One of the great frustrations of discussing monetary policy is that most people buy into the notion that when the Fed is “doing something” we should see changes in short term rates and/or the money supply. In fact, it is expectations of future policy that drive AD.
There are several problems with trying to find links between movements in the money supply, and changes in the economy. First, the Fed is not usually trying to conduct natural experiments. They are trying to stabilize the economy. Thus they will often move the money supply in response to changes in money demand. Markets understand this, and hence often don’t react much to changes in the money supply.
Even worse, an exogenous change in the money supply, even when not done in response to changes in the demand for money, may have little impact if expected to be temporary. If we put these two facts together, then policy will only affect AD when it is not responding to money demand fluctuations, and is not expected to be temporary. But how can the markets know this? After all, the Fed rarely announces “we are setting out on a plan to create the Great Contraction, or the Great Inflation.” Instead, markets gradually become aware of the fact that monetary policy is drifting off course. And it is at precisely this moment, when markets understand that the change in the money supply is no fluke, that we observe market reactions. Of course changes on the broader economy are closely correlated with those market reactions.
Here’s an example. Suppose in the 1960s the Fed ran an easier than normal monetary policy. At first inflation would only rise a bit. Why bid up houses to twice their normal level, if you expected the Fed to soon return to its longstanding practice of low inflation. Then gradually, little by little, people realize that monetary policy is changing in a fundamental way. This time was different. There may be signals, such as the breakdown of Bretton Woods, but even those changes may be partly endogenous. In any case, once the policy is recognized as permanent, the prices of assets such as commodities and real estate will start to rise faster. Stocks are more complicated, rising with the price level, but falling with higher rates of inflation (due to the inflation tax effect.)
This delayed reaction led many monetarists (and non-monetarists) to assume that there were “long and variable lags” between changes in monetary policy and aggregate demand. In fact, the lags are extremely short. The problem was that monetary shocks were misidentified, assumed to begin when the money supply changed; whereas they actually begin from the moment the money supply increase was viewed as permanent. Thus some people argue that the bloated level of bank reserves is a sort of time bomb, waiting to explode into higher prices with a long lag. Instead, markets don’t expect the Fed to ever allow the bomb to explode, they expect the Fed to eventually pull the reserves out of circulation, or else pay higher rates of interest to encourage banks to hold on to the reserves. And if it does explode into high inflation, the “cause” won’t be the current rise in the base, but rather the later decision not to do something about it—the something the market now expects the Fed to do. Even the great Milton Friedman missed this insight.
How can I so arrogantly assert that the economics profession is wrong about long and variable lags? Because there are other types of monetary policy that have an immediate effect on the future expected money supply. And these alternative policies also immediately affect asset prices and AD. Indeed, according to the EMH, the standard view would only make sense if there were long lags between asset prices (which are surely impacted right away) and the broader level of AD. But if that were true then business cycles should be forecastable; which they are not. Hence the “variable lag” cop-out.
Fortunately, there are other ways of doing monetary policy, techniques that can immediately influence the future expected money supply. The one I most often talk about is the 1933 decision to raise the price of gold. This immediately raised the future expected price level (via PPP) the future expected money supply (price-specie-flow), future expected NGDP, and thus the current level of AD. I like to think of higher levels of future expected NGDP as being what Keynes meant in his frequent references to “confidence” in the General Theory. Keynes had great intuition, but wasn’t able to effectively work that intuition into abstract models of nominal shocks.
Of course we don’t need to go back to the gold standard, any currency devaluation will work just as well—assuming the devaluation is an exogenous monetary policy shock, and not the fall in the real exchange rate due to an economic crisis. (The 2002 Argentine devaluation saw a bit of both. NGDP started rising immediately, but RGDP was temporarily depressed by economic turmoil.) But why are currency devaluations so powerful? The reason is that they tend to be highly credible. When a country is operating under a fixed exchange rate regime, it loses enormous credibility and reputation by devaluing. Therefore they often hold off until the bitter end. Once they take that painful step, there would be no reason for the policy to be reversed. Thus when a large devaluation occurs, it is generally the case that the future expected exchange rate (i.e. the forward rate) falls by about the same amount. This immediately raises the future expected price level, and hence current AD. The lags are very short.
In some cases the effect is even more certain. When a country’s nominal interest rate is near zero, the rate cannot fall further. The interest parity theorem shows that if you do a major devaluation, the exchange rate cannot be expected to appreciate back to the old level unless nominal interest rates fall significantly. But that can’t happen at zero nominal rates, and is why people like Lars Svensson argued that depreciation of the Japanese yen was a “foolproof” way out of their liquidity trap.
The US can’t really use the exchange rate as a policy tool, it is too controversial. But there is a price it can use, CPI or NGDP futures. If the Fed sets a higher explicit nominal target, the effect is almost identical to currency devaluation—AD is affected immediately. When I debated Jim Hamilton last year, I recall he was skeptical of my assertion that NGDP targeting would have prevented the economy from falling off the cliff in late 2008. I think he was relying on the long and variable lags concept. That would have been a good argument if my proposed policy tool was money supply expansion or a lower fed funds target. Those changes might have had little effect on future expected policy, and hence future expected NGDP. But an explicit NGDP target would be different. As with currency devaluation, it would immediately change the future expected path of NGDP, and hence prop up current NGDP as well. The Fed may try something new in the next few months. It will either work right away (observable in the market reactions) or not at all.
We have two languages for discussing monetary policy. The fed funds rate/money supply language is full of mysterious long and variable lags, and makes intelligent conversation almost impossible. Literally any outcome can be assumed, depending on your assumptions about the future stance of policy. (On the other hand, the alleged perfect substitutability between cash and T-bills is nearly irrelevant.) If we shift monetary policy talk to prices, we can immediately pin things down. It could be the price of gold, a basket of commodities, a foreign exchange rate, the overall price level, or NGDP. What’s important is that this sort of discussion almost always has implications for future policy as well, and that’s what we need to know to pin down the current stance of policy. Intertemporal arbitrage connects current and future prices. Unfortunately, as we saw in the New York Times story I just posted on, almost all M-policy talk continues to refer to near-term levels of short term rates and the monetary base. Which means we are still talking gibberish.