Unconventional times call for unconventional approaches, which central banks around the world took in response to the financial crisis. In this panel discussion from the Atlanta Fed's 2016 Financial Markets Conference, experts discuss some of those approaches, debate their effectiveness, and examine the long-term implications for policymaking.


Loretta Mester: So this session moves into a consideration of the interaction of monetary policy and liquidity and, in particular, the focus is going to be on unconventional monetary policy, which includes, of course, very low or even negative interest rates and quantitative easing. And we know that people all over the spectrum [hold] different views about the effectiveness of monetary policy. Some say that it is now impotent, others seem to think that monetary policy can fix all the ills of the economy. For those in the first camp, I hope that you'll all agree that today's session provides convincing evidence that QE [quantitative easing] is not impotent. It was the reason that we got this great panel together to talk about it, and so already we have something that we can point to that QE was a catalyst for. And for those who like to think that QE can fix all of our problems, I really think we have a research agenda here I hope that you all will investigate whether it's the source of the beautiful weather that we've had over the last couple of days.

So, in theory in thinking about the conference I wanted to think about liquidity, and it seemed like a natural place to start was with a definition of liquidity. And for people who listened intently to President Lockhart on the first day, he suggested that that's not as easy as one might think it is to actually think of that. So decided that I would look at some experts. Whenever I can't know something, I decide it's a good idea to look at the experts. So John Maynard Keynes—pretty famous guy—I figured, go look there. He says the conception of what contributes to liquidity is a pretty vague one, changing from time to time, and depending on social practice and institutions. Didn't really help me much as much as I thought it might.

So then I thought, next place to look [is] Palgrave's—dictionary, solid reference work. There it starts with, "Liquidity is a highly complex phenomenon." All right, so I figured I'm doomed there. So what I decided to do is, since they didn't help me I decided I would do my own poll of economists. I called friends and colleagues and people at the session yesterday. Barbara Novak hypothesized that if you did such a poll you would get several different answers, and I can tell you I can confirm her hypothesis as being correct. So I'm going to give you the top eight, which was basically all the answers I got. I polled about 15 people, but I did get some congruent answers. This is not quite a David Letterman list, but it kind of works.

They actually fall in three categories—the third category is on the next page. The most popular answers were things about ease of conversion into money. Five people basically said, "The ease to which an asset can be converted into money without causing a loss in value," and one person actually gave me a definition of value as the discounted value of the asset if one had no limit on the time to convert it into money. Two people explicitly mentioned transaction costs, although I would say "ease" is basically, essentially transaction cost. Then somebody said, "the speed with which an asset can be converted into money at a price that is close to market value of the asset," and he defined market value as "what you would get for the asset if everyone in the market bid for it and you could chose the highest bid," which doesn't quite make sense because if you don't want the asset, you'd bid zero so you probably could ignore them, but nonetheless. The third was, "the speed with which an asset can be converted into money in a market with a narrow bid-ask spread." That's kind of getting at what we have talked about over the last couple of days about breadth and depth of the market. And the fourth was, three people said, "the speed with which an asset can be converted into money." I found that interesting because there is like no concept of price there, although one could presumably say that's just an oversight because speed will definitely be affected by the price that's being demanded in the market. And then one answer mentioned only the volatility of the asset price, the stability of an asset's value relative to the unit of the account.

So other people just said it has to do with the spot price versus the future or intrinsic price, and so you can see different ideas up there. The extent to which an asset in the spot market is not dependent on private information. So in other words, less volatility, the extent to which an asset's intrinsic value differs from its expected market price for a given time period. So here we are getting into Pete Kyle's idea of time. And then the last one is the extent to which an asset's price differs from its expected future price, where smaller differences imply higher liquidity. So the lower the transaction cost, the more liquid the asset, but volatile futures prices don't necessarily imply low liquidity.

So there were a bunch of different answers that I got. It does seem like the essence of this did point to ease of conversion of the asset into something like cash or something that could be a good substitute for cash, and then the price at which the conversion occurs. So transactions costs, which we have been talking about. Information asymmetries, which we have been talking about. A to Z problems, which we've been talking about, all came up in this discussion, but as you can see there are a lot of different views out there of what really liquidity is.

The speakers today are going to touch on all aspects of this topic and in particular discuss the role of monetary policy in affecting different aspects of what everyone in the room might have their own personal view of what liquidity is. They are excellent presenters. The bios are in your books, but it would be remiss of me not to mention a few things about the speakers.

Marvin Goodfriend is the Friends of Allan Meltzer Professor of Economics at Carnegie Mellon's Tepper School of Business. He's a very well-known monetary, economics, and policy scholar. And before coming to Carnegie Mellon he was the director of research and policy adviser at the Federal Reserve Bank of Richmond. So he has both the academic experience and knowledge, plus the policy experience. He's going to be an excellent speaker, and I'm always going to appreciate him. Marvin was very, very generous to me when I became research director at the Philadelphia Fed. He showed me the ropes, and one time when my principal was in the emergency room getting stitches he came back from dinner to see how I was doing in the room—and it wasn't Charlie Plosser, it was before that. So don't go up to Charlie and say, "Why were you in the emergency room?" So Marvin is going to lay out a taxonomy of the conventional and unconventional monetary policy. He's going to speak on how those different policies could be expected to work and affect liquidity.

The next speaker is going to Ulrich Bindseil. He's the director general of market operations at the European Central Bank and again, combining a lot of different skills. In addition to being a central banker, he also has been a lecturer and visiting professor at the Technical University of Berlin since 2008. He's published in the areas of central bank market operations, risk management, and European governance. He's going to talk about the ECB's [European Central Bank] unconventional monetary policy actions and, in particular, also the lender-of-last-resort function and their potential to affect liquidity. And then talk a little bit about what the ECB is doing to mitigate those effects on liquidity,

And our third speaker is Manmohan Singh. He is senior financial economist at the IMF [International Monetary Fund], were he studies monetary policy, financial markets, and shadow banking. And he is planning to speak about the effects of unconventional monetary policy on money market plumbing. He has a book out that talks about this—Collateral and Financial Plumbing. And in particular, his focus is going to be on how do we unwind some of these nonconventional policies that the Fed and others have been using.

So each presenter will have 20 minutes, and again, as we have done throughout the conference, use Pigeonhole to put in your questions. Anything sparks an idea, write it down, and we will do Q&A after the speeches. Thanks.

Marvin Goodfriend: Thanks Loretta. My presentation has the distinction of being the first time I have ever been to a conference and I don't think I will mention the word that is prominent in the conference title—liquidity. I'm going to avoid mentioning the word liquidity entirely and talk about Federal Reserve policies, or central bank policies that have bearing on what I think you might mean as liquidity. My job is to give you a taxonomy of what the Fed calls "monetary policy" in an overarching way, which really is not monetary policy but a whole series of different policies, which need to be understood separately in order to understand their potential effect on liquidity. So it's a little bit of a circuitous route to getting to, I think, what would be a useful outcome.

