Friday, June 27, 2014

Where Do Interest Rates Come From?

What determines the level of interest rates? It seems like a simple question, but I don't think economics -- orthodox or heterodox -- has an adequate answer.

One problem is that there are many different interest rates. So we have two questions: What determines the overall level of interest rates, and what determines the spreads between different interest rates? The latter in turn we can divide into the question of differences in rates between otherwise similar loans of different lengths (term spreads), differences in rates between otherwise similar loans denominated in different currencies, and all the remaining differences, grouped together under the possibly misleading name risk spreads.

In any case, economic theory offers various answers:

1. The orthodox answer, going back to the 18th century, is that the interest rate is a price that equates the desire to save with the desire to borrow. As reformulated in the later 19th century by Bohm-Bawerk, Cassel, etc., that means: The interest rate is the price of goods today relative to goods tomorrow. The interest rate is the price that balances the gains from deferring consumption with our willingness to do so. People generally prefer consumption today to consumption in the future, and because it will be possible to produce more in the future than today, so the interest rate is (normally) positive. This is a theory of all transactions that exchange spending in one period for spending (or income) in another, not specifically a theory of the interest rate on loans.

The Wicksell variant of this, which is today's central-bank orthodoxy, is that there is a well-defined natural interest rate in this sense but that for some reason markets get this one price wrong.

2. An equally old idea is that the interest rate is the price of money. In Hume's writings on money and interest, for instance, he vacillates between this and the previous story. It's not a popular view in the economics profession but it's well-represented in the business world and among populists and monetary reformers,. In this view, money is just another input to the production process, and the interest rate is its price. A creditor, in this view, isn't someone deferring consumption to the future, but someone who -- like a landlord -- receives an income thanks to control of a necessary component of the production process. A business, let's say, that needs to maintain a certain amount of working capital in the form of money or similarly liquid assets, may need to finance it with a loan on which it pays interest. Interest payments are in effect the rental price of money, set by supply and demand like anything else. As I say, this has never been a respectable view in economic theory, but you can find it in more empirical work, like this paper by Gabriel Chodorow-Reich, where credit is described in exactly these terms as an input to current production.

3. Keynes' liquidity-preference story in The General Theory. Here again the interest rate is the price of money. But now instead of asking how much the marginal business borrower will pay for the use of money, we ask how much the marginal wealth owner needs to be compensated to give up the liquidity of money for a less-liquid bond. The other side of the market is given by a fixed stock of bonds; evidently we are dealing with a short enough period that the flow of new borrowing can be ignored, and the bond stock treated as exogenously fixed. With no new borrowing, the link from the interest rate is liked to the real economy because it is used to discount the expected flow of profits from new investment -- not by business owners themselves, but by the stock market. It's an oddly convoluted story.

4. A more general liquidity-preference story. Jorg Bibow, in a couple of his essential articles on the Keynesian theory of liquidity preference, suggests that many of the odd features of the theory are due to Keynes' decision to drop the sophisticated analysis of the financial system from The Treatise on Money and replace it with an assumption of an exogenously fixed money stock. (It's striking that banks play no role in in the General Theory.) But I'm not sure how much simpler this "simplification" actually makes the story, or whether it is even logically coherent; and in any case it's clearly inapplicable to our modern world of bank-created credit money. In principle, it should be possible to tell a more general version of the liquidity preference story, where, instead of wealth holders balancing the income from holding a bond against the liquidity from holding "money," you have banks balancing net income against incremental illiquidity from simultaneously extending a loan and creating a deposit. I'm afraid to say I haven't read the Treatise, so I don't know how much you can find that story there. In any case it doesn't seem to have been developed systematically in later theories of endogenous money, which typically assume that the supply of credit is infinitely elastic except insofar as it's limited by regulation.

5. The interest rate is set by the central bank. This is the orthodox story when we turn to the macro textbook. It's also the story in most heterodox writers. From Wicksell onward, the whole discussion about interest rates in a macroeconomic context is about how the central bank can keep the interest rate at the level that keeps current expenditure at the appropriate level, and what happens if it fails to do so. It is sometimes suggested that the optimal or "natural" interest rate chosen by the central bank should be the the Walrasian intertemporal exchange rate -- explicitly by Hayek, Friedman and sometimes by New Keynesians like Michael Woodford, and more cautiously by Wicksell. But the question of how the central bank sets the interest rate tends to drop out of view. Formally, Woodford has the central bank set the interest rate by giving it a monopoly on lending and borrowing. This hardly describes real economies, of course, but Woodford insists that it doesn't matter since central banks could control the interest rate by standing ready to lend or borrow unlimited amounts at thresholds just above and below their target. The quite different procedures followed by real central banks are irrelevant. [1]

A variation of this (call it 5a) is where reserve requirements bind and the central bank sets the total quantity of bank credit or money. (In a world of bind reserve requirements, these will be equivalent.) In this case, the long rate is set by the demand for credit, given the policy-determined quantity. The interbank rate is then presumably bid up to the minimum spread banks are willing to lend at. In this setting causality runs from long rates to short rates, and short rates don't really matter.

