Sunday, January 4, 2015

Solving the ZLB Problem without Eliminating Cash

Should the Federal Reserve should eliminate cash as a way to avoid the zero lower bound (ZLB) problem? Ken Rogoff says yes in a recent paper. John Cochrane, on the other hand, is not ready to give up cash and is convinced that even if we did it would not solve the ZLB problem. Who is right?

Before answering these questions, let us recall the nature of the ZLB problem. It occurs when the market-clearing level of nominal short-term interest rates turn negative while actual short-term interest rates get stuck at 0%. This happens because individuals would rather hold paper currency at 0% than invest their money at a negative interest rate. The ZLB, in short, is a price floor that prevents interest rates from clearing the output market. And like any price floor, the ZLB creates a glut. In this case, it is an economy wide-glut better known as a recession.

So why not get rid of cash, as suggested by Ken Rogoff? John Cochrane gives several reasons why getting rid of cash may not be such a good idea. First, doing so would hurt the people who depend the most on cash: the poor who do not have access to or do not trust the formal banking system, the foreigners who need hard currency (e.g. Zimbabwe), and those wanting anonymity in their transactions. I share these concerns. Are harming these groups really worth beating the infrequent ZLB?

You might take an utilitarian approach and say yes, but even then it would be wrong. For one does not need to eliminate cash to solve the ZLB problem. As Miles Kimball has argued for the past few years, all that is needed is to make electronic (deposit) money the sole unit of account and turn the current fixed exchange rate between cash and deposits into a crawling peg based on the state of the economy. When the economy falls into a slump and the central bank needs to set a negative interest rate target to restore full employment, the peg would adjust so that paper currency would lose value relative to electronic money such that folks would not rush to it as interest rates go negative. This would effectively impose the same penalty on cash and deposits and kick start the monetary hot potato. Once the economy started improving, the crawling peg would start adjusting toward parity.

Now this is where John Cochrane's second objection comes into play. He worries that even if the Federal Reserve did lower short-term interest rates to a significantly negative value (say -5%) it still would probably not work because individuals would cleverly find other assets that would earn a 0% nominal return. Here is Cochrane:
[Q]uiz question for your economic classes: Suppose we have substantially negative interest rates -- -5% or -10%, say, and lasting a while. But there is no currency. How else can you ensure yourself a zero riskless nominal return?    
Here are the ones I can think of:    
(1) Prepay taxes. The IRS allows you to pay as much as you want now, against future taxes.  
(2) Gift cards. At a negative 10% rate, I can invest in about $10,000 of Peets' coffee cards alone. There is now apparently a hot secondary market in gift cards, so large values and resale could take off.  
(3) Likewise, stored value cards, subway cards, stamps. Subway cards are anonymous so you could resell them.  
(4) Prepay bills. Send $10,000 to the gas company, electric company, phone company.  
(5) Prepay rent or mortgage payments.  
(6) Businesses: prepay suppliers and leases. Prepay wages, or at least pre-fund benefits that workers must stay employed to earn.  
Cochrane should be more optimistic here. For all of these options only move the negative interest rate problem from one party to another. Here, the issuers of the 0% yielding assets have inherited the negative interest rate problem. They have effectively borrowed money at 0% and must decide if they want to deposit the funds at -5% in their banks or invest in something with a higher yield like riskier financial assets or capital expenditures. Obviously, the former option is not sustainable so they will opt for the latter one. But the latter option implies more risk taking and spending--the old monetary "hot potato" at work! So even in these cases the negative interest rate is still doing its intended job. Bill Woolsey makes this point in his response to Cochrane:
I am going to start with Cochrane's second example.  People could supposedly get a riskless zero nominal rate of return by purchasing gift cards... Under usual circumstances, when a retailer sells a card it is getting a loan... a zero interest loan. And so, now the retailer has the money. What do they do with it? If the interest rate on money is sufficiently negative, then the retailer will find borrowing money at a zero interest rate and then paying to hold it  unattractive. Of course, perhaps the retailer can invest by purchasing assets that have a positive yield. Or maybe they will accumulate inventory to be prepared for the greater sales when the cards are spent. It doesn't matter. As long as the retailer doesn't hold the money, the lower (below zero) nominal interest rate has done its job. 
And what does Walmart do with the money it receives?  If it holds it, that is a problem. But that is what the below zero interest rate on money aims to deter. If Walmart purchases other assets or purchases inventories of goods, constructs new buildings, or whatever, the problem is solved. 
Consider Cochrane's fourth example--pre-paying utilities... [I]f the utility companies allowed people to [prepay and] have credit balances on their accounts and didn't charge any fee, the question remains, what does the utility do with the money? All the negative yield on money is supposed to do is reduce the amount people want to hold. If the utility spends the money on other financial assets or spends it to construct a new plant, the negative yield on money has done its job. 
Cochrane also says that people could prepay their mortgages or their rent... what are the monetary consequences?   Paying down bank mortgages tends to contract the quantity of money.  However, any single bank receiving such repayments will accumulate reserves.   And the interest rate on that form of money is negative as well... Banks are motivated to purchase other assets due to these negative yields on reserves.
Scott Sumner and JP  Koning make similar arguments. The good news here is the John Cochrane can take solace in knowing the ZLB can be tackled without eliminating cash via Miles Kimball's approach and it can be effective in restoring full employment. Negative nominal interest rates will still get the monetary hot potato going.   

