Saturday, January 4, 2014

Does QE actually reduce inflation?


There's a counterintuitive meme floating around in the blogosphere that quantitative easing doesn't do what we commonly suppose. Somehow QE reduces inflation or causes deflation, rather than increasing inflation. Among others, here are Nick Rowe, Bob Murphy, David Glasner, Stephen Williamson, David Andolfatto, Frances Coppola, and Bill Woolsey discussing the subject. Over the holidays I've been trying to wrap my head around this idea. Here are my rough thoughts, many of which may have been cribbed from the above sources, though I've lost track from which ones.

Let's be clear at the outset. Inflation is a rise in the general price level, deflation is a fall in prices. QE is when a central bank purchases assets at market prices with newly issued reserves.

In equilibrium, the expected returns on all goods and assets must be equal. If they aren't equal then people will rebalance towards superior yielding assets until the prices of these assets have risen high enough to iron out their superior return (and away from low yielding assets until their prices have fallen enough so that their expected return is once again competitive with all other assets).

Central bank reserves are one of the many assets whose yield is included in this calculus of returns. The return on reserves can be decomposed into two specific categories of return: expected capital gains, or price appreciation, and a liquidity return, sometime referred to on this blog as a monetary convenience yield.

Regarding the first return, this is typically negative. People expect the purchasing power of central bank reserves to be lower in the future than in the present—they anticipate inflation.*

The liquidity return exists because reserves are highly marketable. The ability to quickly mobilize reserves to deal with unanticipated events yields a flow of liquidity services, specifically the alleviation of felt uncertainty. The expected return on these liquidity services outweighs the expected capital loss on reserves, providing reserve owners with a combined return that is competitive with other assets like cars, olive oil, education, t-bills or houses.

When a central bank conducts QE, the quantity of reserves in the economy increases so that they are less scarce. All else staying the same, the marginal value that people attribute to the flows of liquidity services provided by reserves declines. With their liquidity return having fallen, reserves now yield a lower overall return than competing assets.

Given these unequal returns, reserve owners will want to rebalance their portfolios into higher yielding alternatives. However, existing owners of these assets will be unwilling to accept this trade since the return they can expect to receive on reserves is no longer competitive with the return on the assets that they would be forgoing. Reserve owners will have to sweeten the deal by offering potential counterparties an improved return on reserves held. The way they can do this is to offer to sell their reserves at a reduced price today relative to their price tomorrow. In doing so, reserve owners are offering counterparties an improved potential for capital appreciation to counterbalance the diminished liquidity return on reserves.

Another way to describe this trade is that reserve owners must create some inflation, or a higher price level, in order to attract interested buyers. From this higher plateau, prices will rise at a much slower rate than before, or, put differently, the purchasing power of reserves will fall much slower than previously expected. The new expected price trajectory of reserves may even be a deflationary one—the market anticipating prices tomorrow to be lower than those today. In any case, only when the expected capital gain on reserves has been sweetened enough to sufficiently compensate would-be owners of reserves for bearing their diminished liquidity return will potential counterparties be willing to trade away their existing assets for reserves.

So back to our initial question: does QE reduce inflation? Not quite. By diminishing the liquidity return on reserves, QE reduces *expected* inflation. This change in expected inflation occurs via a leap in inflation in the present. Subsequent rounds of QE will continues to breed inflation and lower expected inflation until the liquidity return has been reduced to zero, at which point further QE will have no effect.

------

But let's introduce another wrinkle. What happens if other assets also carry a liquidity return? And let's assume that there are different kinds of liquidity returns, so that the liquidity services provided by one asset can't be easily substituted with the liquidity services of another. Thus, when the economy is flooded with reserves and their marginal liquidity return hits zero, the liquidity return on alternative assets needn't also decline to zero.

Let's take up where we left off. QE has reduced the liquidity return on reserves to zero and subsequent rounds of QE no longer cause inflation or reduce expected inflation. Let's assume that short term bills, specifically those issued by, say, Microsoft, provide a unique set of collateralizability services, and therefore yield a liquidity return > 0.

