Thursday, January 5, 2012

Frustrating Responses by the Famous Keynesians

Brad DeLong responds to John Cochrane, and Krugman approves of
DeLong's argument and comments Krugmanly on those who disagree with




I think I understand the model they're thinking about this with, but
it seems to me to be obsessively short-term focused. They take
Cochrane to task on some technical things, but a "best argument
possible" read of Cochrane's fiscal argument, tweaking some of the
stuff he gets wrong, is something they do not think about (and,
reading them regularly, don't ever respond to, which aligns with Alex
Tabarrok's issue with Krugman's argument style at Marginal

This isn't to say that I agree with all of Cochrane's statements. I
think Cochrane misses that some substitution occurs intertemporally in
a world that's not perfectly rational expectations – but Krugman and
DeLong seem to indicate no substitution at all when they take him to
task for this.

Cochrane's argument is best summarized with these two paragraphs:

"[L]et's think of a "fiscal stimulus" in which the government borrows
money and spends it, but with the clear plan that the debt will
eventually be repaid with future taxes, not just by printing money.
Can this kind of stimulus work, and if so how?… First, if money is not
going to be printed, it has to come from somewhere. If the government
borrows a dollar from you, that is a dollar that you do not spend, or
that you do not lend to a company to spend on new investment. Every
dollar of increased government spending must correspond to one less
dollar of private spending. Jobs created by stimulus spending are
offset by jobs lost from the decline in private spending. We can build
roads instead of factories, but fiscal stimulus can't help us to build
more of both. This form of "crowding out" is just accounting, and
doesn't rest on any perceptions or behavioral assumptions.*… Third,
people must ignore the fact that the government will raise future
taxes to pay back the debt. If you know your taxes will go up in the
future, the right thing to do with a stimulus check is to buy
government bonds so you can pay those higher taxes. Now the net effect
of fiscal stimulus is exactly zero except to raise future tax
distortions. The classic arguments for fiscal stimulus presume that
the government can systematically fool people [This is the Ricardian
Equivalence argument].

The government should borrow to finance worthy projects, whose rate of
return is greater than projects the private sector would undertake
with the same money, spreading the taxes that pay for them over many
years, after making sure its existing spending meets the same
cost-benefit tradeoff. Just don't call it "stimulus," don't claim it
will solve our current credit problems, "create jobs" on net, or do
anything to help the economy in the short run, and don't insist that
we have to pass this monstrous bill in a day without thinking about

The pieces of DeLong's response most relevant to these paragraphs can
be summarized:

1) "[C]rowding out only happens "if we are in a cash-in-advance
economy with a technologically-fixed velocity of money. But we

2) "The government purchases $100 billion of goods, issues $100
billion of bonds, and raises taxes by $3 billion a year in order to
amortize the bonds. Government purchases go up by $100 billion this
year. Private consumption goes down by $3 billion this year. Net
fiscal impetus is not $0 but rather $97 billion. Cochrane doesn't
understand the Ricardian Equivalence argument he is trying to make."

My responses to these:

1) I fall into the Scott Sumner camp - fiscal spending is endogenous
to the Fed's reaction function, which means that a perfect Fed is
going to result in a fiscal multiplier of zero outside of supply-side
effects. Or, in lay terms, you can't stimulate the economy with
government spending very well, because the Fed's going to offset any
demand-side stimulus or slowdown and it's hard to find any government
spending on top of what we already do that helps the supply side. Tax
cuts hold slightly more promise, but still, the effects would be small
in the short-term (though potentially bigger in the long-term).

This is less true if you believe in liquidity traps (which mean the
Fed can't stimulate well). Market reaction to unconventional Fed
action makes me skeptical that we're in a liquidity trap, but I know
that Krugman and DeLong believe we are, so for the purposes of this
debate, let's assume they're right, and we are in a liquidity trap.

