[Finally spring break! I will do the Depression piece this weekend, but I am doing this first, as I hope it will give people a better idea of where I am coming from. To non-economists, it may seem weird and off-topic at first, to economists it may seem simplistic and off-topic, but bear with me, there is a method to my madness.]
Lesson 1. Nominal GDP measured in apples:
Before playing tennis, you are supposed to stretch your muscles (although I never do), so this is a mental stretching exercise to help you see monetary theory differently. Let’s take a good with a roughly unit elastic demand, pretend it is apples. I want to estimate nominal GDP measured in apple terms, and see what happens to that total after a big harvest.
Now suppose that initially the price of apples is 50 cents a pound, and then after a harvest twice as big as normal the price falls to 25 cents a pound. In that case if we used apples as our “medium of account,” or what monetary theorists call our “numeraire,” then our $14 trillion dollar GDP would start at 28 trillion pounds of apples, and then quickly double to 56 trillion pounds of apples.
Now let’s consider how the big apple harvest “caused” nominal GDP in apple terms to soar. Did it do so by reducing interest rates? By making consumers more confident? By boosting the animal spirits of investors? By spurring banks to lend more? (Here’s where I have probably lost half my readers already.) OK, it is a silly example. I agree. But it is also useful to think about exactly why it is silly.
Perhaps it is silly because apples are not the medium of exchange. Money is a medium of exchange, and so I am not doing “monetary theory” here at all. Maybe I am really doing microeconomics, or “price theory,” which is the theory of relative prices. But I don’t agree. Consider, the antebellum free banking era. I believe the U.S. was on a metallic standard (whether it was gold, silver or bimetallic, I don’t recall. But it doesn’t matter.) Let’s say it was silver. Make it simple and assume the “dollar” was an ounce of silver. I also read that the media of exchange were private banknotes of dubious quality, and that shops would post sheets listing the discount on each note based on the perceived soundness of each bank. (Good to know we’ve moved beyond that era of reckless banking!)
Now I’m sure I’ve oversimplified and that many of the notes traded at par, but what if none had? The wildcat banking era shows that the media of exchange and the medium of account need not be the same. And if they weren’t, which one was really “money”? Many would say the medium of exchange. But monetary theory says otherwise, all our models of inflation are based on the notion that prices are denominated in a medium of account (which is now Federal Reserve notes, and in 1929 was 1/20.67 oz. of gold.) Monetary theories are theories of the value of the medium of account. Thus in my example the price level is going to be exactly the inverse of the value (or purchasing power) of silver, not dodgy banknotes.
Returning to the apple example, where apples are assumed to be the medium of account for the sake of argument, what makes the doubling of the apple supply seem so different from doubling the supply of FR notes? I think there is really only one plausible answer–nominal rigidity. There are two types of nominal rigidities that matter:
1. Nominal debt
2. Wage and price stickiness
The first is less important, as although it leads to wealth being redistributed under unanticipated inflation, ex post those are sunk costs and benefits and hence normally have little direct impact on output. The second type of stickiness changes behavior at the margin, and is widely viewed as contributing to the business cycle, especially unemployment fluctuations.
OK, so your intuition that the apple example was fishy is right, but maybe for the wrong reason. Before we move on, however, we need to consider one more problem. Wages and prices are not sticky forever, and thus the apple example, believe it or not, is actually a good analogy for the long run effects of monetary policy, for the classical model. So I wasn’t completely joking with my sarcastic questions about what roles do interest rates and bank lending play in the doubling of NGDP in apple terms. They play no role. And I believe they also play no role in the long run effect of monetary policy, although that is more debatable.
Here’s where the mind stretching comes in. It seems to me that most people approach solutions for the problem of raising our $14 trillion dollar NGDP up to $15 trillion, like they are viewing an almost insurmountable task. Atlas with the world on his shoulders. The metaphor that I think is actually more apt is a drunk standing next to a pillowbed. How hard is it to just flop down? That’s the challenge our monetary policymakers normally face if they wish to raise NGDP. Just debase the currency.
