It's Supply and Demand--Cole and Ohanian on the Great Depression

I agree with Cole and Ohanian's arguments that efforts by the federal government to reduce supply and use price floors to keep prices high during the Great Depression were very counter-productive. However, the real problem during the period was falling demand, and then, after a modest recovery, an excessively low level of demand for most of the decade. Reducing supply, or keeping prices high despite surpluses, was a mistake, but reversing the decrease in demand was by far the best solution.

Like Cole and Ohanian, I do not think that large expansions in the size and scope of government spending programs were desirable. While I can see various pathways by which additional government spending might increase demand overall, monetary institutions that keep nominal spending on output on a slow, steady growth path are a much better approach. It is in that context that the opportunity cost of government programs can be better perceived, and further, any benefits of government programs, particularly social spending programs, can be understood.

However, Cole and Ohanian want to argue for a "supply only" account of the Depression. I agree there is some growth path of prices and wages that will allow real expenditures to remain equal to the productive capacity of the economy, given almost any growth path of nominal expenditures. Cole and Ohanian, however, are mistaken when they claim that falling prices and wages were generating a recovery of real output near the trough of the Great Depression.

David Glasner's response shows that rising demand appears to have been causing both higher production and prices in the fall of 1932. Unfortunately, soon after, demand began falling again as did both output and prices. A sustained recovery in demand only began in March of 1933, with both output and prices rising after the suspension of gold payments.

Of course, Cole and Ohanian might respond that prices and wages needed to be lower still, and I would agree--leaving aside the possibility of reversing the massive decrease in nominal expenditure. However, I would like to emphasize that the Hoover and Roosevelt anti-supply-side policies aimed at keeping prices and wages from falling and trying to force them back up were not some kind of bolt from the blue. If money expenditures on output had continued on the growth path of the twenties, it is unlikely that Hoover would have decided that nominal wages should be suddenly hiked to generate even more rapid growth. And Roosevelt might never have been President, much less supported or passed an unconstitutional grandiose scheme of price fixing. How successful would have labor organizers been if they didn't have the argument that the bosses are trying to cut your wages? How much political support would pro-union legislation have had if there had not been surpluses of labor and both the incentive and ability to force pay cuts?

Allowing money expenditure on output to fluctuate while depending on widespread wage and price adjustments to keep real expenditures equal to productive capacity seem to be highly counterproductive. To me, both the Greater and the Lesser Depressions are showing that to be true.

Sumner on a Pure Credit Economy

Scott Sumner has responded to the "hypothesis" of a pure credit money.

Sumner correctly points out that the U.S. certainly does not seem to be in a situation where the demand for base money--currency and reserve balances at the Fed--is withering away. The quantity base money is remarkably high by historic standards. Further, the standard Market Monetarist account of the status quo is that the demand to hold those balances is higher still.

Sumner also correctly points out that if a monetary authority chooses to target interest rates, or anything else, the quantity of money becomes endogenous. It is adjusting the monetary base to hit its target. To respond to a claim that the monetary authority should target base money with detailed explanations that amount to saying that it doesn't, is wrongheaded.

If we did live in a world where the demand for hand-to-hand currency was decreasing and new methods of settling inter-bank clearings were allowing for a reduction in the demand for reserve balances, then this would make holding the quantity of base money constant inflationary. From a Market Monetarist perspective, the necessary course of action would be to decrease the quantity of base money in step with the demand to hold it.

Of course, Market Monetarists have never proposed fixing the quantity of base money and having the price level adjust so that the real quantity adjusts to the real demand. Instead, they favor a target for the growth path of nominal GDP. In this scenario, the nominal quantity of base money would be reduced in step with its falling demand so that nominal GDP would remain on the targeted growth path.

Only when the demand for base money, both hand-to-hand currency and reserve balances, is zero, would controlling the quantity of base money so that it equals the amount demanded be a bit paradoxical. It is in this scenario where Woodford argued that interest rates could be controlled by the relatively conventional channel method. The central bank sets an interest rate at which it will lend to banks by creating balances in their reserve deposits. And it also chooses a lower interest rate that it pays banks for holding those balances. By raising or lowering both, the central bank controls short term interest rates throughout the economy.

Of course, simply paying interest on reserve balances would slow any process that was leading banks to reduce their demand to use them for clearing purposes. Still, if the demand for base money does fall to zero in equilibrium, that doesn't prevent the monetary authority from using open market operations to control nominal GDP, the price level, or some conglomeration of privately-issued deposits that it might call "money."

If the monetary authority wanted to increase inflation, raise nominal GDP, or increase the nominal quantity of some set of monetary assets, it would makes an open market purchase. The quantity of base money rises above zero. The seller's bank would have excess reserves--the desired quantity being zero by assumption. A bank with excess reserves gets rid of them in the usual way. It makes loans or purchases some other asset, or perhaps pays lower interest on deposits. Of course, this simply shifts the reserve balance to some other bank. The "hot potato" continues until whatever the desired nominal target is reached. Notice, that as the banks make loans or purchase securities they are increasing the amount of deposits in the hands of households and firms and as they lower the interest rates on those deposits they are reducing the demand to hold them. The private banking system is creating an excess supply of money.

What is unusual, is that if the demand for base money is zero in equilibrium, the monetary authority must make an open market sale once its nominal target is reached. This reduces the quantity of base money back to zero, the amount demanded in equilibrium.

The opposite scenario is even more peculiar. With the demand for base money being zero in equilibrium, the monetary authority has no assets to sell. How can it make an open market sale? It must borrow a security and then sell it. This is what is called short selling a security. (I am not sure I want to say that base money is negative.)

The buyer's bank then has a reserve deficiency. If the penalty for such a penalty is mild, for example, some penalty interest rate, then the bank with debit balance contracts credit by restricting new loans or selling off securities. Or it could pay higher interest rates on deposits. That just shifts the deficiency to some other bank, and the contraction continues until the monetary authority reaches its target. Again, notice that the banks are contracting the quantity of checkable deposits or raising the interest rates paid on them. The private banks are creating an excess demand for money. Again, when the monetary authority reaches its target, it must make an open market purchase, returning base money back to zero and paying back the security it borrowed.

If the penalty for a reserve deficiency was sufficiently harsh, then the demand for base money would have never fallen to zero. Allowing banks overdrafts at market rates leads to banks desiring to hold no reserve balances in equilibrium. And while the penalty rate suggested above is very similar to a central bank's lending rate in a corridor system, it does not have to be a target. It can be set at a rate higher than some market determined rate. Similarly, if the monetary authority pays interest on reserve balances, that interest rate can float as well, being set below some money market rate.

If the demand for base money falls to zero, a central bank can control interest rates. However, as Sumner explains, it is never sensible to have have an interest rate goal. This leaves the price level indeterminate. The interest rate must be adjusted to control some other nominal quantity. Woodford mostly focuses on an inflation target.

And if the demand for base money falls to zero, treating the growth rate or path of the monetary base as a target would be absurd. Still, there is no need to control interest rates. All interest rates can be left to market forces, and open market operations can be used to control some other nominal quantity--like inflation or the growth path of nominal GDP.

While Sumner's criticism of Ashwin Parameswaran's post were mostly on the mark, he goes too far when he says:

There will probably always be money; a pure credit economy is unthinkable. Without money there is no price level, because the price level is defined as the average price of goods in terms of money.

A pure credit economy has money. It's defining characteristic is that all money is a liability of some issuer. It represents a debt of someone. Checkable deposits, for example, serve as media of exchange, and they are a debt to the banks that issue them. Privately-issued banknotes can serve as hand-to-hand currency. They serve as media of exchange, but they are credit money. They are debt to the issuing banks.

From the point of view of the firms and households making payments, these bank liabilities are assets that serve as media of exchange. They can quote prices in terms of them, receive payments of them, and then make payments with them.

Now, if the monetary liabilities of many banks are to be accepted at par, there needs to be some kind of clearing system between the banks. It is possible to have a settlement system where the quantity of the settlement medium demanded in equilibrium is zero. And so, it is arguable that there is no base money. All the money is credit money.

However, even if banks do hold balances of some settlement medium, that doesn't mean that it is necessarily "outside money," that is, a liability of no one. Suppose inter-bank clearings are handled by a private clearinghouse, and the clearinghouse creates (and destroys) balances by ordinary open market operations. It purchases and sells assets. The balances are liabilities to the clearinghouse. It is another type of debt.

Whether or not it is desirable, a pure credit money system is thinkable. There is money. Prices can be quoted in terms of that money. A price level can be calculated in terms of that money. It is even conceivable that the quantity of reserve balances created by a private clearinghouse would be adjusted to stabilize some other nominal value, like a growth path for nominal GDP. For example, if the clearinghouse was obligated to maintain index futures convertibility.

