Saturday, May 28, 2016

The Paper Gold Fallacy

George Selgin wrote another of installment of his excellent primer on money.   Nick Rowe has written two blog posts about central bank profits.  Here and here.

Selgin introduces what he says is an old-fashioned distinction between money proper and money substitutes.   He explains that this distinction was initially between gold coin and various claims to gold coin.   That would have included transactions accounts issued by private banks but also redeemable currency issued by a central bank or even the Treasury.    In the U.S. today, however, Federal Reserve notes (and presumably any Treasury currency and the token coinage) count as the money proper while transactions accounts count as money substitutes.    Selgin also counts the reserve deposits held by banks with the Federal Reserve as a money substitute--a claim to "money proper" the Federal Reserve notes also issued by the Fed.  (Of course, both now and before various other highly-liquid assets might be counted as money substitutes or not--such as overnight repurchase agreements or noncheckable savings accounts or even Treasury bills.)

Rowe discusses the profits a government earns from its central bank.   He claims that most economists see the profits to be the flow of new currency issued by the central bank.   He says that central banks don't account for it that way but rather count it as the interest earnings on their asset portfolio.   When a central bank issues currency, it purchases assets such as government bonds.   The interest on those assets generate a flow of revenue.    Rowe explains that the present value of the two flows of income are the same.

I don't disagree.   Under the first approach, the government only earns profit now from new issues of currency.    The currency issued in the past provided profit in the past, but no additional profit now or in the future.   Under the second approach, the entire currency issue allows profit to be earned now, but any increase in the quantity of currency only adds to profit by the added interest earnings from the larger asset portfolio it allows, not the entire principal amount of the newly issued currency.

I think Rowe's claim that many economists treat the flow of currency issue as profit has an important element of truth.   For example, there is a tradition of thinking about budget deficits as being financed by either money issue or debt.   While I am not aware of many economists describing this money issue as "profit," it is certainly treated as a current source of revenue.   Of course, this source of revenue is often described as an "inflation tax."   If economists were consistent, then, they would say that government spending is financed by a variety of taxes--on income, sales, what have you, and also by an inflation tax.  If government spending is greater than that amount, then it may run a budget deficit, which would always by funded by debt.   That tradition,  of treating money issue as part of "deficit finance" rather than just one of the many sources of tax revenue, appears inconsistent.

Suppose a nation on a gold standard issues redeemable hand-to-hand currency.   Since it is redeemable, it is clearly a liability--a type of government bond.  According to Selgin, it is not "money proper," but rather a money substitute.   If, as is usual, there is no nominal interest rate paid on this currency, the government is borrowing at a zero nominal interest rate.   Whether or not the government is running a budget deficit or budget surplus is really beside the point.   The government funds part of its national debt at a lower interest rate than otherwise.

Central banks, like the Bank of England, started as private banks set up to provide loans to the government.   The central bank issued redeemable hand-to-hand currency with a zero nominal interest rate.   They borrowed at a very low (zero) interest rate.    Then they shared some of that benefit with the government by lending to the government at a low (though positive) interest rate.  

There were plenty of reformers who complained about the share going to the bankers and insisted that the government should cut out the middleman and issue the currency iself.   When the Federal Reserve was set up, the U.S. national debt was being paid off, and there was concern about the inflationary consequences of money-financed budget deficits.   The redeemable currency was supposed to finance "real bills," short term commercial loans, with the benefit shared according to a formula.  The bank-shareholders of the Fed were to get a share of the income compensating them for their capital investment--something considered essential due to the risk of failure.   The government was left as the residual claimant.    Today, the Fed, like nearly all central banks, is effectively nationalized with the earnings going to the government, though even today, the U.S. member banks do earn something.