When our central bank purchases Microsoft bills, the supply of Microsoft collateral in the economy shrinks, which increases the liquidity return on Microsoft bills. The total return on Microsoft bills, the sum of their liquidity return and expected capital gain, is now superior to all other assets. This spawns a mass effort by investors to sell other assets for Microsoft bills. The only way that existing bill owners will agree to sell away their superior yielding Microsoft debt is if potential buyers offer to pay a higher price. As short term Microsoft bill prices are bid up, the expected capital gain on bills is reduced, counterbalancing the higher liquidity return. At some appropriately higher bill price, the total expected return on bills will be reduced to a level competitive with all other assets, restoring equilibrium.

This process, however, doesn't have any impact on inflation. All that is happening is that the relative price of a certain asset—the short term Microsoft bill—is rising.

Subsequent rounds of QE will further reduce the supply of short term Microsoft bills, increasing their liquidity return and eventually driving their price above par. At any price above par, capital returns on bills are effectively negative—bills, after all, never pay out more than their par value. People will continue to be attracted to a <0% yielding short term bill as long as it sports a sufficiently large liquidity return. The latter can outweigh the negative capital return, providing a total return that is competitive with other assets.

One problem with QE is that it drives the price paid for the liquidity service on Microsoft short term bills above the cost that Microsoft must incur in maintaining those liquidity services. People are effectively paying more to enjoy Microsoft liquidity services than they would in a competitive economy in which prices are pushed down to the cost of production. The artificially high price for bills that has been caused by QE incentivizes people to acquire a smaller flow of Microsoft liquidity services than they would otherwise prefer. This represents a deadweight loss to the economy, or what is termed an allocative inefficiency by economists. The surplus that consumers enjoy is smaller than it would be in a world in which large scale purchases of Microsoft bills had not pushed their liquidity return to artificially high levels.

This loss of allocative efficiency, however, does not equate to deflation. While QE involving Microsoft bills may not be ideal for the economy, it doesn't cause the price level to fall.

Given QE's effect on Microsoft bills, it would be odd if Microsoft did not choose to continually issue new short term bills until the marginal value of liquidity services yielded by bills was driven back down to the cost of maintaining those services. This would goose Microsoft's profits while simultaneously increasing the consumers' surplus, removing all of the inefficiency created by QE.

However, if the issuer of these unique collateralizable bills is the government, not Microsoft, things might be different. Because the government isn't profit-driven, it may be less motivated to issue new bills and reduce the allocative inefficiency created by QE. Is this a big deal? The excess of liquidity's price over cost is similar to any other monopolistic distortion, take for intance the diamond or potash oligopolies that price their products above cost. Situations like these are unfortunate, but I'm not so sure that they have large macroeconomic consequences. The benefit of not doing QE because one might create inefficiencies in a few lone markets for collateral are surely not as large as the benefits of doing QE in order to boost the economy's price level.

So the best I can do in my mental meandering is that QE either produces inflation or is irrelevant. It does not cause lower inflation or deflation. The by-product of any QE-inspired jump in inflation is lower *expected* inflation than before. A few inefficiencies may be created in various markets targeted by purchases, but as interesting as these inefficiencies are I don't see how they produce severe macroeconomic consequences.



*For the sake of simplicity I assume that reserves don't pay interest.

19 comments:

  1. A few questions.

    Why might the liquidity yield on bills exceed that on reserves? Doesn't the liquidity value of bills simply derive from their potential to be converted to reserves by borrowing against them, which must presumably always involve some non-negative cost? It could be said that there is some benefit to banks in holding bills rather than reserves in that bills can be used for repo, so there is a cash saving to the extent that rates for secured borrowing are lower than unsecured. I'm not sure I'd think of that as a liquidity yield, but is that what you had in mind?

    What is the cost to the government of maintaining liquidity services on bills?

    In talking about bill purchases you seem to be focussing on steps which affect the short end of the yield curve. I thought the discussion here was about the affect of the purchase of long-term assets, once short rates were already at the ZLB.

    ReplyDelete
    Replies
    1. "Why might the liquidity yield on bills exceed that on reserves?... It could be said that there is some benefit to banks in holding bills rather than reserves in that bills can be used for repo"

      Yes, the superior repoability of t-bills is the reason that some people say bills have a superior liquidity yield to reserves. For instance, see Stephen Williamson making the point here. I'm skeptical that we are at the point at which t-bills have a higher liquidity yield than reserves, or that we'll ever arrive at such a point. But for the sake of argument, I assumed in this post that it is the case that such a differential can emerge.

      "What is the cost to the government of maintaining liquidity services on bills?"