Now, if we consider the status quo, where an independent government agency issues base money, the only real question is whether the currency and reserve balances are best understood as a special type of debt or not. This very much depends on how serious is the commitment to the target for some nominal variable like inflation or a growth path of nominal GDP. A monetary authority organized on banking principles and subject to index future convertibility, using open market operations to change the quantity of base money, and so matching its issue of base money with financial assets, looks very similar to the private clearinghouse described above.

Is the current Fed's commitment to its inflation target enough? It is an empirical question. And Ashwin Parameswaran was correct about that.

Are We Now in a Pure Credit Economy?

Are we now in a pure credit economy?

Tyler Cowen asked this question, linking to a post by Ashwin Parameswaran.

The post discussed Wicksell and and quoted Leijonhufvud, two of my favorites.

And I have always been interest in a "pure credit," or "pure inside" monetary order.

What is important about a pure inside monetary order is that there is no "pigou" effect from a lower price level. Even if the quantity of money is somehow given, the only real balance effect of a lower price level is a portfolio effect of lower nominal interest rates. Further, if the expected price level is somehow bounded, then a lower price level creates expectations of inflation, and so a lower real interest rate. There is no increase in real wealth, reduction in saving, and increase in real consumption.

However, Parameswaran has another concern:

There is ample reason to believe that reduced real rates across the curve have perverse and counterproductive effects, especially when real rates are pushed to negative levels

What are these perverse and counterproductive effects?

Prolonged periods of negative real rates may trigger increased savings and reduced consumption in an attempt to reach fixed real savings goals in the future...

Market monetarists argue that an expansionary monetary policy can raise real interest rates.

Sumner insists that the way an expansionary monetary policy works is through expectations of increased money expenditures on output in the future. Nick Rowe has suggested a positively-sloped IS curve to show how real interest rates can rise with real output and income in the face of an expansion in the quantity of money.

Naturally, market monetarists complain when the Fed explains the purpose of quantitative easing or operation twist as an effort to reduce long term interest rates. And we insist that the key to solving aggregate demand problems is a target for a growth path for nominal expenditures on output.

However, if the "confidence fairy" is not enough, I do think that lower real interest rates play a key role in raising current real and nominal expenditures. Given current expectations of future nominal expenditures, lower real interest rates raise current real and nominal consumption expenditures and current real and nominal investment expenditures.

More importantly, for improved confidence to raise both real and nominal expenditures, it is important for firms and households to believe that lower real interest rates will raise real and nominal expenditures. And while we can imagine it still working despite a false belief, certainly the more sure approach would be for lower real interest rates to be able to raise real expenditures given current expectations of future nominal expenditure. It is firms and households recognizing this and expecting that the Fed will purchase whatever amount of assets needed to make it happen, that raises expectations of future expenditures, and so motivates increased current expenditures.

If lower real interest rates reduce consumption expenditures, this would foreclose one avenue by which open market operations now and in the future can raise and be expected to raise real and nominal expenditure. However, focusing on the income effect for one segment of the market is an error. While it is possible that the income effect might be larger than the substitution effect so that consumption falls for those currently saving, debtors and borrowers both have opposite income effects that should raise current consumption. (Those living off of accumulated savings will almost certainly reduce consumer expenditures due to the income effect.)

While nominal interest rates may be subject to a zero bound, real interest rates have no such bound. It is difficult to believe that there is no real interest rate sufficiently negative to motivate reduced saving and increased consumption. The persistent saver willing to give up ever greater amounts of current consumption in exchange for less and less future consumption is implausible enough. However, when added to the rentier forced to dissave to maintain some consumption, the borrower tempted by obtaining more and more consumer goods today for a smaller sacrifice of future consumption, along with the debtor able to reduced real debt while consuming all of current income and even more, suggests a quite different result.

More importantly, the notion that investment can be approximated by perfect interest inelasticity becomes difficult to maintain when real interest rates can turn negative.

Shackle was skeptical about the impact of lower interest rates in stimulating business investment. He noted that businessmen when asked rarely noted at the level of interest rates as a critical determinant. In an uncertain environment, estimated profits “must greatly exceed the cost of borrowing if the investment in question is to be made”.

What negative real interest rates usually mean is that the nominal revenues generated by the investment greatly exceed the nominal costs of the project. And the nominal interest rate doesn't offset this large nominal profit.

Do businessmen usually focus on interest rates when making investment decisions? Presumably small changes in interest rates only have a small impact on investment decisions. However, if businessmen were asked if a 50% real interest rate would impact their investment behavior, what would they say? If, like Keynes or Shackle, you are wondering whether a modest decrease in a nominal interest will result in much of an increase in investment in the context of a stable price level, perhaps skepticism is in order. But when large decreases in real interest rates are considered, perhaps a different conclusion is appropriate.

Of course, if real interest rates must become highly negative to significantly expand real expenditure, then a zero nominal bound on nominal interest rates would imply high inflation and rapid growth in nominal expenditure. This is hardly consistent with slow steady growth of nominal GDP.

If significantly negative real interest rates are necessary to generate sufficient real expenditure so that nominal expenditures will return to target, then breaking the zero nominal bound would be a possibility. However, I believe that targeting nominal GDP and committing to purchase whatever quantity of assets necessary to reach that target will generate the expectations needed to put nominal GDP on target. If and when there is a discussion of whether the Fed should purchase blue-chip stocks, then perhaps the view that low long term interest rates result in less consumption and no more investment must be taken more seriously.

Hendrikson on Operation Twist

Josh Hendrikson responded to my post giving conditional approval to "Operation Twist."
He agreed with my basic argument, but claimed that it wouldn't have much effect. He claims:

Now suppose that the Federal Reserve sells $X of T-bills uses the proceeds to buy $X of long term bonds. In doing so, the price of T-bills fall and the yield correspondingly rises. However, this generates an arbitrage opportunity for banks, which will use excess reserves (now paying a lower rate of interest than T-bills) to buy T-bills. They will continue to do this until the price of T-bills is bid up such that the yield on T-bills is equal to the interest rate on reserves. The degree to which the new reserves created by the sale of T-bills are used to purchase T-bills will be determined by the price elasticity of demand.

The transmission of monetary policy works through changes in relative asset prices. The relative price effect generated by this policy is minuscule. One can always argue that the sale of T-bills increase the supply of short-term safe assets by $X, but the only effect is a pure arbitrage opportunity that ultimately changes the distribution of T-bills and reserves, but leaves the total relatively unchanged.

I don't think this is correct.

The Fed sells the T-bills, increasing their supply. Banks use their existing excess reserves to purchase some or all of the T-bills the Fed just sold. This does not decease the amount of excess reserves in the banking system, but simply shifts the reserves from the banks buying the T-bills to the banks whose customers sold the T-bills. The total of T-bills, including those held by the banks, and reserve balances held by banks, is still higher.

Looking at the money holdings of the nonbanking public, if the Fed sold a T-bill to an individual, his or her money holdings, (checking account balance) are decreased. That individual has substituted money, perhaps FDIC insured money, for T-bills. If a bank then buys the T-bills from that person (arbitraging as above,) that person again has the money instead of the T-bill. There has been no increase in the quantity of money plus T-bills held by the nonbanking public. The nonbanking public has the exact same amount of money and no more T-bills than before.

But this, of course, ignores the purchase of the long term bonds by the Fed. The sellers of those bonds have additional money balances. And so, in this scenario, the amount of checkable deposits held by the nonbanking public expands. The amount of reserve balances held by the banking system is unchanged. And the banks have more T-bills.

If it was a bank that purchased the T-bills from the Fed initially, then the Fed would decrease its reserve balance and that bank has just substituted T-bills for reserve balances. But again, the Fed is buying long term government bonds.

If the Fed purchases the long term bonds from a bank, then that bank has additional reserves, perhaps additional excess reserves. If the Fed purchases from someone other than a bank, that person has an added balance in a checkable deposit (rather than a long term government bond) and his or her bank has increased reserves, perhaps more excess reserves.

However, if a bank responds to this scenario by selling a long term government bonds (perhaps because the yields have fallen,) then checkable deposits held by the nonbanking public return to their initial value. The problem isn't that banks might purchase the T-bills that the Fed sells, either directly or indirectly. The problem is that banks might sell long term government bonds that they already own.

This is the notion that banks reduce their private financial intermediation in a way that offsets the Fed's increased financial intermediation. Really it amounts to the notion that long term government bonds are also perfect substitutes for money.

