Friday, August 19, 2011

Liquidity Traps, Money, Inflation, and Bond Yields

What is a liquidity trap? This is really two questions: (i) What is the "liquidity" that is getting trapped and (ii) What is the "trap" all about? In Old Keynesian economics and Old Monetarism, we think of the financial world in terms of two assets: interest-bearing assets and money. Money is the stuff that is used in transactions - liquidity - and private sector economic agents are willing to substitute money for interest-bearing assets in response to changes in the nominal interest rate. The nominal interest rate is essentially a measure of the scarcity of money as a medium of exchange. Monetary policy is about swaps by the central bank of money for interest bearing assets, or the reverse, and the attendant effects of those actions. One of those effects is a short-run liquidity effect. A central bank swap of money for interest bearing assets tends to make money less scarce as a medium of exchange in the short run, and the nominal interest rate falls.

But what if the nominal interest rate were zero? In that case, money is not scarce as a medium of exchange, so that money and "interest-bearing" assets are essentially identical, in which case central bank swaps of money for other assets are irrelevant. That's Grandma's liquidity trap.

Now, as Hicks said,
The General Theory of Employment is a useful book, but it is neither the beginning nor the end of Dynamic Economics.
Modern economists have expanded on Old Keynesian and Old Monetarist ideas and made them more precise. For example Ricardo Lagos, building on earlier work by Charles Wilson, tells you exactly what Grandma's liquidity trap is. It's basically a Friedman rule idea. There are many ways for policy to support a Friedman rule, i.e. a zero nominal interest rate forever. If there are many policies that support a zero nominal interest rate forever, that's also saying that there is a subset of monetary policies, among which the choice is irrelevant, i.e. if the central bank conducts certain kinds of asset swaps, then nothing changes, i.e. there is a liquidity trap.

Why am I calling this Grandma's liquidity trap? During the National Banking era, there were recurrent banking panic episodes, and it is most useful to think of these as currency shortages, i.e. "liquidity" shortages. The way to correct a currency shortage is through conventional central bank actions - open market purchases of interest-bearing assets and discount window lending. This is just the logical extension of standard day-to-day central banking practice: Target a "degree of liquidity scarcity," i.e. a nominal interest rate, and then accommodate shocks to the private sector's demand for liquidity. This view of the world is consistent with how Friedman and Schwartz thought about the Great Depression. Basically, in Grandma's world, the idea is that the central bank does not need to be worried about the sources of fluctuations in the demand for currency. All that matters is that these fluctuations be accommodated appropriately with the standard tools available to central bankers.

I am convinced that this is not the way we want to think about the recent financial crisis, or about how monetary policy should be conducted given current circumstances. We are not in Grandma's liquidity trap.

Here's a better way to think about it. Think of the world as having two kinds of liquidity: currency and other liquid assets. The other liquid assets include bank reserves, Treasury bills, long maturity government debt, asset-backed securities, bank loans, etc. - all the assets that are somehow useful in some form of exchange (retail, wholesale, financial) either directly, or because they can be transformed by financial intermediaries into some asset that can be used in exchange. The "other liquid assets" of course have different characteristics, and are used in different ways in exchange. Risk, including maturity risk, is important in determining the prices of these assets (relative to each other), and different liquid assets will have different liquidity properties, reflected in how they are used in exchange. An important concept here is the "liquidity premium" which we can measure as the difference between an asset's market price, and its "fundamental," i.e. the price it would trade at, based on the stream of future payoffs the asset is a claim to, if the asset were not useful in exchange.

Bank reserves - the accounts of financial institutions with the Fed - are a key liquid asset, as reserves are used daily in the clearing and settlement of a very large volume of transactions. However, it takes a very small quantity of reserves to support this extremely large volume of financial trade - the velocity of circulation of reserves within a day (pre-financial crisis) is typically humongous. Currently, the financial system is awash with reserves, to the point where the marginal value of reserves in financial transactions is essentially zero. The interest rate on reserves (IROR), which is currently 0.25%, is higher than the interest rate on 3-month T-bills, which is currently 0%. Thus, T-bills command a higher liquidity premium than do reserves, as they are actually more useful in financial exchange (inside and outside the US).

The key liquidity trap - the contemporary liquidity trap - the Fed is faced with currently is that all of the other liquid assets are now essentially identical, from the point of view of Fed asset swaps. The Fed cannot make reserves more or less scarce as a liquid asset through swaps of reserves for other assets, and therefore has no hope of moving asset prices. The Fed can swap reserves for T-bills or reserves for long-maturity Treasuries all it wants, but because this essentially amounts to intermediation activities the private sector can accomplish as well, this will have no effect.