      Good question. One of the reasons I used Microsoft is because it allows me to think about these costs better. Microsoft has to hire an investor relations team and pay fees to brokers who in turn make a market in bills and provide research thereon, all of which contribute to maintaining the liquidity on bills. The government is more complicated, but I think the analogy can sort of be made.

      "In talking about bill purchases you seem to be focussing on steps which affect the short end of the yield curve."

      I just used bills as an easy illustration. Substituting in bonds or stocks shouldn't change things.

      Delete
    2. Interesting comment from SW there. I'd agree that the main reason T-bills are trading below IOER is that reserves are limited in who can hold them, but I think that's more a credit risk matter than liquidity. Banks have to pay less on unsecured than the IOER rate to get positive carry, but T-bills present a better credit risk to non-banks than bank unsecured, so they'll accept a lower rate on them.

      Delete
    3. The limits on who can hold reserves make things complicated. If everyone could open a reserve account at the Fed, would t-bills be trading at 0.08%? Also, an FDIC fee reduces the effective IOR rate, but by a varying amount for different banks. All these factors make it difficult to determine if the true interest rate on t-bills is actually lower than that on reserves, or, conversely, if the liquidity return on reserves is indeed lower than that on t-bills.

      Delete
    4. Also, whilst I think the ability to repo bills is a potential benefit to banks through reduced funding cost, it really doesn't seem right to think of that benefit as being a liquidity yield. As soon as you repo the assets, they cease to become part of your liquid assets. You can either have the liquidity benefit or the funding benefit. You can't have both. So I think the funding benefit is a different thing. It's like you say, there are other factors in play here - not just pecuniary returns and liquidity yields.

      Delete
    5. I do think that the ability to repo something adds to its liquidity yield. The liquidity yield, at least as far as I've been using it in this post, is the consumptive value someone enjoys in knowing that some asset they own can be easily marketed should the need arise. It isn't necessary that this asset actually be liquidated or repoed for those consumption flows to be enjoyed, just like we don't actually have to use a fire extinguisher in order to enjoy its services. The easier it is to repo something or quickly sell it outright the larger these consumptive flows, and the better an asset's liquidity yield.

      Delete
    6. I totally agree that the ability to repo increases the liquidity yield of an asset. What doesn't sound right though is that this could ever make the asset's marginal liquidity yield higher than that of money (meaning reserves for banks and transaction accounts for non-banks). After all, what do you get if you repo the asset? Money. So how can the ability to repo an asset make it more liquid if money is less liquid?

      SW's comment that you referred to implies that if T-bills are more liquid generally, then they have a higher marginal liquidity yield for everyone. There are two problems with this. First what might apply generally doesn't necessarily apply individually. So reserves are definitely a more liquid asset for banks than T-bills, even if the argument can be made that T-bills are more a liquid asset as a whole. Secondly, because liquidity yields and cash returns are not the only factors in play, people will not in general all have the same marginal liquidity yield on money, even in equilibrium. This is particularly true when comparing entities which can hold reserves and those that cannot.

      Delete
    7. "What doesn't sound right though is that this could ever make the asset's marginal liquidity yield higher than that of money (meaning reserves for banks and transaction accounts for non-banks)"

      I pretty much agree with you there.

      Empirically, if treasuries did somehow have a higher marginal liquidity yield than cash, this would show up in repo markets as a persistently negative repo rate, since anyone who forgoes treasuries' superior liquidity return would require compensation. But repo rates are positive -- providers of cash still need to be compensated for giving up cash because it is more liquid.

      "...then they have a higher marginal liquidity yield for everyone."

      Let me try and wrap my head around that. Like any consumer good, different people & institutions will attach varying marginal utilities to the liquidity services thrown off by a given asset. But we can still arrive at a market clearing price for those specific services, the point at which the marginal buyer and seller of liquidity services are willing to transact. If the price people are willing to pay for t-bill liquidity services is greater than reserve liquidity services, then the market is signaling that t-bills are more liquid generally, even though individual marginal utilities differ.

      Delete
    8. I think we're on the same page here, except that I'm not sure that it is possible to extract out a market clearing price specifically for liquidity services, because it's assets that are traded not services. Each asset comes with a bundle of features yielding utility or disutility, not just the liquidity aspect. I know that we can often in theory deconstruct asset prices into various components, but I can't see how you would do that here.

