financial system before that time was actually 'bailout proof'. But it did operate in some way.

    1. 1) Whether the lender-of-last-resort buys or lends against risky credits does not make too much of a difference, and that is basically all the Fed did.

      2) Sure, you can have a private lender-of-last-resort (Bank of England in the olde days, J.P. Morgan in 1907.) it worked so well that schoolkids used to memorise the dates of banking panics. Having frequent depressions that hammered the working class was politically sustainable when only rich people voted, but is not going to happen with universal suffrage. Politicians will intervene rather than let the private sector finish the process of liquidating itself.

    2. This 20 page paper discusses The State and National Banking Eras (A Chapter in the History of Central Banking):

      There is a difference between Fed lending against collateral (discount window borrowing by banks under normal operations or special liquidity programs) and Fed purchasing financial assets (Treasuries and/or other agency-backed securities including MBS). When Fed lends to banks there is an increase in bank reserves only. When Fed purchases a security from a bank there is an increase in bank reserves only. When Fed purchases a security from a nonbank, because banks clear the transaction, there is an increase in reserves and deposits in the aggregate bank balance sheet. Early in the financial crisis Fed provided liquidity mostly via discount window lending and attempted to control interest rates by selling Treasuries from its asset portfolio. The problem with this strategy is Fed might run out of Treasuries and lose control of interest rates. By late 2008 Fed had authority to pay interest on excess reserves. When Fed purchases MBS from nonbanks it provides deposits to nonbanks, Congress increased FDIC insurance on such deposits during the crisis, nonbanks and banks get to keep more Treasuries in their respective liquidity cushions when Fed buys fewer Treasuries, banks get excess reserves in their liquidity cushions. So if Fed and Treasury were cooperating to provide liquidity to both banks and nonbanks simple discount window lending might not accomplish the goal. In regard to a write-off and bailouts the government simply needs to decrease its net worth, help banks remain liquid while taking the write-off of bad loans, and help banks recover sufficient profits in the long run to absorb the write-off and continue to hold an equity (loss) cushion. In terms of moral hazard unfortunately the bankers that take down large pay packages and bonuses and cash in stock at inflated prices have successfully "gamed the system" when a bailout occurs unless the government makes it possible to claw back such gains from the bankers after the fact. The incentive to increase risk and shift it to others is an agency problem that could be addressed by regulations but cannot be totally eliminated. If the government insures all bank liabilities except for a mandatory equity cushion and Fed provides all overdraft services required in the economy then bank assets should be strictly regulated and the government should make it clear that it will claw back ill-gotten gains from executives of failed financial institutions in addition to wiping out the equity there. To my knowledge Warren Mosler is the primary proponent of strict regulations and government backing of liquidity of the whole financial system.

    3. 1) Whether the lender-of-last-resort buys or lends against risky credits does not make too much of a difference, and that is basically all the Fed did.

      Why is that? If it lends against risky assets and those assets fail then the bank that holds that failed collateral will fail. If the Fed buys the risky assets then there are two problems- one is it might be impossible to establish a fair market value for the asset if the private marketplace has frozen up, and the Fed is at risk of overpaying for the assets. Two, if the assets do end up failing, then whoever sold them to the Fed might already be off the hook. Maybe I am just vindictive, but I would prefer that the original holder still paid a price for making their bad bet.

    4. If you lend $80 against a bond, you have the economic exposure that is equivalent to buying the bond at $80, with an added guarantee from your counterparty. In a credit crisis, that extra guarantee and $1.95 will get you a cup of coffee. That is, whatever you are willing to lend is the de facto price.

      The whole point of lender-of-last-resort operations is to say that market prices are wrong. During the crisis, anyone could have told you that “market” prices were stupid, but people with cash were not going to pay fair value. They knew that they had the sellers over a barrel, and they exploited them mercilessly. The only way to break that psychology is to have someone big step in, pay something close to a reasonable value, and force people into a buying panic. If the fire sale is ending, you gotta get a piece of the action before things get back to normal.

      Finally, the objective is not vengeance, it is to avoid major failures. Lehman was a second tier investment bank, and its failure wiped out practically every traded market, and hit things like trade finance. The possible loss of a big universal bank is going to cause total devastation in the real economy. You prop up asset prices, and the problem goes away. You needed to regulate better in the first place.

    5. Addendum: playing it safe and lending $20 against the $100 bond does not help. By hypothecating the bond, it is no longer an asset that is useful for subordinated creditors. From their perspective, a loan of $20 against an asset that is being carried at $100 on the balance sheet is equivalent to the firm losing $80. A highly-levered investment dealer cannot do that for very long without breaching loan covenants for subordinated debt.

    6. Then why not purchase the securities with a repurchase agreement from the seller? The situation of being illiquid at that time is solved and the Central Bank is less likely to be accused of buying junk off the banks in order to bail them out. The Central Bank might still get stuck with junk if when at the end of the repo, the security has failed and the repurchase is unable to be made. But the immediate liquidity crisis if that was the problem would be solved, and the now bankrupt issuer does not have to be resolved in the midst of a system-wide crisis.

      Yes, a lender of last resort is not about vengeance, but it is about imposing a certain justice over the market system to avoid unnecessary failures when through no real fault of any one firm, good assets would have to be sold at an unfair discount because of a panic situation. It really should not be designed to save a firm like Lehman which had more than a liquidity problem. And if you just prop up asset prices the problem doesn't go away- it gets worse.

    7. A repurchase agreement (repo) is economically equivalent to lending against collateral; there might be some legal and accounting distinctions that I never needed to care about. The price on the repo is therefore running into the problem described above. Lend at too high a price, you effectively bought it at that price, and if you set the price too low (e.g. $20) the counterparty is not in a position to enter into the agreement.