6. The interest rate is set by convention. This is Keynes' other theory of the interest rate, also introduced in the General Theory but more fully developed in his 1937 article "Alternative Theories of the Rate of Interest." The idea here is that changes in interest rates imply inverse changes in the price of outstanding bonds. So from the lenders' point of view, the expected return on a loan includes not only the yield (as adjusted for default risk), but also the capital gain or loss that will result if interest rates change while the loan is still on their books. The longer the term of the loan, the larger these capital gains or losses will be. I've discussed this on the blog before and may come back to it in the future, but the essential point is that if people are very confident about the future value of long rates (or at least that they will not fall below some floor) then the current rate cannot get very far from that future expected rate, no matter what short rates are doing, because as the current long rate moves away from the expected long rate expected capital gains come to dominate the current yield. Take the extreme case of a perpetuity where market participants are sure that the rate will be 5% a year from now. Suppose the short rate is initially 5% also, and falls to 0. Then the rate on the perpetuity will fall to just under 4.8% and no lower, because at that rate the nearly 5% spread over the short rate just compensates market participants for the capital loss they expect when long rates return to their normal level. (Obviously, this is not consistent with rational expectations.) These kinds of self-stabilizing conventional expectations are the reason why, as Bibow puts it, "a liquidity trap ... may arise at any level of interest." A liquidity trap is an anti-bubble, if you like.

What do we think about these different stories?

I'm confident that the first story is wrong. There is no useful sense in which the interest rate on debt contracts -- either as set by markets or as target by the central bank -- is the price of goods today in terms of goods tomorrow. The attempt to understand interest rates in terms of the allocation across time of scarce means to alternative ends is a dead end. Some other intellectual baggage that should overboard with the "natural" rate of interest are the "real"rate of interest, the idea of consumption loans, and the intertemporal budget constraint.

But negative criticism of orthodoxy is too easy. The real work is to make a positive case for an alternative. I don't see a satisfactory one here.

The second and third stories depend on the existence of "money" as a distinct asset with a measurable, exogenously fixed quantity. This might be a usable assumption in some historical contexts -- or it might not -- but it clearly does not describe modern financial systems. Woodford is right about that.

The fifth story is clearly right with respect short rates, or at least it was until recently. But it's incomplete. As an empirical matter, it is true that interbank rates and similar short market rates closely follow the policy rate. The question is, why? The usual answer is that the central bank is the monopoly supplier of base money, and base money is used for settlement between banks. This may be so, but it doesn't have to be. Plenty of financial systems have existed without central banks, and banks still managed to make payments to each other somehow. And where central banks exist, they don't always have a monopoly on interbank settlement. During the 19th century, the primary tool of monetary policy at the Bank of England was the discount rate -- the discount off of face value that the bank would pay for eligible securities (usually trade credit). But if the discount rate was too high -- if the bank offered too little cash for securities -- private banks would stop discounting securities at the central bank, and instead find some other bank that was willing to give them cash on more favorable terms. This was the problem of "making bank rate effective," and it was a serious concern for 19th century central banks. If they tried to raise interest rates too high, they would "lose contact with the market" as banks simply went elsewhere for liquidity.

Obviously, this isn't a problem today -- when the Fed last raised policy rates in the mid-2000s, short market rates rose right along with it. Or more dramatically, Brazil's central bank held nominal interest rates around 20 percent for nearly a decade, while inflation averaged around 8 percent. [2] In cases like these, the central bank evidently is able to keep short rates high by limiting the supply of reserves. But why in that case doesn't the financial system develop private substitutes for reserves? Mervyn King blandly dismisses this question by saying that "it does not matter in principle whether the disequilibrium in the money market is an aggregate net shortage or a net surplus of funds—control of prices or quantities carries across irrespective of whether the central bank is the monopoly supplier or demander of its own liabilities." [3] Clearly, the central bank cannot be both the monopoly supplier and the monopoly demander of reserves, at least not if it wants to have any effect on the rest of the world. The relevant question -- to which King offers no answer -- is why there are no private substitutes for central bank reserves. Is it simply a matter of legal restrictions on interbank settlements using any other asset? But then why has this one regulatory barrier remained impassable while banks have tunneled through so many others? Anyway, going forward the question may be moot if reserves remain abundant, as they will if the Fed does not shrink its balance sheet back to pre-crisis levels. In that case, new tools will be required to make the policy rate effective.

