A simple macroeconomic model of the current situation is the ISLM with two wedges. The first wedge between the IS and LM curves is the gap between nominal and real interest rates. The second wedge between the IS and LM curves is the gap between government and private interest rates. We add (or subtract) the two wedges to get a theory of aggregate demand. The second wedge is the thing we don't understand very well.
Those who don't need a refresher in basic macroeconomics can skip to the *.
The IS curve shows combinations of real output and interest rates such that the demand for output, at the interest rate and that level of output, equals that same level of output. (In full equilibrium, output demanded = actual output = output supplied, but the IS curve says nothing about supply, so only captures the first half of that condition, just in case you didn't already know that.)
The LM curve shows combinations of real output and interest rates such that the (stock) demand for money, at that interest rate and that level of output, equals the (stock) supply of money. The LM curve is currently very flat, because both the demand and supply of money are currently very interest elastic.
The demand and supply of money (LM) depend on the nominal interest rate, and the demand for output (IS) depends on the real interest rate. The real rate equals the nominal rate minus expected inflation. So we need to place a vertical wedge, equal to expected inflation, between the IS and LM curves, to represent the spread between real and nominal interest rates.
With positive expected inflation, the wedge gets shoved in from the right, and output demanded is higher than the level of output at which IS and LM intersect. With expected deflation, the wedge gets shoved in from the left, and output demanded is lower than the level of output at which IS and LM intersect. This is why expected deflation is worrying; it makes demand for output lower, which makes unemployment higher, which makes deflation worse, which makes expected deflation worse.
* Welcome back.
That was an OK model of aggregate demand until the current financial crisis. The biggest problem with that model, currently, is that it ignores the second wedge: the spread(s) between the interest rate(s) on government bonds and the interest rate(s) at which firms and households can borrow and lend. It was always a problem, of course, because those spreads always existed. But until recently those spreads were seen as small enough, and constant enough, to be ignored.
This problem is not just a problem of simple second-year textbook macroeconomics. TOTEM, the very sophisticated Bank of Canada model, superior in every other way to the ISLM model, has exactly the same problem. TOTEM only has one (Canadian, nominal, quarterly) rate of interest. Right now, given a choice between an ISLM model with two wedges, and TOTEM, I would choose the former. But we need a theory of the second wedge.
By "theory of the second wedge" I mean two things: the first part is defining what that wedge is, so we can measure it; the second part is explaining what determines the size of the second wedge, so we can understand why it varies over time, and how (unorthodox) monetary and fiscal policies might change its size.
Let's start with the first part. In Canada, and for most countries currently, we are fairly safe in assuming that a short-term government (or central bank) interest rate reasonably represents the relevant interest rate for the LM curve. The Bank of Canada can, if it wishes, make the supply of money perfectly elastic at that rate of interest, and currently it so wishes. (If the central bank held the stock of money fixed, so the relevant interest rate were a private rate representing the opportunity cost of holding money, my answer might be different).
The interest rate relevant for the IS curve should be a weighted average of private interest rates, both borrowing and lending (opportunity cost) rates. The weights should depend on both the shares and the interest-elasticities of various components of private consumption and investment expenditures. And so the second wedge should be the difference between the government rate and that composite index of private rates.
For Canada, Ian Pollick and Eric Lascelles at TD Economics have already made an excellent start at measuring the second wedge. More could be done, of course (like weighting by elasticities as well as shares), but it would be hard, (and very hard for me), so I will say no more.
With a positive spread between the private and government interest rates, the second wedge gets shoved between the IS and LM curves from the left, just like expected deflation for the first wedge. We add the two wedges together, vertically, to get the total wedge. Both wedges affect aggregate demand, in much the same way.
The second part is trying to understand the size of the second wedge, explaining why it changes over time, and how government policy might change it.