However, I agree with Hendrickson's conclusion. Operation Twist is not enough. It is really wrongheaded.

The biggest problem with current Federal Reserve policy is that it lacks any coherent direction or policy goal. Expectations matter. (Read Woodford, for heaven’s sake! This is supposed to be mainstream monetary theory.) For Fed policy to be successful, they need to outline an explicit goal for policy in the form of a target for nominal income and the price level and commit to using the tools at their disposal to achieve that goal. Random announcements of specific quantities of asset purchases provide no guidance and will not be effective. Temporary monetary injections are not successful for much the same reason that temporary tax cuts are not successful (see Weil, “Is Money Net Wealth?”, 1991). Without a coherent goal or strategy, monetary policy with all its fits and starts will continue to fail.

Very well said. Look at today's Wall Street Journal headlines, "Markets Swoon on Recession Fears" and then, "Global Distress Rises Over Government, Central Bank Futility." Exactly.

Sumner's Monetary Economics

Scott Sumner has an article in National Affairs defending a target for nominal GDP targeting.

I think he gives a good account of his views, but I didn't find his arguments for 5 percent nominal GDP growth and 2 percent inflation persuasive.

He argues that 2% inflation will allow central banks to better use conventional monetary policy (interest rate targeting) because the zero nominal bound on the target interest rate is a -2% real bound.

He also argues that it is easier to refuse to give workers cost of living raises than to impose pay cuts. This allows employers to cut real wages by 2% per year without explicitly telling the workers that they are getting real pay cuts. (His blog is called The Money Illusion.)

In my view, the best reason to stick to a 5 percent nominal GDP growth path is because that is what we had for 20 years, and shifting to a lower one would not be worth the disruption. The reason I favor such a change now is that we have already suffered massive disruption and when we shift to a new regime, we should get it right.

What about conventional monetary policy? Accommodating the preference of central bankers to stabilize short term interest rates has turned out to be a mistake. If watching short term interest rates provides a good estimate of whether supply and demand conditions for money, (the quantity of money in relation to the demand to hold money,) is consistent with keeping nominal GDP on the target growth path, then perhaps central bankers should take this into account.

But putting all of the monetary policy eggs in that one basket, and insisting that "setting" the current short term interest rate and, more importantly, generating expectations for short term interest rates in the future, is the whole of monetary policy is a proven disaster. Central bankers need to learn that when short term interest rates don't work, they have to go to with some alternative approach. Perhaps a targets for nominal GDP would allow central bankers to shift to looking at a longer term interest rate. Perhaps looking directly at base money or some alternative measure of the quantity of money would help. But in the end, the goal must be to purchase as many assets as it takes to get nominal GDP back to its targeted growth path.

As for the wages, a 3 percent growth path for nominal GDP involves nominal wages being on a 3 percent growth path. In some particular labor market, more rapid increases in labor supply or slower increases in labor demand can slow the growth in equilibrium real and nominal wages. The actual growth path can shift down 3 percent over a year without any absolute decrease in nominal wages. Shifting to a 5 percent growth path of nominal GDP does increase the maximum rate of reduction in growth path given stable nominal wages, but only by an additional 2 percentage points. Of course, that extra 2 percent is a decrease in real wages.

If slower growth in labor demand in some segment of the labor market requires an actual decrease in real wages, and real wages don't decline, then employment in that sector decreases and given the reduced labor input, real output in aggregate shifts to a lower growth path. This implies that with a given target for nominal GDP, the price level moves up to a higher growth path. Real wages do shift down at least slightly. While presumably this reduction in real wages slightly improves condition in the market where real wages need to fall, the larger effect is to enhance profitability in expanding sectors of the economy.

Usually, the appropriate response to a shift in the demand for labor is for workers to abandon contracting industries where real wages are growing less than trend and gain employment in growing industries where real wages grow faster than trend. If real wages must fall to maintain employment in some industry, this almost surely a sign of a needed shift. It isn't entirely clear that making it easier for firms to lower real wages in those circumstances and postpone the adjustment is desirable. Nor is it clear how effective this will be.

It is certainly correct that if there is an excess demand for money, clearing it by having prices and wages shift to a lower growth path is disruptive. Similarly, if there is some economy-wide productivity shock that reduces real wages across the board, having the price level rise to a higher growth path is the least disruptive course. But when what is needed is a shift in the allocation of labor between industries, perhaps layoffs in the shrinking industries matched by more rapid hires in the growing industries is the least bad option.

Rules of thumb that have proven effective in recruiting and retaining a productive workforce in competition with other firms do not apply when there is an excess demand for money or even when there is a adverse supply shock impacting real wages throughout most of the economy. The benefit of these rules of thumb is exactly in situations where there are shifts in demand between firms and industries. Lowering real wages in those situations is apparently problematic for the firms.

And, of course, with nominal income growing, even without inflation, there is room for new firms to enter, hiring workers at lower nominal and real wages. And those firms with growing demand can start their new workers at lower nominal wages. These are not small factors considering the huge turnover in employment in a healthy, growing economy. The very serious problems that develop when there is a need to lower nominal real wages more or less across the board, do not necessarily apply when there is some small segment of the labor market requires lower real wages to maintain current levels of employment.

Finally, the more official and open the commitment to a trend inflation rate, the less we can pretend that the absence of a cost of living pay increase is anything other than a pay cut. Worse, the entire notion that "cost of living increases" are something to be expected is entirely counterproductive when there is an adverse supply shock. Cost of living increases to correct of that sort of inflation greatly reduce the benefit of nominal GDP targeting. In my view, it is better to not get into the habit of cost of living increases than to require firms to explain that they can give the usual 2 percent cost of living increase, but the additional inflation that has reduced your real income can't be compensated because that wasn't part of the normal increase in the cost of living planned by the Federal Reserve.

Monetary Disequilibrium--some ABC's

Monetary disequilibrium refers to an excess supply of money or an excess demand for money. (An excess supply is the same thing as a surplus and an excess demand is the same thing as a shortage.)

If we abstract away from hand-to-hand currency and focus on the checkable deposits that households and firms use to make most purchases, then an excess supply of money is a situation where the quantity of those deposits is greater than the demand to hold them. While there are a variety of possible responses to this situation, spending the excess money balances on some good, service, or asset is obvious and natural.

If this additional spending would be a problem, for example because spending on currently produced output is already pressing against productive capacity, then how can the excess supply of money be corrected? There are two obvious solutions. Either the quantity of money needs to decease, which will reduce the quantity to the existing demand. Or else, the interest rate paid on money (on checkable deposits) needs to increase, to raise the demand to hold money to the existing quantity. Of course, combinations of reduced quantity and increased yields would also be possible.

An excess demand for money, on the other hand, is a situation where the amount of money people want to hold exceeds the existing quantity. While there are many possible responses, the most natural response is to reduce expenditures out of current income to rebuild money holdings to the desired level. Selling off other assets would be an alternative approach.

If the reduced flow of spending on output and assets, or additional asset sales, are seen as a problem, presumably because the flow of spending falls below the productive capacity of the economy, there are two obvious solutions. The quantity of money can be increased or else the interest rate paid on money can be decreased. An increase in the quantity of money raises the quantity to the existing demand to hold money. A decrease in the interest rate paid to those holding money reduces their desired money holdings to the existing quantity. And, of course, a combination of both would be possible.

What about changes in the price of the deposits? A dollar price of a dollar balance in a bank deposit is fixed by redeemability. Short of default, a bank must adjust the quantity or interest rate paid on deposits so that there is no excess supply. If a bank is going to allow its customers to deposit receipts and make payments, it is going to need to provide effective two-way convertibility. It will have to correct an excess demand by raising the quantity issued or lowering the interest rate paid.

However, this still allows for the possibility of disequilibrium between the quantity of base money and the demand to hold base money. Base money is issued by the monetary authority (often called the central bank,) and the deposits of private banks must be redeemed with this base money. If a bank is going to participate in the payments system, and allow its customers to deposit checks (or the electronic equivalent,) then the bank effectively accepts deposits of base money by accepting it in exchange for payments drawn on other banks.

Base money takes two forms, zero-nominal-interest hand-to-hand currency and reserve balances at the monetary authority. Abstract away from hand-to-hand currency.

If there is an excess supply of base money, then the quantity of base money is greater than the demand to hold it. If that is considered a problem, then there are two solutions. The quantity of base money could be decreased, or the interest rate paid on base money increased. A decrease in the quantity brings the quantity down to the existing demand to hold it. An increase in the interest rate paid on base money raises the demand to hold base money up to its existing quantity. Combinations of reduced quantity and higher interest rates paid on reserve balances are possible.

