Central banks are indeed banks. As financial intermediaries, their social role depends on what a central bank can do that some private financial intermediary cannot, or cannot do as well. Fundamentally, financial intermediation is asset transformation - how the intermediary structures itself so that its liabilities have attractive properties as compared to its assets. In our changing world, central banks have to adapt their liabilities to evolving technologies and macroeconomic developments. My first installment on this topic was about new types of central bank liabilities - the idea that a central bank's reach could expand beyond supplying old-fashioned currency, with a focus on liquid overnight liabilities that reach into the upper levels of financial interaction, beyond retail payments.
This installment will deal specifically with currency. Are paper money and coins going the way of the dinosuars? Should they? If we were to eliminate or significantly curb the circulation of Federal Reserve notes and coins in the United States, should the Fed issue new types of liabilities to replace them? What types of liabilities might work, and how should they be managed? You can find plenty of writing on this topic recently. Ken Rogoff and Larry Summers make a case that we should elminate large-denomination currency. Agarwal and Kimball, Kocherlakota, Goodfriend (new version), and Goodfriend (old version) discuss various schemes for either eliminating currency or making it more costly to hold. In contrast to Rogoff and Summers, who think eliminating $100 Federal Reserve notes, for example, will put a big dent in undesirable and illegal activities, the latter group of authors is interested in ways to effectively implement negative nominal interest rates.
What does standard monetary theory tell us about how much currency we should have? The basic logic comes from Friedman's "Optimum Quantity of Money" essay - what's come to be known as the "Friedman rule." If the nominal interest rate on assets other than cash is greater than zero, then people economize too much on cash balances. Cash balances are costless to create, and economic forces will create the optimal amount of real cash balances if the central bank acts to equate the rate of return on cash to the rate of return on other safe assets. There are different ways to do this. One approach is for the central bank do whatever it takes to make the nominal interest rate on safe assets zero forever. Another approach, if feasible, is for the central bank to pay interest on cash at the nominal interest rate on safe assets. There are other approaches, but we'll stop there.
How do we connect the Friedman rule with what people are saying about currency? Those people are saying we have too much currency, but the Friedman rule says that a zero nominal interest rate is just right, and otherwise (with nominal interest rates above zero) we have too little currency. Of course, there's a lot going on in the world that is important for the problem at hand, and is absent in standard monetary models. First, currency is certainly not costless to create. Real resources are used up in designing currency so as to thwart counterfeiters, to print and distribute it (think about guards and armored trucks), and to maintain it (taking in and shredding worn-out notes). Second, currency can be stolen. A virtue of currency is its anonymity - people will accept my Federal Reserve notes without worrying about my personal characteristics - but this also makes currency easier to steal than some other assets. Third, the use of currency implies social costs. Again, a virtue of currency is its anonymity, in that the existence of currency allows us to conduct transactions in private. In some ways, we might think this is consistent with democratic notions of individual freedom - we can conduct whatever transactions we want out of the reach of surveillance. But this anonymity also lowers the cost of carrying on domestic illegal activity, and international activity that might be harmful to domestic interests, which is what Rogoff and Summers are on to. The illegal domestic activity could be various aspects of the drug trade and organized crime, while harmful international activity might involve, for example, the international market in military weapons.
Currency is a remarkably resilient payments technology. As I pointed out in my last post, the ratio of U.S. currency to GDP is as high as it was in the 1950s, and has risen since the beginning of the financial crisis. Though about 80% of the stock of Federal Reserve notes consists of $100 bills, currency still shows up in surveys of legal payments as an important transactions medium. As we might expect, this Boston Fed publication indicates that currency is used much more intensively by poor people than by the rich. The advantages of currency are obvious. It's easy to carry around (except if you want to buy a house or a car with it, for example); it comes in small denominations, making it convenient for small transactions; and it provides immediate settlement without the use of electricity (unless the other party in the transaction needs to open the cash register).
So, any scheme to eliminate currency altogether faces significant potential costs. Unless an alternative low-tech payments medium can be provided, a lot of poor people (and possibly some rich ones as well) are going to be worse off. But Rogoff and Summers have a point - it's hard to see what the benefits of $100 bills are. In fact, there would not be much loss in convenience for legitimate exchange from eliminating $50 notes as well. The U.S. government, however, would lose part of the revenue from the inflation tax. How large is that? U.S. currency outstanding is about $4500 per U.S. resident, so a 2% inflation rate generates about $90 per person in seignorage revenue. If eliminating $100 and $50 bills reduced the stock of currency outstanding by about 70% (assuming that some of the existing demand migrates to other denominations), the loss would be about $63 per person, which perhaps is a small price to pay for putting a serious dent in criminal activity and terrorism. Thus, what Rogoff and Summers are proposing seems like a no-brainer.