So I'm going to deconstruct central bank policies. I want you to forget the word "monetary policy" as it is usually used. It is used as this sort of umbrella term and therefore has very little meaning, so let's dispense with it. I will talk about what I believe is a useful definition of monetary policy in a narrow sense, and I'm going to distinguish it from what I would say is a credit policy, and I'm going to distinguish both of those two policies from what I would like to talk about as "interest on reserves policy." And then after doing those things, I'm going to talk about the fiscal aspects of these three policies, how they work, and whether they are effective, and what the different policies mean for the boundaries between central banks and governments—a little bit; I don't have a lot of time—and also what they mean for liquidity, especially in the last section when I talk about the interplay between central bank policies and money markets. I'm not going to have much to say directly about liquidity in the corporate bond market except for a couple of slides, and I'm hoping to remember exactly where that is so I can point that out.

Monetary policy, as I'd like to define it, would be open-market operations by a central bank that expand or contract central bank monetary liabilities, so I'm going to include in that currency, reserves, or reverse repurchase agreements, term deposits—anything a central bank in these days wants to call a liability. I want to define it that way because I have a clean definition. If you take the central bank balance sheet and you consolidate it with the Treasury—if the Treasury is buying only Treasury securities, then the liabilities cancel if you combine the Treasurys and the Fed's balance sheets. So I'm going to call open-market operations that buy U.S. Treasurys "pure monetary policy" because upon consolidation all that is left are the liabilities of the central bank. Now, prior to the credit turmoil, the Fed satisfied virtually all of its asset acquisition needs to support monetary policy by purchasing Treasurys, a policy that was known as Treasurys only.

What do I mean by credit policy? Credit policy, in my mind, shifted the composition of the central bank's balance sheet holding liabilities fixed between Treasurys and credit to the private sector or other government entities in the form of loans or security purchases. So for example, mortgage-backed securities on the Fed balance sheet today, I would call a credit policy because they do not have the full faith and credit of the U.S. Congress like a treasury does. You can have a combination monetary and credit policy where the money finances credit policy. In in late 2008–09, famously, the Federal Reserve created reserves to a trillion dollars and loaned those reserves to a bunch of private entities, effectively, or bought private securities. That would be a combination monetary and credit policy—after I finish the taxonomy, I'll tell you how I think these work differently.

Interest on reserves policy is, again, a third, completely separate policy. A central bank would change interest paid on reserve balances, holding fixed its balance sheet, holding fixed its monetary policy and its credit policy. Interest on reserves policy frees interest rate policy from monetary policy. The Fed raised interest rates in December without doing anything essentially on monetary policy or credit policy. And interest on reserves would enable monetary policy to finance credit policy independently of interest rate policy. The original reason the Fed got emergency authority to pay interest on reserves, some of you may remember in 2008, was the optimistic hope that it would need to hold interest rates above zero while it used monetary policy to fund credit initiatives to save the credit system. In the event, things collapsed so suddenly that we went to zero and didn't really need interest on reserves, but we do now.

So, what are the fiscal aspects of the policies? Pure monetary policy, as I am describing it, manages the interbank interest rate by influencing the spread between the interbank rate and interest paid on reserves. By maintaining a scarcity of reserves and a positive marginal services yield and a positive interest opportunity cost spread, this jargon sounds odd, but from 1913, essentially, until 2008, the Federal Reserve used this procedure to make sure that the interbank interest rate floated above zero. The Fed had no power to pay interest on reserves. It was essentially creating an artificial scarcity of reserves so that banks trying to borrow these scarce reserves from each other would float the interbank rate up to where the Fed wanted it to be. It's ancient history.

Reserve scarcity imposes a tax reflected in a below market interest rate on reserves, and as those of you who have followed banks know, until 2008 the banks found ever-ingenious ways to do without reserves so they could avoid the tax. So by the time the crisis happened there was just a slim sliver of reserves held in the banking system that created trillions of dollars of deposits. The Federal Reserve, or the central bank, would collect a tax on reserves as interest on the Treasurys that it would acquire in making those reserves available and, more importantly, making currency available. And the Treasurys only policy that is part of monetary policy as I am defining it would transfer all tax revenue, net of interest on reserves and expenses, to the Treasury. So monetary policy as I am defining it is a pure separation of fiscal policy and monetary policy.

What about pure credit policy? Pure credit policy executed by the central bank is really debt-financed fiscal policy. Why? Interest on Treasurys held by the Fed or the central bank is simply returned to the Treasury, so from the Treasury's point of view, these Treasurys held are extinguished. They cost the Treasury nothing if they are held by the central bank. So when the central bank sells a treasury to take the proceeds and make a loan to a private entity, those Treasurys now in the market have to be financed by the Treasury, and the Fed in turn takes on a loan to a private-sector entity or some private-sector security. It gets paid interest, which the Fed will return to the Treasury. But what's happened is the pure credit policy imposes government creditworthiness between borrowers and lenders. It puts the taxpayers' funds at risk, and any losses that occur because the private credits fail to pay or they have a problem are monies would no longer be going to the Treasury. I'm going to come back to this point. This is just my mechanical introduction to the idea.

Many times, the Fed likes to say that the Federal Reserve and other central banks take good collateral. So it's important for me to go through the next slide. Even if the central bank takes good collateral and assumes negligible credit risk itself, it exposes taxpayers to losses if the borrower fails subsequently. So what's happening in many lender of last resort or pure credit policy actions is that the most nervous creditor of an institution, maybe an uninsured creditor, a creditor that would ordinarily would be low in line wants to leave the bank. The central bank finances the exit of that creditor, but the central bank takes good collateral itself. So if that bank fails subsequently the central bank essentially has stripped the institution of collateral that would be available otherwise to cover the cost of more senior creditors, insured deposits, or government guarantees.

Now this is finance conference, so I decided to use a finance term: "novation." I hope some of you know what this is, because I'm proud of using this. In effect, credit policy is a novation by which the central bank inserts the Treasury taxpayers between a private lender and a borrower and takes out the lender and transfers the risk of default to more senior creditors or the Treasury or taxpayers. What I'm defining as credit policy, I realized after reading about finance, is nothing more than a public novation. Maybe that's appealing. I love that word—novation. And I'll come back to that in a few minutes.

What about interest on reserves policy? It utilizes a fiscal instrument—the payment of interest on reserves—to eliminate the tax on reserves, improve the efficiency of the payment system, and could be run with Treasurys only because it says nothing about the asset-acquisition policy. And a relatively small expansion of reserves would be sufficient to push the interbank rate nearly to the interest on reserves floor. In other words, you can satiate the market without having $3 trillion of reserves on your balance sheet. And beyond that monetary policy would be free to finance credit policy with little effect on interest rates.

So the last part of my talk is going to discuss interplay between central bank policies and money markets, and there will be a few times here where I think I will make contact with the issues discussed at the conference. First is, the policy rate arbitrage to money market rates. So given excess reserves at the margin, the interest on reserves—IOR, today in the Fed's case and many central banks around the world—puts a floor on interest opportunity cost of loanable funds for depositories. Now, the U.S. banking system is a net borrower in the money market. In other words, there is going to be an equilibrium—and there has been for quite some time—positive wholesale deposits on the balance sheet of the U.S. banking system. So the money market is a net lender to the U.S. banking system. So if the Fed raises interest on reserves, it's going to pull up money market rates to the degree that the banks take up the marginal money market lending in CDs or RPs or wholesale deposits and use the proceeds to earn interest on reserves. That's how monetary policies are pulling up rates in the money markets and beyond the depository system. Money market rates fall below interest on reserves by enough to cover the arbitrage costs, including shadow balance sheet cost of the supplementary leverage ratio. I'm going to come back to that issue because one of the things that is happening now is a kind of convergence between operational monetary policy and regulatory policy, and the extent to which interest rates actually reach out beyond the depository system effectively.

