Saturday, January 26, 2013

Negative Nominal Interest Rates

Richard Anderson and Yang Liu have a short article on the Federal Reserve Bank of St. Louis website, How Low Can You Go, Negative Interest Rates and Investors Flight to Safety.   They describe the unusual episodes in which real and even nominal interest rates have been negative.   They argue:

Nominal interest rates also, at times, are negative. Generally, each occurrence of a negative rate has its own special story. Most stories involve fear or uncertainty, with investors fleeing to perceived safer assets.

However, they are dismissive of proposals to set policy rates at less than zero.   They state:

 Some analysts have argued that such examples suggest that central banks should consider setting negative policy rates, including negative rates on deposits held at the central bank. Such proposals are foolish for a number of reasons. First, a policy rate likely would be set to a negative value only when economic conditions are so weak that the central bank has previously reduced its policy rate to zero. Identifying creditworthy borrowers during such periods is unusually challenging. How strongly should banks during such a period be encouraged to expand lending? Second, negative central bank interest rates may be interpreted as a tax on banks—a tax that is highest during periods of quantitative easing (QE).3 Central banks typically implement QE policies via large-scale asset purchases. Sellers of these assets are paid in newly created central bank deposits, which, in due course, arrive in the accounts of commercial banks at the central bank. It is an axiom of central banking that the banking system itself cannot reduce the aggregate amount of its central bank deposits no matter how many loans are made because the funds loaned by one bank eventually are redeposited at another. Is it reasonable for the central bank to impose a tax on deposits held at the central bank when the central bank itself determines the amount of such deposits held by banks and the banking system?

"Such proposals are foolish"!  

What a mess of fallacies.  

Banks are financial intermediaries.   They borrow money by issuing deposits and they lend the funds either by purchasing bonds or else making various sorts of loans.   

If the interest rate paid on the reserve balances banks keep with a central bank is reduced, this provides an incentive for banks to hold fewer reserves.   On the asset side of their portfolio they can instead purchase securities or make loans.  On the liability side of their portfolio they can pay lower interest rates on deposits.  

While it might not be wise for banks to make additional loans to less credit worthy borrowers, it is almost certainly sensible to lower the interest rate charged to both more credit worthy and less credit worthy borrowers.    Further, banks can always purchase securities or lower the interest rates paid on deposits.   If the yields on sufficiently safe securities are too low, that still leaves a lower interest rate on deposits.

If banks cannot find credit worthy borrowers willing to pay high interest rates, then the interest rates they pay on deposits should be lower.   That is the fundamental nature of a financial intermediary-- covering costs and earning profit based upon the difference between the interest rates charged and paid.  If banks can only earn less, then they must pay less too.

If the interest rate on reserve balances that banks hold with the central bank falls below zero, this same reasoning holds.  It motivates banks to hold more securities, make more loans, or else reduce the interest rates paid on deposits.   

It is quite possible that the yields on safe assets, like short term bonds issued by credit worthy governments, would be driven below zero.   Similarly, competition among banks for those borrowers who are most credit worthy may allow those particular borrowers to pay negative interest rates.   And most importantly, the interest rates paid on bank deposits, especially government insured bank deposits, might turn negative.   In this sort of scenario, the interest rates the less credit-worthy borrowers must pay should also fall.  The banks' cost of funds and the opportuntity cost of such loans is falling.

Of course, why are economic conditions weak?   If nominal GDP has fallen below a reasonable growth path, then these lower interest rates will tend to cause nominal GDP to recover.   While it is true that an expansion of lending to less credit worthy borrowers who spend the added borrowed funds on consumer or capital goods will tend to cause spending on output and so nominal GDP to rise, that isn't the only possibility.   Those selling securities to banks might use the funds to purchase consumer or capital goods.   Lower yields on existing securities might encourage new issues to fund capital goods.  And lower interest rates on bank deposits might encourage those holding the deposits to spend their funds on consumer or capital goods.

Of course, if everyone (or even many people) know that if nominal GDP falls below target, interest rates will fall to whatever level necessary for nominal GDP to recover, then this will tend to dampen or even prevent "economic weakness" in the first place.   If incompetent central bankers instead insist that interest rates must be kept up because banks might make too many loans to borrowers who are not credit worthy, then not only will decreases in credit demand lead to "economic weakness," self fulling expectations of recessions become possible.