Along these lines, it is worth noting that another way of looking at the ZLB problem is that it creates excess money demand. That is, the ZLB prevents the desired money holdings of individuals from lining up with the supply of money. This is a big deal because, as Nick Rowe likes to reminds us, money is the only asset on every market. Disrupt the supply of or demand for this one asset and you will disrupt every market. In the case of excess money demand you get a recession. This 'monetary disequilibrium' view of the ZLB implies, therefore, that the real reason we may sometimes need negative nominal interest rates is to restore monetary equilibrium.

Miles and Scott's Excellent Adventure


  1. David, I agree with everything you wrote from a technical viewpoint. However I have trouble with the implication of this argument. Is it actually desirable to do all of the things needed to enable negative nominal rates (eliminate currency or introduce the crawling peg)? When there's a better alternative, namely NGDPLT?

    After all, the other way to gain traction at the ZLB is for the CB to create inflation expectations that drive the real rate as negative as needed to clear markets. NGDPLT seems like it should be able to achieve that. So it appears that either NGDPLT or negative nominal rates can provide monetary policy traction at the ZLB. If so, it seems like NGDPLT has two important advantages over negative nominal rates:

    1) NGDPLT requires no changes to the financial system. It is purely monetary policy that fits within the existing Federal Reserve structure. Negative nominal rates, on the other hand, mean either eliminating currency or introducing the crawling peg, either of which requires non-trivial changes to the financial system.

    2) Negative nominal rates appears politically hopeless. Can you imagine the uproar you will hear from the public when you try to explain why their bank is helping itself to their money each month ("negative interest"???) and also charging outrageous fees to deposit cash? I view negative nominal rates as technically possible but politically hopeless, whereas NGDPLT, thanks to the money illusion, is a much easier sell.


    Kenneth Duda
    Menlo Park, CA

    1. There are two issues.

      A nominal GDP level target would be less likely to have problems with the zero nominal bound. It shares that characteristic with a price level target.

      However, that the probability is not reduced to zero.

      Suppose we had a nominal GDP level target, nominal GDP was on target, and a large decrease in credit demand resulted in market clearing interest rates below zero?

      Market Monetarists argue correctly that the quantity of money must needs to rise more than otherwise, but I think all of us recognize that this is going to require the purchase of assets other than those whose market clearing rate is negative. That is, longer and riskier assets.

      That exposes the money-issuer to risk of capital loss. In my view, the monetary authority would just be duty bound to bear that risk. The quantity of money should rise enough to be consistent with nominal GDP staying on its target level.

      If the actual money issuers were private, they would not do this, and the interest rates on money would turn negative.

      if the central bank is not willing to bear the risk it needs to, then allowing negative nominal interest rates is better.

      Also, if this problem is at all persistent, and the government is issuing money, the large balance sheet implies large amounts of government ownership of the economy. At the very least, a government body is going to be allocating a substantial amount of credit.

      An alternative is to raise trend growth rate of nominal GDP. So, we have more inflation always to make occasional problems with the zero nominal bound.

      This tends to rule out my preference for a stable trend price level, much less modest deflation consistent with the productivity norm.

      My view is that yes, nominal GDP level targeting is a good idea. One of its benefits is that problems with the zero nominal bound are less likely.

      But I also think that a shift from the use of an interest rate instrument (if it is used at all) to a quantity of money instrument should be seamless. And there should be no "bills only" rule either explicit or implicit. The opportunity to buy whatever it takes should be there.

      But also, the ability to break the tie to currency should be there as well.

      I have to agree, however, that abolishing currency so that we can keep the Taylor rule is pretty bad.

    2. Ken, yes, a NGDPLT should generally prevent the ZLB from occurring and that is my preferred approach. And I agree it would be a much easier to sell the idea to the public. Bill Woosley makes some good comments on this above.

      Also, given the signs of recovery in the U.S. economy, I believe the ZLB is a fading issue for us. So this is an academic debate here, but in Europe this is still something to think about.

    3. Bill Woolsey

      The issue of capital losses by the money issuer shouldn't exist in a fiat money system. The issuer can simply recognize money as a form of equity instead of a liability.

      If the central bank conducts heli drops of central bank issued emoney then no government ownership of assets problem will exist either. It will also have capacity to independently conduct heli's.

      The idea of emoney is really good and if heli's of emoney are performed you dont need negative rates or to demote physical currency.