Arbitrage keeps other money market rates aligned up to credit liquidity and intermediation cost differentials and so forth, and Treasury bills generally float below many money market instruments because they are unique. In many of the ways we've been talking about today, they are liquid—I promised not to use that word, but I'm going to stick to my slide. Treasury bill rates are marginally below the interbank rate because interbank borrowing is unsecured, T-bills are safe, actively traded, and yield implicit collateral services.

So what about monetary policy on money markets? As I mentioned earlier today, the monetary policy has satiated the demand for reserves, eliminated the scarcity of reserves, driven the implicit marginal monetary services yield to zero. In so doing, monetary policy has eliminated the tax on reserves. Large depository holdings of reserves improved depository...oops, I did use liquidity. It's hard to avoid it. Abundant reserves help substitute for private credit flows among banks in the provision of payment services and for Fed daylight credit extended to banks to facilitate the provision of payments. Now, here is where one of the places Fed policy makes contact with private markets. Reserves are free for the public to produce; they are electronic blips. When you make reserves scarce, you are causing credit to take the place of reserve holdings in settlement and so forth. Credit is expensive. The use of credit in these markets in place of reserves is socially inefficient because it's creating real costs in society and real instability risks and, I think, problems in markets on that side. The elimination of the tax on reserves improves depository efficiency and also competitiveness with money markets.

Now, people sometimes wonder what do I think should be the size of the Fed's balance sheet after we exit in some nirvana in the future? Well, you could have a balance sheet that has currency plus only about a few hundred billion dollars of reserves based on what happened in earlier decades and in other countries. You'd only need a few hundred billion dollars of reserves to satiate the market and make the federal funds rate fall to the interest on reserves floor. Now, the funds rate has been below interest on reserves, and I'll come back to that in a minute.

Central banks could hedge the payment of interest on reserves with short-term Treasury bills. So there wouldn't have to be a fiscal cost to the Federal Reserve of using a smaller balance sheet, satiating the market for reserves with monetary policy, and holding short-term Treasury bills.

The next couple of points also bear on this conference. In general, expansionary monetary policy does a few things—say, expansion beyond the few hundred billion, or even in the few hundred billion, but certainly beyond. It uses up depository balance sheet capacity. It absorbs Treasury collateral from the markets because the Fed is buying Treasurys that otherwise would be available to do things like collateralize trade and facilitate the kind of liquidity—and I'll use this word again—that's being discussed for the last couple of days in this conference. So there's no question that monetary policy deprives private markets of Treasury collateral, which is extremely valuable.

The third point is that both effects would raise depository arbitrage costs to money market rates under the supplementary leverage ratio constraint. So there's given an interest on reserve target. Expansionary monetary policy, or a big Federal Reserve balance sheet does the following: It lowers money market rates relative to IOR because there is a higher shadow value of the supplementary leverage ratio on the depositories. And two, because of collateral being withdrawn from the system, the capacity of money markets through use of collateral to fund illiquid assets is diminished, and that's a point that's been made, especially in the rehypothecation discussion earlier. So there are these two effects. These are opposite effects on the pressure on illiquid rates that come about because of balance sheet expansion by the Fed. So I'll repeat them because they are important and probably one of the main ways in which my comments touch on your concerns.

For a given interest on reserves a bigger expansion in monetary policy will cause potentially a tighter, more binding supplementary leverage ratio, which means the banks will not be willing to arbitrage to money market rates except with a bigger spread. So money market rates fall below interest on reserves by a little more than otherwise. And, two, the Fed drains collateral, which then raises the cost of the use of collateral through a rehypothecation to fund illiquid assets in money markets. These two effects are going in opposite directions for the illiquid rate, and it's a question of which is bigger. I don't know. I think we'll have some discussion of that later.

What about credit policy on money markets? Well, there's no question that expansive Fed credit policy was essential to preserving the integrity of the payment system in the crisis. As I mentioned before, the central bank credit policy could be considered a novation. It worked by the central bank selling Treasurys to entities no longer willing or able to lend in money markets, including in the interbank market, and the Fed or the central bank lending the proceeds to depositories or money market entities no longer able to borrow at reasonable rates in money markets. In part, so depositories could refinance through lines of credit commitments their money market clients. So Fed credit policy was a novation.

Now, the problem with novations executed by the independent central bank is as follows. Independent central bank has the discretion to shift risks to senior creditors to the Treasury, or taxpayers without due process or the appropriations process. This is a fiscal policy action that has been delegated to the independent central bank. I could go into what I think about this, I just want you to appreciate this particular issue. Two, independent central bank is inclined to implement credit novation policy, if you will, whenever monetary collapse seems imminent or otherwise. You may or may not think this is a good thing. It seems like a good thing at the time, but the question is what incentives does it set up over time. And then three, expecting this, depositors, money market investors, and monetary service providers are encouraged to expand leveraged depository and money market finance of illiquid assets beyond what would otherwise be thought prudent. I like those words. I feel more strongly about this than those words, but I'll live with that for now.

If depository and money market entities with explicit or implicit expectation of central bank credit novation policy must be supervised, regulated, and bonded to restrain a fragile leveraged depository and money market finance. Now this is directly regulated to liquidity issues in markets that you all are talking about. I love this conference by the way. I've learned a lot, but I'm a little worried that we may be through these kinds of policies—if you will, subsidizing, and I'll use liquidity again—in markets well beyond the central bank. I'll leave that to further discussion.

Interest on reserves and money markets...in the United States, large lenders such as the government sponsored enterprises [GSE] and the Federal Home Loans Banks are eligible...they hold reserves at the Federal Reserve. They're not banks, the really have no business holding reserves at the central bank, no doubt they're there to make these institutions look "too big to fail," which is probably not a good idea. Nevertheless, everyday they dump their funds at the Fed into the federal funds market and look for take-ups from banks who are eligible to receive interest on reserves. So the federal funds rate is really peopled mainly, or only by the GSEs and the Federal Home Loan Banks. The other banks don't need the federal funds market anymore—they're getting interest on reserves at the Fed. So a number of factors have limited the power of depository arbitrage to bring the fed funds rate up to interest on reserves. Leverage or liquidity requirements, counterparty line limits on unsecured loans to a single bank, adverse perceptions associated with significant overnight borrowing, internal oversight costs, lack of strong competition among banks with access, and finally the deposit insurance fee since the crisis has now been imposed on the entire balance sheet of banks—many of these arbitrages are occurring through domestic branches of foreign banks for reasons that I don't want to go into.

The Fed is prepared to use overnight reverse repurchase agreements to help put a floor under the federal funds rate at or near interest on reserves. With overnight reverse repurchases, the Fed essentially creates a risk-free money market instrument on its balance sheet available to a broad array of money market counterparties. So this is another place where liquidity is being offered beyond the depositories by the Fed. You may or may not like this idea. One of my concluding points is going to pose some questions about how this should be utilized.