My best guess is that no one would be talking about currency much if it weren't for the negative interest rate enthusiasts, which gets us into a very different set of issues. Here's their (Kocherlakota/Agerwal/Kimball/Goodfriend) argument:
1. Take as given that the most important role (at the extreme, the only role) for monetary policy is correcting the distortions arising from sticky wages and prices. That's a key assumption - none of these authors thinks it is necessary to defend that.
2. The real interest rate is low, for reasons unconnected to monetary policy, and is expected to remain low into the indefinite future. It's hard to quarrel with this, but the reasons for the low real interest rate could matter.
3. (1) coupled with (2) implies that, in pursuit of inflation-targeting and output-gap targeting, the central bank will in the future be encountering the zero lower bound (if indeed it is a bound) with greater frequency.
4. Once constrained by the zero lower bound, there are losses in economic welfare due to output gaps and departures from inflation targets.
5. Such welfare losses can be avoided, if only nominal interest rates can go negative - this relaxes the zero lower bound constraint.
6. There are no costs to negative nominal interest rates, only benefits.
The hardline NK (New Keynesian) view, shared by Agarwal/Kimball/Kocherlakota/Goodfriend (AKKG) is that things that get in the way of the downward descent of nominal interest rates are an "encumbrance" that we are better off without. In this respect, currency just gets in the way. As the argument goes, a negative nominal interest rate on reserve balances will tend to make all safe short-term rates of interest negative. But for banks, there's a problem. The bank's assets are close substitutes for assets earning negative rates of interest, but its deposit liabilities are substitutes for zero-interest currency. So banks are squeezed - if they lower deposit rates, they lose small depositors; if they don't their profit margin goes down.
What to do about that? AKKG argue that there should be programs put in place to make currency less desirable, thus giving bank depositors a poor alternative to flee to. In Goodfriend's 2000 paper he suggests a scheme for taxing currency - equip every Federal Reserve note with a strip. Whenever the note returns to a bank, the bank charges the accumulated tax since the note was last turned in, and deducts this from your deposit. Alternatively, Agarwal and Kimball suggest a scheme for a changeable exchange rate between currency and reserves. The current arrangement between financial institutions holding reserves and the Fed, is that reserve balances can be converted one-for-one into currency, and vice-versa (presumably with service fees added for armored trucks and such). Under the Agerwal/Kimball arrangement, during periods of negative interest rates the rate at which reserves can be converted to currency would decline over time to match the negative interest rate on reserves.
The Goodfriend scheme seems problematic as, if the vintages of Fed notes ("vintage" meaning the time since the tax was last paid) are known, then notes should trade at different prices. This of course messes up the whole currency system, according to which the denominations are set up so that it's easy to make change - seems hard to do that if one note marked with a 10 trades at 97% of the value of another note marked with a 10, for example. All of these schemes will not make things any better for banks, who are in part selling a convenience to small depositors - the right to exchange their deposits one-for-one with currency, which these people find useful for making transactions. And people and firms are quite ingenious when it comes to getting around regulations. If a negative interest regime is in place for a long time (something of course the advocates don't imagine, but no one in Japan imagined they would have interest rates close to zero for over 20 years either), this creates profit opportunities for various types of cash hoarding operations. Hoarding cash is an activity subject to increasing returns. A financial intermediary could, in principle, set itself up as a simple currency warehouse, supplying safekeeping services, and make a profit in such an environment. Of course, AKKG, or people like them, could think up many ingenious ways to thwart such arbitrage schemes.
But would all this be worth it? In the process we would put sand in the gears of financial institutions that are serving a useful social function, and for what? We're going to have better monetary policy, apparently. This part of the story gets a bit complicated. Academic Keynesian economics is a well-defined thing - it's clear what it is, and you can evaluate it and determine for yourself whether or not it's useful. Keynesian economics in practice can sometimes be a completely different beast, saturated with myth rather than science. One myth is that, in reducing nominal interest rates, we can have our cake and eat it too. Conventional wisdom seems to be that reducing nominal interest rates increases inflation and reduces unemployment. That's just Phillips curve logic, right? Wrong. Even in basic New Keynesian (NK) models with Phillips curves, lower nominal interest rates make inflation go down - that's basic neo-Fisherism. The Fisher effect is ubiquitous in models, and it's there in the data too. Here's an example. Data for Switzerland on the overnight nominal interest rate and inflation look like this: The Swiss went to negative interest rates at the beginning of 2015. Since then inflation has also been below zero. Of course, monetary policy isn't the only factor affecting the inflation rate. There is substantial variation in inflation in the chart, but you can see the trend. Negative nominal interest rates for 21 months in Switzerland are hardly making inflation explode.