      Maybe it is possible, but even if there is an implicit price for liquidity services, I don't think that's a major factor in the rate differential between bills and reserves, because I think that's mainly caused by the other factors.

      Delete
    9. If two financial assets have the same credit quality, picking out their relative liquidity return is easy since the only axis along which they differ will be liquidity. But if you're right that credit risk helps explain the rate differential between bills and reserves, then that muddies the waters for picking out the liquidity return.

      One of the themes of this blog is trying to fashion financial instruments and techniques for stripping away the price for liquidity services from the rest of an asset. See here, here, here and here among others.

      Delete
  2. JP. Sorry I'm a bit confused. In your example of Microsoft bills are they considered "money" or not (i.e. what does "Microsoft, provide a unique set of collateralizability services," mean?), and if so, do they have higher or lower moneyness than reserves (they have a higher liquidity premium in your example but they may not be higher moneyness)? I *think* SW's original point was if Tsys were considered more money-like than reserves (equal or higher liquidity premium vs. reserves; more widely accepted as payment than reserves), then it could cause deflation. I didn't think either SW or his detractors clearly argued this point, so the whole dustup left me a bit confused. Or maybe I missed the whole point!

    ReplyDelete
    Replies
    1. "I *think* SW's original point was if Tsys were considered more money-like than reserves (equal or higher liquidity premium vs. reserves; more widely accepted as payment than reserves), then it could cause deflation."

      Yes, I'm just substituting treasuries with Microsoft bills in my post. The special collateralizability of MSFT bills means they are more money-like (ie have more moneyness & have a higher liquidity premium) than reserves and other assets.

      My reading of Williamson is that he says that open market purchases always cause deflation, even if Treasuries are not more money-like. It's the old Kocherlakota brouhaha all over again:

      http://worthwhile.typepad.com/worthwhile_canadian_initi/2010/08/why-everyone-should-be-forced-to-take-intro-economics.html

      Delete
    2. Thanks JP. It's not even clear the MSFT-easing would increase hardship if the UOA was still $. If MSFT bills became "MOE/collateral" for a certain economic subculture, but prices remained denominated in $, it shouldn't make any difference, just like your Bitcoin easing example.

      Delete
  3. I think you are missing the part of Williamson's story that pins down the price level with a fiscal policy reaction function. This is strange, but it is in the model. If you think of it as a modeling trick to represent a very sticky price level (which it clearly is not in his case), it is slightly less strange. Meanwhile, your model seems to assume that the reduced convenience yield is independent (remains lower post QE) despite the initial price level rise, thus the lower future inflation. Why do you assume the convenience yield doesn't increase upon the rise in the price level?

    ReplyDelete
    Replies
    1. I stayed away from actually writing about Williamson's story in this post because I don't feel like I'm capable of reading his papers, althoughI have tried a few times. What you mean when you say that the price level can be pinned down with a fiscal policy reaction function?

      "Meanwhile, your model seems to assume that the reduced convenience yield is independent (remains lower post QE) despite the initial price level rise, thus the lower future inflation."

      Am not quite grokking your comment. Give it another shot.

      Delete
  4. A general thought on liquidity premia: can we think of it as an option without a maturity date, such that the drivers of liquidity premi would be the same as the drivers of options (interest rates, volatility)?

    ReplyDelete
    Replies
    1. Yes, I think that's a good point. I've actually written a few post on the idea of liquidity options here and here, and monetary optionality here. (the idea evolved, the older posts aren't so great).

      If you lock in an asset for x years so that you simply cannot sell or repo it over that period, you've effectively sold away the liquidity option on that asset for x years. You should get compensated for that, the size of the payment being the x-year liquidity premium. If you've locked it up into perpetuity, then you've sold an option without a maturity date.

      Delete
  5. Thanks for the post, some interesting ideas to think about.

    I'm bit late in the party but did you consider that the banking sector as a whole is more or less stuck with the reserves. For them, as an aggregate, the only exit is to get the public hold more actual currency, notes and coins.

    Does this change anything? At least this wouldn't hold anymore, would it?

    "Given these unequal returns, reserve owners will want to rebalance their portfolios into higher yielding alternatives. However, existing owners of these assets will be unwilling to accept this trade since the return they can expect to receive on reserves is no longer competitive with the return on the assets that they would be forgoing."

    ReplyDelete