So by setting up an overnight repurchase rate at interest on reserves with full allotment, the Fed could certainly put a floor under the funds rate because it would invite all comers to lend at that rate to the fed balance sheet. But in effect, if the Fed were to put the overnight repurchase rate at the interest on reserves, it would undercut depository arbitrage by taking up loanable funds at a tighter spread than is profitable for depositories, especially under the supplementary leverage ratio, which is designed to make the banking system safe. In doing so, the overnight repurchase would subsidize arbitrage costs otherwise absorbed by lower money market rates that are big enough to cover the shadow value of banking balance sheet.

And my last slide. The question for me is—and I haven't completely made my mind up about this—should the money market be made to absorb the cost of depository arbitrage in commensurately lower money market rates relative to interest on reserves, or should the Fed take up from money markets directly at the interest on reserve rate via the reverse repurchase agreements; that lending? So, two questions. For instance, supplemental regulatory leverage ratio for U.S. depositories has the potential to raise arbitrage costs, especially when it's fully phased in. I'll take for granted, if you think that the SLR [statutory liquidity ratio] is a socially beneficial means of maintaining safe and sound banking, then it seems since the money markets are more lightly regulated, then maybe they should be made to face that same kind of cost at the margin. Otherwise you are subsidizing interest intermediation in the money markets at the expense of banks. And then more importantly, or equally importantly, it's very possible that the shadow value of bank balance sheets with a supplementary leverage ratio could fluctuate over time with financial distress, with the business cycle, with all kinds of reasons. It seems to me you might consider that that flexibility is a useful relative price that would then serve to steer credit either through depositories or through money markets as seems appropriate.

I have to say that I'm inclined to want that price to fluctuate, but for point three I have to say that it would definitely complicate the transmission of interest rate policy to the economy if the Fed is controlling interest on reserves and then the money market rate is fluctuating with respect to the shadow value of the supplementary leverage ratio on the bank. So I leave this for further discussion, and what I hope to do in this presentation is at least give you a way to think through all these policies that the Fed is using and how they effect liquidity in different ways. Thank you.

Ulrich Bindseil: So thank you very much for having invited me to this conference, and I will use the word liquidity frequently. Yesterday, Jeffrey Kessler clarified he is not a German-speaking English way, so I'm the German and I try my best to speak English.

So I try to really cover the topic of impact of nonstandard measures on market and funding liquidity and what nonconventional measures should one try to cover. The first one, lender of last resort; second, LSAPs [large-scale asset purchases], and third, negative interest rates. So, that gives an overview, and obviously the three can all have an impact on markets.

Lender of last resort, I think one should view it as a monetary policy tool. It's about, let's say, stabilizing the funding of banks and thereby supporting the intermediation capability of the banking system and thereby, in particular, if the zero lower bound is not remote, keep financial conditions—funding conditions of the real economy—low and not deflationary. So obviously, the lender of last resort impacts on liquidity, directly on funding liquidity of banks, but also on asset liquidity as it prevents fire sales and the very unbalanced market.

LSAPs. The obvious liquidity impact that comes to my mind is the one on the bond market. So it can be positive; it could be negative. It's probably positive if it has elements of market maker of last resort. So if it addresses an imbalance in the market, it is potentially negative of course, if it removes a lot of high-quality bonds from the market. Another impact of LSAPs is on money market liquidity. On one side, of course, it overfloods markets with liquidity, money markets, but if every bank is overliquid in the definition of liquidity meaning reserves with the central bank, then there's little to trade, so money market turnover will be low and money market liquidity in that sense will be low.

And finally, negative interest rates. I will be pretty short on that because our experience is negative rates have hardly any impact on market functioning—only, let's say specific, unsolved IT institutional or legal issues can make a difference. But those being resolved, there's no impact on market functioning.

So let me give three slides on a little model that tries to capture the lender of last resort. It's a little bank run model. There are lots of bank run models, and so this one, let's say, tries a bit harder to actually integrate into the bank run problem, the lender of last resort. Most of the bank run models assume they'll somehow talk at the end about lender of last resort, but there is very little concrete in those models. So this model just simplifies a lot, but gives a bit more detail to how the lender of last resort functions and what role it can play to prevent bank runs. The trick here is to have an extremely simply liability side with three types of liabilities and two short-term depositors who play a very simple strategic bank run game. Long-term funding is more expensive, but it cannot run away. Equity is assumed to be even more expensive, and it doesn't run away and it can absorb losses, in addition.

Then on the asset side. So the asset side is the length of the balance sheet is one, and the assets have liquidity properties in two dimensions. One is, in principle, all assets are pledgeable with the central bank. So that's where the lender of last resort comes in, and all the assets can be fire-sold. The assumption is—to simplify the calculus—that the ordering of the assets in terms of haircuts and in terms fire-sale losses is identical. So the most liquid asset in terms of fire sales is also the asset with lowest haircut by the central bank. Secondly, there is a very simple function or form imposed on the haircut function and the asset fire-sale loss function. They are both power functions, just with a different exponent that has—the power function has a nice property that maps the most liquid asset into a zero—basically the haircuts/fire-sale loss—and the most illiquid into one. You can manipulate it very simply. Then the model is just basically as this timeline indicates. So at the beginning you have asset-liquidity parameters—so, data and theta—and you have a cost of the different forms of liability...pardon me, it's a non-Modigliani-Miller approach, so there is a fixed kind of funding cost of each of the types of liabilities. So then the banks choose a cheapest sustainable liability structure, then depositors play this run game, and then the bank may default because of illiquidity. Even if it survives, it may then be insolvent because of fire-sale costs in trying to secure the liquidity to finance a bank run may make it insolvent and may create then the liquidation of the bank, at least by the supervisor.

That's, again, just the power function. What banks do in the first stage is to establish what is called here a liquidity strategy. So they assign assets either to be used in the run scenario for fire sales or to be pledged with the central bank. There's a trivial case—if data is bigger than theta, then you would always assign all your assets to be pledged with the central bank. Because central bank pledging has no costs. So the nontrivial cases where you foresee to fire sale a part only occur if haircuts are higher than fire sale losses. That's logical. So you can, in fact then, show that this liquidity stress strategy consists in choosing a zed, meaning that all the assets from zero to zed, the more liquid asset, you fire-sale them, and the less liquid asset, you pledge them with a central bank. So you can show that this makes more sense than the other way around. That's basically, then, what you formulate as a condition for no run. That's that the liquidity that the bank can generate exceeds the deposits of one depositor, and the fire-sale losses generated by this strategy are lower than the equity of the bank. So that's intuitive, and the chart shows that zero to zed would be fire-sold, so you have the red triangle "k," which are losses, and that cannot exceed equity—otherwise it is not stable. And the liquidity is on top of that, and the rest is pledged with the central bank. Haircuts are higher, but you get the liquidity without another loss.