As is well-known, a belief that low nominal interest rates cause high inflation inevitably leads to self-perpetuating low nominal interest rates and low inflation. That's part of the force behind the negative nominal interest rate lobby. Surely, they think, pushing interest rates into negative territory will cure the low-inflation problem. Well, sorry, adding another hole to your shoe doesn't fix your shoe.
But, the other motivation for negative nominal rates is stabilization policy. It may be true that a lower bound of zero gives the central bank less room to move when real interest rates are persistently low, but some people seem to find it difficult to reconcile themselves to the realities of the stabilization policy they claim to believe in. If we really think that moving a short-term nominal interest rate around has large real effects on aggregate economic activity, we have to recognize the need to make intertemporal tradeoffs. That is, we may think that lowering nominal interest rates confers some benefit, but we can't always be moving nominal interest rates down. Sometime they have to go up. Seemingly, a key principle of monetary stabilization is that it's less costly at the margin to increase nominal interest rates in good times than in bad times. Also, that it's of less benefit, at the margin, to lower nominal interest rates in good times than in bad times. Therefore, optimal monetary stabilization is about increasing interest rates in relatively good times, and reducing them in relatively bad times, while on average hitting an inflation target. Every time interest rates go up, central bankers are asking people to bear some short-term pain, in exchange for larger future short-term gains from lowering interest rates. Larry Summers, for example, seems to have a hard time recognizing this. According to him, we're in a long period of secular stagnation, which implies that we're in a relatively good state compared to the future. Further, Summers is very concerned with the fact that interest rates can't go down much if a recession happens. So, one might think he would be in favor of interest rate increases now, but he's not. Go figure.
Just as wrongheaded inflation control can lead to perpetually low nominal interest rates, so can wrongheaded stabilization policy. People who are single-minded about stabilization always see inefficiency when they look at the economy. Things are never quite right, so there's always a reason to lower interest rates. Given these tendencies among policymakers, perhaps a zero lower bound is a good thing. It's a constraint that prevents well-intentioned interventionists from defeating themselves, in terms of their own goals.
Where does this leave currency? Rogoff and Summers seem to have good arguments for getting rid of large-denomination currency. Getting rid of all currency, or making its use more costly seem like too high a cost, given the likely benefits of negative interest rate policy - about which proponents seem pretty confused. Central banks should, however, be thinking about electronic alternatives to currency. Blockchain technologies are potentially useful, and could be tied to the decentralized transfer of central bank liabilities. There are private alternatives to currency, for example Green Dot provides stored-value cards that are not tied to checking accounts. Would it be a good idea for central banks to issue stored value cards? In any case, we're not there yet with these alternative payments technologies so, for the time being, currency is a convenient low-tech payments instrument that we should keep.
What's happening in monetary policy and macroeconomics.
Thursday, September 8, 2016
Friday, September 2, 2016
Central Bank Liabilities: Part I
In this first installment, I'll discuss issues related to Greenwood, Hansen, and Stein's Jackson Hole paper, and leave discussion of the future of currency for installment #2.
An important model for modern central banking was the Bank of England, founded in 1694, which subsequently developed a symbiotic relationship with the British crown. The crown needed to finance spending, particularly on wars, and the Bank was looking to make a profit. The crown granted the Bank an ever-expanding monopoly on the issue of circulating currency, culminating in Peel's Bank Act of 1844, under which the Bank became the sole supplier of circulating currency in the UK (save for some grandfathered Scottish private banks, which are still issuing currency in 2016). Given the Bank's monopoly on currency issue, it could lend to the crown at a low interest rate, and still make a profit. The Bank, through experimentation or accident, discovered crisis intervention. During financial panics the Bank could, through judicious use of information available to it, engage in lending to banks with liquidity problems (banks experience a lot of bank note redemptions). In so doing, the Bank would expand its note issue to fund lending to banks that it deemed illiquid but solvent and, on the flip side of that, give those holding bank notes a safe asset to run to. The Bank did this, driven by its motive to make profits on superior information.