For many years, the Fed's asset portfolio was almost entirely government bonds.   The U.S. government ran budget deficits and had a large and growing national debt.   The Fed collected interest from the government, but nearly all of that money was given back to the government.   It seemed to be little more than a shell game.   If the increase in base money was less than the budget deficit, the notion that the flow of new money was an alternative to debt as a source of deficit finance seemed plausible enough.    That there has been persistent price inflation over the period makes the term "inflation tax" also quite reasonable.   If the quantity of currency had increased no more than the demand to hold it at constant purchasing power, "inflation tax" might seem like a stretch.   If the quantity of currency increased an even smaller amount, so that it appreciated in value "inflation tax" would seem a misnomer.   But that was hardly the issue during the period of high and rising inflation during the seventies.

Since 2008, there have been some problems with this traditional framing.   First, the increase in the monetary base was greater than the increase in the national debt.   The Fed sold of part of its holdings of government bonds and expanded its balance sheet by purchasing private securities--mortgage backed securities guaranteed by the Federal government.   Finally, the Fed began to pay interest on reserve balances and those reserve balances became the larger part of the Fed's liabilities.   The tangible hand-to-hand currency that pays no interest remains large in absolute value, but no longer is the other form of Fed liabilities trivially small.   Perhaps most importantly, the Fed remains committed to an inflation target, so that it is committed to do something to make sure that the large increase in the monetary base has no more than a minimal impact on inflation.   Most obviously, the large increase in the monetary base is temporary, though the Fed could continue to hold a large asset portfolio by paying a sufficiently high interest rate on reserves or borrowing in some other way.

I find it difficult to characterize the issue of hand-to-hand currency under this scheme as anything other than the Fed funding a portion of its asset portfolio at a low (zero) nominal interest rate and negative real interest rate.   This is profitable for the Fed and these profits are almost entirely transferred to the government.    I am not saying that it is impossible to frame this as the flow of newly-issued currency being profit to the government.  I just don't see that as a helpful framing any more.

Selgin's essay explicitly treats government-issued hand-to-hand currency as "money proper" and states that the banks' reserve balances at the Fed are claims to currency.   This is a bit unconventional.  If anything, the more usual view is that because the Fed adjusts the quantity of currency passively to demand, it is bank reserves that are more important.

Of course, traditionally, it was the interest rate on inter-bank lending in the reserve market that was emphasized by the Fed and most monetary economists.  The interest rate paid on those reserves began to receive more attention since 2008.   And the interest rate at which the Fed lends reserves to banks received much more emphasis in years long past and actually played a significant role in Fed policy in 2008 and 2009.

Still, this process of passive adjustment of the quantity of currency to demand very much involves banks redeeming reserve balances for currency and making currency deposits as well.   But is Selgin's distinction economically useful?

I describe both Selgin and Rowe's approach as the "paper gold fallacy."   For a government with an entirely irresponsible monetary policy, then the paper gold framing is useful.    In a gold standard country, gold prospectors can pick up gold nuggets and spend them.   What this does to the purchasing power of money is likely to be of little interest to them.   With a fiat currency and a completely irresponsible government, money is printed more or less for free and then spent by the government on whatever it wants.   The "irresponsible" element of this is that the government doesn't worry about what this does to the purchasing power of money.   (It is the evil capitalist speculators causing the inflation say the socialist leaders of Venezuela as they print up money and spend it.)

With a gold standard, the flow of new gold is small relative to the existing stock of gold.   While the rate of change is unlikely to be constant, it is easy to see how some economists would see intentionally limiting the issue of new money to a slow, stable growth rate as a way of imposing responsibility on a government issuing fiat currency.   Rather than just printing up money and spending it willy-nilly with never a thought about what this might do to the purchasing power of money--exactly like our gold prospector--a responsible government might follow a quantity rule limiting the issue of paper money to something like what would occur with the quantity of "money proper" under a gold standard.   Just as the quantity of gold rises a slow, somewhat steady rate, the quantity of paper money can increase at a slow, perfectly steady, rate.   This will provide the government with a source of revenue, but it is limited in a responsible way.