The sixth story is the one I'm most certain of. First, because it can be stated precisely in terms of asset market equilibrium. Second, because it is consistent with what we see historically. Long term interest rates are quite stable over very long periods. Third, it's consistent with what market participants say: It's easy to find bond market participants saying that some rate is "too low" and won't continue, regardless of what the Fed might think. Last, but not least from my point of view, this view is clearly articulated by Keynes and by Post Keynesians like Bibow. But while I feel sure this is part of the story, it can't be the whole story. First, because even if a conventional level of interest rates is self-stabilizing in the long run, there are clearly forces of supply and demand in credit markets that push long rates away from this level in the short run. This is even more true if what convention sets is less a level of interest rates, than a floor. And second, because Keynes also says clearly that conventions can change, and in particular that a central bank that holds short rates outside the range bond markets consider reasonable for long enough, will be able to change the definition of reasonable. So that brings us back to the question of how it is that central banks are able to set short rates.

I think the fundamental answer lies behind door number 4. I think there should be a way of describing interest rates as the price of liquidity, where liquidity refers to the capacity to honor one's promises, and not just to some particular asset. In this sense, the scarce resource that interest is pricing is trust. And monetary policy then is at root indistinguishable from the lender of last resort function -- both are aspects of the central bank's role of standing in as guarantor for commitments within the financial system.  You can find elements of this view in the Keynesian literature, and in earlier writers going back to Thornton 200-plus years ago. But I haven't seen it stated systematically in way that I find satisfactory.

UPDATE: For some reason I brought up the idea of the interest rate as the price of money without mentioning the classic statement of this view by Walter Bagehot. Bagehot uses the term "price of money" or "value of money" interchangeably with "discount rate" as synonyms for the interest rate. The discussion in chapter 5 of Lombard Street is worth quoting at length:
Many persons believe that the Bank of England has some peculiar power of fixing the value of money. They see that the Bank of England varies its minimum rate of discount from time to time, and that, more or less, all other banks follow its lead, and charge much as it charges; and they are puzzled why this should be. 'Money,' as economists teach, 'is a commodity, and only a commodity;' why then, it is asked, is its value fixed in so odd a way, and not the way in which the value of all other commodities is fixed? 
There is at bottom, however, no difficulty in the matter. The value of money is settled, like that of all other commodities, by supply and demand... A very considerable holder of an article may, for a time, vitally affect its value if he lay down the minimum price which he will take, and obstinately adhere to it. This is the way in which the value of money in Lombard Street is settled. The Bank of England used to be a predominant, and is still a most important, dealer in money. It lays down the least price at which alone it will dispose of its stock, and this, for the most part, enables other dealers to obtain that price, or something near it. ... 
There is, therefore, no ground for believing, as is so common, that the value of money is settled by different causes than those which affect the value of other commodities, or that the Bank of England has any despotism in that matter. It has the power of a large holder of money, and no more. Even formerly, when its monetary powers were greater and its rivals weaker, it had no absolute control. It was simply a large corporate dealer, making bids and much influencing—though in no sense compelling—other dealers thereby. 
But though the value of money is not settled in an exceptional way, there is nevertheless a peculiarity about it, as there is about many articles. It is a commodity subject to great fluctuations of value, and those fluctuations are easily produced by a slight excess or a slight deficiency of quantity. Up to a certain point money is a necessity. If a merchant has acceptances to meet to-morrow, money he must and will find today at some price or other. And it is this urgent need of the whole body of merchants which runs up the value of money so wildly and to such a height in a great panic.... 
If money were all held by the owners of it, or by banks which did not pay an interest for it, the value of money might not fall so fast. ... The possessors would be under no necessity to employ it all; they might employ part at a high rate rather than all at a low rate. But in Lombard Street money is very largely held by those who do pay an interest for it, and such persons must employ it all, or almost all, for they have much to pay out with one hand, and unless they receive much with the other they will be ruined. Such persons do not so much care what is the rate of interest at which they employ their money: they can reduce the interest they pay in proportion to that which they can make. The vital point to them is to employ it at some rate... 
The fluctuations in the value of money are therefore greater than those on the value of most other commodities. At times there is an excessive pressure to borrow it, and at times an excessive pressure to lend it, and so the price is forced up and down.
The relevant point in this context is the explicit statement that the interest, or discount, rate is set by the supply and demand for money. But there are a couple other noteworthy things. First, the concept of supply and demand is one of monopolistic competition, in which lenders are not price takers, but actively trade off markup against market share. And second, that the demand for money (i.e. credit) is highly inelastic because money is needed not only or mainly to purchase goods and services, but first and foremost to meet contractual money commitments.