If there is an excess demand for base money, then the quantity of base money is less than the demand to hold it. If that is a problem, then there are two solutions. The quantity of base money could be increased or the interest paid on base money could be decreased. The first raises the quantity of money to the existing demand, and the second reduced the demand to hold base money to the existing quantity.

With zero-nominal-interest hand-to-hand currency, adjusting the nominal interest rate paid on the currency is difficult. This requires than any excess supply or demand for this type of base money be corrected by adjustments in the quantity.

The ABC's are a good start. But there is one complication. Attempting to increase the quantity of money buy purchasing assets that have a yields equal to or lower than the interest rate being paid on money may easily be ineffective, with the demand to hold money simply rising in parallel with the quantity of money. The solution is to lower the interest rate paid on money or else purchase assets with yields greater than that paid on money.

For private banks, insisting on a positive margin between the interest earned and paid, this should be natural. The competitive interest rate that banks are willing to pay on checkable deposits should remain below the interest rates banks obtain from earning assets.

If zero-nominal-interest currency is taken to be the "model" of money (rather than deposits,) and the financial assets that the monetary authority buys have positive yields, then there is no problem. Only if the financial assets the monetary authority typically purchases have a zero or negative yield (because they are easier to store than currency) would there be a problem.

But if the monetary authority chooses to subsidize those holding reserve balances by paying an interest rate higher than the financial assets it buys, then increases in the quantity of money might be unable to correct an excess demand for money.

A Glass Half Full

João Marcus Nunes points to something positive in the Congressional leaders' letter to Bernanke:

They said Fed officials should avoid further action, “particularly without a clear articulation of the goals of such a policy, direction for success, ample data proving a case for economic action and quantifiable benefits to the American people.”

I zeroed in on their complaints about the exchange rate. Sumner accuses them of "treason," on the view that they know monetary policy would help the economy and they want a weak economy going into the election.

What "glass half empty" thoughts.

Nunes argues that this is the perfect opportunity for Bernanke to explain to Congress that he plans to raise nominal GDP. Now, that is "glass half full" thinking.

Operation Twist--Conditional Support

The Fed is considering a change in the composition of its balance sheet away from short term government securities and towards long term government securities. This "operation twist" is the right thing to do.

In 1956, Leland Yeager explained the "monetary disequilibrium" approach to the liquidity trap. One core principle of his view is that any general glut of goods, in particular, any drop in the flow of money expenditures on output, must be matched by an excess demand for money. People are trying to accumulate and hold more money than exists.

If we think about the possibility of people trying to sell current output and accumulate some other nonreproducible good, like "old masters" or land, and the result is a shortage, then we must ask what the frustrated buyers do. If they purchase some other reproducible good or service, then there is no general glut of output and nominal expenditure on output is maintained. If, on the other hand, they simply hold money, then the result will be a shortage of money and a general glut of goods. Nick Rowe often writes on this issue.

Suppose the good that people want to accumulate are short and safe financial assets. Suppose people are trying to accumulate T-bills. Yeager pointed out that once the interest rate on those bonds become so low that it isn't worth the bother of buying them rather than just hold money, then the shortage of them is leaking over into a shortage of money. It is very much like the frustrated buyers of "old masters" choosing to hold money rather than buy something else.

These days, the liquidity trap is identified with the zero nominal bound on nominal interest rates. So rather than this shifting of a shortage of bonds to a shortage of money at a very low positive interest rate, the leakage supposedly happens at zero. The way I would describe the problem is that if the market clearing interest rate on these bonds is negative, and greater than the cost of storing currency, then of course, the shortage of these bonds is going to shift over to a shortage of money.

When DeLong insists that the problem the economy faces today is an inordinate demand for short and safe assets, he is mostly correct. The reason it causes reduced nominal expenditure on output and a general glut is because it shifting over to an excess demand for money. Further, one key reason why the demand for short and safe assets is high is because of expectations of the consequence of the excess demand for money, lower nominal expenditures in the future, makes purchasing equities, corporate bonds, or even capital or consumer goods less attractive. Still, the shortage of short and safe assets at an interest rate close to zero plays a key role in the process.

An alternative way to see the problem involves the argument that T-bills become perfect substitutes for money when the interest rate is zero. Barro made that argument. And while he used it to suggest that quantitative easing would be ineffective (arguing that it amounts to an "operation twist" by the Treasury which could not possibly be the solution,) the implication is that when the T-bill rate hits zero, the quantity of money immediately increases by the stock of T-bills directly owned by households and firms. Of course, the amount of those T-bills households and firms are willing to hold at an interest rate of zero would be an increase in the demand to hold "money." In that scenario, if the Treasury funds current deficits with T-bills, or refinances the national debt with T-bills, then it is an increase in the quantity of money.

What does that imply regarding "operation twist?" By having the Fed sell off its holdings of short term government bonds, the Fed will relieve that underlying excess demand for those securities and lessen any shift of that excess demand to an excess demand for money. It should help relieve the monetary disequilibrium. Of course, if the Fed reduced the quantity of base money, as would be the usual consequence of an open market sale, then any decrease in money demand would be offset by a decrease in the quantity of money. However, by purchasing long term bonds, the Fed sterilizes the impact of the sale of short term bonds on the quantity of base money.

The other way to look at the issue is that with nominal interest rates on T-bills (nearly) at zero, they are perfect substitutes for money. By selling T-bills and purchasing long term government bonds so that base money does not decrease, the total quantity of money, T-bills held by households and firms and base money, increases. This will tend to relieve the excess demand for money.

A third way to look at the problem, and the one I find most illuminating, is consider a world without government issued hand-to-hand currency. If there is a shortage of short and safe assets at a nominal interest rate of zero, then the nominal interest rate would simply turn negative. The markets would clear at those negative rates. If the quantity supplied doesn't increase, quantity demanded will fall enough to clear the market. Presumably this would occur by some combination of people accepting the interest rate risk of longer term bonds, the credit and interest rate risk of corporate bonds, the risk of equities, or even purchasing capital goods or consumer goods.

If there is a central bank and it chooses to block this process by paying interest on reserve balances at something above the negative market clearing rate, then the shortage of short and save assets would be shifted over to an excess demand for money. (At least if the central bank is considered financially sound. Reserve balances are short, but they might not be safe.) Imposing such a floor on interest rates would be a mistake in my view.

With the financial system fundamentally being based on redeemability in government-issued, zero-nominal interest, hand-to-hand currency that is perfectly short and at least as safe as any other short and safe asset, the zero nominal bound (or the negative at the cost of storing currency bound) applies. Under those conditions, it makes sense for the central bank to charge for holding reserve balances, but at a rate slightly less than the cost of storing currency. Why pay to print currency and make people pay for safes?

But to return to operation twist, the point is to increase the supply of T-bills in order to raise (make less negative) the shadow market clearing interest rate on those and similar assets. This negative interest rate is what would occur if the financial system was not based on redeemability in government-issued, zero-nominal-interest-rate, hand-to-hand currency. And the reason to raise this shadow rate is to reduce the shortage at the current rate, and so the shift over into the demand to hold money.

So, why only provisional approval? As usual, the Fed is explaining the benefits of this proposal not in terms of relieving monetary disequilibrium, but rather in terms of lowering long term interest rates.

While it is certainly correct that having the Fed buy long term bonds should raise their prices and lower their yields--ceteris paribus--there are other, more desirable, possibilities. The whole point of "operation twist" is to relieve monetary disequilibrium and expand nominal expenditure on output. One happy scenario would be that firms expecting this would demand more capital goods now to be able to meet additional consumer demand in the future. To fund those capital goods, the firms would sell off some of their current holdings of long term government bonds. If firms sell more long term government bonds to fund purchases of capital goods than the Fed purchases, then yields on long term government bonds rise.

Now, there are other, less happy scenarios. Some involve expectations of higher nominal expenditure and greater inflation. Fear of inflation results in the sale of long term government bonds. If more are sold for this reason than the Fed buys, then long term interest rates rise.

But one of the worst scenarios is that no one believes that this approach will work, and continue to expect nominal expenditures to be low in the future, but because long term government bonds have lower yields, households and firms will purchase corporate bonds instead. Firms will overcome their concerns about excessive debt and issue new debt to purchase capital good despite their pessimism regarding future sales.

And which scenario is the Fed focusing on? Why? Is it just their mindset? Monetary policy has to be about lowering some interest rate? Lowering some interest rate has to be about expanding credit volumes and so total debt? Maybe, but not necessarily.

Those Republicrats!

