« Negative natural rates of interest and NGDP targets | Main | Whatever happened to black cat toffee? »


Feed You can follow this conversation by subscribing to the comment feed for this post.

"Or, if you prefer, you can continue to think of a downward-sloping IS curve, but allow that a commitment to a more expansionary monetary policy would make people and firms more optimistic about future real income, and this would shift the IS curve right."

The phrase "if you prefer" suggests that this is just a matter of taste. But if you assume an upward-sloping IS curve you have to ask whether the equilibrium is stable or not (if the LM curve is steeper, no problem). I don't think the difference is just a matter of taste and I'm pretty sure Krugman doesn't either.

Just to add: I do think this post highlights an unresolved tension in Krugman's ideas. An Old Keynesian would agree whole-heartedly that "permanently higher inflation isn't a necessary condition for NGDP level path targeting to work." But could a New Keynesian say that in good conscience? I'm not sure. I don't much care either, since I have no shares in NK Plc. However I suspect Krugman wrestles with this now and then, struggling with the fact that OK models seem to work better in a deep slump, but NK models are more respectable. Like most of us, he wants to have his cake and eat it.

Kevin: OK. But if the central bank can move quickly enough to target NGDP, or E(NGDP), we've effectively got a vertical LM curve. It is shifting right at the speed 5%(or whatever) minus inflation.

I prefer (in this context) to define the IS curve as: "that curve which stays put when monetary policy changes in the way I'm considering right now". It's "the *other* curve".

Kevin: agreed. I found myself reaching back to OK ideas in this. A lot of old stuff gets forgotten when people do formal models. And some of that old stuff was alright, even if it wasn't fully formalised.

But the OK's also thought the SRAS curve (or SRPC) was reverse L-shaped. We need to keep that separate from the IS slope.

It does seem more accurate to have your IS curve shifting right rather than sloping up. The question is whether you need "permanently higher inflation" to accomplish this shift. You are saying that a temporary inflation "blip" will shift that curve. I question whether actors' behavior will respond to such a transient expected rise in the price level, especially when inflation expectations are already near-normal.

You seem to view the economy as a place where actors are waiting for evidence of temporarily higher Fed-induced nominal income before they invest in productive projects. I see it as a place where actors are waiting for evidence of permanently higher Fed-induced inflation before they hedge against it in unproductive ways. How do you know, ahead of time, which one of us is right?

David: I like to draw the IS curve with the real interest rate on the vertical axis, and real income on the horizontal. So an increase in expected inflation, and no change in expected future real income, will not shift the IS curve. (It just puts a wedge between the IS and LM curves, since the LM has nominal interest rate on the vertical axis.)

If you put the nominal interest rate on the vertical axis, an increased expected rate of inflation shifts it up vertically by the same amount.

I was really talking about the effect of (current and expected future) real income on saving and investment.

Thanks. I thought the increase in real income expectations resulted from expansionary monetary policy, which means from higher inflation expectations. If not, then I would argue that what you call expansionary monetary policy is really fiscal policy with real effects. For instance, Treasury establishing a Sovereign Wealth Fund to purchase risk assets in exchange for T-bills would just as easily shift the IS curve to the right. That action would have very little to do with price level expectations, which is the only thing monetary policy (might) directly control.

The ECB is currently resisting the forced assumption of a fiscal action: continued purchases of risky Italian bonds in exchange for short term ECB liabilitites. If the central bank loses money on these bonds (quite likely), the losses will be passed on to German voters, whose representatives never appropriated the money for those ECB purchases. In that case, the ECB's extra-democratic fiscal action would certainly come back to haunt it in the form of reduced independence.

Nick: "And if that rightward shift were big enough, the equilibrium real interest rate would actually be higher with a commitment to a more expansionary monetary policy."

In what way do you distinguish between the "equilibrium real rate" and the natural rate? Are you saying that in a recession the natural rate goes up with an expansionary monetary policy?

[OT: In case you missed it I replied to your comment in Matt Rognlie's monetary frictions post.]

"And since the nominal interest rate can't fall any lower, the only way to lower the real interest rate is to increase actual and expected inflation."

Interest rates can still be cut as long as people don't expect the policy rate to remain at zero forever. The limit of conventional monetary policy is zero forever, not simply zero.

The effect of raising the inflation target would be to cut interest rates across the yield curve, since people would expect the policy rate to remain at zero longer.

See http://krugman.blogs.nytimes.com/2011/10/10/sign-of-the-times/?gwh=77BCA97A58BC3CD92F15A1C98F244BBF (says AD curve should be upward sloping)

A simple old Keynesian IS-LM model which formalizes your upward sloping IS curve is rather easy to put together. Instead of insisting that investment is solely a function of r just make it a function of r and Y along the lines of I(r,y)=I0-br+(mpi)y, where b is the interest rate elasticity of investment and mpi is the marginal propensity to investment, or income elasticity of investment, which is the one new addition.