Many economists have had a pretty pessimistic view about the prospects of responsible issue of fiat currency.   There are plenty of examples of wildly irresponsible monetary regimes.   And when the last ties to gold went away in the late twentieth century, the major industrialized countries certainly looked pretty irresponsible as the Great Inflation developed.   And other than the tiny remnant who favored a return to redeemability, most advocates of a constrained monetary policy supported a constant growth rate in some measure of the quantity of money.

In my view, economists who were critics of the gold standard in the nineteenth and twentieth central were never advocating an irresponsible monetary policy nor did they advocate a quantity rule for paper currency.   A much more typical view would be that the quantity of money be adjusted to stabilize its purchasing power.   Clearly, any such regime is inconsistent with the government simply printing money up and spending it without constraint.   Creating money cannot be like gold prospectors picking up gold nuggets and just spending them.

And further, adjusting the quantity of money to keep its purchasing power stable is not the same and at least potentially inconsistent with some notion that "responsibility" involves the government just printing money and spending it at a limited rate, more or less like the stock of gold grows at a slow rate.  

In fact, the economics of gold mining does result in the flow of gold adjusting in a way that tends to stabilize its purchasing power.   A higher relative price of gold makes it profitable to expand the production of gold and a lower relative price of gold makes it less profitable to add to the existing stock of gold.   Economists who were critical of the gold standard claimed that adjustments in the quantity of an irredeemable paper money could improve on that process and provide more stability.

If the demand for hand-to-hand currency is always growing, then the issue of currency could well be treated as a source of government revenue even if it is constrained so that its purchasing power be stable. When the demand for currency grows more rapidly, then more revenue would be generated from this source.   When the demand for currency grows more slowly, less revenue would be generated.

But what happens if the demand for currency decreases?   With the irresponsible monetary policy, understood like the gold prospectors spending the nuggets they pick up, then this means more inflation than otherwise.   In fact, decreases in the real demand for currency is just a step in the predictable process of hyperinflationary collapse.   When a "responsible" monetary policy is understood as a constant growth rate of currency, then a decrease in the demand for currency is also inflationary.   It raises the growth path of the price level.   Under this sort of regime, if we imagine that the demand for currency might fall to zero one day, we are left with puzzles as to why such currency has any value today.

However, there is nothing difficult about handling a decrease in the demand to hold hand-to-hand currency.   Governments and their central banks issued redeemable hand-to-hand currency and the possibility of a decrease in the demand to hold that currency was a constant source of worry.   Central banks hold assets and can sell them off as needed to reduce the quantity of currency (or reserves.)    But even if the government is typically just spending currency at a variable rate according to the rate of increase in the demand for currency, all it needs to do in response to a decrease in the demand for currency is to sell government bonds.

Framing the issue of currency as the government's profit from the central bank might be useful if the demand for currency always grows--even with a responsible monetary policy.  But I don't think that a public finance regime where total government spending changes with the rate of growth in currency demand is sensible.   I am pretty sure that varying other tax rates to stabilize government spending when currency demand grows more or less quickly would be unwise.   Keeping tax rates stable is good policy.  A much more sensible approach would be to sell bonds when currency demand grows more slowly and buy them back when currency demand grows more quickly.   To me, a much better framing of such a scheme is that the government is running a budget deficit all the time and it is funding part of the national debt by issuing a type of debt that has a zero nominal interest rate when it can and funds it with interest bearing debt when it must.   While the interest rate on most types of debt would change to clear the market, the issue of currency would have to adjust with the demand to hold it at a zero nominal interest rate.   (Of course, the nominal and real demand for currency would depend on the nominal anchor.  I favor a stable growth path for nominal GDP rather than a stable price level or modest inflation.)

From this perspective, there is no puzzle at all about what happens if currency demand actually falls.  Just sell more bonds.

And this is why "helicopter money" is meaningless.  If the Treasury issues currency and spends it, it can always withdraw it from circulation by selling bonds.   There is nothing "permanent" about the quantity of any sort of money.  And if the goal is to maintain the purchasing power of money (or have its real purchasing power depreciate at a constant rate), then changes in the quantity of money should not be permanent.

