Tuesday, July 12, 2011

The sum of all right-wing assumptions

Dang you, Matt Yglesias! If I don’t get my dissertation done on time, it’s all your fault!

I am referring to this post, in which Matt points out a paper by John Cochrane of the University of Chicago:
Cochrane argues that...not only is it plausible to think that runaway inflation is right around the corner even though there’s no evidence of elevated inflation expectations, he argues that runaway inflation is likely to be sparked by tax increases and can best be combatted by gutting Social Security and Medicare. 
Here’s a slice from the conclusion of his piece:

Will we get inflation? The scenario leading to inflation starts with poor growth, possibly reinforced by larger government distortions, higher tax rates, and policy uncertainty. Lower growth is the single most important negative influence on the Federal budget. Then, the government may have to make good on its many credit guarantees. A wave of sovereign (Greece), semi-sovereign (California) and private (pension funds, mortgages) bailouts may pave the way. A failure to resolve entitlement programs that everyone sees lead to unsustainable deficits will not help.
When investors see that path coming, they will quite suddenly try to sell government debt and dollar-denominated debt. We will see a rise in interest rates, reflecting expected inflation and a higher risk premium for U.S. government debt. The higher risk premium will exacerbate the inflationary decline in demand for U.S. debt. A substantial inflation will follow—and likely a “stagflation” not inflation associated with a boom. The interest rate rise and inflation can come long before the worst of the deficits and any monetization materialize. As with all forward-looking economics, no obvious piece of news will trigger these events. (emphasis Yglesias')

Basically, if Cochrane is right, then tax cuts actually represent fiscal austerity, and conservatives don’t have to worry about all those silly economists who say that you can’t cut the deficit without tax hikes. Matt drily notes that it’s pretty convenient that this…um…counteruintuitive conclusion just happens to dovetail perfectly with the policy proposals of Cochrane’s drinking buddy Paul Ryan. But Matt stops short of critiquing the paper’s methodology, basically daring me to do it:
 I obviously don’t have the technical chops to wrangle with Cochrane’s model in detail.
Hence, Matt is entirely responsible for my failure to complete my dissertation.

Anyway, on to the paper… 


Cochrane analyzes the effects of fiscal and monetary policy using a system of two equations. The first equates the total real government debt + the money stock with the expected present discounted value of future government surpluses:

This is just the statement that the value of government debt is equal to the government’s ability to pay back the debt by taxing people. The second equation is the “quantity theory of money”:

The first of these equations is pretty uncontroversial. For a critique of the quantity equation, see Matt Rognlie, Matt Rognlie, and Matt Rognlie, but having some kind of relation between money, output, and prices is pretty unavoidable in a macro model.

An analysis of the entire paper is, well, beyond the scope of this blog post. I think I’ll stick to analyzing the claim that Matt Yglesias talks about, i.e. the story that higher taxes can cause stagflation. But first a point. The fact that Cochrane reduces the macroeconomy to two more-or-less uncontroversial equations causes me to invoke Noah's Third Law of Article Reading:
If your model is a subset of the intersection of most existing models, either you didn't get any new results, or you need to check your assumptions.
In other words, don't expect to get Exciting New Results using only the Same Old Stuff. Is John Cochrane really going to show that higher taxes cause inflation using nothing but an intertemporal budget constraint and hoary old MV=PY?

Simplifying your model past the point of full determinacy doesn’t remove the need to make additional assumptions about economic relationships; it just makes those assumptions implicit rather than explicit. If you say that only these two equations matter, you’re saying that other things, like imperfect competition and price stickiness, don’t.