Orthodox monetary policy is the attempt (usually successful but less so now) to shift the LM curve. I see unorthodox monetary policy as the attempt to change the size of the two wedges. If it can influence expected inflation, and change the nominal-real spread, the central bank can change the size of the first wedge. If it can influence the private-government spread, by buying private bonds, the central bank can change the size of the second wedge. If it can change the size of either wedge, it can affect aggregate demand even if it cannot shift the LM curve.
Here are brief sketches of three theories of the second wedge, and how policy might change the size of the second wedge:
1. Financial intermediaries borrow, lend, and earn an income off the spread between borrowing and lending rates (a third spread). That third spread may widen if banks lose some of their capital, and are capital-constrained. Assume the borrowing rate is roughly equal to the government rate, and all private savings is intermediated by banks. If the savings and investment elasticities are roughly equal, the composite private interest rate will be roughly halfway between banks' borrowing and lending rates. So the second wedge will be roughly half the size of the banks' spread. If governments or central banks buy private bonds and sell government bonds, they act as financial intermediaries, reduce the constraint on banks' capital, reduce the banks' spread, and so reduce the size of the second wedge.
2. Liquidity matters for private savers, and matters much less for the government. Private bonds are less liquid than government bonds, and that creates the second wedge. In financial crises, liquidity matters more, the liquidity premium rises, and the second wedge increases. Unless the demand for liquidity is perfectly elastic, if the government or central bank sells government bonds, and buys private bonds, the changing mix of asset supplies reduces the price of liquidity, and reduces the second wedge.
3.Some private bonds are lemons, and this creates the second wedge. In a financial crisis, the proportion of lemons traded on the market increases, so the second wedge increases. If a government or central bank buys assets, it will buy mostly lemons, and will reduce the proportion of lemons traded on the market, and reduce the second wedge.
I read the Pollick and Lascelles piece as well and thought it was excellent. It would be interesting to see how shocks to the TD Effective Measure impact Canadian macro variables compared with traditional monetary policy reaction functions.
Posted by: brendon | February 06, 2009 at 11:53 AM
Brendon: there are a bunch of fun things that macroeconometricians could try by playing around with the Pollick/Lascelles measure. I would like to see a regression of deviations of CPI from target on 2-year lagged P/L spread. If it's negative, that means the BoC should have paid more attention to the P/L information (had it existed). (That's in line with the thinking behind my previous posts on why it's so hard to estimate multipliers, and how to improve policy.)
Posted by: Nick Rowe | February 06, 2009 at 12:11 PM
What about Bernanke and Blinder (AER 1989)?
Posted by: PCLE | February 07, 2009 at 12:03 AM
I'd like to see this TD effective measure calculated for the US market. Lots of stories as to how credit spreads have reached unprecented levels in the US but their measure for Canada shows TD now not as even as high as it was say around 2001.
Posted by: pgl | February 07, 2009 at 06:37 AM
PCLE: "What about Bernanke and Blinder (AER 1989)?" 1988? Dunno. It's a story of intermediaries, therefore under 1? But it's more a story of the transmission mechanism of orthodox monetary policy. What do you (or anyone else) think?
Posted by: Nick Rowe | February 07, 2009 at 11:37 AM
The ISLM model is an equilibrium model. It suffers by presupposing an equilibrium that does not exist in the meaningful time frame.
Posted by: Alan | February 07, 2009 at 03:13 PM
Most economists would claim that the (a) problem with ISLM is that it is NOT an equilibrium model.
Look, given a few minutes, one could think up 101 problems with ISLM: no supply side; assumes 1 good; no stocks other than M; no LR govt. budget constraint; closed economy; ignores demographics, expectations, etc., etc.
But so what? All models are false; there are only models.
What about that second wedge?
Posted by: Nick Rowe | February 08, 2009 at 08:12 AM
There was a tangentially related post on nakedcapitalism:
http://www.nakedcapitalism.com/2009/02/steve-keen-roving-cavaliers-of-credit.html
I don't have the background to evaluate the model (for all I know Keen could be a complete nut), but on the service it looks interesting...
Posted by: Patrick | February 09, 2009 at 10:47 AM