Is the notion that the demand for currency might fall nonsense?  Is it reasonable to follow Rowe's approach and treat the demand for currency as a fixed proportion of nominal GDP?   Perhaps for Canada, but in the U.S. substantial amounts of currency are held in other countries.   What if they start to use Euros?   Further, the "legitimate" use of currency could be substantially replaced by improved electronic payments.   Scott Sumner is always emphasizing that most currency demand is really by criminals (including?) tax evaders.   Suppose some types of vice are legalized or the tax system is improved so that there is less motivation (or ability) to evade taxes even using currency transactions?   It seems to me that substantial reductions in the demand for currency are quite possible.

And, of course, there is the possibility of allowing private banks to issue hand-to-hand currency.   In my view, it really isn't that difficult to pay interest on that currency to those who withdraw it from their banks.   If permitted, this could greatly reduce the demand for government currency.

Treating currency issued by the government as a special type of debt, that allows the government to borrow and a low (zero) nominal interest rate and with our current inflationary policy, at a negative real interest rate, is a much more sensible way to frame this when it is understood that the demand for hand-to-hand currency is not something that is always growing.   And further, a quantity rule for any sort of money, much less currency alone, is not desirable.  And, of course, printing money and just spending it is a recipe for disaster.

Returning to Selgin, I see the identification of currency as "money proper" as simply another aspect of the paper gold fallacy.   There is no reason to expect that the quantity of currency will behave like the quantity of gold.   And it almost certainly shouldn't behave like the quantity of gold.   The quantity of currency should adjust according to the demand to hold it.    If the central bank is treated as autonomous, it is a liability of the central bank.   Just as a private bank can and should adjust the quantity of hand-to-hand currency according to the demand to hold it, so should the central bank adjust the quantity of hand-to-hand currency it issues according to the demand to hold it.   If the central bank's balance sheet is consolidated with the rest of the government, then hand-to-hand currency is a type of government debt and the amount issued can and should adjust with the demand to hold it.   That is, with the desire to lend to the government in that particular way.

In my view, most payments are made with transactions accounts.   When these payments are cleared, they are redeemed with balances at the central bank.   I think that today, the most important element of redemption is with the reserve balance portion of the Fed's balance sheet.   To me, emphasizing hand-to-hand currency is the tail wagging the dog.

I think the paper gold fallacy (framing) has served some economists poorly in recent years.   The huge increase in the quantity of reserves was supposed to cause hyperinflation.  Why?  Because of an implicit assumption that it was permanent.   Just as newly discovered gold nuggets would be permanent.    Well, it might be, but it doesn't have to be and most market participates clearly don't believe it will be.

But more importantly, even though governments and central banks have no redemption requirement for hand-to-hand currency, they should act as if they do.  That is, treating the issue of new currency as profit and not worrying about its purchasing power should be seen by everyone for what it is--irresponsible.


  1. Assuming full employment then current govt spending can be financed by
    - tax now
    - tax in the future (bond sales in the present)
    - inflation tax

    All 3 of these are real costs to the non-govt sector which gets real-income (now or in the future) at the expense of the public sector. The inflation tax method is "helicopter money" and it really doesn't matter (apart from to accountants) if its done by direct money creation, or by the CB buying up a qty of govt bonds.

    When there is less than full employment the situation seems different. In theory at least the govt should be able to fund the buying up of unused resources (or stimulating AD so the private sector buys them up) without reducing anyone's real income.

    Assume it does this with helicopter money. Overall output increases, and no-one has to give up any real income in the present period to fund the extra govt spending.

    Assume that in the next period the recession is over. However the helicopter money injected in the previous period will not become inflationary. This can be addressed either just by letting the one-time price increase take place, or by taxing the money away. Unlike helicopter money at full employment the tax will be more apparent than real. The money raised is just destroyed and no-ones real income is affected. If the inflation option is used again there is just an increase in prices - no inflation tax.

    Seems like helicopter money at less than full employment really is a free lunch (at least in theory - I know that in practice its way less clear-cut.)