Cochrane acknowledges this at the beginning of the paper:
[These two equations] are the beginning, not the end of analysis, and I do not mean to imply otherwise. In particular, monetary models also include a description of dynamics, and price-stickiness or [an]other mechanism that…translates inflation into real output…Additional frictions [that generate] stimulative effects of tax or real debt-financed government spending, and additional financial frictions can easily be added to this style of analysis. 
And again later on:
I am abstracting here from [frictions that] potentially have important effects on the analysis of fiscal stimulus…My goal is…not to deny other channels or try to have a last word on an 80 year old debate. 
So Cochrane is being intellectually honest here. He’s admitting that he just left out all of the things that monetarists and Keynesians believe makes stimulus and quantitative easing effective. They “can easily be added,” but he doesn’t add them. So the conclusion that countercyclical policy doesn’t work is going to follow pretty naturally from the assumption that countercyclical policy doesn’t work.

The Sum of All Right-Wing Fears

Now let’s talk about the scenario that Matt Yglesias picked out. In this scenario, higher taxes – which we normally think of as deflationary – actually create inflation, while still stifling output. To reiterate, the scenario is:

Higher taxes ---> slower growth ---> higher structural deficits ---> higher inflation, capital flight ---> stagflation

As it turns out, this particular sequence of events relies on three assumptions:

Assumption 1: Higher taxes would actually increase deficits, i.e. we’re to the right of the Laffer peak.

Assumption 2: Increased deficits are structural and cannot be reversed.

Assumption 3: Bond investors will not believe that the U.S. will be able to reduce spending in the future, and will thus abandon U.S. Treasury bonds.

Normally, we think tax increases should raise revenue and help close deficits. Not so, say supply-siders; we’re to the right of the peak of the Laffer curve, and thus higher taxes will decrease revenue and increase deficits. Of course, static analyses all agree that the U.S. is well to the left of the Laffer peak, so we’re in no danger of this happening. Supply-siders are forced to assume that the dynamic effects of taxes on future growth are large. Cochrane does in fact assume this:
The top of the Laffer curve is where the elasticity is equal to zero, so higher tax rates raise no revenue. Many economists think the U.S. is comfortably below that point…Trabandt and Uhlig (2009) offer a detailed Laffer calculation with fixed productivity growth and no migration, yielding the result that the U.S. is substantially below the Laffer limit….
The present value of future tax revenues is what matters for the fiscal valuation equation, however...[so] small growth effects can have a big impact on the fiscal limit. Thus, if a rise in [taxes] from 30% to 35% only implies a 0.02 x 0.15 = 0.3% reduction in long-term growth, then we are at the fiscal limit already. 
Note that this is a huge supply-side effect. It implies that the Bush tax cuts raised our trend per-capita growth by 0.36%, or more than one-sixth. It also implies that the Johnson and Reagan tax cuts raised our trend growth far more than that, and that the Clinton tax increases significantly decreased our trend growth. Unsurprisingly (to me, anyway), a long-term plot of U.S. output doesn’t show any of these things.

Does Cochrane really believe in this huge supply-side effect? It appears he does:
[T]he ‘‘supply siders’’ in the Reagan administration advocated lower tax rates, and larger short-term deficits, arguing that lower tax rates would spur growth and lead to larger tax revenues. The lower tax rates of the Reagan administration were in fact followed by two decades of economic growth, leading to strong surpluses as shown in Fig. 1. The present values of the surpluses that occurred validated the quick end of inflation in the early 1980s. 
Here is the aforementioned Fig. 1:

Do you see the “strong surpluses” that followed Reagan’s tax cuts? If so, maybe you should consider a job as a “technical analyst” on Wall Street…because all I see is a deficit, followed by a small surplus after Reagan partially reversed his tax cuts, followed by a very big surplus after the Clinton tax increases, followed by a deficit after the Bush tax cuts…

So, I conclude that Cochrane’s “dynamic Laffer curve” analysis is not supported by any evidence presented here (or any evidence of which I am aware).