[1] See Perry Mehrling's useful review. Most of the text of Woodford's textbook can be downloaded for free here. The introduction is nontechnical and is fascinating reading if you're interested in this stuff.

[2] Which is sort of a problem for Noah Smith's neo-Fisherite view.

[3] in the same speech, King observes that "During the 19th century, the Bank of England devoted considerable attention to making bank rate ‘effective’." His implication is that central banks have always been able to control interest rates. But this is somewhat misleading, from my point of view: the Bank devoted so much attention to making its rate "effective" precisely because of the occasions when it failed to do so.


  1. But why in that case doesn't the financial system develop private substitutes for reserves?

    It may be the "tax" factor. You can't pay taxation in anything other than legal currency (digital or real), and that limits the ability of private substitutes in terms of usefulness unless you somehow create an entirely separate economy using the alternatives which can't be pursued by the authorities.

  2. Chartalism is so hot right now. But no, that's not it.

  3. Great post.

    I'd agree that most of the orthodox and heterodox stuff on interest rates is muddled. It doesn't help that there are different interest rates but people are often not clear which they are talking about. For example, the interest rate in your story (1) must be a real interest rate and, as I understand it at least, it must be a rate that is stripped of any liquidity or credit factors, so probably something that's close to the overnight wholesale rate. This is obviously a very different thing to the interest rate that appears in liquidity preference stories, which is really more of a differential between the rate on less liquid bonds and the rate on money (which need not be zero).

    The only work I know that comes close to setting this out in a coherent way is Godley & Lavoie's Monetary Economics, although in my opinion it could do with new chapters on things like repo to bring it more up-to-date.

    The use of central bank liabilities for clearing is not an essential element of its ability to control rates if it runs a corridor system. If the central bank is paying 20% on its liabilities, why would any bank lend on the interbank market for less?

    1. Thanks for the comment.

      The interest rate in your story (1) must be a real interest rate and, as I understand it at least, it must be a rate that is stripped of any liquidity or credit factors, so probably something that's close to the overnight wholesale rate.

      It's even worse than that. The "interest rate" in standard models is the price of ANY transaction that exchanges spending in the present for goods or income in the future. The ratio of a house price to its rent, the deflation rate (if anyone voluntarily holds money), the ratio of college tuition to the college wage premium, the return on investment of any kind -- all are equally "the interest rate" of orthodox theory. If these are different from each other, there's really no basis for saying which is the "true" intertemporal price.

      This is obviously a very different thing to the interest rate that appears in liquidity preference stories, which is really more of a differential between the rate on less liquid bonds and the rate on money

      Right. Keynes would say (and I would agree) that this latter thing is the only relevant interest rate for positive economics. Of course there are a bunch of these too.

      Godley & Lavoie's Monetary Economics

      Yes, Godley is definitely on the short list of economists who got this right, from my point of view. I'm sorry to say I haven't looked at the book -- I should.

      The use of central bank liabilities for clearing is not an essential element of its ability to control rates if it runs a corridor system. If the central bank is paying 20% on its liabilities, why would any bank lend on the interbank market for less?

      This is the big question. I I don't have a well worked out answer, but here are a couple reasons for doubt:

      1. Anybody can offer an arbitrarily high own-rate of interest on their own liabilities. In particular, the world is full of central banks, many of which also operate corridor systems. Since interest rates don't all converge to the highest central bank rate (and UIP does not hold), a particular central bank's liabilities must play some special role for its own domestic financial system.

      2. Let's imagine a single bank. Suppose the central bank announces its willingness to pay an interest rate on reserves that is greater than the market rate. With only one bank there's no question of needing reserves for clearing, and let's assume there are no reserve requirements. So the banking system will lend all its reserves to the central bank. Now what, if someone is seeking a loan. For the bank, the relevant question is the interest rate they will pay (net of defaults, transaction costs, etc.) relative to the rate the bank must pay on deposits. The latter in turn depends on advantages of bank liabilities for making payments relative to the next best alternative. How is the fact that the bank also happens to be getting 20% on its reserves relevant to this calculation? As far as I can tell, it is not.

      Both of these points go to the larger issue that you need some positive story for why the rest of the world cares about central bank liabilities. Otherwise, you end up with Mervyn King's unintentional reductio ad absurdum, where the central bank has perfect control over the price at which it trades with itself.

