Wednesday, April 14, 2010

Inflation Control

The CPI report for March came out yesterday, so this seems like a good time to discuss inflation control in the US. Where are we today, and what are the prospects for the future? By any measure, recent inflation rates have been low. Year-over-year increases in the CPI are running at a little more than 2% (with monthly increases at an annual rate much lower), and year-over-year increases in the GDP price deflator were lower than 1% in the third and fourth quarters of 2009. Though the inflation rate has been low, rates of growth in conventional measures of the money stock have been very high. Since the fall of 2008, the monetary base has increased by a factor of about 2 1/2, while year-over-year increases in M1 over that period have ranged from 7.5% to 17.5%.

How would an Old Monetarist explain these observations, for example what would Milton Friedman have to say if we could resurrect him? An Old Monetarist might appeal to a liquidity trap argument (though of course the liquidity trap idea is typically associated with Keynesians) and argue that, at a zero nominal interest rate (and 0.25% is close enough to zero) banks are willing to hold any outside money injections as reserves, so the money multiplier is not working to bring about a large increase in media of exchange, which would otherwise have increased the price level. Perhaps our Old Monetarist would argue that the increases in M1 have been somewhat large, but there were also large increases in the demand for M1 as asset holders fled to the quality of insured deposits and currency during the financial crisis, and this mitigated any potential inflationary effects.

An Old Monetarist would have dire predictions for the future, however. He or she might say that, once better lending opportunities arise for banks, and they loan out the reserves on their balance sheets, the money multiplier goes to work, and dramatically large increases in M1 fuel a large increase in the price level. Indeed, there appear to be Old Monetarist voices on the FOMC. For example see this speech by the Philadelphia Fed President, Charlie Plosser. Plosser echoes Milton Friedman in this line:
Ultimately, inflation is a monetary phenomenon and there is no question that current monetary policy is extraordinarily accommodative.
Plosser argues that the Fed should be preemptive, pulling reserves out of the system before inflation rears its ugly head, so as to put a brake on an increase in inflationary expectations.

It is views like Plosser's that are likely behind proposals for new monetary instruments to supposedly withdraw reserves, without the Fed having to sell assets from its balance sheet. As the theory goes, the Fed can have its cake and eat it too. One proposed instrument is a Fed term deposit facility, which I discussed in an earlier post. Another is the reverse repurchase agreement. Reference to this instrument can be found as early as this 2002 press release from the Fed, but the New York Fed was experimenting with this relatively recently, as if it were new and they had to figure out how it worked in practice. From the Old Monetarist point of view, the reserves can still be in the system, but if they are somehow "locked up" in term deposits and reverse repos, they can't get out into the economy through the money multiplier process and cause the price level to rise.

An alternative school of thought, New Keynesianism, also has voices on the FOMC, including that of Janet Yellen. This speech does a good job of capturing her views on how monetary policy works. The following two paragraphs are quite useful, for my purposes.
In light of these continuing headwinds in the financial system, the housing market, and the job market, I expect that the economy will be operating well below its potential for several years. Economists use the term “output gap” to refer to an economy that is operating below its potential. We define potential as the level where GDP would be if the economy were operating at full employment, meaning the highest level of employment we could sustain without triggering a rise in inflation. Obviously, with the unemployment rate so high, we are very far from that full employment level. In fact, the output gap was around negative 6 percent in the fourth quarter of 2009, based on estimates from the nonpartisan Congressional Budget Office, or CBO. That’s an enormous number and it means the U.S. economy was producing 6 percent fewer goods and services than it could have had we been at full employment. In view of my forecast of moderate growth and high unemployment, I don’t expect the output gap to completely disappear until sometime in 2013.

This idea of an output gap has important implications for inflation. We have a tremendous amount of slack in our economy. When unemployment is so high, wages and incomes tend to rise slowly, and producers and retailers have a hard time raising prices. That’s the situation we’re in today, and, as a result, underlying inflation pressures are already very low and trending downward. One simple gauge of these trends comes from looking at the U.S. Commerce Department’s price index for core personal consumption expenditures, which excludes the prices of volatile food and energy products. These prices have risen a modest 1.4 percent over the past 12 months, below the 2 percent rate that I and most of my fellow Fed policymakers consider an appropriate long-term price stability objective. I just predicted that the output gap might not disappear until 2013. If the economy continues to operate below its potential, then core inflation could move lower this year and next.
In Yellen's view, it is the New Keynesian Phillips curve that determines the inflation rate. In the 1960s and 1970s, Friedman and Lucas, among others, dismissed the Phillips curve relationship (a negative relationship between the unemployment rate and the inflation rate) then observed in the data as nonstructural, and therefore not exploitable by policymakers. However, the Phillips curve has been revived by Woodford and others (see Woodford's book, Money, Interest, and Prices) as a positive relationship between the inflation rate and the output gap. As far as I can tell, New Keynesians view this Phillips curve as structural and exploitable by the Fed in the short run.