Cochrane’s second assumption is that stimulus deficits are structural and permanent (i.e. a negative shock to the present value of all future budget surpluses). This just seems unrealistic to me; in the real world, spending from, say, the ARRA has already pretty much finished. In fact, the assumption seems to be basically political in nature. Cochrane mentions (in the quote above) that Reagan-era supply-siders advocated “larger short-term deficits”…by commending this policy, Cochrane appears to assume that deficit spending by Republican administrations can be reversed, but that deficit spending by Democrats never will.

Finally, we have the notion that these permanently higher deficits will lead to capital flight – the so-called “attack of the bond vigilantes.” This assumption is embodied in a posited decrease in the stochastic discount factor, i.e. a rise in expected interest rates. But Cochrane’s model is forward-looking; as he explicitly states (page 3), this should make long-term interest-rates spike as soon as (or actually well before) the higher deficits are announced. As many, many people have pointed out, nothing like this has yet happened so far to the U.S. economy, despite the explosion in deficits described by Cochrane. If a “fiscal inflation” is in our future, either it doesn’t follow Cochrane’s implied model of expectations, or it is, as yet, completely hypothetical to even the most far-sighted observer. 

So to make a long story short, the “Sum of All Right-Wing Fears” scenario is really just a “Sum of All Standard Right-Wing Assumptions.” Tax increases decrease revenue. Democrats are addicted to deficits. Bond vigilantes are ready to abandon U.S. Treasuries.

It seems to me that all three of these ideas have been rebutted (or at least candidly discussed) a million bazillion times in the past. But by cloaking them in math and burying them amid a bunch of reformulations of present-value equations and econ-specific jargon, John Cochrane was able to pour old wine ($370/bottle wine, no less) into a new glass, and keep smart non-econ people from seeing that that was what he was doing.

Well, for a little while, anyway.

At any rate, I invite everyone to go through and read the rest of Cochrane's paper and find more errors or good points that it makes. In particular, I didn't have the time to look at his claim that monetary policy is ineffectual. I’m sure there are a bunch more assumptions about things like causality, stability, expectations, unmodeled behavior of aggregates, etc. that I didn’t have time to go through and identify. After all, I’ve got a dissertation to write…or something…


  1. I'd be curious to hear what your other laws of article reading are.

  2. I've only read about half of this post, and have to go do yard work. But I want to point out that Chochrane is making a 4th unfounded assumption that you do not note.

    higher structural deficits ---> higher inflation

    Looking at 100 years of data has convinced me that over that span there is no meaningful correlation between Federal budget deficits and inflation. To be clear, this doesn't just mean the absence of cause and effect - it means no correlation at all!


    Please continue exploding the nonsense of agenda-driven right wing hacks.

    And good luck with your thesis.


  3. An enviable post. However, I am unfortunately inspired by John Cochrane to make my own pre-baked observation on how the surplus / deficit correlates with tax increases / cuts.

    I suppose a scenario where tax cuts / increases are only marginally related to the surplus / deficit in that higher taxes necessarily leads to greater / lesser revenue. However, it's practically doubtless that these revenue gains / losses are insufficient to explain the full swings of surplus/deficit as exhibited in Figure 1. Like Yglesias, I also don't have the chops to prove this mathematically, but it seems pretty obvious to this casual observer.

    I prefer a narrative for the phenomenon that higher taxes lead to an increase in real surplus (and vice versa) based on the understanding that the public is only able to stomach tax increases when society comes to a consensus, whether outwardly recognizable or not, that things are going to improve economically, and thus that we can afford to give more to the government. Our democratic process works to make this happen. As a result, when taxes go up, revenue increases at a greater rate because the economy as a whole is improving, prior knowledge of which is why we allowed our taxes to be raised in the first place. A similar argument can be made for when taxes go down.

    This supposes that the collective consciousness, expressed through our political system, is more aware of economic conditions than models and concise economic arguments, but then again the collective narrative has always been more powerful, if not quite as summarizable, as individual short stories. I'd challenge you to provide a more coherent argument for why raising taxes leads to a disproportionate gain in revenue.