So we now know that Obama believes that unemployment is caused by improvements in technology and improved labor productivity. (It is just common sense. If it takes less labor to produce the output we need, then employment must fall. Given a growing population, unemployment must rise. We have to think of something new for them to do--high speed rail, green jobs...)

He just refused to believe his economic advisors, who told him that the problem is inadequate demand and that monetary policy could do more even if the federal funds rate was near zero.

Meanwhile, Republican leaders in Congress have proven themselves no better, writing Bernanke to oppose monetary expansion. Their key argument? It will depress the foreign exchange rate of the dollar. If the dollar gets too low, the U.S. will lose gold reserves, and the next thing you know, convertibility will be in danger! Oh, that's right. We aren't on a gold standard.

Can Monetary Policy Work on This Planet?

Steve Waldman, on Interfluidity, criticized a post by Bryan Caplan. Caplan had argued that fiscal policy can only increase employment temporarily whereas monetary policy can do so permanently.

Caplan imagined a "helicopter drop" of money and imagined that this would permanently raise nominal GDP, raise prices, lower real wages, and increase employment. In Caplan's view, an increase in the flow of government spending may employ people to produce the goods the government is spending on, but that this effect will disappear unless the additional flow of government spending is continued forever.

Waldman has some doubts about Caplan's economics, and I could nitpick as well. But his major criticism was of Caplan's assumption of a helicopter drop. The Fed doesn't have the legal authority to do helicopter drops, but rather it must purchase financial assets or make loans.

Helicopter drops by the Federal Reserve are illegal. Helicopter drops by the Treasury happen all the time. Every law ever passed that overpays for anything, that has some manner of a transfer component, is a helicopter drop. I think all of us, left and right, delightful-smelling and stinky, can come together in a big Kumbaya and agree that most Federal spending has a transfer component, and is therefore a helicopter drop to some degree. It would not be a big deal for Congress to pass some law with even bigger, badder transfer components. They love to outdo themselves.

I think on the technocratic right, monetarists tend to think that helicopter drops by the Fed would be fairer than fiscal policy that launders transfers through expenditures. On the left and hard-crank right are people who don’t see the Fed as very fair at all, and emphasize the institution’s inclination to support certain interest groups when it does find ways of sneaking transfers through its legal shackles.

I don't find the "helicopter" thought experiment very helpful either. It is most useful in understanding the effects of a fully unconstrained monetary policy, like Zimbabwe. Create money because we want to spend it, and who cares about what happens to the price level.

The other way it is somewhat useful is if the only constraint on the issue of money is the quantity created by the monetary authority. The monetary base must be at a certain level or stay on a constant growth path. In the case of the constant growth path, the "seigniorage," is a source of government revenue that is presumably spent along with all the other government revenues.

If, on the other hand, monetary policy is constrained by some other nominal target, from maintaining redeemability with gold at a fixed price, to keeping some broader measure of the quantity of money on a target growth path, to targeting inflation, the price level, or nominal GDP, then the issue of the monetary base is a type of borrowing and it is necessary to stand ready to reduce the quantity when the demand to hold it falls. There is no philosopher's stone, creating gold that can be spent or given away.

The Federal Reserve is targeting inflation, more or less, so helicopter drops are not the best way to analyze its situation. Even if the Fed were closed and the Treasury took over the role of monetary authority without pretending to operate on banking principles, it would still need to stand ready to reduce that part of the national debt funded by treasury currency by selling other, interest bearing debt--as long as it wanted to keep to some nominal target other than a growth path for the quantity of treasury currency.

Waldman insisted that economists should start to include the payment of interest on reserve balances as the Fed in their models.

The Fed has signaled (ht Aaron Krowne) they may pay interest on reserves at the overnight interest rate indefinitely under a so-called “floor” regime. I wish more economists would update their models for a world in which interest is paid on reserves as a matter of course. Interest on reserves represents a permanent policy shift that had been planned since 2006. It was not an ad hoc crisis response that can be expected to disappear. If interest is paid on reserves at the overnight rate and short-term bond markets are liquid, then short-term bonds and base money are perfect substitutes and a helicopter drop performed by the Tim Geithner dropping bonds from an F-16 would be as effective (or ineffective) as Ben Bernanke dropping dollar bills from his flying lawnmower.

Ignoring the payment of interest on reserves would be a mistake. It is related to the "paper gold" mentality of the helicopter drop. Gold pays no nominal interest. Hand-to-hand currency generally pays no interest, and for obvious reasons, a monopoly issuer likes to keep it that way. However, money can pay interest, and not all money takes the form of hand-to-hand currency. In particular, reserve balances at the central bank are a form of money, and they can pay interest and they do pay interest today. If there is a shortage of money, there are two ways to fix it. Increase the quantity or reduce the yield paid.

On the other hand, treating the interest rate paid on reserve balances as a constant is also a mistake. When the Fed's program was initiated, there were different rates for required reserves and excess reserves. And the initial rate was much higher than the current rate.

Setting the rate "paid" on excess reserves so that holding reserves is slightly less costly than storing currency would allow the interest rates on short and safe assets to fall to their true lower bound if market conditions merited such low rates. And while I suppose a "floor" is appropriate in that situation to save on the cost of printing and storing currency, until that point is reached, the interest rate paid on reserves should be less than the interest rate on other safe and short financial assets. Why should the Fed provide intermediation services for free?

Anyway, there is no legal requirement that the Fed pay interest rate on reserves in order to keep the target for the Federal Funds rate from falling too low. The authority of the Fed to pay interest on reserve balances doesn't mean that the rate paid cannot be zero. And I don't see why charging banks for "storing" their money in a perfectly safe form is either a tax or a violation of the Federal Reserve act.

Waldman also complains:

If you think the Fed’s existing toolkit of running asset swaps and controlling the rate of interest on reserves would be enough if only they set expectations properly, then we still need new law. The Fed is not going to target NGDP or a price level path over any relevant time frame without a change in governance structure or mandate.

I will grant that the new Keynesians at the Fed who propose targeting a higher inflation rate so that real short term interest rates will be more negative, and so real demand will rise and the output gap close, and unemployment fall, may have some legal problems. The Fed's willingness to interpret "price stability" as 2 percent inflation forever seems like a bit like a stretch of the dual mandate. But intentionally raising that even higher, without even a fig leaf about how it is really price stability, and that the entire point is to erode people's savings, might result in average people wondering why the Fed has been intentionally raising their cost of living for the last twenty years.

More importantly, the failure of the Fed to complete the disinflation that began in the early eighties all the way to zero is probably more tolerable than heading off in the wrong direction. Failing to make further progress, perhaps for a time, is not the same thing as purposely and openly choosing a destructive course.

On the other hand, shifting to a target for the growth path of the price level is not quite the same. It is just a different interpretation of "price stability." Leaving aside the planned reduction in the purchasing power of money, a stable growth path of prices certainly seems closer to price stability than having the growth path of price level on some kind of random walk.

Of course, laying out that 2 percent growth path and explaining a commitment to reverse any deviations does mean that rather than waving hands about and saying "price stability" the progressive intentional increase in the cost of living is there for everyone to see. Then who could continue suffer under the delusion that inflation just happens and as hard as the Fed might try to control it, slow inflation is all that it can accomplish? They are doing their best, and keeping it low! Perhaps no average voter suffers from that illusion. Perhaps.

Further, if the Fed does go with a price level target, it would need to be symmetrical. And that means that adverse supply shocks in the future must be met with a contractionary monetary policy that reverses the inflation, exacerbating what will be already be depressed real output and employment. If that is not the intention, and it is a price level target now, but back to an inflation target later, then it is really just a fraud.

Nominal GDP targeting is not just an excuse to raise inflation now and lower real short term interest rates, raise real aggregate demand, close the output gap, and reduce unemployment. It is an alternative monetary regime that happens to be very consistent with the Fed's dual mandate. Leaving aside the choice of trend growth rate for nominal GDP and so the price level, it keeps real and nominal demand growing at a level consistent with prices remaining on a constant trajectory. It keeps excessive spending from causing excessively high, much less rising, inflation. On the other hand, it does not require the Fed to cause monetary disequilibrium and further depress output and employment when an adverse supply shock (like higher oil prices) results in temporarily higher inflation.

Perhaps targeting the growth path of nominal GDP doesn't promote high employment subject to price stability exactly, but it does have the consequence of promoting price stability in way that avoids sabotaging high employment.

I would prefer that Congress change the Federal Reserve Act and impose a target for the growth path of nominal GDP on the Fed. But I believe that it is entirely reasonable for the FOMC to open their eyes and see that the success of the Great Moderation was in keeping nominal GDP on a reasonably stable growth path for 25 years, and that the large and growing gap in the last three years means that they have failed to meet the legal mandate, and that getting nominal GDP back up to that trend is doing their duty.