So what does it allows us to calculate or get us insights? The cheapest liability structure the cheapest stable liability structure will be a function of the two liquidity parameters and the costs of the different liabilities. And in a crisis, obviously what happens is that theta—the liquidity parameter—will go down. Liquidity will go down, and if banks in competition move towards the cheapest stable funding structure, then a sudden deterioration—an unanticipated deterioration of asset liquidity—makes the system switch from a no-run to a run situation. And the lender of last resort, I would say, is about then adjusting the parameter data, because of course you can counterbalance a decline of theta with an increase of data so that the no-run equilibrium is maintained, and I guess that's what central banks tended to do in 2008, when many of them expanded the lender of last resort function as they did. So the lender of last resort is not necessarily about providing funds, but it ideally, it is about preserving the conditions that there is no run, and doing so by adjusting this lender of last resort.

Let me now move to other effects of nonstandard measures, and maybe a very simple one is the one I mentioned—the impact of excess liquidity injected by large-scale asset programs on money market volumes. And obviously, if every bank is in excess liquidity —in the definition of excess reserve with the central bank—there is no opportunity to trade in the interbank market, and this is what you can see a bit here. So the blue area is the excess liquidity, and the line is the EONIA [euro overnight index average] volume—that's the overnight interest rate in the euro area. And the volumes, you can see, decline when we have large excess liquidity, so it's what you'd expect. We recently—our volumes are very low, so EONIA has volumes of around 15 billion [euros] recently, which is, of course, not a lot. That is supposed to illustrate the nonimpact of negative rates per se on money market volumes. And you could do the same with capital market trading volumes. So what you can see here is that when the overnight rate moves into negative territory, nothing happens on the interbank market. So there was no effect whatsoever of moving into negative territory. Neither moving the deposit facility into negative territory nor moving the actual EONIA rate into negative territory—there was no effect on money market activity because there were no specific IT or legal constraints, which didn't allow it to go smoothly into negative.

Moving to bond markets, and here we have a data set, and I will match this data set with different things. So the data set is basically the data set of the number of trades of bonds in a given month, and this data tracks trade data, so collected from all major dealers in fixed income in the euro area. So it's a data set with a total of 4,500 bonds, and it classifies those bonds on a monthly basis into number of trades per months, and here you can see the distribution. So there are around 300 ISINs [International Securities Identification Number] that are traded at least 250 times a month. And you have 1,000 that are traded less than five times a month. So that gives you an overview of actual trading activity and fixed income. Once you have this data set, this transition metrics looks at the migration. It is monthly data, so you can easily calculate how securities migrate over time, and you can see they migrate in an asymmetric way. They tend to migrate towards less trading frequency because there weren't more absorbed by long-term investors who don't trade them any longer. The little table at the bottom matches the trade date data with a bit of our spreads in Bloomberg, and here the conclusion is bid-offer spreads have a limited information content, in particular, for the very infrequently traded assets. But our spreads go down so that cannot be—that's counterintuitive. So I would read from that that once you move into really, let's say, illiquid territory bid-ask spreads mean less and less.

The top of this table shows the ECB [European Central Bank] liquidity categories and haircuts. So this goes a bit back to the lender of last resort model where we had this—the model assumed that you can order the assets in terms of haircuts and in terms of market liquidity in the same way. So here, we matched this number of trades per month measure with the liquidity categories of the ECB. Our first liquidity category is, basically, central government bonds. There you can see the haircuts we imposed on single-A to triple-A [bonds], then depending on maturity and coupon type, and a triple B. And what you can then see in the table below is that in terms of trade frequency, this liquidity category one has securities everywhere. And also this category H in the table—the last column before the total—is securities that were never traded in the months. So the tracks data had only securities which showed up because they were traded once, but we have 33,000 ISINs, which are literally collateral, so the last column gives you all those who never showed up in the monthly date set of trades. So even the most liquid of our collateral categories you have the largest amounts of ISINs actually are never traded. So the positive reading of the table is that if you move down our liquidity categories and our collateral framework, you have more and more assets on the right side of the table. But on the other side, you have to notice that if you build institutional categories—you say government bonds are the most liquid assets, and I impose the lowest haircuts—then in reality you have a widespread distribution. Of course, what is in [category] H could be off the run bonds, could be bonds from Luxembourg, issued by the government of Luxembourg, of the government of Malta. You have lots of things, I would say, from the point of view of possibilities to liquidate if necessary. It's not very worrisome, but it is very liquid nevertheless in terms of trade frequency.

That's the ECB asset purchase program, as we do it. The last month we did even 80 because we increased the volumes, and what you can see is we have the total volume, 60, and it is distributed so far in three asset categories: ABS [asset-backed securities], covered bonds, and public sector. Basically, the logic is we buy as much as we can in the two private pillars, and then the rest is taken by the public-sector program, such as to reach the 60 billion on average. Let me hurry up.

So normally if you think about asset liquidity the worry is, "Can I sell as if I need to sell them?" We, with our purchase program, we worry about asset liquidity in the sense of can we buy enough to fulfill our monthly operational target? Quotes are not very useful to really assess liquidity in many cases, but this is a kind of real liquidity measure where when we buy securities we request several quotes, binding quotes—normally at least three—and this measure gives a spread between the winning offer at the end and the second-best one. And that is a measure of liquidity, because in a super-liquid market you would expect them to be very close to each other, and in an illiquid market the second offer may be far away from the best one, so that's the statistics. I think it is a nice data set, and you can see that over time, so far deterioration of liquidity throughout the program is moderate. There are no fundamental issues. OK, I just have to finish.

In the program, of course, we tried to minimize negative effects on bond market liquidity, so we have an initial share limit of 33 percent. We are liquidity providers in the sense that we buy what gets offered to us. We avoid buying bonds trading, especially, in repo markets, and we have a securities lending program. Thank you.

Manmohan Singh: This is the last talk, and so I will try to make colorful, but there are some messages here. Again, most of these views are not of the IMF [International Monetary Fund].

I will try to spend some time on the first two charts because this is data not readily available. This is from annual reports and footnotes of the annual reports. They don't end up in the call report, they don't end up in the floor funds. Let me see if I can do a decent job here. If you see the blue columns, that's the pledge collateral received by the large banks who have a global footprint in the pledge collateral market. When I mean the pledge collateral market, this is the market where a large bank picks up collateral with title transfer. Some of this was covered yesterday, but basically, I'm trying to put some flesh on the discussion, which was theoretical yesterday. These numbers are very large, but once a bank gets pledge collateral, it can reuse it in its own name. The sources for this collateral could be the repo market—it comes in as reverse-repo, it could be the sec lending market—it comes in as a sec borrow, from the old deliberative margins—when you are in the money, you pick up the collateral—or from a prime brokerage clients. So some laws are more amenable, and I'll talk about that. But, basically, these numbers are which allow you to reuse collateral, and these numbers are not on the balance sheet. Especially 2007 before Basel III, there was not much balance sheet constraint, so pledge collateral would flow off balance sheet. It did not have to come on balance sheet. Many of the new rules will force a lot of this to go on balance sheet. Not all, but there is good attempt that a large part of balance sheet will be used for this type of business.

