Berkeley Faculty Play with Their Toys...
Doug Elmendorf Thinks He Has It Right...

Liveblogging the Berkeley Finance Seminar: John Cochrane (2010): "Understanding Policy in the Great Recession: Some Unpleasant Fiscal Arithmetic"

Liveblogging the Berkeley Finance Seminar: John Cochrane (2010), "Understanding Policy in the Great Recession: Some Unpleasant Fiscal Arithmetic":

Very much a Fischer Black-style macro talk...

Important things that I had not known before that I learned from the paper:

  1. That if you add in a Phillips-curve friction--so that production and employment fall in deflation--and a credit-channel friction--so that the interest-rate spread and thus the interest rates on Treasurys depend on the capitalization of the banking sector--then the fiscal theory of the price level strongly militates for a helicopter drop of money to banks as the best policy to fight a financial crisis and the resulting recession: a TARP, a TALF, a PPIP, or its left-wing version of bank nationalization is exactly what the government can do to most effectively stem and cushion the deflation.

  2. How large is the fall in the primary surpluses Peter Orszag and his successors will have to raise in order to amortize the debt without accelerating inflation. The marginal investor has long believed that the U.S. debt is very safe--that no matter what happens, Peter Orszag and his successors will be able to raise the primary surpluses needed to amortize the debt without a big rise in the price level. As a result of the financial crisis, the required future primary surpluses have fallen roughly in half, so that we now have much more fiscal headroom than we had three years ago (if, that is, the long end of the Treasury yield curve can be taken as rational forecasts of the interest rates at which the debt will be rolled over and amortized).


Armed with the fiscal-theory-of-the-price-level identity:

(1) P = B((1+r)D/S)

where:

P: consumer price level
B: nominal government debt
D: duration of the primary surpluses to be used to amortize the debt
r: real interest rate on government debt
S: expected real primary surpluses in the future required to amortize the debt

John Cochrane sets out to analyze the depression of 2007-2010...

He says:

Finance rules!: "This essay requires ways of thinking about the world that may not be familiar to macroeconomists, so I have to spend half an hour in macro seminars going over basics like P = E/(r-g)..."

We have learned that most changes in valuations are the result of changes in discount factors...

Updating Sargent and Wallace...

Only one friction allowed in the model today: a Lucas-supply-function Phillips curve...

Fiscal-theory-of-the-price-level identity: nominal government debt is equity, with future primary surpluses being its dividend...

(1) P = B((1+r)D/S)

Everything jointly determined, but this identity must hold always.

Conclusion: expansionary fiscal policy and lots of other things expansionary--it is a way of producing more B, and more B raises P, and there is some Phillips Curve in the background that raises Y--unless it is offset by:

  1. Expectations of higher real primary surpluses in the future.
  2. Declines in the real interest rate on government debt.

Since there is always some chance that some inflation will be applied to the debt, and since additional debt should not lower but raise the real interest rate on government debt, full crowding-out appears to be a non-starter.

Cochrane: Ricardian equivalence? Requires that issuance of debt B (a) not affect r, and (b) be reflected one-for-one proportionately in changes in expected S...

Apply this identity to fall 2008: Back-of-the-envelope: Big changes in discount factors. The quantity of nominal debt which we read off the monthly Treasury Report is predetermined... because debt is of short duration... The consumer price level drops relative to trend by maybe 2%... The real discount factors we tread off of these daily Treasury real yield curve emails... The right side is of long duration so these discount factors D(1+r) go down by roughly 50%.... (.98) = (.5/S) => That means that expected primary surpluses go down by 48% in the fall of 2008...

Cochrane: In this model there is only expansionary monetary policy--ineffective except to the extend that it can change r, and fully ineffective once r hits its zero-bound floor--and fiscal policy. But lots of things qualify as expansionary fiscal policy in this model: infrastructure, aid to states, tax cuts, tax rebates, helicopter drops of cash, TARPs, AIG nationalizations, bank bailouts, etc. There is a presumption that fiscal policies that shrink credit spreads and so raise interest rates on government debt are more powerful, but which is most powerful depends on second-order frictions not modeled here.