An important point to note is that the contemporary liquidity trap, in contrast to Grandma's liquidity trap, has nothing to do with the zero lower bound on the nominal interest rate. Remember that the short-term safe nominal interest rate tells you how scarce currency is relative to other liquid assets. The IROR could be 2%, 5%, or 8%, so that currency is scarce, but there would still be a contemporary liquidity trap if reserves were not scarce relative to other liquid assets.

As I have discussed in earlier posts, the only relevant policy instrument the Fed has, so long as the stock of excess reserves is positive, is the IROR. Thus, the only lever the Fed has is the ability to make currency more or less scarce relative to other liquid assets, and that is the avenue by which the Fed can control the prices of goods and services. Here is where the zero lower bound on the IROR comes in of course. At the zero lower bound, the Fed cannot achieve a higher price level, except through talk about the future.

Now, to put this in context, let's look at some traditional monetary measures. The first chart shows the stock of currency over the last five years. You can see that the currency supply grew significantly during the financial crisis; this is likely driven by overseas demand, but it is hard to know - we don't know exactly where US currency resides or what it is doing. Over the last two years, and particularly this year, currency has been growing at a reasonably brisk pace (more on growth rates later).

The key thing to note here is that, under the current regime (positive stock of excess reserves), the currency stock is not directly under the Fed's control. The Fed determines the total stock of outside money (currency + reserves) and financial institutions, firms, and consumers determine how that is split between currency and reserves. The stock of currency could be rising because the demand for it is rising, or because reserves are looking less desirable for financial institutions. This of course matters - in the latter case this would be inflationary, in the former case not.

The next two charts show M1 and M2, again for the last 5 years. These charts are very interesting. Both M1 and M2 show recent large spikes. Further, if we look at year-over-year growth rates in currency, M1, and M2, these are all getting large, as we see in the next chart. The twelve-month growth rate in M1 now exceeds 20%, and growth rates in M2 and currency are at or close to 10%. Allan Meltzer should be having a cow.

Of course, this is where Grandma's liquidity trap comes in. Monetary aggregates are constructed to include private and public liabilities that are widely used in exchange by firms and consumers, under "normal" circumstances. But circumstances are not normal - a checking account is now a convenient short-term store of value with no associated opportunity cost. Thus, these spikes in M1 and M2 need not be associated with more inflation. However, I don't think we know enough to tell.

But why the spikes in M1 and M2 in the last weeks? That may have something to do with what you see in the last chart. This one shows, again for the last five years, nominal and real (i.e. TIPS) yields on 10-year Treasuries. These yields have also fallen significantly in the last weeks. This reflects a scarcity of other liquid assets, presumably because of a general "flight to quality" in world asset markets.

Now, if other liquid assets are becoming more scarce, it would make sense that we should see growth in financial intermediation, reflected in growth in M1 and M2, as the private sector creates more assets in the "other liquid assets" category. However, it could also be that there has been a decrease in financial intermediation not reflected in M1 and M2. For example, there could have been a shift out of "shadow banking" into conventional commercial banking, but unfortunately we do not measure shadow banking activity.

The scarcity we are observing is not a traditional currency scarcity. As such, we can't correct the scarcity by using conventional central banking tools - open market operations in short-term government debt and discount window lending. Neither can we correct it through "quantitative easing." We cannot ease anything through swaps of reserves for long-maturity debt, as that cannot make reserves relatively less scarce under the current circumstances. But the inability of monetary policy to correct the liquidity scarcity problem has nothing to do with the zero lower bound on short-term nominal interest rates, as the key problem is a contemporary liquidity trap, not Grandma's liquidity trap.

How can government action mitigate the liquidity scarcity? If monetary policy cannot do it, that leaves fiscal policy. But there is a tendency, particularly in the blogosphere, to frame the problem in Old Keynesian terms. In this view, we are facing Grandma's liquidity trap, the LM curve is flat, monetary policy doesn't work, so shift the IS curve instead. Further, unemployment is very high and persistent, so it might seem natural to have the government employ people directly by spending more. But the problem here is financial, and it's not a Keynesian inefficiency associated with real rates of return being too high; in fact real rates of return are too low given the scarcity of liquid assets, which produces large liquidity premia.

One way to solve the problem would be to have the Treasury conduct a Ricardian intervention, i.e. issue more debt with the explicit promise to retire it at some date in the future. If the future arrives, and we still have a scarcity, then do it again. This requires a transfer, or a tax cut in the present, and leaves the present value of taxes unchanged, but the result is not Ricardian because of the exchange value of the government debt issued.