It wasn't like some economist was promoting the idea that a monopoly on currency issue and financial crisis intervention were the natural province of a central bank, or that this central bank should be have some form of public governance. Those ideas came much later. Indeed, the Bank of England remained a private institution until 1946. However, the Federal Reserve System was founded as a public institution, albeit with semi-private governance, in terms of how the regional Federal Reserve Banks are run. The work that led up to the Federal Reserve Act of 1913 included, for example, the National Monetary Commission, created by Congress to study the U.S. financial system, and systems elsewhere in the world, and to come up with recommendations for reform. The Commission produced volumes of useful stuff, which you can find on Fraser. The basic ideas in the Federal Reserve Act are that the role of a central bank in the United States was to: (i) through its monopoly power, provide a safe currency; (ii) make the currency "elastic," i.e. make the supply of currency respond to aggregate economic activity, and to financial panics, by way of discount window lending.
Basically, the framers of the Federal Reserve Act had in mind an institution that would play the same role as the Bank of England, but the motivation was different. A currency monopoly for the central bank was seen as socially beneficial, as experiments with private currency issue in the U.S. had not turned out well. And crisis intervention was seen as a means for preventing the disruption to the aggregate economy that had occurred during the banking panics of the National Banking era (1863-1913).
Fast forward to 2016. In some ways things haven't changed. U.S. Federal Reserve notes are basically the same stuff (with some anti-counterfeiting features added) they were in 1914, and currency is still an important Fed liability. Here's the stock of currency relative to U.S. GDP: The ratio of currency to GDP has risen to close to 8% from less than 6% before the financial crisis, and is at the same level as in the mid-1950s. Retail payments using currency have fallen, but perhaps not as much as one might think. For example, surveys by the Boston Fed show a declining use of currency, but currency is still important in consumer payments, accounting for 26.3% of payments in 2013, as opposed to 31.1% for debit cards. But, the quantity of Federal Reserve notes per U.S. resident is about $4,500 currently. I don't know about you, but I'm not holding my fair share of that. In this chart you can see that, by value, about 80% of currency outstanding is in $100 notes, and studies indicate that about half of the stock of U.S. currency is held outside U.S. borders. There are some legitimate questions about the role played by currency, and whether its use should be curtailed, but more about that in the next installment.
The key point is that, if we took large-denomination currency - which is likely serving no useful social function - out of the mix, only a small part of the existing Fed portfolio would be funded by what was once thought to be the primary central bank liability. As it is, even with all those $100 bills included, interest-bearing Fed liabilities are greater than the quantity of currency outstanding: The key Fed liabilities are, in order of magnitude (from small to large):
1. Reverse repurchase agreements, or ON-RRPs, used in "temporary open market operations." This is overnight lending to the Fed, with securities in the Fed's asset portfolio used as collateral. ON-RRPs are roughly reserves by another name - overnight interest-bearing liabilities of the Fed that can be held by a wider array of financial institutions than are permitted to have reserve accounts with the Fed. In particular, money market mutual funds cannot hold reserves, but they are key participants in the ON-RRP market. ON-RRPs play an important role in post-liftoff Fed implementation of monetary policy. The idea is that the ON-RRP rate, currently 0.25%, puts a floor under the fed funds rate, so that fed funds will trade between the interest rate on excess reserves (IOER), currently at 0.5%, and the ON-RRP rate.
2. Reverse repurchase agreeements held by foreign government-related institutions. This is somewhat mysterious, and explained here. Just as there is a foreign demand for Treasury securities, there is a foreign demand for very short-term liabilities of the Fed.
3. Currency. We know what this is about.
4. Reserve balances. As is well-known, this quantity has grown substantially, as this stuff financed the Fed's large-scale asset purchases post-financial crisis.
We're now in a world in which interest-bearing Fed liabilities have become very important. Is this a temporary change, or is it permanent? Should it be permanent? This gets us to Greenwood/Hansen/Stein (GHS). Basically, their answer to the last question is yes. GHS first argue that short-term safe assets are useful in financial markets and that, by looking at market interest rates, we can find evidence of "moneyness" for these assets. If assets are used in facilitating some kind of exchange, or they are in wide use as collateral, they bear a liquidity premium. That is, people are willing to hold these assets at lower rates of return than seem consistent with the actual payoffs on these assets. For example, Gomme, Ravikumar, and Rupert calculate real rates of return on capital in the U.S. since 2007 of from 5%-7% (after tax, for all capital). But here are some short term nominal rates of return:Before December 17, 2015 (liftoff date), the IOER was 0.25% and the ON-RRP rate was usually 0.05% (with some experimentation). After that date, IOER was set at 0.5% and ON-RRP at 0.25%. In the chart, you can see that, after liftoff, the fed funds rate has typical fallen in the range 0.35%-0.40%, except at month-end and quarter-end (for technical reasons). Also, four-week T-bills trade in the same ballpark as ON-RRP, with some variation.