Later we will consider whether, and how much, a liquidity trap makes their job harder. But unless you have a sense of why it is normally so easy to raise NGDP, even in an economy beset by all sorts of real problems that are unrelated to liquidity traps, then I am afraid we cannot have a meeting of minds, a productive conversation. Raising NGDP? What we are talking about is basically just debasing the currency (except NGDP includes both prices and real output.)
So don’t provide me with any argument that an expansionary monetary policy cannot work, that is not somehow tied to the liquidity trap. Housing market is overextended? Businesses don’t want to invest because sales are weak? Exports falling because of a worldwide drop in demand? Credit market risk spreads are high? Those are often true in non-liquidity trap recessions, where monetary policymakers are perfectly capable of debasing the currency.
Lesson 2. Humean Monetary Theory, 1752-1968:
Hume developed much of the macro that we teach to our EC101 students. Later we will see that there is something interesting about the endpoint of Humean macro, as 1968 is both the date it was replaced by another model, and (not entirely coincidentally) the date it stopped working.
Hume did the obvious thought experiments that many of us do in class; “what if a helicopter suddenly dropped lots of cash on a kingdom, and doubled the money supply.” (Well, without the helicopter.) I don’t think he needed any data to come up with an answer. Common sense was enough, and Hume had plenty of that. (Although since he was a student of history I presume he also got his model through “data mining.”)
But Hume actually went beyond the simple Quantity Theory of Money, and also discussed the transition period (before wages and prices had fully adjusted.) In doing so, he came up with roughly the version of the Phillips curve used by Irving Fisher in the 1920s and Keynesians in the mid-1960s. In the short run, a nominal shock (such as an increase in the money supply) will increase both prices and output, but in the long run only prices will rise. And by the way, I use the term “nominal shock” because Hume’s theory wasn’t just about money, he understood that changes in velocity had the same impact as changes in money (as you saw in the quotation that started my second blog post a month ago.)
How can I be so presumptuous as to dismiss 216 years of steady progress in monetary theory? Well I am not the only one to do so; Milton Friedman argued that:
“As I see it, we have advanced beyond Hume in two respects only; first, we now have a more secure grasp of the quantitative magnitudes involved; second, we have gone one derivative beyond Hume.” (1975, p. 177.)
I hope David Laidler is not reading this as he has written several fine books that are sympathetic to the overlooked pre-WWII monetary theorists. I enjoy reading that stuff, and there are clearly many interesting insights that modern young DSGE theorists would benefit from. But I still think that most research into monetary economics, which has obviously been focused on the transition period before wages and prices have fully adjusted, has never really resolved anything. Here are just a few ideas for the transition mechanism:
1. Nominal interest rates (Keynes)
2. Market rates relative to the natural interest rate (Wicksell)
3. Real interest rates (Fisher)
4 Real wages (Pigou)
5. Relative asset prices and investment (the Austrians?)
6. Excess cash balances—>spending (lots of people)
And I am sure that there are many more. But if they have never really resolved anything, then I’m not so sure if we had actually gone much farther than Hume by 1968.
Lesson 3. Milton Friedman, the prophet of fiat money:
During the 1960s, Friedman developed a model of money that is appropriate to a fiat money regime, not a commodity money regime. To understand why, we first need to briefly discuss the price level under a commodity money regime. Note: The following is not exactly true, but is close enough for my argument. Barsky found that price levels follow a roughly random walk under the classical gold standard. He argued that the rational expectations estimated rate of inflation was roughly zero, even when, ex post, the eye discerns negative trends (1879-96) or positive trends (1896-14). I strongly agree with this as a rough approximation of reality, which explains why Keynes refused to accept the idea that the Fisher effect (inflation expectations affect nominal rates) was of practical importance outside of a fiat money regime.
