Stabilising deflation under the gold standard

We can learn from history, but we need to be careful about the lessons we learn. If the historical monetary policy regime is different from today's monetary policy regime, stabilising deflation in history might be destabilising deflation today.

Put the price level P on the vertical axis, and real output Y on the horizontal axis. Draw an Aggregate Demand curve. Does the AD curve slope down? Will a sufficient fall in the level of prices (and nominal wages) ensure there is sufficient demand for goods (and labour)?

That depends. It depends on what is being held constant when you draw that AD curve. It depends on the monetary policy regime.

Suppose the central bank is targeting the price level. Suppose it wants to keep the price level constant at 100. It does whatever it needs to do to make the AD curve horizontal at 100, and it responds to shocks to stop the AD curve moving up and down too much.

Now suppose that same central bank falls asleep at the switch for (say) 70 years. But everyone knows the central bank will wake up 70 years from now, and put the price level back to 100. What does the AD curve look like now?

That depends on precisely what switch the central bank falls asleep at. Let's suppose it falls asleep at the nominal interest rate switch. It sets the nominal interest rate at 5%, which was the natural rate of interest when it fell asleep, and then falls asleep and leaves it at 5% for 70 years.

The AD curve is no longer horizontal at 100, but it does slope down. Suppose the natural rate of interest falls to 4%, immediately after the central bank falls asleep. There is deficient demand for goods, and so the price level falls. If the price level falls to 50, people will expect 1% inflation for the next 70 years, so the real interest rate is now 4%, and equal to the natural rate once again. The price level target, even 70 years in the future, acts as an automatic stabiliser. The fall in the price level today creates expectations of a rising price level from today onwards. The fall in the price level today means a fall in the relative price of goods today compared to goods 70 years from today, so people buy more goods today. (If the price level is sticky, and takes time to fall, there will be a recession until it finishes falling and starts rising at 1%. Falling prices means that the cure of low prices is not yet complete, and the incomplete cure even makes things worse.)

But if people instead expect the central bank will wake up in 70 years and target whatever the price level is when it wakes up, this won't work at all. Today's inflation targeting is like that, except today's central banks don't usually fall asleep for that long.

I think that the gold standard was a bit like that fixed price level target. Not exactly like that, but a bit like that. It creates a nominal anchor for the future price level, based on the marginal cost of production of gold and the industrial demand for gold. It's certainly not a fixed nominal anchor, because those things do and will move around. And it's not a very good nominal anchor either. But it is a nominal anchor nevertheless.

When we draw an AD curve, we are asking what would happen, counterfactually, to output demanded, if the current price level were at various different levels, whether or not those are equilibrium levels according to the model. The AD curve is a statement about counterfactual conditionals. When we draw a downward-sloping AD curve, we are saying that if the current price level were counterfactually lower than it is, current output demanded would be counterfactually higher than it is. And the nominal anchor of the gold standard, bad as it is, ensures that the expected future price level would not follow, one-for-one or more, those counterfactual disequilibrium changes in the current price level.

Under the gold standard, if sellers were to offer goods at lower prices than they do, that would mean those prices would be lower relative to what prices will be in the future. Deflation would mean today's goods would be a bargain, so we would buy them now. It would not mean that goods will be equally cheap or cheaper still in future.

[I was reading ProGrowthLiberal's and Barkley Rosser's good posts (HT Mark Thoma) on the 1921 recession kerfuffle, and thought I would add my twopenceworth. But I'm not very good at history.]

Comments

[Roger Sparks]

It seems to me that we should think of the AD curve as being driven by two forces. Draw a curve beginning at zero, zero, then moving up in price and up in production to a peak, then falling in price and becoming closely tangent to the production line.

Demand reigns on the zero to peak side. Supply reigns on peak to ever-greater-production side.

If we assume that production has been greatly stimulated so that supply is far onto the supply side of the curve, can we reach a supply level that becomes less responsive to monetary policy?

[Nick Rowe]

Roger: Whoah!

Start with micro, the market for apples. There are 3 different quantities: the quantity that buyers *want* to buy (Qd); the quantity that sellers *want* to sell (Qs); and the quantity actually bought-and-sold (Q). The demand curve tells us what determines Qd; the supply curve tells us what determines Qs; and it is reasonable to assume that Q=min{Qd,Qs} (which we call "the short side rule").