Now, from Yellen's point of view (in contrast to Plosser's), inflation is not a problem. Her view of the recovery from the current recession is quite pessimistic, and she expects the currently large output gap (as measured by the CBO apparently) to persist for several years, thus subduing inflation for a long time. I don't have a direct quote on this, but my impression is that the prediction for current inflation coming out of New Keynesian Phillips curves estimated on previous data is negative and large in absolute value. An extreme New Keynesian might even argue, in the manner of Krugman for example (see here), that the immediate risk is not inflation, but deflation, due to the large output gap.

Mike Woodford might say that New Keynesianism somehow encompasses both Plosser and Yellen, i.e. as in the 1960s when some macroeconomists argued that there was a neclassical synthesis, now there is a New Keynesian synthesis that has absorbed all the key advances in macroeconomics (Old Monetarism, the Lucas critique, real business cycle theory, or whatever) in the New Keynesian framework. Then we can just view Plosser and Yellen as Taylor-rule policymakers, but with different weights on inflation and the output gap in their quadratic loss function. However, it appears that Plosser does not really want to be absorbed - he seems not to buy the Phillips curve logic, and sounds like an unrepentant Old Monetarist.

Now how would a New Monetarist think about the current observations on inflation, money growth, and the prospects for inflation control in the future? In case we haven't been speaking loudly enough for you to hear, Randy Wright (University of Wisconsin-Madison) and I are self-proclaimed New Monetarists, and we lay out our case for New Monetarism here and here. Also see this working paper of mine. First, let's throw out the money multiplier - a misleading load of tripe that remains an integral part of most undergraduate money and banking and intermediate macroeconomics textbooks (excepting this one of course).

In my opinion no one takes currency seriously enough. I use currency very little, and my friends and colleagues appear to do the same, but the quantity of U.S. currency held by the average US citizen was about $2900 in 2009, or about 6% of annual GDP in the aggregate. Now, some of this currency is clearly held abroad (and by various nefarious characters), but the last I checked, Canadians held about 3% of annual GDP in currency at any one time, and the Japanese held about 15%. As Fernando Alvarez pointed out to me, the stock of currency per capita in the world has not been falling, in real terms. Further, in normal times most of the liabilities of the Fed are currency - i.e. outside money is typically mostly currency. We are now in an unusual situation where currency outstanding is $935 billion, or 46% of total outside money, with reserves accounting for the other 54%.

What determines the price level? It's the supply and demand for currency. As I told someone the other day, currency is where the rubber hits the road. The price level has to be determined by the rate at which outside money trades for goods and services, which is the rate at which currencytrades for goods and services. Now this is not so helpful if one wants to be a quantity theorist (particular an American quantity theorist), as (i) we have only a vague idea where U.S. currency resides, so (ii) it seems pretty hard to uncover all the important determinants of the demand for the stuff. However, important determinants of the demand for US currency would have to include the costs of using alternative transactions technologies (checks and debit cards, for example), nominal interest rates, the ease of exchanging other assets for currency (in part determined by access to and costs of using ATM machines), the potential for loss and theft, and the role of counterfeiting.

Now, note that year-over-year percentage increases in the stock of US currency outstanding rose to a peak of about 11% in early 2009, and have since fallen to a little more than 3%. This coupled with an increase in the world demand for US currency (which is probably much more popular in places like Iceland) seems easily consistent with the low inflation rates we have been seeing. One can be a Monetarist - but it helps to be a New Monetarist who is looking at the right monetary quantity.

Should we take the Phillips curve seriously as a guide for monetary policy? Absolutely not. First, Harold Cole and Lee Ohanian long ago provided convincing evidence here that Phillips curve relationships are of no value in forecasting inflation. Second, particular in the current context, we have no idea what the output gap is. Given that we cannot fully understand and quantify the causes of the current recession, we cannot measure the output gap. Maybe the drop in output was all due to inefficiencies, and the output gap is huge. Maybe there were no inefficiencies and the output gap is zero. Further, even if the output gap is huge, it's not a sticky-price-inefficiency output gap, it's a financial-frictions-inefficiency output gap, which has to mean something different.