Aside from failing to understand the basic economics of  banking, what about this argument that negative interest rates are a "tax" on banks and that the banking system must hold whatever amount of reserve balances the Fed chooses to create?

Suppose the central bank wants banks to reduce their lending and instead hold balances with it, so that it can direct credit  to where it is most needed.   In other words, the banks would make loans or purchase securities according to their criterion for risk and return.   This might leave some worthy projects unfunded.   If banks can be discouraged from lending, then that will free up funds for the central bank to direct according to the public interest.

For example, suppose banks determine that housing is overbuilt and overpriced and cut back on their holdings of mortgage loans and mortgage backed securities.   This results in less credit being provided for new mortgages and makes it difficult for worthy borrowers to obtain homes, for realtors to make money, for construction firms to make profit, and for construction workers to find jobs.   And so, the central bank can encourage the banks to reduce their purchases of safe securities, such as government bonds, or their loans to safe businesses, and instead get them to hold reserve balances.   The central bank can then purchase mortgage backed securities.   Instead of credit being directed by imperfect market forces, it would be directed according to national priorities (which frequently means funnelled into housing.)

Surely, if that it the central bank's goal, then charging banks to hold reserve balances would be foolish.  Instead, the central bank should pay banks to hold reserves.

I think it is fair to say that most economists considering negative nominal interest rates on the balances banks hold at a central bank consider having the central bank allocate credit according to the priorities of the politicians to be foolish.  

Now, suppose instead that the situation is quite different.   There is a flight to safety.  Banks share the now greater aversion to risk, and reduce their loans to relatively-risky borrowers and instead hold reserves.   The total quantity of reserves (or base money, anyway) is determined by the central bank.   If the quantity of reserves remains unchanged, then as each bank tries to accumulate reserves by reducing lending, the impact is a reduction in the quantity of checkable deposits created by the banking system.   If the demand to hold checkable deposits were unchanged, then the resulting shortage of money would result in asset sales and reduced spending out of current income.   The result would tend to be higher interest rates and reduced sales of output.  

Of course, the scenario is a scramble to safety.    At the same time this problem would be developing, firms and households would be selling off relatively risky financial assets and purchasing safe assets.   The monetary contraction might tend to raise "the" interest rate, but the shift in relative yields might result in relatively safe assets having lower yields than before.  

Further, government insured checkable deposits would be another "safe" asset, so rather than there simply being a decrease in the quantity of checkable deposits as banks reduce lending to risky borrowers, there would also be an increase in the demand to hold government-insured deposits, including checkable ones.

To avoid a contraction in the quantity of money, a central bank must expand the quantity of reserves to match the increase in the demand to hold them.    If the demand for checkable deposits is also growing at the same time, this increase in reserves must be redoubled so that the quantity of checkable deposits will expand to match the demand.  

So, rather than the central bank trying to discourage banks from lending or holding securities, and instead encouraging them to hold reserve balances so that it can direct credit where it thinks is best, we have a situation where the central bank is expanding reserve balances in order to offset the impact of banks choosing to lend less and instead hold reserve balances, and further, to accommodate any increase in the demand by households and firms to hold larger balances in checkable deposits.

To the degree that banks can be encouraged to hold additional securities or make additional loans to whatever borrowers they still find credit worthy or even to lend more to those borrowers that they had decided were not so credit worthy, the amount of reserves the central bank must create is smaller.   Further, to the degree banks lower the interest rates they pay on deposits, especially government insured deposits, it will deter households and firms from accumulating larger balances and so reduce the needed increase in the quantity of both checkable deposits and reserves.

But then, why not just have the central bank create more reserves?   Open market purchases are not difficult to do.   But in the scenario where there is a scramble for safety, the sort of government bonds central banks typically purchase are just one such asset that everyone else is trying to buy.   For every such bond the central bank buys, it simultaneously creates an additional unsatisfied demand for similar safe assets.   The result could easily be that banks selling short and safe government bonds to the central bank will simply expand their demand to hold reserves and firms and households selling them will expand their demand for government insured deposits.