What can we conclude? (i) Reserves are a less liquid asset than ON-RRP or short-term Treasury debt. Reserves can be traded among a smaller set of financial institutions than these other assets, and you have to pay banks more to take reserves than to take T-bills or ON-RRP. (ii) Short-term Treasury debt and ON-RRP seem to have roughly the same liquidity properties.
So, once the case has been made that short-term safe assets bear liquidity premia, reflecting their usefulness in financial markets, who is going to supply these assets? There are three options: (i) the private sector; (ii) the Treasury; (iii) the Fed. GHS argue that the private sector does not do a very good job of this. Why? First, there are some "externalities" involved, according to GHS. This is somewhat vague, but GHS seem to have in mind that private sector production of short-maturity assets involves intermediating across maturities, which is risky. And maturity transformation makes these private financial intermediaries sensitive to market stresses, according to them. Second, regulatory changes, for example the Supplementary Leverage Ratio requirement, gums up the private sector's ability to produce safe short-maturity assets. GHS also argue that the Treasury does not issue enough short term debt, because it is worried about the risk of auction failure if it has to roll over a lot of short-term debt.
The heart of the argument is that the Fed has advantages over both the private sector and the Treasury in issuing short-term debt. But what kind of short-term debt should the Fed issue? Currently, the choice is between reserves and ON-RRP, though in principle we can think about the possibility the Fed could issue Fed bills - short term circulating debt that looks exactly like Treasury bills. The Swiss National Bank can issue such securities, for example. But, if wer're constrained to considering only reserves and ON-RRP, it seems clear that ON-RRP is much more successful in serving the liquidity needs of financial markets than are reserves. After all, you have to give the institutions that are permitted to hold reserves a 25 basis point inducement to get them to do so.
So, it seems that, in general, ON-RRP is a better instrument for the purpose than reserves. As GHS point out, if the Fed were to expand its ON-RRP program and shrink reserves, this would amount to savings for taxpayers, given the level of short-term market interest rates. This is becasue the Fed could have the same effect on market interest rates, but be paying lower interest on its liabilties, thus handing over larger transfers to the Treasury. What expands the ON-RRP program? One approach would be to set the ON-RRP rate equal to IOER.
But, how large should the Fed's ON-RRP balances be, and what's the optimal size for the Fed's balance sheet? Prior to the financial crisis, the size of the Fed's balance sheet was essentially determined by the demand for currency (in real terms), given the level of market interest rates, as the quantity of reserves was very small, and almost all of the Fed's asset portfolio was financed with currency. If we take the GHS proposal at face value, there should be a similar natural limit for ON-RRP. If the Fed sets an ON-RRP rate, and conducts a fixed-rate full-allotment auction, with IOER equal to the ON-RRP rate, what could happen? (i) the Fed reaches the upper bound on securities that it can use as collateral in ON-RRPs - it gets more bids than it can satisfy. (ii) the Fed does not reach the upper bound on available collateral, and there are some reserve balances outstanding. In case (i) it seems the ON-RRP program is too small, and in case (ii) it's the right size, but the Fed's balance sheet is too large. The bottom line, if we accept GHS's hypothesis that a Fed supply of short-term interest-bearing liquidity is a good thing, is that the market can determine how much of this stuff it needs. I should make it clear that this is my conclusion, not theirs. But I think this is the logical implication of their analysis.
Comments:
1. How does this relate to quantitative easing (QE)? GHS seem to think this is a different issue - that QE is aimed at some short-run problem, and the role for ON-RRP is long-run. I don't think so. What GHS is putting forward is indeed a theory of how QE works. They are assuming that the Fed has a special role in maturity transformation, and when the Fed does this it matters. And this matters all the time, not just in unusual circumstances, so the implication is that QE should be an ongoing thing. But it seems there is a role for long-maturity debt in financial markets too - presumably the Fed shouldn't be sucking up all the long-maturity assets in existence, and GHS don't seem to be thinking about that.