2) Let's grant, for a moment, DeLong's point that crowding out of
private savings and investment don't matter here. I have trouble
intuiting how relevant the "cash-in-advance economy" part is because
generally speaking, you do pay cash pretty quickly – within a couple
months – but the velocity point is important, and what DeLong doesn't
say is that for the purposes of this crisis, credit intermediation is
clearly having problems, hence lots of (seemingly) excess reserves.
There's also the possibility that you're selling part of that 97
billion in bonds to foreigners who would otherwise sit on the cash
rather than spend it immediately (hello, political ramifications of
seizing foreign currency held at the central bank in China / other
Asian growth model countries). So we'll give DeLong the crowding out
argument, for purposes here.

3) For Cochrane's Ricardian equivalence argument to happen, you need a
world that is very, very rational expectations – and while I have no
papers off the top of my head to confirm this, the empirics probably
don't back up the fact that if the government threatens to tax you 100
billion in the future in the private sector, you pull back 100 billion
of spending into savings right now. DeLong ignores this rational
expectations argument entirely in his response about Ricardian
Equivalence, and I think the right answer is probably somewhere in the
middle.** This, itself, is also not damning of DeLong's response; it
weakens the argument but doesn't ruin it.

Which finally brings me to my big issue with how Krugman and DeLong
are approaching Cochrane's arguments.

To an extent, the multiplier of spending is going to be based on the
productivity of whatever you're spending on. This is pretty simple.
Your unproductive spending is not going to multiply as much because it
won't have a positive impact on the economy outside of the first order
employment boost and associated consumption spending. Building a
subway from New York to New Jersey allows trade and labor mobility
that stimulates further economic growth; digging a ditch doesn't do

If you impose large future taxes to pay for unproductive investments
today, you certainly do create some fiscal impetus today (ignoring Fed
endogeneity from point 1 and assuming a liquidity trap). But you're
doing this by reducing economic productivity tomorrow with tax
increases. This is what Cochrane is alluding to when he talks about
governments funding projects with a rate of return above the market's
- you're going to harm the economy if you transfer projects from
market-rate productivity projects to below-market-rate productivity
projects. Instead of pandemic protection or useful highways, you're
building a bridge to nowhere, or a high speed rail in areas that won't
use it, or solar panel factories that aren't efficient enough to

So if the items you're buying today are unproductive (say, ditch
digging, or at least a portion of the fiscal stimulus passed), and the
activity you're taxing is productive (like the US private economy),
then you're creating a cocktail of bad long term outcomes:

1) You've replaced future productive activity with unproductive activity today.

2) You've created crappy incentives in the future with high tax
rates, hurting the supply side.***

3) There's also a temptation to reallocate funds to the places that
"need it most", which may be the areas that overbuilt and misbehaved
the most (in this recession, housing-related areas). This is very
Austrian, but on the supply side, the Austrians could have a point
(even if I disagree with them on the demand side). Look at GSE
funding, for example. This creates not only long-run moral hazard but
also inefficiently continues the misallocation (credit for housing
remained cheap throughout the downturn. Even if you believe, as I do,
that household formation is the big housing issue now and excess
housing supply is no longer the problem, it certainly was the problem
for a long time). I don't know how applicable this Austrian argument
is by magnitude but it's at least plausibly a measurable factor.

4) You reduce skill-damaging long-term unemployment today, but unless
you believe the economy will get a LOT more dynamic, it's going to be
hard not to increase long-term unemployment by even more tomorrow as
overall cyclically-adjusted (ie, structural) unemployment goes up
during deleveraging but unproductive investments haven't paid for

Basically, you're trading more than 1 job in the future for 1 job
today. Maybe the economy is in such rough shape right now that it has
no choice but to be way more dynamic tomorrow, but you still think
long-term unemployment is going to be super damaging for each
individual****, so you're willing to trade 2 jobs tomorrow for 1 job
today because you think long-term unemployment (and misallocation)
will disappear in the more dynamic economy. But you need a really,
really large effect of long-run unemployment to make that happen, and
you need to be a major optimist about the effects of technology and
policy on the supply side. Because otherwise, fiscal stimulus may be
stimulating in the short-term, but the intertemporal substitution
means you're damaging the economy in the medium or long term. In my
opinion, that's the "best possible interpretation" of Cochrane's
argument, and in my opinion, it's a persistent weakness in Krugman's
and DeLong's arguments for fiscal stimulus. I wish they'd address it
instead of consistently brushing it aside like it doesn't matter.