For a closed, government free economy with your standard consumption function, the IS curve is just y=C0+(mpc)y+I0+(mpi)y-br or y=(C0+I0-br)/(1-mpc-mpi). If mpi is greater than mps, the marginal propensity to save, then a rise in y will lead to a rise in r along the IS curve, because mpc+mps must always equal 1, so the negative b cancels with the negative denominator. If it's less then you have your standard downward sloping IS curve, which makes sense. If a rise in y leads people, at the margin, to want to save more than they would like to invest more money flows into the loanable funds market than is taken out through investment so r falls, if it's the reverse it rises. If mps equals mpi then r is independent of y and will always equal some value no matter what y is since at the margin, as much money flows into the loanable funds market as flows out.

It also provides a framework for thinking how a central bank can cut interest rates now which leads to an investment boom which leads to greater interest rates later if you assume that interest rates adjust much quicker than gdp, which I would say is reasonable.

The biggest issue in my mind is that this all terribly conflates the traditional realms of IS and LM, Not sure if it also double counts some effects in the IS curve which are already accounted for in your usual LM curve formulation.

The way that I see market monetarism in the context of the IS-LM model is that the IS curve is a function of expected future income E(Y). So it’s similar to the Woodford-Bernanke "expectations-based IS curve" in the model. In a similar manner, you could have changes in the expected future demand for money and expected future supply of supply money shift the LM curve.

A QE policy that shifts LM to the right that is not credible will lead to lower interest rates because the IS curve will not shift. A QE policy that is highly credible (perhaps because of a stated commitment to a particular level of NGDP) will shift the IS curve to the right substantially. The impact on the LM curve is ambiguous as both expected future demand for money and expected future monetary base will be lower. So the result will almost certainly be higher interest rates (rightward shift in IS and an ambiguous impact on LM).

So it seems to me that you don't need to argue that the IS curve slopes upward to show that the result of an expansionary monetary policy leading to higher interest rates.
What am I missing?

K: "In what way do you distinguish between the "equilibrium real rate" and the natural rate? Are you saying that in a recession the natural rate goes up with an expansionary monetary policy?"

Different people use those terms differently. In the context of this post, here's what I mean:

The "equilibrium" rate of interest is the rate of interest at which desired investment = desired saving at the *actual* level of income. It's the height of the IS curve, at the current actual level of income.

The "natural" rate of interest is the equilibrium rate when actual income equals the long-run equilibrium level of income. It's where the IS curve cuts the LRAS curve.

So I'm saying that the IS curve slopes up, which means that in a recession the equilibrium rate will be below the natural rate.

And an expansionary monetary policy causes Y to increase, so the equilibrium rate increases towards the natural rate. Which sounds very weird to Wicksellian ears.

Max: "Interest rates can still be cut as long as people don't expect the policy rate to remain at zero forever. The limit of conventional monetary policy is zero forever, not simply zero."

You are right. I ignored the whole term structure business. A commitment to keep "interest rates too low for too long" can lower expectations of future short rates, and so lower current long rates, even in nominal terms, quite apart from lowering real rates by increasing expected inflation.

David: yes, within a New Keynesian/neo-Wicksellian context, there may be reasons for an upward-sloping AD curve. But that's different from an upward-sloping IS curve. The AD curve is in {P,Y} space. The IS curve is in {r,Y} space.

Joseph: that's very much how I think of the upward-sloping IS curve.

One subtlety: I think it matters *why* Y increases. If Y increases because demand increases, so the economy comes out of a recession, then I can see a big effect on investment demand. Because firms will be able to sell the extra goods the extra capital can produce. But if they can already sell as much as they want, and Y increases further, for supply side reasons, I don't see that effect any more.

And, Y may adjust slowly (if e.g. it takes time to hire more workers), but expected future Y could adjust quickly, if the central bank sends a strong signal.

Gregor: "So it seems to me that you don't need to argue that the IS curve slopes upward to show that the result of an expansionary monetary policy leading to higher interest rates.
What am I missing?"

I don't think you are missing anything. It all depends how you define an IS curve.

In the simplest ISLM, with no expectations, the IS curve didn't shift when monetary policy changed. That was a nice feature of the old ISLM. Monetary policy shifted LM; and fiscal policy shifted IS. If you like that nice feature, and *define* the IS curve as "that curve which doesn't shift when monetary policy changes", it's gonna have to slope up.

I think there was some argument that the IS curve in a country like China sloped up. (I'm a bit out of my element here...) The concept that rising interest rates would increase consumption seems to have some empirical backing. Not sure if that's the same thing as you're describing.


"But for many years I have strongly disagreed with this claim that raising rates is a way of combating inflation. If it is the wealth effect, and not the consumption-postponement effect, that really drives changes in savings and consumption rates, then raising rates would only reduce consumption if there were a negative correlation between interest rates and wealth, as there clearly is in the US.

Is there a negative correlation between the two in China? Probably not. Most Chinese savings, at least until recently, have been in the form of bank deposits. In a financial system in which deposit rates are set by the central bank, the value of bank deposits is positively, not negatively, correlated with the deposit rate.

Chinese households, in other words, should feel richer when the deposit rate rises and poorer when it declines. In that case, rising rates should be associated with rising, not declining, consumption and with higher, not lower, inflationary pressure."

The comments to this entry are closed.

Search this site

  • Google

Blog powered by Typepad