2. I've got doubts about whether there are limitations on the private sector's ability to intermediate across maturities. Turning long Treasury debt into overnight debt is risky, but there are ways to manage such risk. It's not clear that the Fed and Treasury are any better at bearing such risks than are private financial institutions.
3. If the problem with private sector liquidity transformation is what can happen in crises, why shouldn't the Fed's intervention be confined to crisis times? For example, expand the ON-RRP program in a crisis, and contract it when the crisis is over.
4. A concern of GHS is a possible flight to safety during a panic. The argument is that, during times of financial stress, financial market participants could abandon private sector liquidity for ON-RRPs, and that could be bad. The "cure" for this is caps on the ON-RRP program. I've always found this idea puzzling. In the ON-RRP auction, the Fed can either set the quantity, or the price, but they can't set both. If there's a binding cap on the ON-RRP program, the price is too low, i.e. the ON-RRP rate is set too high. Presumably, in a crisis the correct policy is to lower the ON-RRP rate.
5. GHS's explanation for why the Treasury did not, or could not, issue more short-term debt did not make sense to me. This seemed more like a call for the Treasury to do a better job of auctioning their securities. Further, if a Treasury auction "fails" on a given day, I'm not sure why that's the end of the world. For example, the Treasury has a reserve account with the Fed, and the balance looks like this: You can see that the balance in this account fluctuates a lot - it's an important buffer for the Treasury in managing cash inflows and outflows. Further, the average size has increased substantially, to around the $300 billion range. If rollover risk is greater with more short debt outstanding, why can't the Treasury have a larger average balance in that account?
In general I found this paper very useful. It's the first coherent story I've seen about a legitimate role for a central bank in what we would typically call "debt management." But much more research and thought needs to go into these questions.
An important model for modern central banking was the Bank of England, founded in 1694, which subsequently developed a symbiotic relationship with the British crown. The crown needed to finance spending, particularly on wars, and the Bank was looking to make a profit. The crown granted the Bank an ever-expanding monopoly on the issue of circulating currency, culminating in Peel's Bank Act of 1844, under which the Bank became the sole supplier of circulating currency in the UK (save for some grandfathered Scottish private banks, which are still issuing currency in 2016). Given the Bank's monopoly on currency issue, it could lend to the crown at a low interest rate, and still make a profit. The Bank, through experimentation or accident, discovered crisis intervention. During financial panics the Bank could, through judicious use of information available to it, engage in lending to banks with liquidity problems (banks experience a lot of bank note redemptions). In so doing, the Bank would expand its note issue to fund lending to banks that it deemed illiquid but solvent and, on the flip side of that, give those holding bank notes a safe asset to run to. The Bank did this, driven by its motive to make profits on superior information.
It wasn't like some economist was promoting the idea that a monopoly on currency issue and financial crisis intervention were the natural province of a central bank, or that this central bank should be have some form of public governance. Those ideas came much later. Indeed, the Bank of England remained a private institution until 1946. However, the Federal Reserve System was founded as a public institution, albeit with semi-private governance, in terms of how the regional Federal Reserve Banks are run. The work that led up to the Federal Reserve Act of 1913 included, for example, the National Monetary Commission, created by Congress to study the U.S. financial system, and systems elsewhere in the world, and to come up with recommendations for reform. The Commission produced volumes of useful stuff, which you can find on Fraser. The basic ideas in the Federal Reserve Act are that the role of a central bank in the United States was to: (i) through its monopoly power, provide a safe currency; (ii) make the currency "elastic," i.e. make the supply of currency respond to aggregate economic activity, and to financial panics, by way of discount window lending.
Basically, the framers of the Federal Reserve Act had in mind an institution that would play the same role as the Bank of England, but the motivation was different. A currency monopoly for the central bank was seen as socially beneficial, as experiments with private currency issue in the U.S. had not turned out well. And crisis intervention was seen as a means for preventing the disruption to the aggregate economy that had occurred during the banking panics of the National Banking era (1863-1913).