Now, if you look Lehman—that is the second column—2007, that's about 800 billion [dollars] of pledged collateral received by Lehman to be reused in its own name. That's funding. Now, the last balance sheet of Lehman was 700 billion [dollars] assets, 700 billion [dollars] liabilities. Clearly, these numbers are big. Big enough that when you add them—or add up the 10 to 15 key players in this market—they resonate with money metrics. So, for example, if you add the blue bars of the key U.S. banks and the blue bars for the key European banks in 2007—I mean, UBS was way out of kilter. UBS model has changed significantly, but if you do add both of them, the U.S. and the European, that's 10 trillion [dollars]. Now the M2 metric—and not all M2 flows through Wall Street—the M2 metric for U.S. was about 7.5 trillion [dollars] in 2007, the European numbers are about 8.5 trillion [dollars] in 2007. A 10 trillion [dollar] market is something to think about.

Now, what has happened since then? Again, just to clarify, these blue bars or this pledge collateral is the fair market value of collateral you receive. It doesn't have to be triple A. There is a whole literature coming out on triple A and safe assets, and demand as a public good to supply safe assets. I'm not saying there is no triple A here. There's triple A, there are equities over here. Whatever is liquid can be priced and acceptable delivery, it will move. Now a large part of this is high-quality liquid assets—or what I call good collateral—but it's not all AAA. Now, if you look at this market since 2007, the numbers have come down, and they are not coming back. So if you look at the outermost column for each bank—Morgan Stanley, Goldman, the combined BofA, the pink bars at the very extreme, and if you add the Europeans, the pink bar, it's about 5.6 trillion. The numbers have not picked up in the last three to four years. Regulation has to do with some of this. Regulation is still not fully on board. Many of the Basel III regs will be enforced in 2018, 2019, but I believe QE also—when you take off triple As from the market, it does have an impact because to move a pledged collateral portfolio, having triple As in the portfolio helps in pricing, helps in the movement of this. And if you have less triple A relative to all others, it makes collateral movement relatively more difficult.

So that's a market which has come down to just about over half the size, and this is a market which had nothing to do with tri-party repo, but this is a genuine, bilateral collateral market. Not repo market—this is bigger than the repo market, but with title transfer, once the collateral is pooled with the large bank it moves, and these numbers are a lot less. Now whether this is good or bad, I'll talk about it. But just to show that these numbers are not coming back, and I believe given that the rules will be more enforced, and some countries are still doing QE, I believe that these numbers are unlikely to come back in the near future until the day we see the Fed and some other central banks unwind the balance sheet. And I don't know much about the regulations, but it is unlikely that some of the leverage ratios and SLRs are going to be compromised in the near future.

OK, where does this collateral come from? What is the biggest pocket? What are the bigger pools which supply it? And I will give you a short summary. Hedge funds are about half this market; they provide half this type of collateral because they churn a lot. They do trades and they have portfolios; they move on, they need financing. And an interesting aspect, which was touched on yesterday, was that English law—or outside New York law—is more friendly towards collateral reuse. New York has regulations. If you know the SEC, Rule 15-3C doesn't let you fund your books off client collateral—there is cap. You also have Regulation T. You don't have these issues in London. That means it is a little more friendly towards leverage. Many more leveraged transactions can happen in English law, which is, in my opinion, a more mark-to-market-based collateral. They take the risk, but there is also collateral friendly. Out here you might be willing to put your collateral on the table, but it will not be used beyond what's allowed by the regulation.

I have some numbers—I mean, this is not science, but the hedge fund supplied roughly around the time of Lehman about 1.7 trillion in collateral if you look down at the prime brokerage sources and the repo sources, you have to adjust for the leverage in this business. FSA [Financial Services Authority] used to, and now FCA [Financial Conduct Authority] comes up with pretty good statistics on how the hedge fund industry looks like, and you can extrapolate a lot from that. You also look at how the market has rebounded—assets under management are up to 3 trillion at the end of 2015. Leverage is not the same level as 2007, but in general hedge funds are suppling roughly the same amount of collateral as it did in 2007. However, the other pocket which supplies title transfer collateral, which I would call sovereign wealth fund, central banks, insurers, pension funds; that has not picked-up. That was also about 1.7 trillion in 2007. Those numbers have been flat around a trillion, inching towards 1.1 trillion, and this has to do with balance sheet space. So I will try to talk a little bit about that.

This is the sec lending business not bouncing back, and if you see the last three bullets—balance sheet space—and the opportunity to invest cash from sec lending in a zero-rate environment is not that impressive; the incentives are different.

So I put some numbers together. I haven't seen anybody else do that, although a recent paper is coming up on these lines. I will make a remark, although it is not public. In 2007, if you look at the first row, hedge funds provided some collateral, and the other bucket did. Roughly the amount of collateral coming into the banks, if you see the first circle, that was 10 trillion. Well, the collateral needs to be connected to the supply source and the demand source, and the banks have a global footprint. So, for example, if a hedge fund in Hong Kong has Indonesian bonds, which is in demand with, say, a Chilean pension fund in Santiago, UBS is a prime broker for this hedge fund. Hypothetically, UBS books everything out of London, Citibank knows that someone there through the New York office has a demand for a client in Santiago. So it takes a few steps for collateral to move. If a collateral is not very liquid, etc., it may take a few more steps, but on average, I found that the reuse rate—what I call velocity—is about 3. This is not much different than some of the money metrics we are aware of, which we have seen a lot. If you look at the St. Louis website for the FRED, they still publish M1, M2, M3.

Now what's happened? Fast forward six, seven years—in 2015, the reuse rate is down to about 1.9/1.8. The numbers haven't changed much from 2014, but I will make a few comments. The volume in this market is about 5.6, which is lower than 2014's, and basically hedge fund is supplying the collateral, but the balance sheet is not elastic enough to move it. The balance sheet space is constraining it and also, as I said, when the Treasurys and the bonds get siloed, it makes it difficult to move. And some of the markets are not unwinding, the gilts are unlikely to unwind. FRED's timetable of unwind is not really there, it's mostly reinvestment. So those things do impact the lubrication in this market. And this is the bilateral collateral market. This is very unlike triparty repo—a lot of research here is done using triparty repo data, which in my opinion is a market for funding for the collateral which did not get a bid during the day. Now, if the plumbing works—if this market is deep and liquid—you will get approximate prices in the triparty repo. But if this market is not liquid and this market keeps contracting, then the signals are not that liquid to be extrapolated.

This is a typical collateral chain. We talked about it, and so I will skip it. But once you receive title transfer, you move it. You can debit or credit the way you want to—you use that. So in this type of thinking, the demand for collateral is equal to supply times the reuse, and reuse is falling down. Hence, many pockets and academic literature are asking for more supply from some fiduciary agent who can do this as a public good. But as you all know, the U.S. Treasury and the Treasury borrowing committee has to issue in regular and predictable intervals at least cost. So, it's not very easy to just supply this as a public good. But the reuse rate is important. I think yesterday the words were mentioned "bond heaven." A lot of collateral goes in and never comes out. The Reserve Bank of India picks up, say, a Treasury, picks up two coupons a year and that's it. It doesn't see the light of the day until maturity. Many central banks do not reuse collateral. Many other pockets do not reuse collateral.

