January 1, 2016I Win My Inflation Bet with Robert Murphy
December 31, 2015Thoughts on "Almost Wholly Negative"
December 30, 2015Tina Rosenberg on Kidneys in Iran
December 30, 2015There's no taste for accounting
December 30, 2015Recession Bet
December 29, 2015Coercive Priors
Entries by author
Frequently Asked Questions
Here is the long-awaited section on monetary theory. If you ask me the question, what will happen if the Fed stops paying interest on reserves, my answer would differ from Scott Sumner's. I would say, "Not much will happen in the overall economy." Bank profits will be a little lower, and Federal Reserve profits (which are rebated to taxpayers) will be a little bit higher, and that is about it. This installment, below the fold, eventually spells out why I believe that. (Actually, some of the justification for my belief will come in a later installment, on finance theory and macroeconomics.) At this stage, I do not expect to convince anyone. It will have to be later in the book, when I review history, that I convince others (or change my own mind).
Most economists use some form of monetary theory to explain economic fluctuations. The simplest model of fluctuations, alluded to earlier, is that the money supply determines prices, prices determine the real wage rate (because nominal wages are sticky), the real wage rate determines labor demand, and labor demand determines employment and output. Alternatively, one version of the Keynesian model is that when savings are channeled into money, this causes an overall shortfall in demand, because money is a non-produced good. The Austrian model of the business cycle has a role for money in that the central bank can use the power to print money in a way that distorts interest rates, leading to booms and busts.
A Sparse Financial System
The easiest way to do monetary theory is to posit a primitive, sparse financial system, with only real assets and currency. Think of real assets as fruit trees. Currency is created by the central bank and is the only medium of exchange.
Classical monetary theory is an argument for why in this type of financial system the supply of money will determine the overall price level. With the level of real output given, this is equivalent to saying that the supply of money will determine the level of nominal output, or nominal GDP.
Prices in the economy will be quoted in terms of currency units. If what circulates are various denominations of dollar bills, then prices will be quoted in dollars.
When a brain surgeon comes into my restaurant, she can either offer me currency or brain surgery in return for a meal. Most of the time, I will prefer to take the money, which I can spend on other things. Even if I think I might need brain surgery years from now, I would prefer to take the currency and invest in fruit trees until the time comes when I need the surgery.
Between the time the brain surgeon pays for her meal in currency and the time I spend it or invest it, this currency sits in a drawer. If it sits in a drawer a long time, we say that the velocity of money is low. If I dispose of the currency quickly, we say that the velocity of money is high.
Suppose that the central bank has put $1 billion of currency into circulation. With this amount in circulation, there is a set of prices at which there will be no excess demand or supply for any good or service. For example, it might be that with a price of a meal of $10 at my restaurant, there is neither excess supply or demand.
If the central bank instead had put $2 billion of currency into circulation, it stands to reason that the price of a meal at my restaurant would settle at $20, rather than at $10. If the price of a meal were still at $10, and all other prices were consistent with that $10 price, then the ratio of currency to the price level, M/P (also known as the level of real balances), would be twice as high as when there was $1 billion in circulation, which means that velocity would be halved. But there is no reason for anyone to want to keep more real balances in a drawer, so it would seem better to presume that the rate of spending would rise, driving up prices until they had doubled.
Suppose that there is now $1 billion in circulation, but we know that the government plans to add another $1 billion to currency in circulation next year, and in fact that it plans to double the amount of money in circulation each year. This means that I know that the value of currency is going to depreciate rapidly every year. Because of that, I become a lot more focused on disposing of money quickly, before it can depreciate. If I expect inflation to be low, I get pretty relaxed about leaving money in a drawer for a while. If I expect inflation to be high, I try to exchange money for real assets or for other goods and services as quickly as possible. I increase the velocity of money and lower my average holding of real balances.
The inverse relationship between inflation and the amount of real balances people are willing to hold is sometimes called the Cagan money demand function. Phillip Cagan examined the way money demand behaved during hyperinflations. When inflation is low, it is difficult to observe the effect of inflation on money demand.