The uptrend of inflation under the last gold standard (1934-68) was mostly a one-time adjustment in the price level to the devaluation of 1934. After 1968 the free market price of gold rose above $35/oz. and we were clearly in a fiat money world. But even before this time Friedman (and Phelps) developed a model that features:
1. Highly potent monetary policy with an almost unlimited ability to influence the time path of nominal spending.
2. A Phillips curve that shifts as people change their inflation expectations.
3. Criticism of old Keynesian theorists for ignoring the Fisher effect.
This is what Friedman meant by the “one derivative beyond Hume” remark. Friedman saw that Hume only looked at changes in the price level, whereas now we look at changes in the (expected) rate of inflation. But of course there was really little need for such a model until 1968, when the last vestige of commodity money was abandoned (not 1971 as many believe.)
Lesson 4. The Rational Expectations Revolution:
immediately after 1968 there was a brief period of speculation about what would happen if you kept raising the inflation rate. But that silliness could not last long, and by the mid-1970s Lucas and others showed that what really mattered is not actual changes in inflation, but rather (rationally) unanticipated changes in inflation.
The key insight of rational expectations (which should be called consistent expectations) is that if you model the economy in a way where policy X produces result Y, you should not assume the the rest of the public believes policy X produces result Z. This is especially true of public policies. It is very unlikely that a policy regime will be effective if it is based on the assumption that the public will respond foolishly to your policy. They might behave foolishly, but you can’t count on it.
A finance equivalent would be to set up a regulatory structure that assumed the SEC could predict stock prices better than investment banks. (Many don’t realize that that dubious assumption is implied in arguments that the SEC should have stopped investment banks from buying subprime mortgage securities.)
The rational expectations revolution also showed that:
1. Today’s AD will be heavily influenced by changes in tomorrow’s expected AD, and thus by changes in the expected future path of monetary policy.
2. Changes in the expected future path of policy show up immediately in the auction-style commodity, stock, and bond markets.
I think point 2 is under-appreciated, even today. Point 1 lies at the heart of right wing macro, but also at the heart of modern new Keynesian macro. The only difference I have with Woodford, et al, is that they look at the future expected path of short term rates (relative to rate consistent with macro equilibrium) and I look at the expected path of the monetary base (relative to the real demand for base money.) They use the interest rate transmission mechanism, I use the excess cash balance mechanism.
Woodford justifies his approach on the grounds that real world central banks target short term rates. That is true, but in my view it is a cognitive illusion to assume that this means short term rates transmit the impact of monetary policy. I see them as merely a symptom of the disequilibrium period associated with sticky wages and prices. Many people think that interest rate cuts speed the expansionary impact of monetary policy. They may help it impact real GDP, but they actually slow down the impact on nominal GDP (by depressing velocity.) If interest rates didn’t change at all, the excess cash balance effect would work almost instantly—raising wages and prices. But of course the fact that wages and prices are sticky means it cannot work instantly—so short term interest rates move, and appear to be more important than they really are.
My other problem with the interest rate approach is that I find it hard to picture the stance of monetary policy by looking at interest rates. During the Great Inflation we often saw charts (like the one in Barro’s macro text) that showed the extremely close correlation between the long term average growth rates for money and prices in the high inflation countries. But what kind of policy data would a Keynesian use in a “monetary history”? Nominal interest rates are also positively correlated with inflation for those countries, but the causation goes from inflation to interest rates. So the rates tell us nothing about policy. Of course this isn’t a flaw in the Woodford model, which allows for the classical result in the long run. It’s just one reason why I prefer the quantity of money/excess cash balance approach. It’s easier for me to see what’s going on.
For me the big insight from the rational expectations revolution is that future expected policy affects AD today (which we already saw in the George Warren post.) Thus if the Fed is expected to increase the money supply 20% next year, and the increase is expected to be permanent, the expected future NGDP three or four years out may rise about 20% (assuming all wages and prices have adjusted by then.) But that expected increase will sharply boost all sorts of (relative) asset prices today (while wages and prices are still sticky) and dramatically impact AD today. (I believe this last part is similar to the Austrian view.)