What's the difficulty here? Well, the US government apparently has a very difficult time making decisions on fiscal matters, and seems not to like commitment, so what I am proposing is just not feasible politically. You might think it convenient if the Fed could conduct limited types of fiscal policy, but that requires giving the power to tax to unelected officials, and that seems a bad idea.

So where does that leave us? The key financial problem facing us is a scarcity of other liquid assets, not a traditional currency scarcity. The Fed is powerless to solve that problem; the Treasury could in principle solve it but cannot. For now, the Fed can only monitor the economy for signs of a more serious inflation. Some of those signs may already be there, for example in currency growth, though it is hard to tell what is driving that.


  1. Stephen:

    The Fed can swap reserves for T-bills or reserves for long-maturity Treasuries all it wants, but because this essentially amounts to intermediation activities the private sector can accomplish as well, this will have no effect.

    Let's say the Fed committed to buying up as many treasuries, foreign exchange, stocks, etc. as needed until inflation hit some permanently higher target inflation rate like 6%. And assume this was clearly understood and believed by the public. It hard to believe the Fed would have no effect on investors' portfolios and that such a regime change in monetary policy would not lead to a change in the velocity of broad money (e.g. M3). Explain why this would not be the case.

  2. "Let's say the Fed committed to buying up as many treasuries, foreign exchange, stocks, etc. as needed until inflation hit some permanently higher target inflation rate like 6%."

    The Fed can buy all the government debt it wants right now, and that will be irrelevant, for inflation or anything else. Once you get into private assets - stocks, for example, as you mention - that's another story. But there are good reasons why having the central bank purchase private assets is a bad idea.

  3. Let's put aside your contention that Fed buying up all the treasuries won't matter for now. If, as you concede, the Fed should be able to make a difference for inflation by buying up private assets, then why shouldn't the Fed be able to pack a punch by simply signaling a commitment to do so via a price level target?

  4. "If, as you concede, the Fed should be able to make a difference for inflation by buying up private assets, then why shouldn't the Fed be able to pack a punch by simply signaling a commitment to do so via a price level target?"

    No, I didn't "concede" something. I said that's a different story. If the Fed buys private assets, the result depends on what they are. In extreme cases where the Fed is not very good at screening the assets, it could run into adverse selection and moral hazard problems, i.e. the Fed could be a bad banker. But suppose that the Fed is as good a banker as the private sector. Then, if it buys private assets on the same terms as the private sector under the current circumstances, this is neutral - same result. If it buys assets on better terms than the private sector is offering, then some credit reallocation is going on, and the Fed will incur a loss on its portfolio which has to be made up somewhere. Some people benefit, some lose, and it's not clear if you get more inflation or less. The credit reallocation illustrates why the private asset purchases are a bad idea. If you start buying private debt on a regular basis for example, then everyone learns that they can benefit if the Fed buys their debt. United Airlines gets in trouble for example, and lobbyists start making the case that the Fed should by United Airlines debt. Where does that stop. Not sure what you are thinking with the commitment and price level targeting.

  5. One of the key trends of the past quarter century, besides the growth of shadow banking, was the erosion of transparent price discovery on exchanges in favour of ill-transparent, decentralised, fragmented OTC trading markets. The assets that grew fastest during this era (OTC derivatives, RMBS, etc.) were all traded through intermediaries in these decentralised markets. Of course, as confidence in the intermediaries eroded, the liquidity of these markets collapsed. Investors holding OTC assets find they cannot price or trade them.

    Perhaps it is time to consider structual market reforms which would encourage more exchange trading, concentrating liquidity and price discovery without manipulation by credit intermediaries with access to central bank liquidity. In essence, this was the 1934 Securities Exchange Act and Glass Steagall Act objective.

    Liquid assets are traded in size, with transparent pricing. Time to nudge the market back toward issuing and trading liquid assets?

  6. Steve,

    How does your policy proposals square with the Del Negro, Eggertson, Ferrero and Kiyotaki paper? That paper was meant to be about QE1, which is closer to a fiscal policy in the sense that the government was swapping illiquid IOU's to liquid gov't paper. By doing so it was not so much running a deficit as it was changing its own portoflio and the portfolio of the private sector.