Fast forward to 2016. In some ways things haven't changed. U.S. Federal Reserve notes are basically the same stuff (with some anti-counterfeiting features added) they were in 1914, and currency is still an important Fed liability. Here's the stock of currency relative to U.S. GDP: The ratio of currency to GDP has risen to close to 8% from less than 6% before the financial crisis, and is at the same level as in the mid-1950s. Retail payments using currency have fallen, but perhaps not as much as one might think. For example, surveys by the Boston Fed show a declining use of currency, but currency is still important in consumer payments, accounting for 26.3% of payments in 2013, as opposed to 31.1% for debit cards. But, the quantity of Federal Reserve notes per U.S. resident is about $4,500 currently. I don't know about you, but I'm not holding my fair share of that. In this chart you can see that, by value, about 80% of currency outstanding is in $100 notes, and studies indicate that about half of the stock of U.S. currency is held outside U.S. borders. There are some legitimate questions about the role played by currency, and whether its use should be curtailed, but more about that in the next installment.
The key point is that, if we took large-denomination currency - which is likely serving no useful social function - out of the mix, only a small part of the existing Fed portfolio would be funded by what was once thought to be the primary central bank liability. As it is, even with all those $100 bills included, interest-bearing Fed liabilities are greater than the quantity of currency outstanding: The key Fed liabilities are, in order of magnitude (from small to large):
1. Reverse repurchase agreements, or ON-RRPs, used in "temporary open market operations." This is overnight lending to the Fed, with securities in the Fed's asset portfolio used as collateral. ON-RRPs are roughly reserves by another name - overnight interest-bearing liabilities of the Fed that can be held by a wider array of financial institutions than are permitted to have reserve accounts with the Fed. In particular, money market mutual funds cannot hold reserves, but they are key participants in the ON-RRP market. ON-RRPs play an important role in post-liftoff Fed implementation of monetary policy. The idea is that the ON-RRP rate, currently 0.25%, puts a floor under the fed funds rate, so that fed funds will trade between the interest rate on excess reserves (IOER), currently at 0.5%, and the ON-RRP rate.
2. Reverse repurchase agreeements held by foreign government-related institutions. This is somewhat mysterious, and explained here. Just as there is a foreign demand for Treasury securities, there is a foreign demand for very short-term liabilities of the Fed.
3. Currency. We know what this is about.
4. Reserve balances. As is well-known, this quantity has grown substantially, as this stuff financed the Fed's large-scale asset purchases post-financial crisis.
We're now in a world in which interest-bearing Fed liabilities have become very important. Is this a temporary change, or is it permanent? Should it be permanent? This gets us to Greenwood/Hansen/Stein (GHS). Basically, their answer to the last question is yes. GHS first argue that short-term safe assets are useful in financial markets and that, by looking at market interest rates, we can find evidence of "moneyness" for these assets. If assets are used in facilitating some kind of exchange, or they are in wide use as collateral, they bear a liquidity premium. That is, people are willing to hold these assets at lower rates of return than seem consistent with the actual payoffs on these assets. For example, Gomme, Ravikumar, and Rupert calculate real rates of return on capital in the U.S. since 2007 of from 5%-7% (after tax, for all capital). But here are some short term nominal rates of return:Before December 17, 2015 (liftoff date), the IOER was 0.25% and the ON-RRP rate was usually 0.05% (with some experimentation). After that date, IOER was set at 0.5% and ON-RRP at 0.25%. In the chart, you can see that, after liftoff, the fed funds rate has typical fallen in the range 0.35%-0.40%, except at month-end and quarter-end (for technical reasons). Also, four-week T-bills trade in the same ballpark as ON-RRP, with some variation.
What can we conclude? (i) Reserves are a less liquid asset than ON-RRP or short-term Treasury debt. Reserves can be traded among a smaller set of financial institutions than these other assets, and you have to pay banks more to take reserves than to take T-bills or ON-RRP. (ii) Short-term Treasury debt and ON-RRP seem to have roughly the same liquidity properties.
So, once the case has been made that short-term safe assets bear liquidity premia, reflecting their usefulness in financial markets, who is going to supply these assets? There are three options: (i) the private sector; (ii) the Treasury; (iii) the Fed. GHS argue that the private sector does not do a very good job of this. Why? First, there are some "externalities" involved, according to GHS. This is somewhat vague, but GHS seem to have in mind that private sector production of short-maturity assets involves intermediating across maturities, which is risky. And maturity transformation makes these private financial intermediaries sensitive to market stresses, according to them. Second, regulatory changes, for example the Supplementary Leverage Ratio requirement, gums up the private sector's ability to produce safe short-maturity assets. GHS also argue that the Treasury does not issue enough short term debt, because it is worried about the risk of auction failure if it has to roll over a lot of short-term debt.
