Lender of Last Resort

Some of the discussion of the sovereign debt crisis in southern Europe has centered on the lack of a national central bank that can serve as a "lender of last resort." The European Central Bank, (ECB) is not playing that role, providing liquidity to government borrowers by purchasing, or standing ready, to purchase their debt.

In my view, one of the traditional roles of central banks has been to serve as "fiscal agent" of the government, and managing the national debt is one aspect of that role. "Lender of last resort," on the other hand, involves lending to banks, rather than to the government.

In my view, the role of the central bank/monetary authority should be to adjust the quantity of base money to accommodate any change in the demand to hold it. Base money is made up hand-to-hand currency issued by the central bank and reserve balances that banks hold with the central bank.

Members of the nonbanking public, households and firms other than banks, obtain hand-to-hand currency through banks. The demand for that part of base money often manifests as deposits or withdrawals of currency from banks. And it is banks themselves that demand reserve balances. The demand to hold base money is closely related to the banking industry. Perhaps the more fundamental reason is that banks borrow through issuing monetary liabilities, financial instruments like checkable deposits, that are close substitutes for base money.

The ordinary process of payments results in withdrawals and deposits of currency and the deposit and clearing of checks. This results in changes reserve balances at different banks--banks with adverse net clearings lose reserves to banks with favorable net clearings. Banks with adverse clearings can borrow from banks with favorable clearings, shifting reserves and currency to where it is needed. In an economy with liquid financial assets, straight sales and purchases of securities can serve the same purpose. Often, short term lending between banks is secured by relatively liquid securities, with repurchase agreements formalizing this hybrid between security transactions and lending.

If, on the other hand, there is an increased demand for base money, either currency or reserve balances, borrowing and lending between banks becomes imbalanced. For example, an increase in the demand for currency results in a withdrawal of currency from some banks that is not matched by additional deposits in other banks. Similarly, if a bank chooses to hold larger reserve balances, it can sell off securities or restrict new lending. Other banks will suffer the balancing reserve deficiencies, but the usual practice of borrowing to meet temporary imbalances doesn't work because the bank accumulating the reserves does not want to lend.

Because the central bank/monetary authority has a monopoly on the creation of base money, it is the only one able to accommodate this increase demand for its "product." Because these demands for base money often show up through the banking system as a demand to borrow without a matching supply of base money to lend, meeting the demand for base money involves serving as "lender of last resort."

In the situation of a bank run, or even a prospective bank run, where many banks seek to accumulate reserve balances to pay out if the run does materialize, the added demand for base money is large, and the consequences of a failure to meet this additional demand can be catastrophic. The market process that eventually returns the economy to equilibrium requires a reduction in the price level, including resources prices like wages. The real quantity of base money rises to meet the demand. While this could occur instantly (leaving aside the near impossibility of conceiving of a smooth and instant resolution of defaults caused by the transfer of wealth from debtors to creditors,) in reality, putting the economy through a deflationary wringer results in unnecessary disruptions in production and employment. Having the central bank/monetary authority serve as "lender of last resort" seems like an essential practice.

What exactly does this have to do with a government that needs to borrow? Generally, governments are not borrowing to accumulate base money, much less to meet the demands for base money by the public. Governments are borrowing to spend. Serving as "lender of last resort" to the government isn't about accommodating an increase in the demand to hold base money, but rather printing up money for the government to spend.

It is true that governments that borrow short to fund a large national debt must constantly borrow to repay debt as it comes due. If lenders loose confidence in the government's ability or willingness to repay, then they will stop lending. The government cannot repay debt as it comes due, and defaults.

Banks are in a similar situation in that their liabilities come due frequently, and banks must borrow again to repay these debts. With monetary liabilities issued by banks the process is more or less continuous. If depositors lose confidence in the ability (willingness is less of an issue with banks bound by contract to redeem in base money,) of banks to repay, then they will no longer lend. The bank defaults.

If, from the economist's God's eye point of view, the bank is really solvent, then going through the costs associated with default is a waste. Similarly, if General Motors is really a sound business, going through reorganization would be a waste, and government planners should fund the business. If the government as a whole really is able and willing to repay its debts, then someone should lend to the government to avoid the wasteful disruption caused by default.

In my view, there is no "God's eye" point of view, and when lenders lose confidence, default and bankruptcy is the proper response. For governments, this means giving existing creditors new bonds to replace the old bond at whatever interest rate the government believes it can pay, and bringing current expenditures in line with current receipts.