  4. Anonymous4:45 PM

    Read Cochrane's Econometrica (2001) piece and you'll have a better idea of where he is coming from.

    The paper you critiqued appears in line with his prior publication on the fiscal theory of the price level.

  5. "the conclusion that countercyclical policy doesn’t work is going to follow pretty naturally from the assumption that countercyclical policy doesn’t work"

    Isn't this a classical Milton Friedman trick (as defined by Joan Robinson) ?

    Thanks for the excellent post. Good luck with the dissertation.

  6. Awesome.

    I really learned a lot from some of Cochrane's papers and his book Asset Pricing.

    To bad such a smart man turned out to be such a crank.

  7. "Thus, if a rise in [taxes] from 30% to 35% only implies a 0.02 x 0.15 = 0.3% reduction in long-term growth, then we are at the fiscal limit already. "

    And he probably also assumes that the tax money is spent digging diches or 100% on consumption with zero investment value. But, hmm, what if it were spent on alternative energy, education, basic scientific and medical research, instead of yachts, mansions, and $10,000 suits, hmm, how would that affect long term growth?

  8. Or what if it were spent on this:


  9. Noah, not to further distract from your dissertation, I have a question.

    I agree that we are to the left of the maximum in the Laffer curve. But political constraints mean that upper marginal rates above about 30% are impossible. The only event likely to change that is a full-scale crisis and a major shift in political power.

    So, replace Cochrane's assertion that high taxes --> slower growth with the assertion that higher taxes --> not going to happen.

    In that case, there are only two choices left to policy makers: keep taxes the same, or lower them. Even if they keep them the same, at some point there will be a recession, tax revenues will fall, and budget deficits will inexorably rise.

    There are four possible outcomes: a) investors, looking at constantly rising deficits will insist on higher interest rates, and therefore there will be lower growth, b) the government will maintain rates interest rates artificially low to stimulate growth, c) a rise in inflation would permit growth at the expense of asset accumulation by creditors, or d) the currency will fall, in which case export growth might be good, but the standard of living might well be falling, as Americans take jobs making textiles and Chinese take jobs in nanotechnology. Falling wages + rising materials prices feels the same as inflation even if the nominal value is unchanged.

    As we see in the last two years, low interest rates, high budget deficits, and falling currency don't necessarily trigger inflation, at least not in the short term. The situation, however, could be different in the absence of an Asian savings glut and global recession. In that case, the US couldn't realistically hold interest rates artificially low, so either interest rates would rise

    Could it be that Cochrane is implicitly assuming that it is impossible to increase tax collection (and that the rights of creditors are sacrosanct)? If so, couldn't he be right by accident?


  10. Anonymous9:32 AM

    Don't bother with your dissertation Noah, you aren't getting a job. Nobody wants to hire a know-it-all that doesn't actually know it all.

  11. Don't bother with your dissertation Noah, you aren't getting a job. Nobody wants to hire a know-it-all that doesn't actually know it all.

    Oh come on now, regular anonymous troll dude. That's exactly the type of person that people want to hire, and you know it. ;-)

  12. Please do finish your dissertation. But you might spare yourself the trouble of applying for jobs at the University of Chicago ;)

  13. Noah said...

    " Don't bother with your dissertation Noah, you aren't getting a job. Nobody wants to hire a know-it-all that doesn't actually know it all.

    Oh come on now, regular anonymous troll dude. That's exactly the type of person that people want to hire, and you know it. ;-)
    2:32 PM "

    Hahaha! So very well said Noah. Thanks for the post. I'm still in the intimidated by systems of equations phase, but you took the fear right out of that. Please continue.

  14. Anonymous3:08 PM

    There is something wrong with Cochrane's equation for the value of the outstanding bonds. It makes no sense if we start to take limiting examples like: the country runs a small continuing deficit forever while the economy grows so quickly that the debt to gdp ration falls forever. The equation would appear to say that the bonds have no value under that circumstance.