Money And Asset Prices (The Liquidity Effect As Epiphenomenon)

by: Scott Sumner

We've already seen how money affects prices and output in previous posts. I'd guess that 99.9% of economists would do money and asset prices first, and use those effects to explain the relationship between money and inflation and/or output. But then if I was 99.9% of economists, you wouldn't be reading this blog.

Because wages and many consumer goods prices are quite sticky, the first place monetary shocks show up is in the flexible price asset markets; particularly stocks, bonds, commodities, real estate and foreign exchange. Let's start with stocks, commodities and real estate.

Monetary stimulus has no long run impact on real stock, commodity and real estate prices, but does affect all three in the short run. When the economy is at full employment, a one time increase in M should cause a roughly proportional increase in expected future nominal asset prices. The impact on current prices depends on the effect of the monetary stimulus on nominal interest rates. Because one-time money supply increases often reduce nominal interest rates, current asset prices might rise by even more than expected future asset prices (as inflated future cash flows get discounted at a lower interest rate). And because consumer prices are sticky in the short run, real and nominal asset prices move in tandem in the very short run.

If the economy is depressed, then a one-time money supply increase raises asset prices for the reasons just cited, and also because it raises real output, as we saw in the previous post. The real prices of stocks, commodities and real estate tend to be somewhat procyclical. However increases in the trend rate of inflation can reduce real stock prices, as it raises the real effective tax rate on capital income from corporate investments. Nonetheless, if the economy is deeply depressed then monetary stimulus is likely to be bullish for stocks, commodities and real estate.

Monetary stimulus will raise the price of foreign exchange for two reasons. One is obvious; money is neutral in the long run and hence any increase in M shows up as higher prices in the long run, and this includes the price of foreign exchange. In addition, a one time increase in M will often reduce the nominal interest rate. The interest parity condition says that the expected change in the price of a foreign currency is the domestic interest rate minus the foreign interest rate. Thus if the US interest rate is 7% and the Swiss interest rate is 4%, then the SF would be expected to appreciate by 3% per year against the dollar. If this was not the case, then arbitrageurs would take advantage of any discrepancy in interest parity.

If monetary stimulus causes the domestic nominal interest rate to fall, then the expected rate of appreciation of the domestic currency must increase. Paradoxically, an expansionary monetary shock will cause the expected future value of the domestic currency to drop, but the expected rate of appreciation of the domestic currency will actually increase. This can only occur if the current value of the domestic currency plunges sharply on the news of monetary stimulus, so much so that it's expected to end up lower than before the stimulus, despite being expected to appreciate over time. A good example of this phenomenon occurred in March 2009, when the dollar plunged 6 cents against the euro on the day QE1 was announced. Rudy Dornbusch called this "overshooting."

The most complex and misunderstood asset market reaction occurs with bonds. Most people are so convinced that monetary stimulus causes bond prices to rise that they equate bond price changes with monetary policy itself. Thus people will talk about "the Fed cutting interest rates" as if they are describing a monetary policy, whereas they are actually describing an asset price change that may or may not be attributable to monetary policy. Even worse, bonds are the most unpredictable of all the asset markets.

We can be fairly confident that monetary stimulus will boost the prices of stocks, commodities, real estate and foreign exchange, but it's not at all clear how it will affect bond prices. Most asset prices rise in response to monetary stimulus due to higher inflation and/or higher real growth expectations. But both of those factors would tend to reduce bond prices, or raise interest rates, which is just another way of saying the same thing. Why then do people often equate monetary stimulus with rising bond prices and falling nominal interest rates?

It seems that bonds are viewed as being special because they are particularly close substitutes to cash. When people (and banks) are faced with excess cash balances, they initially try to get rid of those unwanted dollar bills. In the long run this is done through higher prices. But in the short run prices are sticky, so people attempt to buy close substitutes, assets such as T-bills and bank CDs. The price of those close substitutes rises, which means that short term interest rates tend to fall. This is called the "liquidity effect." (It doesn't occur because the Fed buys T-bills, the same effect would occur if they bought zinc. It's the excess cash that creates the liquidity effect.) The effect on longer term bond prices is much more uncertain, as the inflation/output effect plays a much bigger role in the long term bond market.

Short term nominal interest rates have very little effect on the economy. The liquidity effect is a sort of epiphenomenon, something that happens, but which doesn't have an important impact on the variables we really care about. The other asset markets are far more important.

Now let's consider the impact of an unexpected, one-time, 5% boost in the money supply:

1. Short term nominal interest rates usually fall. This depresses velocity, and prevents any significant immediate rise in NGDP.

2. The real price of stocks, commodities and real estate rises, which boosts business investment in the construction of corporate assets, farm equipment, mining equipment, home building, etc. Over time these factors tend to gradually raise interest rates, and also velocity. As velocity rises, so does NGDP.

3. As foreign exchange prices rise, exports increase, which may also tend to gradually boost real interest rates, velocity, and hence current NGDP.

4. NGDP also rises because the monetary injection increases expected future NGDP. If the higher future NGDP is expected to show up in the form of higher prices, then that increases current inflation expectations, velocity, and hence current NGDP. If the higher future NGDP is expected to show up in the form of higher RGDP, then it will boost current investment and raise the real interest rate over time, which also boosts velocity and current NGDP. Either way, higher future expected NGDP tends to boost current NGDP. Even if real asset prices did not respond to monetary shocks, this channel would be enough for monetary policy to be effective.

So the asset market channel is another way in which monetary stimulus can boost NGDP, above and beyond the simple hot potato effect that causes a proportional rise in NGDP in the long run. To summarize, monetary stimulus raises NGDP for several reasons:

1. It raises expected future NGDP via the hot potato effect, which boosts current NGDP.

2. It raises various real asset prices, which may gradually boost real expected returns on capital, and hence velocity. On the other hand any liquidity effect that shows up will reduce velocity, but only in the short run. The liquidity effect goes away once wages and prices have fully adjusted.

The best way to study monetary economics is to first focus on the hot potato effect in order to understand why monetary policy impacts nominal aggregates in the long run. Then study the musical chairs effect to understand why shocks to nominal aggregates impact employment and output in the short run. The reaction of asset prices to monetary policy (while important) is a distraction that does more to confuse than illuminate for a student first trying to understand monetary economics. It's the icing on the cake.

To summarize the entire model:

1. M -> NGDP (hot potato)

2. NGDP -> RGDP (musical chairs)

3. M and NGDP and RGDP all affect real asset prices (liquidity effect, inflation effect, real income effect)

4. Asset prices also affect NGDP

PS. There is an interview of me in the Brazilian financial paper Valor. (HT: Marcus Nunes.)