    2. If there's onl a single commercial bank, then it may under certain assumptions be able to trump any central bank action. More generally, where there's a number of banks, then if the central bank raises the rate it pays on reserves, commercial banks will raise their deposit rates in step. There is relatively little cost in taking in wholesale deposits and holding the funds as reserves, so banks will compete for deposits to the point where wholesale rates are only slightly below the reserves rate. Retail deposit rates differ from wholesale rates due to operational costs and funding stability, but that's all just differential, so other things being equal, you'd expect a 1% rise in the IOR rate to lead to a 1% rise in deposit rates.

      Obviously anyone can offer to pay an excessive interest rate. However, if I do so, and assuming I'm good for it, then the price of my liabilities will be bid up, since no-one is offering par conversion into my liabilities. Commercial banks do offer par conversion, so if one decided to set its rates at an excessive level, then it would find itself with a much greater level of deposits which it would be unable to re-invest at a comparable rate. It would rapidly go out of business. The central bank does not face this problem because it does not accept deposits in the form of claims on other banks. (This is the point that Nick Rowe calls asymmetric redeemability.)

      In your reference to foreign central banks, did you mean whether changes in the ECB's rate for euro, say, might affect the interest rate at which US dollars trade? This is the par conversion point again. It would require that someone was committing to convert between the two currencies at par (or at least at a fixed rate) in unlimited size for the term of any deposit.

    3. So it seems like the control of the central bank over short rates depends on two factors: first, the legal obligation of commercial banks to redeem deposits in reserves at par. In other words, if a depositor at Bank A makes a payment to a depositor at Bank B, Bank A is legally required to transfer reserves with an equal face value to Bank B, and Bank B is legally required to accept them, regardless of prevailing interest rates. Right?

      So that's step one. And it's important to be clear it's the result of a specific set of regulations, it's not a generic feature of markets. If my gas company, say, decides they would prefer to collect from you instead (perhaps not a bad idea from their point of view), they can't unilaterally transfer the obligation. You and I have to mutually agree on an asset I will give you to take over the bill.

      Second, payments (and intermediation) have to stay within the banking system. The higher you raise IOR, the more arbitrage opportunities you create for units that cannot hold reserves but can expand their balance sheets. For example, businesses might increase trade credit to liquidity-constrained buyers. (There is evidence for this during the 2008-2009 crisis.) More dramatically, firms might pay workers or suppliers in kind, or in effect reinvent bills of exchange by accepting payment in endorsed checks from third parties, as happened in Russia during shock therapy. In the extreme case, we can imagine all the cash in the economy sitting in deposits collecting very high returns, and no loans being taken from banks. While meanwhile payment and lending happened in the real economy using some entirely different medium. I don't mean to suggest that's likely to happen in conditions less extreme than 1990s Russia -- again, the case of Brazil is instructive. But I think it's worth thinking carefully about why it doesn't.

      (One other amendment I would make to what you wrote -- to the extent that banks have monopoly power vis a vis their retail customers, they can pay them less than the policy rate on deposits, and so might be willing to lend to them at lower rates also, especially if the loans might be used to make payments to other retail customers of the same bank. This is another way that I would think very high rates -- equivalent to very scarce reserves -- would encourage a shift away from the centralized payments system.)

      It would require that someone was committing to convert between the two currencies at par (or at least at a fixed rate) in unlimited size for the term of any deposit.

      So in your opinion, why doesn't interest parity hold?

    4. On the whole I'd agree with that.

      Regarding disintermediation in the crisis, I think this has more to do with other issues. It's partly because cash-rich entities were nervous about depositing with banks. But more importantly, banks were needing to reduce assets due to sudden changes in their capital position, so where they were lending at all, it was only at high margins. So the cost of intermediation was much higher.

      Absence credit issues, if you can get nearly 20% just putting money on deposit, why get involved in advancing credit? Non-banks can't hold reserves, but there are plenty of banks who will take their deposit to fund a reserve holding for a tiny spread.

      On interest parity, I tend to see it like this. A dollar based investor might compare two returns - the dollar deposit rate and the dollar equivalent of the euro deposit rate. The latter is the euro rate adjusted for the expected exchange rate movement over the term. The latter also has more risk to the dollar based investor, due to the exchange rate not moving as expected. So the dollar investor needs a greater expected return to tempt him out of dollars into euro. And the more capital flow is needed from dollar based investors to euro based borrowers, the greater that expected return needs to be. (And, of course, you also need to take into account euro based investors holding dollar assets). But with the Fed and the ECB each fixing their respective domestic rates, it's the expectation of exchange rate movements that has to move to make this happen.

    5. The broader issue here is, why does anyone outside the banking system hold bank liabilities? I don't think it's logically possible to use any account of equivrium in an asset market to explain why ownership of that asset gives a claim on real goods and services.