    4. Bill and David, thanks for taking the time to reply.

      I hope we can push the Fed towards NGDPLT even if we begin to escape the ZLB. It's just better, even when not at the ZLB. For example, suppose there is a negative supply shock and a commodity price spike, eventually causing prices generally to rise. One way or another, real wages must fall because there's less supply. With NGDPLT, the CB will create some extra inflation to aid in the adjustment. With inflation targeting, the CB will tighten to keep prices down, causing a spike in unemployment while we wait for people to accept lower nominal wages or we wait for 1.5% inflation to eventually erode the real value of prevailing wages to the point where supply and demand are in sync again.

      Thanks again,


      Kenneth Duda
      Menlo Park, CA

  2. I think its questionable to assume that these various issuers of zero interest liabilities will not respond with some form of proactive liability management - instead of passively allowing themselves to be stuffed with funds they can only invest at a negative spread. Banks for example factor mortgage prepayment risk into their pricing (and charge prepayment penalties for it at least in Canada), and negative interest rates won't suddenly pre-empt that pricing strategy. Other examples offered by Cochrane similarly could be subject to pricing or quota controls.

    The hot potato theory seems to assume that such evasive liability management responses will not be forthcoming.

    Liability management fends off the hot potato.

    It still ends up being about interest rates.

    1. JKH, if I understand you correctly I think we agree. I was just working with the scenarios Cochrane gave as if they actually happened. But I suspect they wouldn't as you argue. If so, that means the issuer of liabilities are just passing the negative interest rate problem (and thus the hot potato) back to their customers. For example, retails stores might start charging their customers a fee on prepaid cards that is commensurate to the negative interest rate the retailers face from their banks.

  3. MV=PY

    Negative interest rates reduce M, increase V, leaving PY unchanged.

    1. M can be kept constant (or increased) by ordinary open market purchases.

      Consider a situation where interest rates on money are positive. A lower interest rate on money reduces the demand to hold money and raise V. But does that mean that M rises less than it otherwise would because less money is being credit to checking accounts as interest? Only if you assume that the only way the quantity of money can increase is by crediting interest to checking accounts.

      You can analysis the effect of a lower interest rate on money (all the way to negative) with a constant quantity of money.

      As a practical matter, if there is a shortage of money (MV is too low,) then increasing the quantity of money and reducing the demand would ordinarily be the best approach. That woudl raise both M and V. Or, at least, it is the way one would expect market driven money to work.

  4. This “problem” and the types of solutions available is actually a transformation of something that already exists for banks at the zero bound without negative interest rates.

    The monetarist prescription (simplified) as I understand it is that the Fed should have done more QE or better yet should have committed to permanent QE. This presumably is hinged to belief in the hot potato.

    But the hot potato thinking seems to assume that banks will not take action in their own liability management. It is all a question of degree. Whether banks earn 25 basis points, 0 basis points, or minus some number of basis points on reserves, they will respond with liability management pricing techniques (which they already have) that will shift the potato back in the other direction. What they will not do is force their own asset allocation and capital deployment rules in the direction of taking on new risk that requires an additional allocation of capital, other things equal.

    Liquidity management does not drive capital management in banking.

  5. The zero bound is a intellectual creation, not necessarily one that is rational.

  6. I agree with Ken Duda. I actually wonder if not only should QE be permanent, but continually--the Fed commits to $40 billion of monthly QE for the duration of subpar growth....

  7. The problem is remunerating excess reserve balances - IBDDs, during OMOs. Always, between 1942 and 2008 (i.e., before the payment of interest on reserves), the commercial banks, lacking bankable opportunities, bought short-term gov'ts from the non-bank public (thereby expanding the money stock), pending a more profitable disposition of their legal lending capacity (i.e., the CBs invested their funds in existing securities). This private-sector, counter-cyclically operation, reduced interest rates, and reduced the need for some gov’t infusions coming out of recessions, like TARP (which bolstered primary liquidity reserves), as bank managers sought to minimize their non-earning assets (whenever there was a lack of credit worthy borrowers).

    And remunerating excess reserve balances inverts the short-end segment of the yield curve allowing and incentivizing, the CBs to outbid the NBs for “specials” (e.g., T–bills), during LSAPs. It collapsed the NBs wholesale funding sources where short-term borrowings cost less than the policy rate. The CBs monopolized this sector of the money market as they could always pay more (and receive higher returns), than a non-bank counterparty.

    The IOeR policy changed the counterparty in which the CBs conducted their transactions. Transactions between the Central Bank and the Reserve bank are asset swaps. But transactions between the Central Bank and the non-bank public create both reserves and new money (pre-1961 procedures). And transactions between the CBs and the non-bank public (pre-Oct 2008 procedures), created new money.

    Excess reserves were never considered a tax. But even required reserves aren’t a tax. Reserves aren't a tax because under fractional reserve banking: an injection of clearing balances (legal lending capacity), allows the member banks to create a multiple volume of new money & credit & acquire a concomitant volume of additional earning assets. We should all have such a privilege. Every man a King. Share the wealth - Huey Long

  8. This analysis requires a reconciliation of the factors that contributed to money stock growth. You have to separate Reserve bank credit from commercial bank credit, etc. E.g., you have to back out the cash drain factor, increase in bank capital accounts, change in the composition of the CBs assets, etc.