The heart of the argument is that the Fed has advantages over both the private sector and the Treasury in issuing short-term debt. But what kind of short-term debt should the Fed issue? Currently, the choice is between reserves and ON-RRP, though in principle we can think about the possibility the Fed could issue Fed bills - short term circulating debt that looks exactly like Treasury bills. The Swiss National Bank can issue such securities, for example. But, if wer're constrained to considering only reserves and ON-RRP, it seems clear that ON-RRP is much more successful in serving the liquidity needs of financial markets than are reserves. After all, you have to give the institutions that are permitted to hold reserves a 25 basis point inducement to get them to do so.
So, it seems that, in general, ON-RRP is a better instrument for the purpose than reserves. As GHS point out, if the Fed were to expand its ON-RRP program and shrink reserves, this would amount to savings for taxpayers, given the level of short-term market interest rates. This is becasue the Fed could have the same effect on market interest rates, but be paying lower interest on its liabilties, thus handing over larger transfers to the Treasury. What expands the ON-RRP program? One approach would be to set the ON-RRP rate equal to IOER.
But, how large should the Fed's ON-RRP balances be, and what's the optimal size for the Fed's balance sheet? Prior to the financial crisis, the size of the Fed's balance sheet was essentially determined by the demand for currency (in real terms), given the level of market interest rates, as the quantity of reserves was very small, and almost all of the Fed's asset portfolio was financed with currency. If we take the GHS proposal at face value, there should be a similar natural limit for ON-RRP. If the Fed sets an ON-RRP rate, and conducts a fixed-rate full-allotment auction, with IOER equal to the ON-RRP rate, what could happen? (i) the Fed reaches the upper bound on securities that it can use as collateral in ON-RRPs - it gets more bids than it can satisfy. (ii) the Fed does not reach the upper bound on available collateral, and there are some reserve balances outstanding. In case (i) it seems the ON-RRP program is too small, and in case (ii) it's the right size, but the Fed's balance sheet is too large. The bottom line, if we accept GHS's hypothesis that a Fed supply of short-term interest-bearing liquidity is a good thing, is that the market can determine how much of this stuff it needs. I should make it clear that this is my conclusion, not theirs. But I think this is the logical implication of their analysis.
Comments:
1. How does this relate to quantitative easing (QE)? GHS seem to think this is a different issue - that QE is aimed at some short-run problem, and the role for ON-RRP is long-run. I don't think so. What GHS is putting forward is indeed a theory of how QE works. They are assuming that the Fed has a special role in maturity transformation, and when the Fed does this it matters. And this matters all the time, not just in unusual circumstances, so the implication is that QE should be an ongoing thing. But it seems there is a role for long-maturity debt in financial markets too - presumably the Fed shouldn't be sucking up all the long-maturity assets in existence, and GHS don't seem to be thinking about that.
2. I've got doubts about whether there are limitations on the private sector's ability to intermediate across maturities. Turning long Treasury debt into overnight debt is risky, but there are ways to manage such risk. It's not clear that the Fed and Treasury are any better at bearing such risks than are private financial institutions.
3. If the problem with private sector liquidity transformation is what can happen in crises, why shouldn't the Fed's intervention be confined to crisis times? For example, expand the ON-RRP program in a crisis, and contract it when the crisis is over.
4. A concern of GHS is a possible flight to safety during a panic. The argument is that, during times of financial stress, financial market participants could abandon private sector liquidity for ON-RRPs, and that could be bad. The "cure" for this is caps on the ON-RRP program. I've always found this idea puzzling. In the ON-RRP auction, the Fed can either set the quantity, or the price, but they can't set both. If there's a binding cap on the ON-RRP program, the price is too low, i.e. the ON-RRP rate is set too high. Presumably, in a crisis the correct policy is to lower the ON-RRP rate.
5. GHS's explanation for why the Treasury did not, or could not, issue more short-term debt did not make sense to me. This seemed more like a call for the Treasury to do a better job of auctioning their securities. Further, if a Treasury auction "fails" on a given day, I'm not sure why that's the end of the world. For example, the Treasury has a reserve account with the Fed, and the balance looks like this: You can see that the balance in this account fluctuates a lot - it's an important buffer for the Treasury in managing cash inflows and outflows. Further, the average size has increased substantially, to around the $300 billion range. If rollover risk is greater with more short debt outstanding, why can't the Treasury have a larger average balance in that account?
In general I found this paper very useful. It's the first coherent story I've seen about a legitimate role for a central bank in what we would typically call "debt management." But much more research and thought needs to go into these questions.
Subscribe to:
Posts (Atom)