As an aside:

Just so I'm not criticized for ignoring the rest of DeLong's
arguments, DeLong also notes: "There is nothing in "traditional
Keynesian" thinking to say that you ought to boost consumption rather
than, say, infrastructure investment. Nothing at all." I suspect,
though of course cannot confirm, that Cochrane and DeLong are talking
past each other a bit on investment/consumption. In the private
sector, if financial intermediation isn't working, then government
policy stimulating private consumption pumps more into the economy
than government policy stimulating private investment, because money
gets saved and not lent out. I'm pretty sure that's the (in my opinion
incorrect) Econ 101 argument for government spending rather than tax
cuts as fiscal stimulus. The government doesn't need to worry about
this, so for direct government spending, investment and consumption
should be equivalent.

DeLong also criticizes Cochrane's interpretation of monetary policy.
Largely, I agree with DeLong's criticisms in this arena and will thus
let them go, although I think it's ludicrous to use that as evidence
against his fiscal arguments just via "he doesn't get how the world
works". Argumentation doesn't work like that.

*I've redacted this: "Second, investment is "spending" every bit as
much as is consumption. Keynesian fiscal stimulus advocates want money
spent on consumption, not saved….". It distracts from the main point
but I address it in an aside at the end.

** DeLong assumes that there's no rational expectations in effect at
all – if 100 billion is spent and 3 is paid for, that doesn't mean
you'll get 97 in stimulus right now, if there's some additional
savings. Plenty of people worry about what their future tax rates are
going to be when making savings decisions. I save more knowing that
I'll probably be taxed more on my income in the future than today's
rates. Maybe I'm the only one, but even if I am the only one… you're
no longer at a perfect 97 billion. Less on topic, there's a pretty big
debate going on in the financial planning community about whether to
use Roth IRAs or normal IRAs for young people – if you trusted the
government not to need to raise tax revenue, the long-horizon tax-free
nature of a Roth is wonderful, but for that you have to trust the
government not to remove their tax advantage, which not everyone does.
Rationally, we should be splitting between Roth and Traditional for
planning purposes. Megan McArdle wrote about this recently, as well.
But fine, let's say people don't fully consider future tax rates in
current savings decisions and allow only partially rational
expectations to say there's some fiscal impetus today. I'd agree with
that, even if it's not 97 billion worth in that example.

I'll note that I think the $100 billion argument (DeLong point two)
was directed at the Ricardian Equivalence argument but I think it's
actually an (incorrect) response to crowding out (incorrect because
you're also putting $97 billion into bonds). But again, I grant him
crowding out on velocity and credit intermediation grounds.

***and, following up on the second footnote, possibly mitigated some
of your own current fiscal stimulus' effectiveness through rational
expectations channels – which should be exacerbated by unproductive
investments because people know that what you're spending on will not
grow the economy in the long run much to offset the spending.

Also worth pointing out (or not) for students reading this that you
could create toy economies with tax systems where increased tax
revenues don't impact labor supply decisions, but that economy is not,
should not be and will never be America.

**** also worth noting that there's an argument to be made that
long-term unemployment is less damaging to an individual's prospects
to get back into the labor force when there are lots of people in the
same boat, because a) in a better economy, employers will need to hire
someone, and if long-term unemployed are their only choice, that's who
gets hired, and b) there should be more training programs available
with a critical mass of people who need them.

This isn't to say that long-term unemployment isn't catastrophic to
the economy in a way short-term unemployment isn't – but on a
skill-erosion level, it's probably the same to have 10 additional
units of long run employment today as to have 5 today and 5 tomorrow,
and on a job prospects/retraining level, I'd much rather not be the
only one screwed, which kind of happens if you redistribute long-term
unemployment between today and tomorrow from just today. Thus, it's
slightly better from a labor supply perspective to have more long-term
unemployment in one period than to have the same human-years of
unemployment stretched over two periods, as long as you see an overall
economic recovery. I hope this is clear… if it's not, I'm happy to
follow up.

No comments:

Post a Comment