One solution to this problem is for the central bank to purchase risky assets.   It is thereby creating safe assets--directly reserve balances for banks to hold, but indirectly government insured deposits at banks for households and firms to hold.    By purchasing enough risky assets, the central bank should be able to expand the quantity of reserves and checkable deposits enough to match the increased demand for them, and so prevent any decrease in spending on output.

But, of course, the central bank is bearing more risk.   Assuming that the politicians will bail out the central bank, this means that the taxpayers are bearing additional risk.   (Of course, the expansion of government insured deposits is another avenue where the taxpayers are bearing additional risk.)

Why should the central bank, and the taxpayers, bear the added risk?  Or, perhaps, we might say, why should the taxpayers provide that service for free?

If a central bank charges banks for the reserve balances they hold with it, then it is charging the banks for the service of sheltering them from risk.   To the degree that this added cost is passed on to bank depositors, so that they earn lower (or even negative) interest on government-insured deposits, then the depositors are paying for the service of being protected from risk.

If the central bank simply adjusts the interest rate it pays on reserve balances with the yield it can earn on safe assets, then when there is a scramble for safety, the interest rate on reserve balances will fall with the yields on other safe assets.    That banks could and would  reduce the interest rates paid on deposits when there is a scramble for safety is also desirable.    The result would be that the central bank would not need to increase the amount of reserves as much to keep spending on output on a slow, steady growth path.

And that is the answer to the supposed puzzle about banks being compelled to pay to hold reserve balances when the total amount that they hold depends on the central bank.

The total amount they have to hold depends on the amount that banks chose to hold given the central bank's nominal anchor.  

Of course, any single bank can reduce its holdings of reserve balances (leaving aside reserve requirements) by purchasing securities, making loans, or paying lower interest on deposits.   This does, of course, shift the reserves balances over to other banks.   But when all banks purchase more securities and make more loans, that increases the quantity of money.   When all banks pay less interest on checkable deposits, that reduces the demand to hold money.   When the quantity of money rises and the demand to hold money falls, this results in more spending on output, which requires the central bank to reduce the quantity of reserves to avoid excessive growth in spending on output and inflation.

It is really quite simple.   The lower the interest rate paid on reserve balances, the lower the quantity of reserve balances the central bank needs to create to keep spending and inflation on target.  And the higher the interest rate paid on reserve balances, the more reserve balances the central bank needs to create to keep spending and inflation on target.

What is the problem with a negative interest rate on reserve balances and negative interest rates on deposits?   It is that it will tend to result in an increase in the demand for vault cash by banks and currency by households and firms.   The problem is that hand-to-hand currency has a zero nominal interest rate.

The lower bound on the interest rates the central bank charges banks on reserve balances and banks charge their depositors depends on the cost of storing currency.   Efforts to lower nominal interest rates below the cost of storing government issued currency is foolish. 

That is why the alternative of having the central bank purchase risky assets is necessary unless some provision is made to break the tie between deposits and government-issued currency.   A simple suspension would be possible, with currency rising to a premium and then falling gradually back to par with deposits.    Kimball's approach of managing a depreciating exchange rate so that currency has a negative yield would be possible too.    That approach works by reducing the price at which currency is accepted for redeposit.   And finally, full privatization of currency (my preferred approach) would also provide more room for negative nominal interest rates, though when interest rates become too negative, there would be shortages of currency.   All of these are ways to reduce the risk that the central bank (and the taxpayers) must bear when there is a scramble for safety. 


  1. We probably don't actually need negative rates. But we desperately need the possibility of negative rates.

    Once the Fed demonstrates that it's willing and able to set an arbitrarily negative rate, the equilibrium rate will surely increase.

    And so we shouldn't worry about how currency users will suffer. They won't suffer, or at worst the suffering will be brief.

    But what if the economy really, truly needs negative rates for the long haul (given a 2% inflation target)? Seems unlikely, but if it's a choice between making currency users suffer a little and making everyone suffer a lot, the latter is the lesser evil.

  2. Regarding the "tax" on reserves: this is equal to the difference between IOR and Fed Funds. And the Fed can't simultaneously set this "tax" and the quantity of reserves. It's impossible, since with IOR < FF, an injection of reserves (unrelated to demand) would cause FF to fall.

    Nothing about a negative Fed Funds rate is a tax, since banks aren't obligated to pay a positive rate of interest on deposits.