Friday, June 28, 2013

Russ Roberts response, part 2

I'm beginning to wonder if Russ's "questions" were just statements that happened to have question marks at the end of them, but what the hell. Let's talk a little about his first question (his third question was answered yesterday in Part 1... yes I'm trying to confuse you guys). Unfortunately this will have to be more of a sketch of an answer but feel free to jump in in the comment section.

He asks:
What empirical evidence can you provide that suggests that these factors (debt to GDP ratio, supply side factors, crowding out) are not relevant right now?
"Not relevant" needs to be qualified. I may have used language like that in the past, but the real point is that it is not the primary concern.

If he's looking for empirical evidence on labor supply the relevant cites for extended UI are Card and Levine (2000) and Valletta and Kuang (2010) more recently (and a little more quick-and-dirty). Each of these have negative labor supply shocks associated with extensions of UI, they just don't appear to be nearly big enough to be driving this. They are also each the impact of extended UI on individual behavior, not the general equilibrium impact of extended UI (which may very well be positive).

I also don't understand how the significant-supply-shock story makes any sense. We would expect to see much more substantial wage and price growth if this were critical. We don't see any of that in the data. Surely we don't think there was a productivity decline that would have been sufficient to mute the supply shock's effect on wages and prices. And what would the supply shock even be? Usually when we point to supply shocks its nice to know what that entails. The demand shock, alternatively, is quite obvious and the price data seems to support the dominance of the demand shock.

One strategy is to point to Obamacare as the supply shock. I don't find this especially plausible given the sequencing. It can't be an explanation of the recession itself. You can cite it as something that is prolonging the recession, but one wonders why you would need to invoke that - presumably a liquidity trap, austerity, global crisis (THAT wasn't caused by Obamacare), and continued problems in credit markets are more than enough. If Obamacare was really a relevant supply shock I would have expected us to do a lot worse after it passed and especially as provisions have been implemented. Instead we've basically been plateauing after the initial shock, whether you're looking at E/P ratios or NGDP. So we have two options: (1.) all the demand shocks let up just about as much as Obamacare imposed a new supply shock, or (2.) Obamacare actually wasn't much of a supply shock. Occam's razor leads me to (2.) even in the absence of all the other signs of a supply shock mentioned above.

If crowding out was the problem through the crisis we would see rising interest rates. If people thought they had something better to spend on - something where the demand was there - they would dump Treasuries. We have been seeing some of this very recently, and for all I know that's a sign that we're getting to that point in time when crowding out is a problem and calls for stimulus should be moderated... but a much more likely explanation (and the one that everyone I've seen comment on it is saying) is that markets are responding to signals that the Fed will end stimulus. These are long-term rates rising after getting news that future short term rates will rise. Again, I'm siding with Occam unless Russ gives me something else.

Debt-to-GDP ratio would be a much bigger concern for me if stimulus were permanent. It's not. Of course we have debt problems looking forward with the entitlements, but that's there whether or not we do stimulus (so it doesn't hurt my case any more than it hurts Russ's - we just need to deal with it). This is tied to crowding out, if it's a problem we'll see it in bond markets, but we know what kinds of debt levels advanced industrial economies can bear. The Reinhart and Rogoff research shows it erodes growth rates, and that's bad, but depressions and unemployment hysteresis erodes growth rates too. So the sensible answer, it seems to me, is Romer's answer: short run stimulus, long run debt reduction. Short run debt reduction is counter-productive, and I'd rather not be dealing with worse depression and worse debt. The real issue with debt-to-GDP is entitlement reform and the sooner we get out of this the sooner serious movement on that will happen, I think.

I have to toss this back to Russ though (maybe before I answer the last question?). Exactly what empirical evidence is there that these things ought to be the guiding concerns right now? The most sophisticated case on this is the on provided by Casey Mulligan and we've talked about that on here before - it's not especially strong as outlined at the New York Times or the presentations he's delivered (or the free material from his book that's available... maybe he's hiding the good stuff in the rest of the book but I would have expected to see some of that in his seminar). So bracketing off Casey Mulligan for the time being, exactly where is the evidence against what I've said here?


  1. R&R do not really show debt erodes growth rates. See Arin Dube's analysis, for example.

    1. That was not my read of Dube - there is still a modest erosion of growth rates from debt. The point is just that most of the association comes from the causality running the other direction.

      But definitely - the dominant point here is that these concerns are not what we should be concerned about now.

    2. Right. His analysis suggests causality from debt to growth for countries with relatively low debt. Nothing significant for higher levels of debt, so even that evidence of causality is irrelevant to the U.S.

  2. I was about to write what Anonymous said in his (her) first comment. I recall that Noah Smith made similar points as you in his blog in response to (I think) Casey Mulligan on the Great Vacation hypothesis of the current depression.

  3. "Debt-to-GDP ratio would be a much bigger concern for me if stimulus were permanent."

    Are their cases where "temporary" isn't the plan?

    "...but that's there whether or not we do stimulus"

    In the current example, it's entitlements, but anytime there *is* a "debt problem", we should expect that "the problem" will be there regardless of the stimulus plan.

    My point is that these two criteria will always be satisfied.

    We have a fuzzy idea about debt contributions to growth erosion. We don't really know how much debt-gdp advanced economies can bear, but it is more than likely not a strict ratio and instead based on complex factors.

    Do you have a ballpark number for a debt to gdp ratio that may matter (due to unsustainability)? or an interest rate effect that would tell us to end a stimulus?


All anonymous comments will be deleted. Consistent pseudonyms are fine.