      There were only a few assets that went to zero. Most of the problems were securities that no one was sure about their valuation, and they had an extremely ugly tail risk. The estimated tail depended on the level of spreads. Someone had to stabilise spreads, and that someone was the Fed. That is exactly what lender-of-last-resort operations are, i have not looking into the data, but I assume that the Fed made a fortune on its risky asset purchases. If you end up making money, you paid too little, not too much.

    8. This speech said that the Fed made $30 billion - ka ching!

  2. “Any attempts to make the system bailout proof would have far-reaching consequences into the structure of the economy.” So any proposed change that has “far reaching consequences” should be rejected out of hand? The steam engine and internal combustion engine had “far reaching consequences”. Clearly all car and truck engines should be scrapped, and we ought to go back to horse-drawn carts.

    Next under “The logic of my argument…” you claim that because the risks involved in various types of short term debt are difficult to estimate, ergo the central bank (i.e. taxpayer) should act as ultimate guarantor. What about loans to your local Mafia boss – let’s say for an entirely legitimate purpose? I imagine they’re pretty risky. Should the taxpayer have to back those as well?

    An alternative, and I think more logical argument is that if a debt is particularly risky it should quite possibly not be incurred at all, or it is incurred, there is no good reason for taxpayers to have to carry any of the risk.

    That alternative is what full reserve achieves: i.e. loans are funded via equity, which probably raises interest rates. But the deflationary effect of that is countered by simply having the state print and spend more money into the economy. Net result: everyone has more money, thus there is no need for so many households or businesses to go into debt.

    Plus I suspect that employers in the early twentieth century earlier would have laughed at the way many corporations are run nowadays: i.e. corporations in 2018 often don’t have enough cash to pay next week’s wages, thus they have to go running around money markets with the commercial paper begging for money.

    1. 1) If a reform has obvious negative consequences, and accomplishes nothing, it’s a reform you do not want to do.

      2) Stop funding risky loans. So you want to wave a magic wand and eliminate risk-seeking behaviour in capitalism? How quaint.

      3) You can label units in a fund that holds short-dated debt as “equity,” but people who work in finance are numerate, and classify assets based on their economic role. The units woukd be classified as “cash” in any portfolio analysis tool, and treated as such.

  3. Thanks Brian. I'm more optimistic about private sector solution. But I have zero argument to back up that :)

    I think there is a conflict in the root of this issue. As you say: "There is a large desire in the private sector to hold cash instruments, and there is a large desire by the business sector to borrow short term.". That IMO summons instability as maturity and risk transformation is needed because the business backing the cash desire are inherently unstable. Business wants short term lending but it doesn't have matching revenues. Someone needs to bear the risk. I wonder if the system based on mostly equity would be better. Usually the mix of less informed investor and high level of leverage is the problem.

    Which is more problematic: excess of long nominal promises (loans i.e. leverage) or excess dependency of short term loans? Probably both? So would you argue that floating rate notes would create more stability?

    1. Well, since the failure of repayment is the problem, the borrowing is technically more dangerous. But the desire to lend short is the enabler. I don’t know whether you can quantify supply/demand, but the system has no problem funding people who want to borrow short, so long as they look credible from a credit perspective. Very limited capacity for riskier short-term lending, which is what non-financial firms would like.

      More equity would help, but nobody wants to issue equity to finance working capital. Furthermore, investors are already loaded up on equity in their portfolios; they need cash.

      Floating rate notes do not really help anyone. They eliminate interest rate risk, but they are still not cash; you have to wait for maturity to get your principal back. If anything, there is a shortage of duration in the current environment.

    2. Thanks Brian.

      Yes, the repayment is the problem but a system as a whole seem to collapse due to over leveraging even if the triggering event is a repayment problem. Like enough leveraging is need to sustain the negative chain reaction, falling of the house of cards.

      Need of cash, lack of animal spirit is IMO the evil here. There needs to be someone bearing the inherit risk. Now it is ultimately the lender of last resort but that is a problematic solution. I guess you are saying there is no good/better alternative.

      For me it is a bit of a mystery to me why capital markets prefer nominal bonds over floating rate notes. It seems to me that the latter has the better risk characteristics for both parties.

    3. People need to save for retirement, and so there is a huge demand for taking duration risk.

      Yes, there is a large demand for safe money market assets (cash), but that is only a small percentage of most portfolios. (Businesses use money markets for cash management, but otherwise generally do not mess around with financial asset portfolios.) It’s just that financial asset portfolios are large.

  4. This is extremely interesting. I need to start reading your blog regularly again!

    One question: Is it clear that nonfinancial firms really need short-term lending? My impression is that most nonfinancial corporations have very little short-term debt, and of course the vast majority of commerical paper is issued by financial businesses. Obviously without access to short-term borrowing you'd have to hold a bit more cash, but it's far from obvious (to me anyway) that the amounts involved would be prohibitive, or even particularly costly.

    I totally buy your argument that the short-term lending market doesn't work without a central-bank backstop. But how do we know that this market is actually providing some socially useful service, as opposed to being just a self-contained activity of the financial system?

  5. This comment has been removed by a blog administrator.

  6. This comment has been removed by a blog administrator.


Note: Posts are manually moderated, with a varying delay. Some disappear.

The comment section here is largely dead. My Substack or Twitter are better places to have a conversation.

Given that this is largely a backup way to reach me, I am going to reject posts that annoy me. Please post lengthy essays elsewhere.