Hyperinflations are situations where the government wants to spend far more than it can collect in tax revenue. It tries to print money to make up the difference, but this causes prices to rise, requiring more money to be printed to meet budget shortfalls, leading to more inflation, and so on. Although Milton Friedman famously said that inflation is, anywhere and everywhere, a monetary phenomenon, hyperinflation is actually a fiscal phenomenon. It is the fiscal imbalance that causes the government to print money in an out-of-control manner.
With a Cagan money demand function, to know the market-clearing price for a restaurant meal, it is not sufficient to know that there is $1 billion of currency in circulation. We also have to know the rate of monetary growth. If money growth is high, the demand for real balances will be low, and the price of a meal might be $12. If money growth is low, people may leave money in a drawer longer, and the price of a meal might be $10.
However, even with a Cagan money demand function, the price level depends on the behavior of the central bank. If we know the amount of money in circulation as well as the rate of money growth, we know what the price level ought to be.
So far, we have assumed full employment and allowed money to affect only the price level. How does money explain fluctuations in employment and output? There are a number of possibilities.
Recall that in the story of Joe, who worked in the GDP factory and faced a cutback in hours worked when he saved some of his paycheck. I pointed out that if Joe had been paid in GDP rather than in money, his savings would not have subtracted from demand.
In general, if workers were paid in output rather than in money wages, there would never have to be layoffs. Assuming that there are diminishing returns to labor, firms could always afford to pay workers a share of output equal to the marginal product of labor times the amount of labor. However, this would just create a problem for workers of how they can sell their output for the goods they desire. Indeed, workers are not better than the firms they work for at finding a market for output, so we know that in-kind payments are not a solution. Having said that, Martin Weitzman once proposed having workers paid in part on the basis of the firm's revenues or profits, which he argued would reduce unemployment by making wages procylical.
Suppose that consumers decide one day that they want to hold more money. In Keynesian terms, there is an increase in liquidity preference. When consumers keep more money in a drawer, the velocity of money goes down, and this puts downward pressure on prices. Ultimately, when prices fall sufficiently, people have enough real balances in the drawer, and the economy can operate at full employment.
As prices fall, should we not see the Cagan effect go into reverse? Indeed, as consumers observe falling prices, they become more willing to keep money in the drawer, leading to more deflation, and so on. However, unlike a hyperinflation, which is self-perpetuating, the deflation caused by an increase in liquidity preference has a natural stopping point, because the demand for real balances is finite. Once the demand for real balances stops rising, for any given level of nominal money supply the price level will stabilize.
We have just argued that, in the case of a large increase in liquidity preference, if the government sits back and does nothing, eventually prices will fall to the point where the level of real balances is high enough to satisfy the demand. However, Keynesians would argue that this non-interventionist approach has tragic consequences. The process of raising real balances by deflation will take time, and along the way output and employment will decline, in part because of sticky wages. Moreover, falling prices will raise the real interest rate ex post, making borrowers unable to repay loans. This is another source of adverse movements in output and employment.
If the public wants to hold more real balances, then the easiest way to accommodate that is for the central bank to increase the money supply. Why, then, does Keynes emphasize fiscal policy?
Keynes argued that when the economy is in recession, the nominal interest rate is likely to be low. In that situation, people are likely to have a large willingness to keep money in the drawer. Accordingly, even a large increase in the money supply will not spur spending or investment. This is the "liquidity trap." Also, as we have seen, Keynes did not think that investment is very responsive to changes in interest rates, because of the importance of long-term expectations. Because of these pathologies, he argued that only deficit spending by the government could restore full employment.
Overall, the monetarist view and the Keynesian view of economic fluctuations have similarities and differences. Think of the monetarist equation, MV = PY, where M is the nominal money supply, V is velocity, P is the price level, and Y is nominal output. Both Keynesians and monetarists see unemployment as caused by a drop in MV. Sticky wages play a role in both stories.
In the monetarist story, fluctuations in MV are typically told as fluctuations in M. There is an assumption that velocity is stable, and problems are caused by monetary authorities causing over-expansion or contraction. In the Keynesian story, fluctuations in MV are typically told as fluctuations in V, which changes as a result of liquidity preference or changes in the "animal spirits" of entrepreneurs.
In the monetarist story, a recession can always be cured with expansionary monetary policy. In the Keynesian story, sometimes the responsiveness of investment to interest rates is so low or the demand for real balances is so high (the liquidity trap) that deficit spending is needed to restore full employment.
What is Money?