To summarize. I see monetary policy as follows. Start with the long run and work backward. We know the apple example applies in the long run. Ceteris paribus, long run NGDP will rise one for one with monetary base increases expected to be permanent. Then work back to get the transmission mechanism. Real wages, interest rates, asset prices, etc., will be affected in a way we don’t fully understand, but the key is that higher expected future NGDP tends to push up current NGDP, just as in microeconomics expectations of future price changes affect the current price of a single commodity.
Lesson 5. Short run monetary policy under a gold standard:
Under a gold standard, changes in the price level are exactly (inversely) proportional to changes in the value of money (in terms of purchasing power.) So it is very simple to model the price level; increases in the supply of gold (from mines) raise the price level. And increases in the demand for gold (monetary or non-monetary) reduce the price level. Monetary policy, if it works at all, generally works through changes in the demand for gold. I titled this lesson “short run policy” as in the long run the price level will be determined by the marginal cost of gold production. But the short run is very important.
There are different types of gold standards. Interestingly, under the two extreme cases, full-bodied money 100% backed with gold, or a system with minimal reserves that uses open market operations in T-securities to peg the nominal price of gold, monetary policy is nearly ineffective. This is because in each case the central bank has almost no impact on gold demand. (They do have one option under full-bodied money, however, by raising reserve requirements they can increase the demand for base money, and hence the derived demand for gold.)
The interesting case (to be considered tomorrow) is when central banks hold substantial but still fractional reserves of gold, and then adjust that ratio for policy purposes. This was never more true than in the interwar period. The basic idea is that if a central bank has a lot of gold (the U.S., France, or even Britain) it can affect the world demand for gold by reducing or increasing its ratio of gold to base money—i.e. by violating the”rules of the game.”
Liquidity traps can occur under a gold standard (as in 1932), but they are not what Keynes thought they were. They occur when large increases in the demand for gold have produced deflation and near–zero interest rates. Under these conditions a central bank may not be able to boost the price level without losing all of its gold reserves. But this is how gold standards are supposed to work, they are supposed to sharply curtail the discretionary power of central banks. This has nothing to do with the liquidity trap as envisioned by Keynes. The gold standard limits discretion just as much when interest rates are 5%, as when they are 0%.
Because year to year fluctuations in the (flow) supply of gold and industrial demand are fairly minor, as a practical matter modeling short run movements in the price level under a gold peg becomes a problem of modeling central bank demand, and private hoarding of gold bars, both of which were very unstable. And central bank demand has two components, the real demand for base money (mostly determined by the public and banks, except for reserve requirements) and monetary policy—changes in the central bank’s gold ratio (the ratio of gold stocks to the base.)
To the extent that PPP holds, small central banks have little power. To the extent that it doesn’t hold, small central banks can have some control over their price level (as in a closed economy model.) In practice they didn’t have much power, partly because of rational expectations—investors understood that PPP tends to limit discretion in the long run, and would attack the currencies of central banks behaving recklessly.)
Lesson 6. Monetary policy under a fiat money regime:
There are many possibilities here. One is that central banks are not independent, and are forced to monetize the public debt. Then you have the “fiscal theory of the price level.” This theory might also apply if money and public debt are perfect substitutes. If fiat money injections are temporary (as in colonial American during the French and Indian wars) they won’t cause much inflation. This can be explained with a fiscal theory (Bruce Smith), or with a more conventional monetary theory with rational expectations (me, in 1993, and Krugman, in 1998.) Or you could have one fiat currency pegged to another, in which case you would first need to model the monetary policy of the dominant currency.
But let’s focus on the most plausible case for the US; a freely floating currency produced by an independent central bank, where currency injections are often permanent. We will first consider the case were cash and bonds are not perfect substitutes, then finish up with the case where they are.
This is where the helicopter example is most often used. You drop some money out of a plane, and per capita holdings double from $100 to $200. At first everyone has more cash than they prefer to carry in their wallets. So they bring it to their bank. But even in a recession when no one is borrowing, banks would prefer to hold T-bills earning 3% rather than reserves earning zero percent. So they shovel the cash out just as fast as they receive it (buy buying interest-bearing assets.) This game of “hot potato” continues to accelerate velocity as people desperately try to exchange the excess cash balances for goods, services or assets. AD rises.