Similarly, while the monetary authority/central bank should always accommodate changes in the demand to hold base money, if depositors loose confidence in a bank, then it should be reorganized. With a very concentrated banking system, the loss of even a single large bank could be very disruptive. And it is also possible that many banks could have similar problems at the same time, leading to a loss of confidence by depositors, independent of an increase in the demand to hold base money. An increase in the demand to hold base money, not accommodated by an increase in the quantity of base money, will almost certainly cause difficulties for all banks. But it has become apparent that many banks can get into difficulty by treating sovereign debt or mortgages as if they are (nearly) risk free. Provisions for a rapid reorganization of many banks is important.

In my view, having a central bank/monetary authority create money because someone needs to borrow, is a mistake. "Lender of last resort" is the wrong perspective. Supplying the quantity of base money demanded is the proper role of a central bank/monetary authority.

Nominal GDP Targeting in 1982?

Marcus Nunes posted some FOMC minutes discussing nominal GDP in 1982. They were found by Ryan Sanchez.

Morris explains that the Fed needs to be looking to nominal GDP, but that an explicit target for nominal GDP would cause political problems. The argument for looking to nominal GDP is that M1 velocity was shifting, and so a target for M1 would be a mistake--it would result in the wrong level of nominal GDP.

What is the political problem? It had to do with the Fed creating less nominal GDP than the President proposed. If the President were to propose 12 percent nominal GDP, and the Fed said that it would only create 9 percent, this would put the Fed at cross purposes with the President. This is very closely related to Sumner's view that the central bank acts last. At least in 1982, the Fed was not willing to openly take this role. On the other hand, they did seem to be willing to really play the role--just not openly.

Interestingly, targeting interest rates was also considered unacceptable. Perhaps the problem was setting interest rates at historically high levels. The quantity of money was considered beneficial because it wasn't controversial. However, from context, that "benefit" apparently was that no one understood the disinflationary impact of low money supply growth targets.

Volcker downplays the ability of the Fed to control nominal GDP (and interest rates, of all things,) and suggests that the Fed watch exchange rates as well as inflation. Volcker also sees "obvious dangers" to targeting nominal GDP. What are those dangers exactly? Volcker doesn't say, but his next remark is "that there is great overemphasis on what monetary policy can do." To me, that looks like nominal GDP targeting creating too much accountability for the Fed.

Nominal GDP in the Seventies

Was nominal GDP growth stable in the seventies?

The following chart shows nominal GDP and the trend.

It doesn't seem that much worse than the Great Moderation. The growth rates are substantially different. The trend growth rate of nominal GDP in the Great Moderation was 5.4 percent. In the seventies, it was 10 percent.

Using the CBO estimates of potential output, the GDP gaps were quite large during the period.

There is a pattern between the gaps between nominal GDP and the trend of nominal GDP and real output and potential output.

While the NGDP gap between nominal GDP and trend and the RGDP gap between Real GDP usually move together, it looks like treating the seventies as a period with a constant trend growth rate of nominal GDP is a mistake. Irresponsible out-of-control monetary policy is probably a better description.

Inflation Expectations: A Two Edged Sword

New Keynesians (and sometimes Scott Sumner) have argued that the U.S. economy needs higher expected inflation. For the new Keynesians, the rationale is that real expenditures are below the productive capacity of the economy and lower real interest rates will help to raise real expenditures. Firms will respond by expanding production and employment, bringing real output towards potential. Usually, given expected inflation, lower nominal interest rates would have the same effect. With the Federal Reserve's target interest rate near zero, however, lowering nominal interest rates by lowering that target rate is no longer possible. The argument for higher expected inflation is an argument about aggregate demand.

New Classical economists have sometimes responded to these arguments by claiming that higher expected inflation will have no effect on real output. They aren't making an argument about aggregate demand. They are making an argument about aggregate supply. If inflation is equal to the expected level, then real output will equal potential. It is only when inflation deviates from the expected level that there is some transitory effect on the willingness to produce goods and so some impact on real output and employment.

New Keynesians generally have a different approach to aggregate supply. Inflation depends on past and expected future levels, and also on the output gap, the difference between real output and potential. Even so, an increase in expected inflation implies that more current inflation will be associated with any expansion of real expenditure necessary to close the output gap.

From a market monetarist perspective, both real output and inflation respond to more rapid growth of nominal expenditure on output. Higher expected inflation should result in a division more weighted towards current inflation and less towards real output. And so, generating higher inflation expectations to reduce real interest rates, or perhaps, increase the expected cost of holding currency (which amounts to the same thing,) in order to raise aggregate demand has undesirable effects on aggregate supply.