How do we control inflation over the medium to long term? Is inflation control possible without somehow "draining" reserves from the system? As I discussed in an earlier post, as long as there are positive excess reserves in the banking system overnight, the relevant policy rate is the interest rate on reserves (IROR). Due to the fact that some important players in the system do not receive interest on reserves (principally Fannie and Freddie), and some lack of ability to arbitrage, fed funds currently trade at 10 or 12 basis points below IROR. There appears to be no reason to believe that this gap will not persist as IROR rises, and it should get smaller, as reserves will tend to shift to the banks that earn interest on reserves as IROR gets larger.

Now, inflation is not hard to control, even with a huge quantity of reserves in the system. As returns on other assets increase as we pull out of the recession, banks' willingness to hold reserves will lessen. Given the quantity of assets on the Fed's balance sheet, if IROR remains fixed at 0.25%, the quantity of reserves will fall, the quantity of currency will rise, and so will the price level. To prevent inflation from happening, the Fed will have to increase IROR, which will increase interest rates on all short maturity liquid assets.

Given that a positive quantity of excess reserves are held overnight, and fixing IROR, what would be the effect of an open market sale of assets (treasuries or mortgage backed securities - MBS) by the Fed? Given the current level of reserves, very little. Since the marginal liquidity value of reserves is essentially nil (the system is awash with the stuff) in daylight transactions, reserves are no better than T-bills, and arguably worse, since T-bills are useful both inside and outside the banking system as collateral. However, as the level of reserves falls, open market operations start to bite. With a fixed IROR, an open market sale of assets replaces an asset with a high marginal liquidity value (reserves) with an asset with a lower marginal liquidity value, and this should increase the T-bill rate, but would have no effect on IROR (obviously it's fixed by the Fed) or the fed funds rate. Thus, with a low enough level of reserves, tightening can be achieved with two instruments - asset sales and increases in IROR. The more asset sales there are, the lower IROR needs to be to prevent serious inflation, and that will be important.

It seems imperative that asset sales (MBS or treasuries) begin as soon as possible. The case for selling MBS is strong - holding private assets is just unseemly for a central bank. The goal of the MBS purchase program was to favor the housing sector, and I think that is both inefficient, and a dangerous political game for the Fed. The Fed should aim to get rid of MBS holdings within 2 or 3 years.

What about the term deposit facility and the reverse repos? Both are silly ideas. The Fed can do everything it wants with either outright sales of assets or by moving up IROR. For example, increasing IROR is more efficient than trying to tie up reserves as term deposits or executing a costly reverse repo.

Finally, what are the potential risks as we move out of the recession? First, there is the possibility that the Fed will not have the stomach to increase IROR as necessary to squelch an escape of reserves. Second, it is possible that the Fed has boxed itself in by mismatching maturities on its balance sheet. The Fed is holding a large quantity of long-maturity MBS and long-maturity treasuries. As the Fed moves up IROR, the Fed's profits fall, though of course it still will pay zero interest on currency. At some point IROR becomes large enough that the Fed has to start printing outside money to pay for the losses on its intermediation activity. I'm sure someone at the Fed knows what this level is, but I don't. Alternatively if the Fed sells its long maturity securities, it will be moving asset prices against itself. Given that the Fed sells the assets at a lower price than it bought them, the outside money that financed the original purchases is in circulation forever. I don't know what the potential quantitative effects would be, but this is something the Fed needs to ascertain.

In conclusion, I see no reason for a preemptive strike on an inflation that has not yet made itself apparent. Once we see more inflation, there is plenty of time to respond to it. However, inflation control should be done in a context where we dispense with output gaps and money multipliers.

1 comment:

  1. Stephen,

    A couple points:

    1.) Milton Friedman and Anna Schwartz actually recommend using currency as the definition of money (at least in cross-country analysis) in Monetary Statistics of the United States.

    2.) Regarding the money multiplier, Brunner and Meltzer actually emphasize the same themes for the multiplier as you do for currency above. Specifically, their 1976 paper that develops a closed economy (Old) monetarist model, the money multiplier matters to the extent that the demand for currency changes.