      (And I don't think you would say this, but the naive answer that the state mandates the use of its currency will work. After all, there are lots of purchases that you cannot make in cash.)

    6. I presume you mean why anyone would hold any claim denominated in a abstract unit (i.e. dollars) rather than in real goods, not just liabilities of banks? And I agree, that is a bigger issue. I have my own thoughts but there not conclusive.

      It's interesting about the uses of cash, isn't it. Personally, I think I make many more payments where I can't use cash than I do where I can't use bank transfer or credit card. In fact, the only reason I would accept cash in payment for anything is that I know the bank will accept it at par value in deposit. This rather turns on its head the idea that people will hold a deposit only because they know they can withdraw the cash. It also suggests that cash may be less liquid than bank deposits.

  4. I’ve had experience with interest rates, sometimes in real life, sometimes in fantasies.

    1. I recently came into a little money, so I paid off my credit card balance. It was $3500 at 11.99 percent APR, so I eliminated about a $40 monthly payment. I used the extra in my budget to switch from generic cornflakes to Kelloggs cornflakes. (But although Kelloggs tastes better, it’s not really better enough to outweigh the anxiety about overspending, so I’ve switched back to generic.) I actually kept $100 on my balance because I’m worried that if I don’t keep paying a trickle of interest to Chase Manhattan they will cut off my credit card.

    2. With the new money I wanted to move up from no-interest checking to an interest-paying checking or savings account. But a savings account, even like a thousand-year CD, pays 0.5 percent now, and a so-called interest-bearing checking account pays all of 0.05 percent, so I didn’t bother. I thought, Geez, what’s the point, it’s not even worth an hour talking to the bank rep.

    3. I play the lottery occasionally. Not PowerBall with the crazy $350 million payouts—I’m not greedy that way—just Lotto, which will get up to a $3-20 million prize, less if you split the pot. The way I used to see it, if I won just $1 million, I could put it in the bank at 5 percent and live off the interest—50 grand a year, which is a lot. Then I could relax, like a drowning man who finally reaches shore, and just sit on the rocks and watch the other poor bastards still struggling in the waves. Maybe throw a line every now and then.

    But with interest rates so low that’s just a pipe dream. I’d have to keep dipping into capital and watch helplessly while it washed away. To get a better return I’d have to become one of those people who buy stocks and spend their time managing their portfolios, which I would probably make a hash of and end up broke.

    So I may start playing PowerBall after all, or MegaBall.

    1. Yes, the dream of living off interest may be obsolete. Keynes:

      "I should guess that a properly run community equipped with modern technical resources ... ought to be able to bring down the marginal efficiency of capital in equilibrium approximately to zero within a single generation; so that we should attain the conditions of a quasi-stationary community... If I am right in supposing it to be comparatively easy to make capital-goods so abundant that the marginal efficiency of capital is zero, this may be the most sensible way of gradually getting rid of many of the objectionable features of capitalism. ... A man would still be free to accumulate his earned income with a view to spending it at a later date. But his accumulation would not grow. He would simply be in the position of Pope's father, who, when he retired from business, carried a chest of guineas with him to his villa at Twickenham and met his household expenses from it as required."

      A chest of guineas may be the most you can hope for.

  5. Such an awesome post and when I find the time I may comment on it much more. Here is a short one:

    " In cases like these, the central bank evidently is able to keep short rates high by limiting the supply of reserves. But why in that case doesn't the financial system develop private substitutes for reserves? "

    The CB is the sole supplier of currency. With rising inflation people will demand more currency. For that banks need to sell bonds to the CB for reserve deposits that get converted into currency. Hence, banks will adjust loan rates accordingly.

    Alternative substitutes actually often develop in high inflation countries which usually is increased demand for foreign currency with the typical effects on exchange rates.

    1. I don't think it's about currency. For one thing, there's nothing about high inflation in this scenario. For another, the private sector has done a fabulous job inventing substitutes for currency. How much of your total spending do you do in bills? In my case, it's about 10%. If currency were the basis of central bank control, I think banks would have a very easy time encouraging people to use other forms of payments when rates rose.

  6. Regarding number 5, I took a look at the Federal Funds Rate and the 3 month T-Bill rate data from 1954 on, and found that when rates change, they either move concurrently [per the data I could find], or the market-auction-determined T-Bill rate leads the policy-determined FFR.

    I could not find a single instance of the market following the Fed when the direction changes. When there are major changes, the FFR can lag by as much as 4 months.

    In your mid-2000s example, T-Bill leadership is obvious on a chart. [Graph 5 at the link.]

    My conclusion was that, contrary to conventional wisdom, the Fed is a close follower, not the setter of short rates.