Eventually prices double, and people are again content with their cash holdings; what $100 used to buy, now costs $200. So each group gets its way—the Fed determines nominal cash balances, and the public determines real cash balances. The price level adjusts to make this happen.
Some, will say that the preceding example is really fiscal policy. I don’t agree, as I haven’t mentioned any government debt, so neither the public nor the government are necessarily any better or worse off in real terms. But let’s suppose they are right. It doesn’t matter because as long as interest rates are well above zero, bonds and cash are not close substitutes and OMOs will produce an almost identical outcome as the helicopter drop. The extra AD generated by having a bit more cash (i.e. disposable income) is trivial compared to the extra AD generate by having the expected future price level double.
So let’s move on to the really interesting case where T-bills earn a zero yield. In that case if you exchange cash for T-bills nothing may happen to AD. In a sense, T-bills have become, de facto, part of the monetary base. So while an open market purchase does raise the official base, it doesn’t budge the more relevant aggregate of “non-interest bearing government paper.” This is the case that Keynes and Krugman worry about.
And it’s also the problem we face today, (or would be if they stopped paying interest on reserves, and rates fell all the way to zero.) Here’s my problem with discussions of the liquidity (or expectations) trap. There is a view that “it all boils down to X.” You might think from my preceding (highly simplistic) course in monetary economics, that I like boiling a problem down to its essence. But in the case of liquidity traps I believe great mischief has been done by oversimplifying the problem. Here are just some of the real world issues:
1. Can the central bank limit the demand for base money (as I proposed with my interest penalty scheme?)
2. If the interest penalty on reserves works, are cash held by the public and T-bills really perfect substitutes? More importantly, are cash and government bonds with positive yields perfect substitutes? What about other relatively safe bonds? Some call this option “fiscal policy” as the government might lose money. But the expected loss is zero if markets are efficient.
3. Can the central bank influence the public’s expectations of future monetary policy by signaling as follows:
a. Are massive purchases of government debt (even T-bills) a signal of future policy intentions? As you have already heard me say, ad nauseum, I think Krugman, et al, drew the wrong implications from the Japanese case, thinking that the BOJ wanted to escape deflation but could not credibly promise to be irresponsible. I see no evidence that this was the case and lots of evidence that it was not.
b. Can a central bank signal future policy intentions with an explicit nominal target and a forceful promise to make up for any shortfalls in the future? The BOJ never made any such commitments.
4. Can a central bank in a liquidity trap target the price level by pegging the currency to a commodity, and then gradually adjusting the peg to control prices (as Fisher and Warren argued, and FDR showed?)
5. Can a central bank peg their currency to another currency, and then adjust that peg until the price level target is hit (as McCallum, Svensson, etc, argue is a nearly “foolproof” policy.)
6. Can a central bank target the price of CPI or NGDP futures contracts (as Sumner, Dowd, Woolsey, et al, have argued.)
I could obviously go on and on. In a sense, my earlier “multifacted” proposal for monetary stimulus was aimed squarely at this endlessly complex and confusing phenomenon called the liquidity trap. It doesn’t boil down to any one thing. Just as AIDS patients do best with a “cocktail” featuring many drugs, the financial markets are desperately crying out for a credible, multifacted policy mix that would address the liquidity trap from all sorts of different angles at once. Once we scare this ghost away, we’ll see how weak it was all along (as FDR found out in 1933.)
I keep searching for the perfect metaphor for how I envision monetary policy. Let me try this one out: I am an Archimedean. Archimedes claimed that given a fulcrum and a long enough lever (and a place to stand), he could move the world. I believe that the Fed is that fulcrum and the control of the supply (and perhaps demand) for base money is the lever. Give me (or anyone with similar views) control over that policy and we could stop the world’s nominal GDP from falling, and lift it back up. And do it surprisingly quickly.