He writes:
What does a negative real rate signify? If you consider a simple one-good economy in which the output is costlessly storable, a negative real rate could never happen-- people would simply hoard the good rather than buy such miserable assets. You're better off storing a can of tuna for a year than messing with T-bills at the moment. But there's only so much tuna you can use, and many expenditures you might want to save for can't really be stored in your closet for the next year. It's perfectly plausible from the point of view of more realistic economic models that we could see negative real interest rates, at least for a while.
Even so, within those models, there's an incentive to buy and hold those goods that are storable. And in terms of the historical experience, episodes of negative real interest rates have usually been associated with rapidly rising commodity prices.
There are two issues here, theory and evidence. First let's reconsider the theory.
Storage is tied not only to interest rates, but also (and more importantly, I believe) to anticipated future changes in prices, as well as physical storage costs. If people start to hoard commodities then current prices will go up. This happens because storing commodities means not consuming them, and as we consume less, the marginal value of consumption goes up. That can of tuna on the shelf starts looking real tasty the hungrier one gets! At the same time, storing commodities means consuming more in the future, which means the future price--the future marginal value of consumption--goes down. All the more reason to consume that can of tuna today if you're hungry and the large quantity you expect to eat tomorrow will make you sick.
Now, at the margin, interest rates certainly do affect the decision to store. The question is how much. The answer, I think, is very little. With commodities, that demand curve is very steep. The marginal value decreases sharply with consumption, so it is a big part of the calculation. If interest rates are the only thing driving growth in inventories, this will cause the expected price change between the present and the future to go down. Also, as hoarding increases, the marginal costs of storage likely increase. So, for small interest rate changes, the relative tradeoff is small, even if interest rates go negative.
My point here is that it's hard for me to see large-scale hoarding even in an environment with negative real interest rates. Hoarding can only happen if markets irrationally believe that price increases will be sustained indefinitely--i.e., a speculative bubble. Hopefully markets will be wary of such things after recent experience in housing and stock market prices.
Besides, when has a speculative bubble occurred with commodities? I know price spikes have occurred when some have tried to corner certain markets, but I think that's different. We've also had price spikes when inventories were depleted, there was real worry of impending supply shocks, or trade policies interfered with international commodity trading. But these are all very different stories as well.
Jim doesn't hint at a bubble. Instead, he's assuming the demand curve is flat. In reality, it's really very steep.
Second, let's consider the evidence. Professor Hamilton presents two graphs, one with a proxy for real interest rates, the second a plot of real wholesale price changes. I lifted those graphs from Econbrowser and pasted them below:
Jim doesn't hint at a bubble. Instead, he's assuming the demand curve is flat. In reality, it's really very steep.
Second, let's consider the evidence. Professor Hamilton presents two graphs, one with a proxy for real interest rates, the second a plot of real wholesale price changes. I lifted those graphs from Econbrowser and pasted them below:
Top panel: S6-month T-bill rate (secondary market from FRED) minus CPI inflation rate (annual rate) over subsequent six months, 1959:M1-2010:M3. Bottom panel: annual percentage change in producer price index for all commodities for 12 months ending at indicated date.
The top panel is a proxy for real interest rates: the 6-month t-bill rate minus realized inflation over the subsequent six months. The bottom graph is a percentage change in the producer price index for a 12 month period.
There does appear to be a correlation, one that is stronger during some periods as compared to others (particularly the 70s and 80s, something I'll come back to). Jim Hamilton's implicit assumption is that causation goes from the top panel to the bottom panel. But is it possible that causation goes the other way, from producer price changes to consumer price changes, and thus reduction in the real rate of return?
Well, yes.
One clue is the tremendous volatility of real interest rates. True real interest rates are not that variable. Consider, for example, rates of return on inflation indexed treasury bills. Those rates don't exist for as long a time series as the top panel above, but for the time period that does exist, they are nowhere near as variable as Jim's series. The problem is that short term inflation measures are noisy, often driven by fluctuations in food and energy prices, and these fluctuations are not anticipated in advance. Only expected changes in inflation should get priced in the real rate.
Second, much of volatility of in the producer price index--whatever the fundamental source--gets passed along to the consumer price index. So, say there's an embargo of oil from Iran and oil prices spike. We'll eventually see that in retail prices and the CPI. The price spike in turn causes the real interest rate to decline. So, there's good reason to think causation goes from the bottom graph to the top graph, rather than vice versa. That is, almost surely, what's going on in the 70s and 80s, which is where the link is strongest. I'd venture to guess that's the main source of correlation throughout.
So, in a nutshell, I'm not worried about the effect of QE2, or generally low or negative interest rates, on commodity prices. If commodity prices go up, I think that will mainly be a symptom of demand growth, which would be a good thing--a sign of recovery--for aggregate economies. Alternatively, prices could go up from real supply shocks and/or trade restrictions, which would obviously be bad things. But whatever may happen to prices, I think QE2 will be a very small deal.
"Besides, when has a speculative bubble occurred with commodities?"
ReplyDelete2008.
Anonymous: I and many others have pointed out that 2008 was not a bubble. A commodity price bubble can only happen with an accumulation of inventories, and that didn't happen. We had a price spike for the opposite reason: inventories were drawn down amid rapid demand growth.
ReplyDeleteMichael,
ReplyDeleteIf actors bid up prices in a futile effort to accumulate inventory, prices will rise without inventories necessarily rising.
In the inflationary period in Brazil, it was typical to see such "forward buying" behavior. Prices at supermarkets rose as consumers arrived after payday to hoard goods that they expected would be in short supply later. The problem was they couldn't stock up on much because everyone else was also trying to hoard.
Also, "inventory" of many commodities includes reserves that are extractable at little marginal cost. Producers can decide to hold back this production in expectation of rising prices.
Rising inventory is but one sign of hedging behavior in expectation of rising prices.
Perhaps another way to present the above is with supply and demand curves. In an inflationary economy, prices rise without quantities changing. This implies supply and demand curves shifting in opposite directions. That shift is merely producers facing higher input costs or extracting rent from holding back production; and consumers wanting to buy more before prices rise. Where is the increase in inventories?
ReplyDeleteAnonymous (first after my response above): It sounds to me that what you're describing is an accumulation of inventories that may not have been captured by official statistics. Yes, some of that may have occurred in 2008. I think it was a small deal. Obviously we can't know for sure as there is no data on private hoarding. But, I don't think people were hoarding oil. People may have hoarded rice. It's so much easier to point to other factors: low inventories in places where they were observable, rapid demand growth, and, in the case of food, export bans. Prices fell fast when demand fell with the world economy.
ReplyDeleteAll of this is textbook commodity price behavior when inventories are low. There's no bubble about it.
Anonymous #2: Inflation doesn't shift supply or demand curves. To a first approximation, it changes nothing fundamental. What Jim Hamilton was describing was a change in inventories driven by negative real interest rates--storage as a form of investment. I'm arguing that this is going to be small potatoes.
I also happen to think odds of inflation, even the Fed's target inflation of 2% or a little less, are extremely slim for the next several years. Deflation still seems the much larger risk, especially given how small QE2 turned out to be. The Fed would need to print 2-4 trillion just to accomplish the Taylor rule equivalent of lowering interest rates (which they cannot do).