Now, turn this argument about. Currently, nominal GDP is on a much lower growth path compared to the Great Moderation. My measures show nearly 14 percent lower. If potential output is on the same growth path as before, then for real expenditures to return to potential, both prices and nominal wages need to fall to a 14 percent lower growth path as well. Wage rates have hardly budged and the price level is on a 2 percent lower growth path.

The CBO estimates of potential output show a growth slow down for the last decade. Using those measures, real expenditure is about 7 percent below potential. Given the current level of nominal GDP, a 7 percent decrease in the growth path of prices and nominal wages would raise real expenditures to close the output gap.

The natural rate hypothesis suggests that this should be occurring. The 2 percent drop in the growth path of prices is but a drop in the bucket. One possible explanation is that the CBO estimates are too optimistic, or more accurately, not pessimistic enough. It must be that potential output shifted to a 14 percent lower growth path.

However, suppose that there really is an substantial output gap. If there were a reasonably prompt decrease in the growth path of prices and wages, this would involve transitory deflation. For example, if the price level needed to shift down another 12 percent (14% minus the 2% already accomplished,) if this were done over 4 years, there would be 3 percent deflation each year. Unfortunately, this would imply higher real interest rates, at least for the short and safe assets with nominal rates currently near zero.

Of course, the Chairman of the Federal Reserve, Ben Bernanke, loudly proclaims that the Fed can and will prevent any such deflation. One reason for his insistence is to keep expected deflation from raising real interest rates and reducing aggregate demand. However, at the very same time, efforts to keep expected inflation up interfere with the deflation in prices and wages need for the current growth path of nominal expenditures to generate sufficient real expenditures to close the output gap.

Lower expected inflation, and the short run aggregate supply curve increases, raising real output and employment, given aggregate demand. But, lower expected inflation, raises the real interest rate, lowers aggregate demand, and so lower real expenditure results in reduced real output given short run aggregate supply.

Suppose the output gap is entirely ineffective in motivating disinflation. Past inflation rates and expected future inflation rates determine the actual inflation rate. The Fed, with all of its credibility and record, insists that the inflation rate will be 2 percent. Believing this, firms and workers set prices at 2 percent. Sure, the output gap implies surpluses of labor and even output, especially if we imagine nominal wages being forced down to market clearing levels, but even so, prices and wages just continue on the path they would if the credible, 2 percent target were consistent with market clearing. If prices (and wages) do rise more slowly, this is followed by actions by the Fed to convince people that this is just a one time fluke, and prices will again return to rising on target.

Could the Fed's effort to prevent disinflation from raising real interest rates and so reducing aggregate demand be interfering with the market process by which short run aggregate supply shifts and brings real output and unemployment to natural levels? Could this be the reason for the pattern of the Great Moderation where recoveries in employment were so gradual? Sure, the other two recessions were mild, but they were very persistent.

Could inflation targeting be responsible? Does inflation targeting (a growth rate target) generate expectations that make it difficult to close an output gap, (a problem with levels?) The gold standard fixed the level of the price of gold. A money supply rule fixed (I guess,) the growth path of the quantity of money. A price level rule fixes the growth path of prices. Of course, market monetarists advocate targeting the growth path of nominal GDP.

Nominal GDP Targeting and Supply Shocks

Adam P. continues his rather uncivil ways, claiming that my post is a "confused mess."

No doubt, this is due to my failure to communicate clearly. Adam P., think back to your principles of economics course, or, perhaps, to the many times you have taught the course over the years. An adverse aggregate supply shock shifts the aggregate supply curve to the left. The result is a higher price level and lower real output. To prevent the increase in prices, it is necessary to shift the aggregate demand curve to the left as well. While this partially or fully reverses or prevents the increase in prices, the decrease in real output is larger--at least if the short run aggregate supply curve has a positive slope.

He goes on about how "supply shocks" involve reductions in output that cannot be corrected by monetary policy. The standard approach would be that using monetary policy to prevent real output from falling in the face of an adverse supply shock requires that the aggregate demand curve be shifted to the right. This will at least temporarily dampen, prevent, or reverse the reduction in output. At least if the short run aggregate supply curve has a positive slope. Unfortunately, prices will rise by even more than would have occurred if there had been no monetary policy response.