    My reasoning is that time travels in one direction only, and if A precedes B, then the contention that B causes A has a steep burden of proof.

    I got a lot of push back in comments, but as near as I can tell it was all either: "But, but, but -- EXPECTATIONS!", hand waving, and naked assertions. I say that getting it right every single time over decades is expecting rather a lot from expectations, most especially so during the Greenspan years, when he spoke more or less in tongues.

    There are coherent arguments that the Fed leads, but, as far as I can tell, nothing that clinches it in the face of contrary empirical data.


    1. Under normal circumstances, the Fed never surprised the market - fed funds futures almost always gets the decisions correct on the day. The market usually shifted ahead, and can often be traced to "open mouth operations" - Fed speeches. (Conspiracy theorists can insert leaks to favourites as a cause.)

      This behaviour means that it looks like the Fed could just look up what the market is pricing each day, and announce that as its rate.

      The only time that the Fed sort-of surprises the market is when they cut rates during crises on dates without meetings, but even then, market rates are typically dropping like a rock.

      Other central banks operate differently (they can have 5-4 vote splits on the rates commitee), and so you might see the central bank leading market rates.

  7. It's not obvious on my screen with these text colors, but in the above comment the words "took a look at" are a hot link to my AB post.


  8. JzB-

    Bob Pollin (my dissertation advisor) has several papers doing more formal statistical tests of these relationships, including this one and this one. (Granger-Sims tests, for whether changes in one variable carry information about future changes in the other.) And he comes to the same conclusion as you do visually -- the policy rate seems to follow the market rates as much if not more than the reverse. However, in this particular case I'm not willing to dismiss the expectations story. After all, empirical evidence does not just mean statistics -- we can also go out and look at what people are doing, or say they are. And participants in credit markets do seem to be trying very hard to predict future movements of the policy rate, while monetary policy makers do seem to be responding to a range of factors beyond current market rates.

  9. Thanks.

    That's a coherent explanation.

    And I'm not suggesting that expectations aren't important.

    I just find it unlikely that over decades, and in the context of numerous other market failures, the markets can use expectations to anticipate Fed actions with essentially unfailing accuracy in an otherwise uncertain world.

    I presented my AB blog post as it if were a conclusion. But it's really more of a probing question. And I'm not satisfied that the answer has been securely nailed down.



  10. This is a very interesting and curious post, since "what determines the interest rate" seems a very important question and it is weird that there doesn't seem to be any clear explanation. I have three - related - observations:

    1) If we speak of an "equilibrium" interest rate, we can speak of 3 different things:
    a) The interest rate that doesn't create a bubble (an excess growth of credit that goes on until the interest that must be paid exceeds what the "real economy" can pay);
    b) The interest rate that " can keep the interest rate at the level that keeps current expenditure at the appropriate level", as per your (5).
    c) The interest rate that is generated by the market between "supply and demand" for credit, whathever it means.
    I think that there is no reason to think that a, b, and c are the same interest rate. Not only c might differ from a and b, but a and b might be different. If this is true, this is a big problem.

    2) If we ask what determines the interest rate, we are probably speaking of my "c" above. I don't think that money (or credit) can be described as a commodity, but if we use the "supply and demand" analogy, there is this weird thing:
    The market price determined by supply and demand is by definition the price that maximises the number of sales of the commodity. If the sale price is higer than the "equilibrium" price, less people will buy it; on the other hand if the sale price is lower, more people would buy it but less people will produce it, and thus less effective sales will happen.
    However, when we speak of credit, it seems to me that the consensus view is that when the interest rate falls, the quantity of loans increases. This is very different from what is supposed to happen for commodities, however might make sense if we think that something is keeping the interest rate higer than its "supply and demand equilibrium" rate.

    3) If my (2) is correct, we could explain the role of the central bank as an actor that puts a floor under the interest rate. If I'm a bank, and the fed is willing to hold my reserves for 2%, I will never lend to anyone for less than 2%.
    This means that, if for any reason the "supply and demand" interest rate is lower than what the central bank thinks is optimal, the central bank can lock it at an higer level, thus reducing the quantity of total credit outstanding. However if for some reason the "supply and demand" is higer than what the central banks wants, the central bank is powerless.

    1. A big part of the confusion here, I think, is that both (b) and (c) get mixed up with the question, "in an idealized market economy, how many goods tomorrow would exchange for one good today?" Which is a different (and in my opinion, irrelevant) question.

      when we speak of credit, it seems to me that the consensus view is that when the interest rate falls, the quantity of loans increases.