  7. RV,

    I have not read the paper, which is posted here:

    However, I read the abstract, and I think I see where it's going. It's basically Kiyotaki-Moore (2008) with New Keynesian sticky wages and prices, and then they look at QE at the zero lower bound. I'm familiar with Kiyotaki-Moore. That model gives the central bank an advantage in producing liquidity by assumption. There are two assets. Capital is illiquid by assumption - it's hard to sell existing claims to it, and there are limitations on issuing new claims to it. This produces liquidity constraints that firms face when they want to invest. There is a straightforward policy which relaxes those liquidity constraints, which is for the central bank to intermediate all the capital - i.e. issue money to purchase the entire capital stock. For me, that makes the model very unappealing. It does not get at what advantages and disadvantages the central bank has as a financial intermediary, and those things are critical for evaluating policy.

  8. Steve,

    How do you square the success of Sweden's monetary policy with your view outline above?

  9. Steve, your summary broadly agrees with my understanding of the model and of the paper. What confuses me is that I don't see how your implicit model is so different from theirs.

    In your post is that you seem to be making the following arguments:

    1) Money/Reserves and Treasuries are perfect substitutes insofar as liquidity is concerned. So swapping one for the other does nothing.

    2) What you want is to increase the total amount of liquidity, i.e., Money AND Treasuries.

    3) The government is able to do this by issuing treasury notes. The financial sector is unable to do this. So the government should do it.

    It seems that in step 3 you are making an assumption very similar to Kiyotaki and Moore '08. The difference is that while they are talking about the gov't doing financial intermediation, you are talking about it issuing debt and backing it with future taxes. I can see merits and demerits in both proposals, but I don't see how they are so fundamentally different in the main mechanism through which they work. Unless somehow you think that the liquidity of treasuries is intrinsically linked to its ability to raise taxes, but this is not in your post.

  10. Hi David,

    I don't know enough about Sweden to say. You would have to figure out whether the "success" was actually tied to monetary policy, or there was something else going on. The fact that the Riksbank was effectively taxing reserve balances I thought was interesting. I discussed that here:


    Unfortunately this is me blogging vs. Nobu and company with an explicit model. In my loose model of words, the two key Fed liabilities - currency and reserves - each have specific roles in particular kinds of transactions. Those roles give the Fed its particular advantage in financial intermediation. In Kiyotaki-Moore, the Fed seems to have an advantage in intermediating private assets.

    "Unless somehow you think that the liquidity of treasuries is intrinsically linked to its ability to raise taxes, but this is not in your post."

    Yes, good point. The power to tax is critical.

  11. Steve,

    The only other thing going on that I know is they have a larger social safety, which makes for a larger automatic stabilizer. However, it is not clear how important this was to their success.

    What we do know is that the Riksbank grew its balance sheet to 25% of GDP, versus the Fed's 15%. Also, they had an explicit, well understood inflation target unlike the Fed. Finally, the negative interest rate on reserves was apparently more symbolic than real. So that leaves the large expansion of Riksbank's balance sheet and the better anchoring of inflation expectations via their inflation target. It is hard to reconcile this with the claim that QE is ineffective.

  12. Steve,

    One more thing. Their success can be seen in the fact that they returned nominal spending to its pre-crisis level. In effect, they did what amounted to NGDP level targeting. See the figures in these two posts:

  13. David,

    But you could think of Canada as a success too. There was no serious expansion in the Bank of Canada's balance sheet. They are now back to "normal" monetary policy with essentially zero overnight reserves.

  14. "What we do know is that the Riksbank grew its balance sheet to 25% of GDP, versus the Fed's 15%."

    i think the BOE expanded their bs even more (as % of GDP) yet growth there has been disappointing (relative to expectations, relative to pre-crisis trend, etc.). so it would seem bs expansion is not the (only) explanation (and nor is the exchange rate).

  15. not an economist, but quote "The nominal interest rate is essentially a measure of the scarcity of money as a medium of exchange" where is the psychology, eg can the interest rate can be zero while you have scarce money if people are scared ?

  16. Steve, thanks for the clarifications. I should look up your papers with models to have a better idea of what you have in mind. But I would enjoy if you were to write a post sometime on why it is that the power to tax allows the government to provide liquidity better then private agents.


  17. Debt does not have to be backed by promises of new taxes - there is an alternative.
    Govt spending on tangible productive assets can be backed by a promise to sell the assets in due course, hopefully a politically more acceptable strategy.
    The 'loan' is thus for a capital expenditure, rather than a running cost.

  18. Many high Street banks who sold so called Bank Swaps are now bracing themselves for a possible yet another wave of claim similar to those brought by people with miss sold PPI.

  19. This is just the logical extension of standard day-to-day central banking practice.

    purchase cialis