It is possible that communicating such a commitment would be consistent with increases in the interest rates on short and safe assets and would make it unnecessary for the Fed to bear unusually large amounts of interest rate or even credit risk. It certainly would reduce what it must do along those lines. And regardless of what is believed initially, when negative interest rates on reserve balances and a heroically large Fed asset portfolio actually begins to impact nominal expenditure despite market skepticism, then short term interest rates can rise somewhat, and the Fed can shed at least some of those assets off of its balance sheet.

Personally, I believe that the target for nominal GDP should be consistent with a stable price level in the long run -- 3 percent growth. Of course, I also favor privatizing hand-to-hand currency and index futures convertibility. Maybe this planet isn't quite ready for those reforms. But a target for the growth path of nominal GDP is a reform whose time has come.

Marcus Nunes reviews some empirical data from the Great Moderation:

Almost 16 years ago I wrote a paper: “Are_analysts_missing_the_point“; that inter alia tried to probe explanations for the increased economic stability since the early 1980´s. A surprising result was to discover, contrary to the idea mentioned in Taylor´s “The Long Boom that after 1983 the Fed reacted more strongly to inflation, that in fact the opposite is evidenced, with the FF rate showing no significant response to inflation.

The reason, according to Nunes, is that inflation showed less persistence during the Great Moderation:

One plausible explanation for the result that reconciles the (apparently contradictory) absence of response of the federal funds rate to inflation after 1982 with a postulated increase in the Fed.’s resolve to fight inflation is that the behavior of inflation changed after the 80/82 recession.

In fact, inflation after 1982 exhibits substantially less persistence than in the previous years (see figure below) so that increases in inflation in one month are viewed as temporary. In other words, inflation is much less auto correlated so that lagged values of inflation provide little information about future inflation. As a result, unexpected movements (or innovations) in inflation no longer require a monetary policy response (which sits well with the argument that the fed funds can be a poor indicator of monetary policy).

Why was inflation no longer persistent? Because it was due to supply shocks. Nunes explains and even quotes Bernanke:

One of the dangers associated with the absolute pursuit of “price stability” is the occurrence of supply shocks. This quote is from a Bernanke and Getler paper (my bold):

Macroeconomic shocks such as oil price increases induce a systematic (endogenous) response of monetary policy. We develop a VAR-based technique for decomposing the total economic effects of a given exogenous shock into the portion attributable directly to the shock and the part arising from the policy response to the shock. Although the standard errors are large, in our application, we find that a substantial part of the recessionary impact of an oil price shock results from the endogenous tightening of monetary policy rather than from the increase in oil prices per se.

Symmetrically, in the case of a positive supply (productivity) shock, a substantial part of the expansionary impact of the shock would result from the endogenous easing of monetary policy.

And what is the answer? Nunes explains:

According to the views of Market Monetarists, since a nominal GDP target ignores aggregate supply shocks it dominates an inflation target. It sees movements in the price level as a symptom of whatever underlying shock is taking place while regarding movements in nominal spending as an underlying shock itself – an aggregate demand shock – over which the Fed has direct influence and can respond to much more effectively. In essence, by not reacting to the “symptoms” and striving to keep AD stable, Fed policy would result in overall stabilization.

Bernanke should know better! Great post by Marcus.

(Thanks for the spelling suggestion Benjamin!)

Negative Interest on Reserves

Former Fed Vice-Chair Alan Blinder speaks out on negative interest rates on reserves:

The rate could then be pushed into negative territory. The notion is strongly opposed by banks, who view it as a tax. Blinder doesn’t disagree with that characterization. “The whole notion is you should tax things you don’t like people to do, and subsidize things that are essential,” he said. “One thing we don’t like is banks just piling up idle reserve,” he said. “We would like to push that money out of the banks and have them do something with it.” Although some money will undoubtedly go into super-safe money funds, “the hope is that some fraction goes into increased lending,” he said.

This has been one of the proposals of market monetarists (as Lars Christensen proposes we be called.)

I don't really agree that it should be called a tax. It is more like a service fee. The Fed is providing a safe and short financial asset for banks to hold. If supply and demand conditions result in people being paid to hold such assets, then that is great. But if market conditions result in an equilibrium where people pay to be willing to store wealth in such a way, then that must be accepted as well.

It is true, of course, that banks are required to hold some reserves, and charging them to hold required reserves is more a tax. The simplest solution would be to reduce the interest rate to zero on required reserves and only charge banks for holding excess reserves.

The ideal solution is to not have required reserves at all.

HT to Tim Duys.

Larry White on the International Gold Standard

Over on Free Banking, Larry White linked to an article where he advocated the international gold standard in combination with free banking. The article began with a focus on problems with the Euro and the role of possible currency devaluation:

Soros hopes that with one pan-European government, financially conservative Germany would no longer rule the roost. The ECB could then pursue looser monetary policy, which he supposes would cure the ills of countries with weak economies and mounting public sector debts. This view is widely shared.

By contrast, when retired Dutch central banker André Szász says that the euro was flawed from the start, as he did earlier this year, he means that it is a mistake to have “a monetary policy of one-size-fits-all.” Such a monetary policy will be too loose for some countries and too tight for others, or, as he puts it, interest rates will be “too low” for some countries and “too high” for others.

This criticism is linked to the so-called optimum currency area analysis, which holds that to share a single currency, two or more economies should have harmonized business cycles so that a single monetary policy (interest rate) fits them all. Absent harmonized cycles, devaluation or exchange rate depreciation is supposed to help an economy in recession reduce its unemployment rate by lowering real wages or by stimulating real output through greater real exports.

Both of these diagnoses arise from false premises. They both rest on the wishful thinking of Keynesian economics, in particular, that an artfully timed discretionary monetary policy will improve or stabilize an economy’s real performance by improving or stabilizing real variables. That is to say, these arguments take for granted an ability to exploit the Phillips Curve (to lower unemployment by cheapening the monetary unit), alternatively known as exploiting the “money illusion” of the workforce.

In fact, the real illusion here is our supposed ability to exploit the money illusion. A policy regime of printing more money and devaluing does not improve real economic performance or dampen business cycles. It does just the opposite. The historical evidence on that question is clear.

While I favor free banking, I don't favor an international gold standard. Rather, I favor a monetary regime that stabilizes the growth path of nominal GDP. This sometimes requires printing money, and sometimes the floating exchange rate depreciates, resulting in more exports.

If we consider a situation where some error has resulted in inadequate money creation so nominal GDP falls below target, the result would likely be slower growth or even reduced real expenditure, real output, and employment. In other words, this error would lead to a recession.

Fixing the error, and returning nominal GDP back to its target growth path, would involve expanding the quantity of money, which could involve a depreciation in the exchange rate, and an expansion of exports (and in the demand for import competing goods.) This would be one avenue by which the expansion in the quantity of money would increase nominal expenditure back to target, which would lead to a recovery of real output and employment.

More importantly, it is quite possible that preventing the exchange rate depreciation would require that less money be created, so that nominal GDP would remain below its targeted growth path. Given that lower growth path for nominal GDP, recovery would require that prices and wages shift to a lower growth path, so that real expenditure and output can return again to their previous growth path.

So, what about White's claim that both approaches are based upon an ability to exploit a phillips curve or involve the exploitation of money illusion?

White describes exploiting the phillips curve as cheaping the monetary unit to reduce the unemployment rate. I don't see it that way and instead favor the natural rate hypothesis, with a focus on nominal GDP.

More rapid growth in nominal GDP--the flow of money expenditures on output--results in rising sales for firms. The firms respond by raising prices more quickly as well as expanding output more quickly. The more rapid growth in output results in more rapid growth in employment, and so a reduced unemployment rate.

That firms are raising their prices more quickly is a cheapening of the monetary unit, but that isn't what causes unemployment to fall. It is rather another consequence of more rapid growth in nominal GDP. Oddly enough, to the degree that firms meet growing sales by raising prices by less and expanding production by more, unemployment falls by more. Given the growth rate of nominal GDP, the larger the increase in the inflation rate, the smaller the decrease in the unemployment rate.

The natural rate hypothesis is the view (which I hold) that the more rapid growth in output and employment are temporary. Worse, output and employment have temporarily moved to a higher growth path inconsistent with the productive capacity of the economy--which depends on "supply-side" factors like technological improvement, the population and willingness to work, and saving, investment, and the accumulation of capital goods. Not only will the growth of real output and employment slow back to its initial rate, it will slow below its long term trend growth rate, to return to its previous growth path. The unemployment rate will rise again to its initial level as employment returns to its lower, equilibrium growth path.