In the case of the sparse financial system, in which money consists of currency, we have seen that money is a medium of exchange, money in circulation is controlled by the central bank, and the price level is determined by the level and the growth rate of money. In the real world, this becomes complicated.
In the real world, there are many possible definitions of money.
--In textbooks, money is defined as the medium of exchange, and traditional this has meant M1, which is the amount of currency plus the amount of demand deposits (money in checking accounts).
--The money that is controlled by the central bank in the United States is called M0, or high-powered money, which is the sum of currency plus reserves held by banks.
--The money that is best able to predict the price level (according to Milton Friedman around 1970) is M2, which is currency plus demand deposits plus savings accounts at banks and thrift institutions.
In fact, researchers who focus on an empirical definition of money have come up with what is sometimes called a Divisia Index, which is a weighted average of currency, demand deposits, savings deposits, money market fund accounts, and other financial assets.
In a sparse financial model, it is easy to think of money as currency, in which case there is no conflict between the definition of money as medium of exchange, money as the nominal aggregate controlled by the central bank, and money as the determinant of the price level. How important is it that in the real world it is not so easy to come up with a definition of money that has all three properties? Does the complexity of real-world monetary arrangements matter?
I believe that this issue matters a great deal. I think it creates some fundamental doubts about the impact of monetary policy. One version of these doubts is known as Goodhart's Law. Charles Goodhart says that the tighter the central bank's control over a particular monetary aggregate, the less predictable will be the relationship between that aggregate and the overall economy. In other words, as a central banker, you can go on for years observing a nice relationship between a monetary aggregate and the economy. However, the minute you attempt to manipulate the economy by altering the supply of that monetary aggregate, the relationship will break down.
For example, suppose that the central bank were focused on the quantity of nickels in circulation. Over the past twenty years, let us say, the velocity of nickels (that is, the ratio of nominal GDP to the number of nickels in circulation) has evolved along a predictable trend. One day, the economy sinks into recession, and the Fed decides that it wants to use its control over the quantity of nickels to try to boost nominal GDP. It undertakes a massive open market operation, in which it exchanges vast quantities of nickels for pennies, times and dollar bills. Based on the trend velocity of nickels, the Fed expects this to have a dramatic effect on the price level and on nominal GDP.
In fact, it is hard to imagine that a vast open market operation to exchange nickels for other currency denominations would have much effect on GDP. People would probably keep more nickels in the drawer and fewer coins of other denominations. The velocity of nickels would plummet, and otherwise life would go on.
In today's economy, it is difficult to imagine the Federal Reserve exerting tight control over the medium of exchange. Many transactions use credit cards and debit cards. Many mutual funds allow checks to be written against the assets in the fund.
In my view, this means that an increase in the supply of money will have neither the consequences predicted by Keynesians or by monetarists. In fact, the central bank does not even control the rate of inflation. Markets on their own can increase or decrease liquidity through innovation and adaptation. In terms of the equation MV = PY, whatever M the bank chooses to control, markets will take steps to move V in the opposite direction. The combination of autonomous changes in V and changes in V that are inverse reactions to changes in M means that the relationship between the money supply and price level is not nearly as close in practice as it is in the theoretical model with a sparse financial sector.
I am not saying that Federal Reserve interventions are as neutral as an open market operation that exchanges nickels for dimes. On a day to day basis, the Fed either makes or liquidates loans in the market for short-term repurchase agreements, or repo. That works as follows. Suppose that a securities dealer has an inventory of Treasury bills that it needs to finance. The dealer sells the securities to a bank and simultaneously agrees to repurchase them at a slightly higher price in a week. The dealer is borrowing money from the bank, using the securities as collateral. The interest rate is the repo rate. In the market, the repo rate is usually close to the Federal Funds rate, which is the rate at which banks lend to one another over night. When the Fed makes repo loans, it drives very short-term market interest rates, such as the Fed Funds rate, down. When the Fed liquidates repo loans, it drives these interest rates up.
There is no question that the Federal Reserve can manipulate the short-term nominal interest rate on risk-free loans. The big question is how much these manipulations affect overall interest rates, the overall supply of liquidity, and the economy as a whole. For me, this is a question of how the central bank affects the equilibrium in financial markets. This in turn takes us down another path, the path of finance theory.