Nominal GDP targeting approximates the "do nothing" response. The aggregate supply curve shifts to the left, and aggregate demand stays the same. The price level rises and real output falls.

I didn't mean to suggest that an aggregate supply shock won't lead to lower real output. My actual view is that a recession is a negative output gap, and so, if potential output falls, and real output was initially equal to potential output, then a recession would involve output falling by more than potential output. Further, it is possible, and perhaps likely, that supply shocks would raise the natural unemployment rate. A recession, on the other hand, would involve the unemployment rate rising above the natural unemployment rate. I believe this is exactly what would happen if aggregate demand is reduced in response to an aggregate supply shock to prevent inflation.

I think the best response to an aggregate supply shock would be one that leaves real output equal to the reduced level of potential output and the unemployment rate equal to the perhaps increased natural unemployment rate. It is almost certain that this would involve a higher price level.

I don't think that nominal GDP targeting will exactly accomplish this. It will if the demand for the particular market facing the decrease in supply is unit elastic. Otherwise, the situation is a ambiguous.

Adam P. shows some charts where he claims that the record of the seventies proves that a stable nominal GDP growth path would have adverse effects due to supply shocks. Of course, part of this was due to his odd assumption that I was claiming that aggregate supply shocks wouldn't impact unemployment if nominal GDP stayed on a stable growth path. He shows that the unemployment rate increased in the recession of 1974.

In reality, the growth rate of nominal GDP shows an upward trend during the period, but it is highly irregular. In particular, the recession of 1974 shows a clear reduction in the growth rate of nominal GDP. It looks like a 4 percentage point decrease out of 12%, and so that is about a 30% drop in growth. (When I get a chance I will look at growth paths.)

What is more incredible, is that he counts the sixties as a period of inflation targeting, where unemployment held stable at 5 percent due to the stable inflation rate. What? The U.S. was targeting real GDP growth and unemployment, trying to exploit a long run Phillips Curve to raise the first and lower the second. The resulting increase in inflation becomes quite pronounced at the end of the period. This was one of the most serious macroeconomic disasters of the twentieth century.

Is this supposed to be a joke?

Nominal GDP Targeting and Supply Shocks: Again.

Adam P. has kindly responded again. After bringing up instability of prices and output initially, an absurd result, he now insists that it is irrelevant. Now, the only question is optimality.

Of course, he is giving up on explosive results or perpetual oscillations, and wants to insist that dampened oscillations to equilibrium is the consequence of nominal GDP targeting. Why? Because if prices are set based upon past inflation rates, overshooting is inevitable. In other words, a macro cobweb based upon myopia.

This result still requires periods of stagflation alternating with deflationary booms. Stagflation is something that has been observed, though I don't think that firms have been especially profitable during those periods. On the other hand, these deflationary booms in output where firms suffer losses--how likely are those?

I don't agree that myopic cobwebbing that eventually settles down to equilibrium is a likely result of a regime of nominal GDP targeting. I think it is entirely possible that the introduction of the new regime might include some undesirable errors in pricing and production.

Adam P. accepted my argument that the equilibrium price level is the target for nominal GDP divided by potential output and the equilibrium for real output is potential output. (Which leads to equilibrium reverting price expectations.) He then claims:

I certainly don't disagree but the problem is that his argument actually implies that an inflation target is superior. The reason is that potential output is neither constant nor observable, as it varies around the value of NGDP that is optimal changes as well unless you want inefficient, and welfare reducing, volatility in inflation. On the other hand, if you stabilize relative prices so that price/wage stickiness doesn't distort the real outcome then you get exactly the outcome Bill describes. If you target NGDP with any variability in all in potential output then you don't get the outcome Bill describes.

The implications of "supply shocks" play a key role in market monetarist thinking. If potential output changes, then firms will need to adjust their prices to clear markets. Adam P. argues that this will create welfare reducing fluctuations in inflation.

Consider what Adam P.'s argument implies. If there is an adverse supply shock, with potential output falling or growing less than the expected amount, then nominal GDP targeting will imply excessive inflation. With nominal GDP remaining on target, the equilibrium level of the prices of output rises. (Nominal incomes, including nominal wage rates, continue to grow at trend.)

How would inflation targeting fix this problem? The monetary authority would slow nominal GDP growth when there is an adverse supply shock, so inflation rate would remain at trend.