Unfortunately, the inflationary consequences of the more rapid growth in nominal GDP persist. In fact, during the short run adjustment--when real output remains on a higher and unsustainable growth path, the inflation rate is lower than its final value. During the adjustment to the long run equilibrium, the inflation rate must rise above its long run rate, and that is when unemployment rises. In this situation, there is a sense in which the higher inflation is depressing real expenditure and causing the higher unemployment. (Of course, this is simply an increase in unemployment from an unsustainable low level.

So why support more rapid growth in nominal GDP if any benefit in terms of output and employment is transitory? It is because a slowdown in nominal GDP growth is subject to a similar analysis.

In the short run, slower growth in nominal GDP (including negative growth) results in slower or reduced sales. Firms respond by slowing their price increases and the rate at which they expand production. With production growing more slowly, employment grows more slowly. The unemployment rate rises. In more extreme cases, prices, production, or employment may fall rather than grow more slowly.

Again, it isn't that the lower inflation causes higher unemployment, it is rather that the slower growth of nominal GDP--the flow of spending on output--is causing both slower inflation and higher unemployment. Further, if the firms responded to the slower growth in sales by raising their prices a smaller amount, or even cutting them, the slow down in money expenditures would be consistent with a smaller slowdown in the production of goods and services, and so employment would be less depressed and the unemployment would rise by less.

And in the long run? Real output returns to potential--to levels consistent with supply-side factors, as does employment, while inflation slows further. As real output and employment recover--moving to a higher growth path--inflation must slow even further, temporarily falling below its long term growth path. Again, the slower inflation results in more rapid growth in real expenditure, resulting in increased sales along with the recovery in production and employment. The slower inflation is causing the falling unemployment.

So, how does this analysis of the Phillips curve fit in with nominal GDP targeting?

Suppose some error result in slower growth of nominal GDP. The slower sales result in slower inflation and slower growth in production and employment. The unemployment rate rises. A monetary regime that reverses such a deviation in nominal GDP, so that it grows at a higher rate to reverse this downward deviation results in more rapid sales and firms reverse the decreases in production and employment. But this doesn't involve a shift in production and employment above potential--beyond the level determined by supply-side factors. On the contrary, it simply reverses the downward deviation. The unemployment rate rises above the natural rate and then falls again.

As long as some of the initial slowing and then recovery in the flow of money expenditures results in changes in at least some prices, then inflation slows and then "recovers" as well. However, describing the reversal of the temporary disinflation as somehow exploiting money illusion seems inappropriate. What it is doing is forestalling the long run adjustment that would result in even more disinflation.

The relevant issue is whether the recovery of nominal GDP to its previous growth path will hasten the recovery of real output, employment, and unemployment, or rather will firms and households that have already fully, or partially adjusted to the slower, or lower growth path of nominal expenditure, respond to the recovery of nominal GDP by raising production beyond potential, along with an associated temporary expansion in employment and reduction in unemployment.

Presumably, the answer to that question very much depends on expectations. A regime of targeting the growth path of nominal GDP would likely have only weak disinflationary forces allowing for a recovery of output. However, confidence by firms and households that a monetary regime will reverse any deviation of GDP from the targeted growth path should both dampen any actual deviations and hasten their reversal.

As for devaluations, adverse changes in international competitiveness are a type of adverse productivity shock. As with other such shocks, a regime that targets the growth path of nominal GDP generates a higher growth path of prices and reduced real output. The decrease in the market prices for foreign exchange results in higher import prices and a reduction in the real value of exports.

A lower growth path of nominal (and real) wages is an alternative approach to adjusting to this adverse shock without any change in the exchange rate. In my view, the situation that foreign goods are more difficult to obtain is better signaled by an increase in the prices of imported goods than reduced employment opportunities at the trend growth rate of money wages.

It is true, of course, that nominal GDP targeting naturally leads to a question of optimal currency areas. What area's nominal GDP should be targeted? (Admittedly, I tend to take the parochial view that the nominal GDP of the U.S. should be targeted.)

I think the key requirement for an optimal currency area is factor mobility--particularly labor. It would seem to me that a shift to a higher growth path of money wages in one region and a lower growth path of money wages in another region is only useful if it serves as a signal for migration. If such migration is difficult, impossible, or even not desired, then changes in product prices seems like the more appropriate signal of the necessary change in real incomes and the allocation of resources between local and distant production.

However, I also believe in competition in currencies, so that individual households and firms should be free to adopt the currency used in some other "area" regardless of what seems optimal to me. I doubt, however, that currency competition will ever result in any kind of gold standard, much less an international one.

Krugman Critiques Quasi-Monetarism

Krugman explained what he thinks is wrong with quasi-monetarism.

He starts with a point of agreement:

So: an overall shortfall of demand, in which people just don’t want to buy enough goods to maintain full employment, can only happen in a monetary economy; it’s correct to say that what’s happening in such a situation is that people are trying to hoard money instead.

Good.

Then he goes on to bring up the liquidity trap.

But we’re not in an ordinary situation here, we’re in a liquidity trap in which short-term interest rates have been driven to zero, yet the economy still languishes.

One key tenet of quasi-monetarism is a rejection of the view that monetary policy necessarily involves using short term interest rates as an instrument. If a central bank usually expands the quantity of money an amount that lowers a short term interest rate some particular amount, and there remains an excess demand for money when that short term interest rate hits some lower limit, then it is time to quit looking at that short term interest rate. It is still the central bank's responsibility to relieve the monetary disequilibrium, to expand the quantity of money to match the demand to hold money.

Krugman then continues:

What that means is that when people are hoarding money, they’re no longer doing so because of its moneyness — the liquidity it provides, which makes money different from other assets. They’ve already got all the liquidity they want, since liquidity is free — you don’t have to sacrifice interest earnings to get more, so people are saturated. So at the margin, they’re holding money simply as a store of value.

I have a quibble with this statement. Liquidity isn't "free" just because T-bills have a yield near zero. People wanting to shift from T-bills to money don't sacrifice any interest, but people who don't own any T-bills must reduce consumption to accumulate money or else sell some asset that has a non-zero yield.

More importantly, it remains an excess demand for money even if the reason firms and households accumulate the money is that they find it an attractive store of wealth. If people began to make dollar bills into paper jewelry, that would still be a demand for money.

But here is the real problem with Krugman's argument--

Now, what monetary policy ordinarily involves is open-market operations: the central bank increases the supply of money by purchasing and removing from the market non-money assets. And this has traction because money is different from these other assets. In a liquidity trap, however, money isn’t different: at the margin an open-market operation just exchanges one store of value for another, with no economic effect.

Krugman equivocates. He was explaining that short term interest rates were zero, and since money is being held as a store of wealth on the margin, it is no different from other short term assets. Money is just like T-bills on the margin, because money and T-bills are both being held as a store of wealth. We would know this because T-bills have a yield of zero.

But now, it is all assets that are no different from money. Are we supposed to ignore that he dropped "short term?"

Monetary policy ordinarily involves open market operations. True.

The central bank increases the supply of money by purchasing and removing from the market non-monetary assets. True.

But in a liquidity trap, money isn't different: at the margin, an open-market operation just exchanges one store of value for another, with no economic effect.

No, an open market operation using zero interest T-bills might well have no economic effect, but a central bank doesn't have to purchase T-bills.

The Fed, at least, has long included a variety of longer term to maturity government bonds in its asset portfolio. Their yields are not zero, and so they are not perfect substitutes for money.

The policy of having the central bank purchase longer term government bonds has been called "quantitative easing." This is a policy that quasi-monetarists have generally supported. However, I think many quasi-monetarists see the policy as being a matter of increasing the quantity of money, without specifying the term to maturity of the bonds being purchased.

My view is that there is little point in making open market purchases using assets with interest rates that have turned negative and reached the cost of storing currency. (If the Fed insists on paying interest on reserve balances, then there isn't much point in expanding the quantity of money by purchasing assets with a lower yield than the Fed is paying.)

If central bank purchases of any particular asset drives its yield below that on money, then there is good reason to believe that purchases of those specific assets will raise the demand to hold money to match the increase in the quantity of money.

However, the quasi-monetarist approach to quantitative easing is not about targeting a particular quantity of base money now or in the future or having the the Fed hold a particular quantity of bonds--short or long. The central bank must commit to expanding the quantity of money however much is needed to get the expected value of nominal GDP to target.

If a central bank prefers to purchase short term bonds, and those purchases are ineffective, or more importantly, perceived to be ineffective, then the central bank will end up purchasing all of them and will need to purchase some other type of asset--for example, long term bonds.

