Follow by Email

Tuesday, April 25, 2017

Dynamic scoring, interest rates, and “crowd out” [feedly]

Dynamic scoring, interest rates, and "crowd out"
http://jaredbernsteinblog.com/dynamic-scoring-int-rates/

I've got a piece over at the WaPo on why, when Treasury Sec'y Mnuchin says that Trump's tax plan–which I suspect does not exist yet outside of some broad principles–"will pay for itself," you should…um…disbelieve him.

 

Such claims abuse "dynamic scoring," the process by which analysts estimate macroeconomic feedback effects from tax cuts. I provide a number of reasons why dynamic scoring abuse–DSA, or claims that tax cuts get anywhere close to paying for themselves–is a serious disease:

—The Tax Policy Center, known for careful, state-of-the-art analysis, finds that one recent version of the Trump tax-cut plan lost $6.15 trillion in revenue over 10 years without dynamic scoring and between $5.97 or $6.03 trillion with it. For the record, I don't believe these either, but at least I can understand the assumptions and these are not abusive scores.

—The reason I don't believe them is because I don't think we (economists) can accurately even "get the sign right" on such estimates, meaning we can't tell whether they'll add to or subtract from growth. All we can know, based on history, is that they'll be "small," à la the Tax Policy Center effects just cited.

—How can this be? Isn't it always the case that tax cuts boost growth? No! There's no such correlation either across countries or across time. Team DSA argues, for example, that tax cuts will lead people to work harder, but economic theory is ambiguous on this point, as some people will maintain their same after-tax income while working less. And, of course, most people can't tweak their work schedules like this anyway when the tax code changes.

—That's microeconomics, but at the macro level, if the revenue loss means less investment in public goods, including both productivity-boosting physical and human capital, growth will be slower.

 —Because of all this uncertainty, dynamic scoring, if done honestly, generates a wide range of estimates, depending on the different assumptions plugged into the model. For example, a dynamic score of a Republican tax plan from a few years ago by former Rep. Dave Camp estimated that it would raise the level of GDP over 10 years by between nothing (0.1 percent) and 1.6 percent.

—Now, guess which end of that spectrum fans of his plan would choose to highlight(see figure below)? Recall how Trump "dynamically scored" the crowd size at his inauguration or the popular vote count from the election. Given a range of dynamic scores, do you trust this team to give an honest answer? In fact, Mnuchin's already teed up his answer: The tax cuts will pay for themselves.

At least one person (OK, exactly one person) has asked me why, in arguing against large growth effects of tax cuts, I left off the impact of higher budget deficits and debt. In conventional models, they lead to higher interest rates and thus lower growth.

I left this off because I believe economists have an increasingly limited understanding of this relationship. To be clear, that's not the same as saying that larger deficits and debt will never crowd out private borrowing and raise rates, which in turn slow growth. It's saying there are numerous links in that chain, and intervening developments for which we must account.

EG, the Federal Reserve is clearly in the mix. The figure below plots the deficit/GDP against the composite nominal interest rate on gov't debt and the Fed funds rate (most of these data come from Kogan et al; I added the funds rate which is on fiscal, not calendar year basis–doesn't matter).

Soruce: Kogan et al

In the first few decades of the figure, the deficit/GDP ratio drifts down and the interest rate drifts up, as per the conventional model! Except that towards the end of the series, the deficit gets a lot more negative and so do interest rates. In fact, the interest rate tracks the Fed Funds rate a lot closer than the budget deficit.

You can plug in the real rate, try lags, whatever–the correlation's not there. The deep recession led to large deficits and very low interest rates, the latter a function of stimulative monetary policy. Forecasters and deficit hawks kept warning that any day, the bond market would punish our profligacy and interest rates would rise, but as the figure below shows, they were repeatedly wrong.

Source: Obama CEA

The historical record doesn't help either. Using Kogan's data, the correlation between the interest rate and the deficit since 1792 (!) is 0.12, i.e., small and the wrong sign. Using the real rate, it's 0.30. Using changes (in the nominal rate) it's 0.17. Using a more recent sample (1970-2016), the correlation is 0.10; using the real rate, it's at least negative, but at -0.04, far from statistically significant.

These are not models, of course–just correlations. But let me put it this way: given the extent to which much deficit hawkery references "crowd out" effects, the evidence for that dynamic doesn't jump out of the data, to say the least, which is why I didn't include it in the WaPo post.

Finally, I liked economist Jason Furman's discussion of these issues under the rubric of "fiscal space" in his recent piece on the "new view" for fiscal policy. First, he notes that the sovereign debt crisis was not obviously deficit driven: "there is no correlation between countries whose debt-to-GDP ratio rose prior to the crisis and those that saw their sovereign spreads spike during 2011. The spikes in debt in places like Ireland and Spain were far more a result of the crisis than a cause."

When fiscal multipliers are robust, say in weak growth periods when the Fed will accommodate fiscal expansion (implying a low interest rate environment), deficit spending that boosts growth can lower the debt/GDP ratio, the opposite of the "crowd out" story.

I've argued in various places that capital flows are in play here as well. As other countries buy safe US assets (like Treasuries), or, for that matter, any US financial assets, from securities to dollars, that too contributes to low interest rates and thwarts the simple crowd-out story.

Still, I wouldn't wholly discount the connection between higher public indebtedness, higher interest rates, and slower growth. I wouldn't, as I believe the Tax Foundation does in their dynamic scores, simply assume away this connection.

We just don't know, which, at the end of the day, underscores my broader point here, which is: I have little faith in dynamic scores to start with, and zero, nada, zip, no faith at all in claims that tax cuts pay for themselves. That's DSA, plain and simple.


 -- via my feedly newsfeed