Now, consider a specific example--a bad harvest for corn. According to Adam P., what should happen is that the monetary authority slow nominal expenditure in the economy, to keep the inflation implied by the higher price of corn from "reducing welfare" due to its arithmetic impact on the price level. What that really means is that all the other prices in the economy, including nominal incomes like wages, must grow more slowly. The price of corn, then, rises slightly less than it would with nominal GDP targeting, and all the other prices are slightly lower than they would otherwise be. If those other prices, for example, wages, are sticky, then the effort to reduce the "welfare loss" from inflation causes a recession. Not only is potential output lower because less corn is being produced, but the falling sales in the rest of the economy that are needed to signal the necessary reduction in prices (or more moderate price increases) reduces output in those markets too. Real output falls below potential.

George Selgin's "Less Than Zero," discusses these matters, though most Market Monetarists believe he worries too much about the optimal trend growth rate in nominal GDP. While Selgin advocates a trend growth rate of nominal expenditure to match the trend growth rate of factor supplies, the more general aspects of his arguments regarding productivity shocks apply to any trend growth rate in nominal expenditure.

Consider a bumper harvest of corn. With nominal GDP targeting, nominal wages continue to grow at trend, and the price of corn falls. Arithmetically, the price level falls (or grows more slowly.) Real wages rise. Workers have higher real incomes and can consume more corn, but they are suffering a welfare loss from "inflation volatility." With inflation targeting, the central bank should expand nominal GDP growth so that other prices rise enough so that the inflation rate does not fall. Now, if this results in nominal wages growing more quickly, then real wages rise and workers can consume more corn. But if wages are sticky, then real wages don't rise. Other product prices rise enough to offset the deflation (or disinflation) in the corn market. Nominal wages grow at trend. Relative prices do change and so workers presumably purchase more corn and fewer other goods. And real and nominal profits expand. The owners of the firms can purchase more corn and more other goods. And what do the workers get out of this? They don't suffer welfare losses from inflation variability.

In the real world, supply shocks are microeconomic events where specialist firms respond to changes in their particular supply conditions by changes in their prices and production. These changes have an impact on the price level/inflation rate and aggregate output. A model that ignores this, and treats price level variance and potential output variance as independent stochastic processes is fundamentally flawed. While a negative covariance between price level shocks and potential output shocks helps add an element of reality, there is no alternative to considering the microeconomics. If the good with the supply shock has unit elastic demand, then nominal GDP targeting is perfect. Otherwise, a sufficiently clever monetary authority might do better by allowing nominal GDP to change. And while stabilizing nominal expenditure in the rest of the economy appears the least disruptive approach, the elasticity of supply for the good with the supply shock matters as well. The allocation of resources should change.

Unfortunately, in the real world, we have monetary authorities who lack this God-like knowledge. Worse, they respond to negative supply shocks with worries about inflation expectations becoming unanchored. At best, the monetary authority just ignores goods and services that have historically suffered from supply shocks, But they don't always stick to their "core inflation" target. And, of course, supply shocks can occur in markets for goods other than food and energy.

My view is that slow steady growth of money expenditure on output is the least bad environment for microeconomic coordination--in the real world. Inflation targeting was tried. It worked pretty well--until there was a massive decrease in nominal expenditures on output. It is time for a new monetary regime.

Milton Friedman on Nominal Income

Lars Christensen has an excellent post discussing Milton Friedman's A Monetary Theory of Nominal Income. Christensen quotes from Friedman:

In … "A Theoretical Framework for Monetary Analysis," I outlined a simple model of six equations in seven variables that was consistent with both the quantity theory of money and the Keynesian income-expenditure theory…The difference between the two theories is in the missing equation the quantity theory adds an equation stating that real income is determined outside the system (the assumption of "full employment"); the income-expenditure theory adds an equation stating that the price level is determined outside the system (the assumption of price or wage rigidity)…The present addendum to my earlier paper suggests a third way to supply the missing equation. This third way involves bypassing the breakdown of nominal income between real income and prices and using the quant ity theory to derive a theory of nominal income rather than a theory of either prices or real income.

In my view, "the Phillips curve" is not a theory of how prices influence output or output influences prices, but rather how nominal income influences output and prices. Generally, it is output and prices together in the short run, and then prices only in the long run.

There is nothing unusual about that approach, and having money determine output in period t+1 and then inflation in period t+2 seems like a simple way to illustrate this concept. Of course, if the model shows that a situation where nominal income remains unchanged (or on a constant growth path,) so that there are no fluctuations in nominal income, the process by which the short run/long run relationship is modeled generates fluctuations in real output and prices, simply shows that the model is flawed.