Scott Sumner is probably the best known quasi-monetarist. He has argued that the reason the yields on short and safe assets are low today is that firms and households have a well founded fear that nominal expenditure on output will remain low for at least the next several years. If the Fed were to commit to expanding the quantity of money however much is needed to return nominal GDP to a higher growth path, then the yields on short and safe assets would rise.

What is Krugman's view? It is unclear. One possibility is that given current expectations of nominal expenditure, real interest rates must turn negative in order for nominal expenditure to rise. After nominal expenditure does rise the necessary amount, then this higher level of nominal expenditure will be expected, and then interest rates on short and safe assets can rise again.

Another possibility is that even if nominal expenditure was expected to be on target, the yields on short and safe assets would remain low. For Krugman, it would have to be zero. For example, the increase in desired saving implied by a general deleveraging would tend to lower all interest rates, including those that are short and safe. Or perhaps it is a matter of risk. Rather than being worried that low nominal expenditure is likely persist for some the time, the worry of loss for any particular investment project leads people to flee stocks and corporate bonds and flock to T-bills and FDIC insured deposits. This is the sudden increase in risk premia theory.

For a quasi-monetarist, in the first scenario, a willingness by the central bank to purchase whatever assets are necessary in the needed amount will result in an equilibrium where the quantity of money rapidly falls back to historical levels and the central bank can hold short term bonds if it prefers.

In the second scenario, equilibrium will require a quantity of base money that remains high by historical standards and a central bank that holds some longer term to maturity and riskier bonds. (As I have explained before, an alternative solution to this problem would be to privatize the issue of hand-to-hand currency and allow the nominal interest rate the central bank pays on reserve balances to fall with other short and safe assets--if necessary, as negative as needed. Monetary equilibrium can be maintained even if the central bank holds only short and safe assets.)

Krugman continues:

Now, in principle you can get traction by making money a less attractive store of value. In particular, if you can credibly promise future inflation, that will make the real return on money negative.

And adds:

Nor does focusing on nominal GDP instead of M2 or whatever really bridge the gap. The point about M2-based monetarism was that it was supposed to give the Fed a target it could clearly control — although in a liquidity trap it turns out that even that isn’t true. Whatever else it is, and whatever virtues it may have, nominal GDP isn’t that kind or target.

One advantage of focusing on nominal GDP rather than inflation is that an increase in nominal GDP doesn't require an increase in expected inflation. Consider two scenarios.

In the first scenario, nominal GDP rises, and real GDP stays the same. The price level rises in proportion, the real return on holding money (or currency anyway) turns negative, and nominal and real expenditure rises.

In the second scenario, nominal GDP rises and real GDP rises in proportion. There is no change in the price level, or at least in the trajectory of prices. However, the improved profitability of investment due to the real increase in production raises the opportunity cost of holding money, and nominal and real expenditure rises.

The second, noninflationary scenario, is better.

And, of course, an expansion of nominal GDP is consistent with intermediate scenarios as well. More inflation and more growth in real output. A lower return from holding currency and a higher real opportunity cost of holding money because of improved real profitability.

Krugman's paradigm completely ignores the ability of expectations of real growth to raise real and nominal expenditures on output. It completely ignores the possible, and likely, scenario where real interest rates on short and safe assets are low because of expectations of persistently low nominal and real expenditures. His framework points solely to the unfavorable scenario where even if real GDP returns to potential and is expected to remain there, persistently high inflation is necessary to keep real returns on short and safe assets highly negative.

What about the difficulty of controlling nominal GDP? According to Krugman, a supposed benefit of M2 monetarism (his term for the policy of keeping the M2 measure of the quantity of money on a slow and steady growth path,) was that at least a central bank could control M2. While Krugman doubts whether that is really true, he correctly argues, I think, that targeting nominal GDP is more difficult than targeting M2.

The most difficult tenet of quasi-monetarism to grasp is that the appropriate target is for the expected future level of nominal GDP. What anchors nominal GDP now--the current flow of expenditure on output--is expectations about what nominal GDP will be in the future. The quantity of money isn't set today in order to keep the current flow of nominal expenditure on target directly, (an impossibility,) but rather to influence the expected flow of money expenditure on output in the future. And it is that expectation that will indirectly influence and stabilize the flow of money expenditure on output today. It is this lesson that Scott Sumner constantly emphasizes.

Is it realistic to expect that nominal GDP would remain on target each quarter? No. But it is the least bad approach. How well can a monetary regime keep the expected future flow of expenditure on output on a stable growth path? That is the proper standard of comparison.

Consumption and Investment

Over on The Coordination Problem, Steve Horwitz pointed to a post by Robert Higgs. The argument is that real consumption has "recovered," reaching the level of its previous peak. Real investment, on the other hand, remains well below its previous peak. Higgs takes this to be evidence for his view that "regime uncertainty" is causing inadequate investment and responsible for the recession.

However, in a growing economy, real consumption is increasing all the time. To say that real consumption has recovered to where it was four years ago tells us little. The key question is where real consumption stands relative to its previous growth path.

The diagram below shows that real consumption remains well below its growth path of Great Moderation from 1984 and 2007. While it is .8% above its previous peak in 2007, it is now nearly 12% below its previous growth path.

However, it is true that investment is especially depressed. The broad measure that Higgs reports, real gross domestic investment remains 21% below its peak in 2006 and 36% below its growth path of the Great Moderation.

This measure of investment, however, includes residential investment. Real nonresidential investment also remains 15.5% below its previous peak in 2007 and 19% below its trend from the Great Moderation.

Nonresidential investment is approximately $327 billion below trend. Real consumption expenditure is $1.26 trillion below trend.

According to the Congressional Budget Office, real GDP is $977 billion below potential. This suggests that a full return of consumption to trend is not possible. Will nonresidential investment return all the way to trend? Will it supplant part of residential investment, surpassing its previous trend so that the total moves towards its past trend? Or will real investment, either total or nonresidential, remain below trend so that real consumption will recover more?

I have no idea, and am content to allow market forces--adjustments in interest rates and preferences--determine the allocation of resources.

Inflating Away Debts

On Coordination Problem, Peter Boettke is skeptical about Rogoff's suggestion that 4% to 5% inflation for a decade would be desirable because it would reduce the real value of debt. Boettke suggests that Rajan's criticisms of Rogoff are more appropriate.

While I agree that shifting to a new inflation target for a decade would be undesirable, I think returning the price level to its growth path of the Great Moderation would be justified. To return to that growth path over the next year would require an inflation rate of 4.7%. Of course, to remain on that growth path, the inflation rate would then return to 2.4%. That is hardly 4% to 5% inflation for a decade.

The diagram below shows the price level measured using the GDP deflator and its trend from the Great Moderation.

I favor nominal GDP targeting. When there are adverse productivity shocks, the price level moves to a higher growth path, which results in higher inflation. According to the Congressional Budget Office, the U.S. economy has suffered a productivity slowdown since 2000.

In the diagram below, the blue line shows the growth rate of potential output estimated by the CBO. The red line shows the 3% trend growth rate of real GDP (and real potential GDP) from the Great Moderation.

With nominal GDP targeting, the productivity slowdown will result in higher inflation. Using the trend growth path of nominal GDP from the Great Moderation, and dividing by the CBO estimate of GDP, the resulting GDP deflator would imply substantially higher inflation since 2001.

In the diagram below, inflation rises above the 2.4% trend in 2000, rises to 3% in 2005 and peaks at nearly 4% in 2009. Even in 2011, the inflation rate consistent with nominal GDP remaining on its growth path of the Great Moderation would be 3%.

Of course, in reality, the Fed allowed nominal GDP to fall far below that growth path. If nominal GDP was returned to its growth path of the Great Moderation, and real GDP returned to the CBO estimate of potential output, then the price level, shown as P** in the diagram below, would be 6% higher than the trend of the Great Moderation. All of those quarters with inflation greater than 2.4% add up.

The current price level is 8% below the level consistent with nominal GDP targeting--assuming the CBO estimate of the productivity slowdown is correct. For the price level to return to that level over the next year, the inflation rate would need to be 12.6%.

It would be possible to adjust to that growth path over a longer time horizon. The CBO provides estimates of potential output until 2020, and 5% inflation for 5 years would result in a GDP deflator of 144.3, very close to the GDP trend value from the Great Moderation of $22.9 trillion for the second quarter of 2016 divided by the CBO estimate of potential output of $15.9 trillion.

With nominal GDP targeting, after this gradual catch up, the inflation rate would need to slow. However, the CBO expects the potential output slowdown to continue through 2016, so that inflation would remain 3%.