The lesson is that the sharp fall in r demonstrates that the market wants more safe assets, so in order to cushion and soften the deflation the government should create more safe financial assets for the market to hold. Thus pretty much everything seems to work to stem the recession to some degree--except for quantitative easing via open market operations: that does not work because it simply trades one set of safe assets for anothers.

Cochrane: investigate the possibility that investors believe that we are already expected to run the largest-possible politically sustainable primary surpluses, and thus that any more bad news will turn our financial crisis into a sovereign debt crisis.

But this possible future is not priced--not priced at all, not priced to any extent--in the long bond. That it is not priced at all in the long bond puzzles me: it should be priced to some extent. And maybe it is rational to think that there is a substantial chance of a U.S. sovereign debt crisis even though the fact that the long bond is not expecting a U.S. sovereign debt crisis. But it isn't...

Cochrane: Reasons to fear that the term structure is not giving us a read on rational expectations of the future:

  1. Perhaps China--or some other non-market actor--is the marginal holder of U.S. Treasury debt.
  2. Perhaps some important friction here: lots of implicit government guarantees floating around, and borrowing cash short for nearly free, buying long-term Treasurys, collecting coupons, and hoping to get out before the crash comes an attractive strategy. See Rogoff, "Bubbles Lurk in Government Debt"), "From Financial Crisis to Debt Crisis"...

By my calculations from equation (1), the fall of 2008 saw the value of future primary surpluses S fall in half: at the start of 2007 we were expected to run primary surpluses in the future twice as large as we are now expected to. So the hypothesis that our primary surpluses are right now at some politically-constrained maximum seems odd... (But I would change my mind quickly if the long bond and the TIPS spread began telling me it was time to do so...)


Other stray unvoiced thoughts:

I understand the P in equation (1) is the price level that we read off the Bureau of Labor Statistics CPI reports. But suppose we add Philips-curve frictions. Is it still the price level in the denominator, or is it something like nominal demand. And if P is not the price level but rather nominal demand, then the lambdas on the right hand side are no longer discount factors we can calculate off of the Treasury's daily real yield curve e-mailsbut are instead something else. What else they become to keep the equation an identity is not clear to me...

We know what it looks like when the market expectations are that expected future primary surpluses are as high as they can go, and cannot go any higher. It looks like Argentina in 2000. In times like that, the discount factors on government debt are low because government debt is perceived--rightly--as being very risky, and the current consumer price level moves essentially one-for-one with additional debt issue. That is not where we are right now. We are not there at all. The discount factors on U.S. Treasury debt are not low because such debt is perceived as risky given the inability to finance more of it by raising future primary surpluses. Instead, the discount factors on Treasury debt are very high as the market perceives U.S. Treasury debt to be extraordinarily safe. And the price level is definitely not moving proportionately with additional debt issue...

Is inflation in this model supposed to be a good thing or a bad thing right now? If P is something like nominal spending, then it strikes me that more "inflation" right now would be a very good thing: I don't know anyone who wouldn't rather have an economy in which nominal spending were 10% higher and so back on its long-term growth trend...

If Mark Gertler were here, he would say that there was no big huge increase in risk aversion on the part of investors leading them to seek a much higher quality and low variance portfolio. He would say that the problem in late 2008 lay not on the demand-for-risk side but on the supply-of safe-assets side. A large amount of assets that people had figured on as pretty good even if not as good as Treasuries turned out not to be so, but to be risky indeed. And the collapse of the net worth of financial intermediaries meant that they no longer had sufficient skin in the game, that they were likely to be gambling for their resurrection, and thus that projects they attempted to induce you to invest in were likely to be much less safe than normal--and it is those things happening in the lending channel that spill over into the Treasury market and drive the huge increase in Treasury discount factors. From Gertler's perspective, focusing on the large rise in the discount factors on Treasury debt is not an alternative to but rather a consequence of all the lending-channel stuff that he likes to talk about...

Comments