I think you are talking about the short side rule, because as P rises from 0 it follows first the supply curve, then kinks where supply and demand curves cross, then follows the demand curve. And you are misusing the word "supply". It does not mean what it normally means in English. And macro is more complicated than that: first because aggregate Qd also depends on aggregate Q; and second because the "aggregate supply" curve is not really a supply curve, if you assume monopolistic competition, which most of us do.

[Roger Sparks]

Hmmm. Perhaps I am thinking of an 'aggregate demand curve'?

A very high price but only a few items sold results in a small market. I would say that the aggregate demand is small.

As the price falls, the number of items sold grows resulting in a larger market. I would say that the aggregate demand is larger and will approach a peak.

As the price falls still more, the number of items sold continues to increase but aggregate demand begins falling because the price-times-item number (P x I) is past the aggregate demand peak. Perhaps we would say that the item is now entering the commodity pricing range.

If we speak of monopolistic competition, and limit ourselves to that restriction, we would avoid allowing the item to become a commodity. I would think that monopoly conditions would only be good for those fortunate enough to be in the monopoly loop.

[Nick Rowe]

Roger: no. You seem to be describing some sort of total revenue curve. I have real output Y on the horizontal axis, not nominal output PY. But you are totally lost. And this is not the place for me to explain it to you. Time to crack open that textbook, so you can at least understand my language, and understand where I'm coming from, whether you agree or disagree with what I am saying.

[marcus nunes]

Nick, a bit of history!
[Link here NR]

[Lorenzo from Oz]

based on the marginal cost of production of gold and the industrial demand for gold

Not monetary demand for gold?

Accepting that discovery of new sources of gold lowers its marginal cost of production, going back onto/expanding a gold standard is deflationary. Both in theory and in practice. So, I am not sure what is so special about specifically industrial demand.

[Nick Rowe]

Lorenzo: good question. You are probably right. But I'm not sure it matters much if I add monetary demand. My brain won't give you a clear answer at the moment. It's been a long day.

[Nick Rowe]

Thanks marcus! (I just found your comment in the spam filter. Sorry about that.)

[Miguel Madeira]

"Lorenzo: good question. You are probably right. But I'm not sure it matters much if I add monetary demand."

In the case of a product with a largely inelastic supply (like gold), I imagine that probably the monetary demand will be the main factor; it will be diferent if the commodity-money was a commoditie with a elastic supply (like cocoa seeds, as in Aztec Empire) - in these case the cost of production will probably be the more important factor

[Adam P]

Hi Nick,

I think your point about the gold standard is correct but I'm nor sure I agree that under an inflation target things are so different.

If an inflation targeting central bank fell asleep for 70 years but the fiscal policy rule was unchanged then the long run price level should be determined by the Government budget constraint (valuation equation as Cochrane correctly calls it).

Thus, the AD curve would still slope down by exactly the same mechanism. The argument still works.

[Min]

"Under the gold standard, if sellers were to offer goods at lower prices than they do, that would mean those prices would be lower relative to what prices will be in the future. Deflation would mean today's goods would be a bargain, so we would buy them now."

If we could. Wasn't one feature of the gold standard periodic or chronic insufficiency of money? As for stability, there was less stability under a gold standard than between 1940 and 1990, wasn't there?

[Nick Rowe]

Miguel: yes. If the stock of gold is fixed, then the elasticity of the non-monetary demand for gold relative to the unit-elastic monetary demand for gold will determine their roles in the elasticity of the AD curve. If the non-monetary demand has an elasticity greater than one, then the AD curve is more elastic than in a model with a fixed stock of money. And adding in production of new gold makes the AD curve more elastic in the long run.

I think that also answers Lorenzo's question.

Adam P: Hi! Yes, we could imagine a world in which the fiscal authority fixes the Present Value of future primary deficits to fix the flow of seigniorage revenue to fix the price level. But I don't see fiscal authorities behaving like that. Instead, they tend to spend the seigniorage revenue the central bank gives them. Plus, even if they did behave like that, nothing prevents the central bank doubling the price level by doing a 2-for-one stock split, which has no real fiscal consequences.

Min: yes. The gold standard was a "stable" equilibrium in the sense that P returned to equilibrium P*, but "unstable" in the sense that the equilibrium P* kept moving around a lot.