    1. I agree that correcting monetary disequilibrium can increase real income. Your helicopter money allows the government to capture this gain. I suppose if they give the new money away to households, it will be used to fund the consumption of those who get the money. With standard open market operations or private money issue then consumption and investment increase optimized by the interest rate. I don't favor you "let the price level increase" solution. Changing tax rates to adjust the quantity of money is a bad idea--keeping tax rates stable is a good idea. Selling bonds can reduce the quantity of money. The issue, which I didn't really discuss explicitly, is the notion that temporary changes in money don't necessarily impact spending on output and helicopter money is no more permanent than any other type. The solution is level targeting.

  2. 2 typos:
    "All 3 of these are real costs to the non-govt sector which gets real-income (now or in the future)" = "All 3 of these are real costs to the non-govt sector which gets LESS real-income (now or in the future)"

    "However the helicopter money injected in the previous period will not become inflationary" = "However the helicopter money injected in the previous period will NOW become inflationary"

  3. I think the 85 per cent of this that I believe I understand is quite brilliant.

    A great number of objective points made that I've rarely seen made in blogosphere writing. I also happen to agree with your general argument about profit calculation.

    The other 15 percent may well be equally right, but I'm puzzled by it.

    Here is my core example:

    Consider the pre-2008 Fed.

    I think you say that the Fed should sell bonds if the demand for money falls - which I think means velocity has increased with corresponding inflationary pressure. Which means the quantity of money supplied is too high - hence the bond sales.

    The problem I see there is that the quantity of bank reserve balances was quite minimal in that pre-2008 regime - one of the very good points you make.

    In fact, it was sufficiently minimal that any material drain of reserves from bond sales would cause the Fed to loose control over the effective trading rate for fed funds - I. E. rates would skyrocket.

    (You are probably familiar with the numbers, but just check out the data on excess reserve balances starting at 2008 and going back to the 1950's at least. ER balances rarely exceeded $ 2 billion - a consequence of the extraordinary leverage that a relatively small amount exerted on the clearing rate for the effective fed funds rate.)

    So that interest rate response to a meaningful quantity of bond sales would immediately require reserve creating OMO purchases - such as bond purchases - to return the fed effective rate to target and ER balances to the original level.

    So I'm not sure how a desired decline in currency would result from that process.

    1. Under the regime you describe, a decrease in currency demand directly results in the deposit of currency in banks. This creates excess reserves and downward pressure on the overnight lending rate. Given the target, this triggers sales of bonds. What has really happened is that the central bank has sold bonds and reduced the quantity of currency. Your analysis is exactly what I meant when I said that the conventional view focuses on reserve balances rather than currency, which is contrary to the spirit of Selgin's framing where currency is the "money proper," though not logically incompatible.

    2. "a decrease in currency demand directly results in the deposit of currency in banks"

      you're right of course

      not quite sure why I overlooked that


  4. Does the price inflation target apply to the lender of last resort function of the central bank to the commercial banks?

    This matters to keep the fixed exchange rate between currency and demand deposits.

    1. Does the price inflation target apply to the lender of last resort function of the central bank? I think that it can and should. (Though I favor a target for the growth path of nominal GDP.) I am not sure that your question is based upon a sound understanding of the lender of last resort function.

  5. "Suppose a nation on a gold standard issues redeemable hand-to-hand currency. Since it is redeemable, it is clearly a liability--a type of government bond."

    Let's assume that means fixed exchange rate and a permanent one. Also, assume a 100% reserve requirement. Gold is probably considered a commodity. Currency could be considered a commodity. It looks like two commodities with a fixed exchange rate that just happen to be MOA and MOE.

    Where is the gov't bond part?

    1. If hand-to-hand currency is backed 100% by gold bullion or gold coin, it isn't much different from a 100% gold coin standard. I don't think such a policy is desirable. A bad policy, in my view and not one with much real world relevance. This was attempted on a marginal basis in Great Britain.