There's another aspect, which is—don't put this in Pigeonhole, but think about it—this is a question from my side. The policy for money has been ultra-loose. We have seen QE in almost every aspect. Japan and ECB are still doing it. On the other hand, the regulatory side is very tight on lubrication so when you look at the regulations they do not like long collateral chains. There are things like no rehypothecation of initial margin. So on one hand you have a desire for money to go out, and on the other side, something very akin to money, as I call both are HQLA [high-quality liquid assets]—a large part of this market is high-quality liquid asset—is restricted, and this is something to think about because you have lubrication in the markets from both cash and collateral. One is abundantly available, shows up as excess reserves, is HQLA, while the other—also HQLA—is not looked upon as favorably.

Just a snapshot of something which was said by my panel before—that deposits in the banking system at about 3.9 trillion...I'm sorry, they are 1.9 trillion, they have doubled now to 3.9 with the top four banks. With the top 50 banks, again, the deposits have doubled. This has been a very different last six to seven years relative to the growth of deposit, which would have been without QE, because the printing of money and the buying of—that has put a structural change in the amount of deposits which needs a home.

What was the world before QE, excess reserves, etc.? You look at the blue figures, you have social money markets, which supplies the bulk of the money, and you have hedge fund supply and collateral and some of the nonhedge fund to the custodians, they supply collateral. And this market was about 10 trillion, as I mentioned. These are the blue boxes before the central bank, QE, etc. I'm trying to give a global snapshot because Europe is very similar to the U.S., so is the UK. The JGB [Japan government bond] market—I don't talk much about because it's mostly a Japan-specific issue. Not much of JGB goes outside of Japan.

Now fast forward six to seven years, we have a lot of changes over here—especially, you know, the Fed has a reverse repo program started in September 2013 and has kept up with the Fed's lift-off. So today, you would have money market funds having direct access to the Fed. With the lift-off, the rate is 25 bps [basis points]. GSEs and some of those who are eligible counterparts can short-circuit the whole plumbing. Here it is where you have to earn through the blue squares, you have to earn the repo or to whatever mechanisms you park your money somewhere. But if the money short-circuits the plumbing either through the RRP [recommended retail price] there are some other programs. More recently the CCPs [central counterparty clearing] can deposit money directly with the Fed. There's a central bank, and international financial institutions also can deposit money directly with the Fed. But if a lot of these money pipes short-circuit the plumbing and go directly to the Fed, which is possible—I just gave you three examples—then what will fund the plumbing? The plumbing, as I said, is down from 10 trillion to 5.5 trillion. It does give useful market signals. And if more and more pipes go directly to the central bank, you do get a little bit of what I call "the rusting of the plumbing." This is something which I think should be looked at going down the road.

This is the change in the balance sheet. Basically, asset purchases, the Treasurys, the MBS [mortgage-backed securities] when they were about 3.4. This is just the change, and it shows up mostly as excess reserves on the liability side. A remark I would like to make is this excess reserves and the reserve repo program. Basically, the reserve repo program doesn't release collateral to the market. It stays within the triparty repo framework.

Let me see if the next slide—here is the picture of the fed funds rate and the GCF [general collateral financing]. I would say that GCF has more information content. It's still from the market side to the extent plumbing works. Fed funds, as people have discussed before, is mostly the Fed's RRP [reverse repo program]. But I think these two—the wedge between these two was never much. It was basically plus or minus 3 basis points. The sum-up book at the New York Fed would make sure that the gap wasn't much, but this is something to sort of look upon as the Fed lifts off into higher and higher rates. The plumbing works, this is the bund GC [general collateral] without EONIA, which is the counterpart at the ECB. Below zero also, the relationship holds.

Now, this is something which if I have time I can show. Let's say we get down to the middle of December—if you are at eight today and this is where you have a lift-off with a very large RRP, which actually happened, the reverse repo program given the sum-up book is up to two trillion. Now, if you hold the collateral—because as I said, the reverse repo program doesn't allow the collateral to be released to the market—you get it in name, but the market cannot slice and dice it. So there is no velocity with the RRP. It stays within the triparty repo structure. Now, what this does is you want more money so there is less money in the market domain, but you don't release collateral. So in relative terms the price of money is more than the price of collateral, collateral up to B. So, in effect, by having a large RRP you can control the movement of GC to some extent as you lift off from A to B. Similarly, you could do this with some sales. You may not reach B, but this basically means that the market can slice and dice it, but if a 10-year goes out, slicing and dicing means there is some escape of collateral velocity. You may not exactly have the fed funds rate at B because here there are some market signals over here. But think about what happened—let's say, in the first part of January. A lot of markets, sales saw from emerging markets—China, and some of the Middle East—and that shows up in some of the GC repo. So, GC repo will always reflect things beyond what the central banks do. If there are significant sales in the market, it will be picked up in one way or the other. Thank you.

Mester: Anyway, so what people want to know is, they understand that asset purchases take collateral out of the market, but they are also interested in knowing whether low interest rates and negative interest rates also can contribute to the decline in market liquidity. They want to know whether given that—Ulrich, you said that it doesn't seem to be that the negative interest rates in Europe have had much adverse effects so far. How low should they go, and would that work in the U.S.? And then there is another question about—can you talk a little bit about what the effect of zero and low interest rates have been on the wealth of savers? When we think of income distributional sects, how should we think about that? So, again, a number of questions on negative interest rates or very low interest rates. And I think these are for anyone on the panel here.

Bindseil: So are low rates negative for trading activities? Maybe it's not so easily seen in the data because the time series...let's say the move to lower rates, of course, correlated with the crisis context, and so I wouldn't be able to easily disentangle what can be attributed to low rates. My intuition would be—but I'm not so close in markets, of course—that it should not make a big difference where your absolute level of rates is for trading opportunities. That would be my intuition.

How low can interest rates go? OK, that's a big, big topic. It is clear that—let's say institutional constraints, legal constraints kick in, and one important one is that banks are issuing liabilities close to money, so retail-based deposit funding is a bit like a central bank. If you are close to zero interest rates, your profitability is low. Central banks' profitability relies on positive interest rates, so let's say we easily understand the argument that a bank which has a business model to issue deposits to households at something close to money will suffer in terms of its business model. And if, let's say, the inability of it to impose negative rates on depositors is an absolute one, then this is relevant for the transmission mechanism of going lower. It is already a relevant effect if you move from 3 percent to 1 percent, but if you move further it continues. And in a bank-based system like the euro area, has one it is something that you must take into account in your, let's say, modeling of the transmission mechanism. But the conclusion of the ECB has been so far also looking at bank profitability effects that we are not yet there at the level of the rate where effects would reverse. And so far, all the steps, the easing measures, have been transmitted to a lowering of bank lending rates. But of course, to extrapolate that and to say at minus 1 percent exactly we will have the reverse is very difficult.

And on ways of savers, that's a popular topic in Germany, in particular, where media like to tell us that savers are expropriated. I think it's mixing up different things. I think this comes mainly from retail-based banks. This argument—the saver who is very unsophisticated and is really unable to diversify its ways is typically one who holds, maybe, a large part in the form of deposits. And I would say the opportunity costs of holding deposits are particularly low those days. In normal times you have higher opportunity costs. So this whole expropriation of the saver story—I don't believe it particularly.