      This is a good point. There is an implicit assumption that most variation in the volume of lending is determined by the supply of credit rather than the demand for it. But I don't think this assumption is unique to credit markets, or necessarily unjustified.

      if for any reason the "supply and demand" interest rate is lower than what the central bank thinks is optimal, the central bank can lock it at an higer level, thus reducing the quantity of total credit outstanding. However if for some reason the "supply and demand" is higer than what the central banks wants, the central bank is powerless.

      In some situations this may be true; in other situations (like the Bank of England in the 19th century) it's the other way round, and the central bank interventions put a ceiling rather than a floor on market interest rates. Some people think it is possible for the central bank to intervene on both sides of the market and keep interest rates within a narrow corridor. I am not so sure.

  11. Great post, I have been thinking a lot about this lately after reading some of your earlier posts on interest rates.

    I don’t know enough about 19th century BOE operations to be sure, but it seems to me that their problem in hitting interest rate targets was that they didn’t take both sides of the market, unlike modern open market operations, (repos and reverse repos in the US). In the modern era the interest rate will quickly converge on the CB’s target because of arbitrage. It just occurred to me that this is why Perry Mehrling says the fed is now the “dealer of last resort”.

    I agree that the 6th story is the one best grounded in reality. This is interest rate risk and is well understood in financial markets. It also effectively explains the different shapes the yield curve can take. For example, there will be an inverted yield curve when people expect interest rates to decline and interest rate risk becomes negative.

    Liquidity preference is a completely muddled concept. It doesn’t refer to what people usually mean by liquidity, which is ease of selling, because government bonds are very easy to sell compared to things that are normally said to be illiquid like private equity investments. When people use it they seem to mean a few different things. The first is that liquidity preference is the convenience yield of holding money rather than an interest bearing financial asset. So you would like to have some cash in your pocket and have a deposit account at your broker so you don’t have to go through the extra step of selling an asset when you want to buy something. This convenience yield probably does exist, but I doubt it is very important or that it would change enough over time to explain changes in the interest rate.

    The other thing people seem to mean when they say liquidity preference either means credit risk or interest rate risk. This is what you describe here,

    “In principle, it should be possible to tell a more general version of the liquidity preference story, where, instead of wealth holders balancing the income from holding a bond against the liquidity from holding "money," you have banks balancing net income against incremental illiquidity from simultaneously extending a loan and creating a deposit.”

    But credit risk and interest rate risk are added on top of the current risk free short term rate, so the mystery is shifted to what sets that rate.

    1. I don’t know enough about 19th century BOE operations to be sure, but it seems to me that their problem in hitting interest rate targets was that they didn’t take both sides of the market

      But that is not trivial. On the one hand, the Fed can't sell more securities than it holds. And on the other, while it can in principle buy an unlimited amount with its own liabilities, that raises the problem of what securities exactly are eligible. If the Fed stands ready to buy any quantity of security X at a given price, the result may be simply to make X a close substitute for Fed liabilities and leave the rest of the term structure unchanged.

      Liquidity preference is a completely muddled concept.

      I think it's fairly clear. It means the importance you assign to being able to make payments at short notice. Government bonds are indeed very liquid. But they are not perfectly liquid, since if you buy a bond for $100 today you have not guaranteed your ability to make a payment of $100 at some future date, since interest rates might rise in the interim.

      But credit risk and interest rate risk are added on top of the current risk free short term rate, so the mystery is shifted to what sets that rate.

      In the liquidity preference story, this is a difference without a distinction. The spread between short loans and cash, and between long loans and short loans, and between risky and "riskfree" loans, are all determined by the same factors, and it's a matter of indifference which of them we call "the" interest rate.

  12. Josh -- I loved reading this, because I felt extra queasy discussing classical/Keynesian/etc theories of the interest rate in the sections of intermediate macro that I taught last semester. I'm curious how you've done it, or if you intend to do it differently, after this evaluation. There is always the: "theory says this, practice is different" approach, but...

  13. I am not sure I understand your point about interest rate parity. The interest rate parity relationship is mainly just a means to price currency (at least for the floating major currencies). I never saw any evidence that would suggest that beliefs about future currency movements influencing interest rates. (I could see that happening in places with currency pegs.)

    - Banks cannot get around reserve requirements since they are are integral part of the regulated (non-shadow) banking system. They would have to break the payments system ("counterfeit") in order to get around them.

    - The point why parallel private currencies do not become important came up in the comments. Within the current economic system, things like income taxes would be an important institutional factor. You have to convert every transaction to the official curency for tax purposes, and you have to withold income taxes on salary in the official currency.

    More generally, entities hold money against uncertain monetary claims. Holdings of private currencies may be just as useless a hedge against those claims as fixed assets or inventories.