Goodfriend: I'll give a sort of broader context for this. On the saver issue, I like to go back to more fundamental things and say if we were back in the good old days and governments had reasonable fiscal policy and some price stability—say, under the gold standard, the long-term government bond rates would fluctuate, but they would have roughly a 3 percent real interest rate. And I'm talking about the period in the '50s in the United States to the mid-'60s, and in Great Britain through a lot of the 19th century. You had good fiscal policy and price stability. And so savers could hold long-term government bonds and they could hold with little spreads the longer-term bonds of well-functioning companies, and so those rates would be roughly stable. And so people wouldn't be thrown into interest rate risk scenarios because of the inflation risk in the long-term bond.

I like to tell this story because my family was a victim of this, and many of you probably. My father retired at the worst time you could possibly retire, in 1984, and he saved most of his money in the '70s, and by the time he died my mother thought that he had another family because he had all his money in government bonds, and the nominal value of government bonds was such that his savings were wiped out by inflation over a decade. So what did my mother do? She learned from her investment adviser: don't hold long-term bonds. And what did she do? She held short-term bonds. She's a saver who was creamed by interest rate risk. So this issue of saving is misplaced. The reason it's so hard for savers to get protection is because governments have been irresponsible about either their monetary systems or other things, and I think we need to start there instead of kind of criticizing where we got in the moment, which is kind of a mish-mash of irresponsible longer run government management of fiscal policy, as well as inflation policy.

Singh: I would just like to make a comment that the low interest rate environment is not the deterrent to collateral and collateral move. It's a very small margin game. So basically, you need high volume in this market, and since balance sheets are no longer elastic, there is a rationing of the markets. So the hedge funds, generally, are the top clients of these banks, and they have a bid at the balance sheet first. The others, they get a second shot at the balance sheet. So I don't see the sec lending market even coming back because you are deprived the opportunity from the balance sheet space. Hedge funds do. So it's not surprising hedge fund numbers are back to what they were and they do have access to those, to the JPs and Barclays, and the others don't. But I think you need volume for this market to come back. In a margin sense, many of these tickets are 5, 6, 7, 8 bps, they are very small margin numbers; it's just a spread. So whether it's 0, or 10, or 25 bps, you still make 5, 6, 7 bps.

Mester: Here's one. It's phrased in terms of the Fed, but I think it can be for all central banks. Will the Fed ever return to a pure monetary policy, or is it inextricably tied to using credit policy to a greater extent going forward? And if so, what's the implications for market functioning and liquidity and other aspects that we have talked about throughout the whole conference? And I'm making you all answer that one.

Goodfriend: I'll start. So these days the question is about, I think, mortgage-backed securities, and that's the major credit policy the Fed is involved in, as I understand it. I haven't looked at the balance sheet lately, but I think that's mainly it. I think they are not talking about emergency credit policy so much. It's a problem—there's a chicken-and-egg problem. To make it profitable for the markets to move outside the Fed, the Fed has to give ground. There's going to be a necessarily a stress point where costs go up, profit opportunities are there, the Fed has to step aside. It's very hard to get out of these markets when you've dominated them. So I can understand the Fed's point of view, I can understand the markets point of view. Give it up, let the market do it. I happen to favor giving it back to the market. I think the Fed could do that, but will it happen? I can't say.

Bindseil: The question says "ever" return, so I would say the transition is an issue, but the long-term objective should be a normalization of the monetary policy framework. Of course, everything may depend on secular stagnation—do we have a long-term low real interest rate issue? And if we have, are central banks bumping on the zero lower bound all the time and need very often to reinject monetary accommodation with nonconventional measures? I mean, hopefully not. And if not—if we have normal growth, normal rates, then we should return to an absolutely normal precrisis policy.

Singh: I will just add that when you look at the roles so far for good measures, QE, etc., but down the road, I think, having the plumbing do as bare minimal as important. Those signals are important. Out here we have flowed the repo, first at 3, 4, 5 bps now at 25 bps, but I think those signals are very useful, especially as the market feels that the Fed funds is the Fed RRP rate, etc. So you need to get signals, and I think the less interference with the plumbing will be conducive down the road.

Mester: Looking at the Fed like a regulated bank, it looks like giant carry-trade; long-term assets funded overnight. Should we care about that risk?

Goodfriend: So yeah, the Fed is essentially running a bond market carry-trade. It's got—the last time I looked—a couple of years ago, we had ten-year average maturity assets and funding on overnight money taking subtracting the currency, which is free money for the Fed. So I think, whatever you think about this policy's effectiveness, it has two kinds of implications. One is, if you think it's effective, it suggests to me that the Fed would have to raise short-term rates higher than otherwise in its own history when it hadn't had a balance sheet this big, if you think it has any effect at all. And so that effect would have to taken into account in models of where so-called natural rate has to go in order to stabilize the system out five years or so—or ten years, if the Fed holds its portfolio. If you don't think the policy was effective—well, that puts you in another camp. And then the third thing I would say is, whatever you think about the policy, the carry-trade exposes the Fed to expected costs because it apparently wants to hold these things out to maturity. Well, then it's already locked in its interest rates. Every time it buys a long-term bond it's got the coupons, it got the price it paid. It knows for the next 20 years what the rate of interest is on that long-term stuff. It's got to worry about carrying that trade with short-term interest on reserves. The irony for me is the Fed's best argument for "don't worry about the costs" is we have a tremendously pessimistic future ahead of us. Interest rates are going to stay below 2.5 percent, which is roughly what the Fed's yield is on its longs, as far as the eye can see. But I don't see the Fed ever making that case in public.

The worst case for the Fed is, the Fed says, "Well, we're getting out of this thing and we need to raise rates and we might have to at some point raise them considerably, and that causes negative interest income, and we might have to take some money that otherwise would go to the Treasury to fund our position." I guess I would—and I'm glad this question came because I didn't bring this up, but I do think transparency in what the Fed is doing would help. The Fed in any case should come clean about the fiscal situation on its balance sheet. And let me make one other point. The Fed had turned over $600 billion on the front end of its carry trade as profit to the Treasury over the past seven years. Well, I don't think any bank would turnover as profit unencumbered profit of a trade that's not unwound yet, and the Fed expects to take this trade for another ten years. At best it should have retained these earnings, or at least some of these earnings. It retained not a dime of these earnings. I would have felt better if the Fed had come clean about the trade and retained something, but it didn't do it. And by the way, if you follow this stuff in detail, on the transportation bill that just was passed sometime in November or December, the Congress has capped the surplus that the Fed is independently able to retain at $10 billion against the balance sheet. The last I saw was $4.5 trillion. So, so much for independent central banking.

Bindseil: If the question is, should we care about that risk ex ante before deciding on measures, I mean, obviously we should not, because the nightmare of the central bank in terms of profit is a long-term, let's say, zero lower bound situation, like in Japan, basically. And the horizon of that almost absorbing state can be way beyond what you have as average maturity in an asset-purchase program. Therefore, getting your program effective, lift-off from the zero low bound makes always more sense from even the point of view of central bank profitability, and therefore clearly we should not care about this risk.

Singh: I think I agree with Ulrich. Should we care? No. And I think the key word over there—if rates rise enough. I mean, that's very hypothetical. If you look at Japan, I don't know where the advanced economies are heading, so this issue is too premature for this question.