Why Bother to Quote Somebody Out-of-Context?

Why Bother to Quote Somebody Out-of-Context?

I have never understood why a bunch of people in Washington DC behave the way they do.

In the comments section of this weblog, there is a quote of Kevin Hassett citing Doug Elmendorf (of the Federal Reserve) and Greg Mankiw (of Harvard) as concluding that the academic economic literature "has typically supported the Ricardian view that budget deficits have no effect on interest rates."

It is necessary to point out that Hassett's quote of Elmendorf and Mankiw is deliberately, massively, and deceptively out-of-context: it does not convey what Elmendorf and Mankiw believe. A fairer precis--a less blatantly out-of-context--short quote of what Doug and Greg mean would be something like:

This literature has typically supported the Ricardian view that budget deficits have no effect on interest rates... [but o]ur view is that this literature...is ultimately not very informative.... [T]he results are simply too hard to swallow...

Or:

This literature has typically supported the Ricardian view that budget deficits have no effect on interest rates... [but o]ur view is that this literature... is ultimately not very informative.... [T]he estimated coefficients on the policy variables must be viewed with skepticism...

Or:

This literature has typically supported the Ricardian view that budget deficits have no effect on interest rates... [but o]ur view is that this literature... is ultimately not very informative.... [T]his [analytical] framework has little power to measure the true effects of policy...

In Elmendorf and Mankiw's view, the failure of Plosser and Evans to find evidence that deficits affect interest rates tells us much more about the weakness of the analytical tools used than about the way the world works.

Why would anyone want to convey the impression that Elmendorf and Mankiw think something different? Why would Kevin Hassett want to quote Doug Elmendorf and Greg Mankiw out-of-context in testimony before congressional committees? People don't like to be quoted out-of-context: they tend to think that people who quote tham out of context are fools. Greg Mankiw is very, very influential indeed: why would anyone want Greg to think that he is a fool? Doug Elmendorf is less influential, but still quite influential: why would anyone want Doug to think that he is a fool?

It's not as though any senator or congressman's vote is going to be changed by including the sentence, "As Elmendorf and Mankiw conclude: 'this literature has typically supported the Ricardian view that budget deficits have no effect on interest rates.'" It's not as though any reporter covering the issue is going to be fooled--it is their job to call Greg (or Doug). They will soon take your measure, right?

---

And here is the videotape:

Here is what Kevin Hassett, citing Elmendorf and Mankiw (Douglas Elmendorf and Gregory Mankiw (1998), "Government Debt" (Cambridge: NBER Working Paper W6470, March)), says:

As Professor Paul Evans of Ohio State University pointed out in his careful studies of links between deficits and interests in several countries, even the large deficits produced by wartime spending had no discernible effect on long-term interest rates. Other studies published since Evans’ papers on this topic have reached similar conclusions. As Elmendorf and Mankiw conclude: 'this literature has typically supported the Ricardian view that budget deficits have no effect on interest rates.'

And here (with Hassett's quote in bold) is what Elmendorf and Mankiw actually wrote:

As with the literature concerning the consumption effects of fiscal policy, research into interest-rate effects appears straightforward, but numerous problems quickly arise. Indeed, some of the problems in the two literatures are quite similar.

One problem is that interest rates depend on expectations of fiscal policy and other variables and those expectations are hard to measure. A number of studies use forecasts from vector autoregressions as a proxy for expectations, but the quality of those proxies is unclear. Vector autoregressions assume that variables follow a stable time-series process, and they do not incorporate non-quantitative information. Both of these points are likely to be important, especially for fiscal policy variables, which are the outcome of a political process. Measurement error in the proxies for expectations biases the estimated coefficients toward zero and, thus, toward the null hypothesis of Ricardian equivalence.

A second problem with this approach as a test of Ricardian equivalence is that there is no natural metric for gauging the size of interest-rate effects. For the effect of taxes on consumption, there are natural Keynesian and lifecycle benchmarks as well as the Ricardian benchmark. Indeed, this features was critical in assessing whether tests of Ricardian equivalence had any power against alternative descriptions of the world. But no such alternative benchmarks exist for interest rates, becuase the size of the movements expected under non-Ricardian views depends on a host of elasticities. In particular, if international capital flows have an important effect on the domestic financial market, interest rates may not respond much to fiscal policy even if Ricardian equivalence is invalid.

With these caveats in mind, it is worth noting that this literature has typically supported the Ricardian view that budget deficits have no effect on interest rates. Plosser (1982) pioneered the approach of measuring expected policy using vector autoregressions. Further work in this vein by Plosser (1987), Evans (1987a, 1987b), and Boothe and Read (1989) has confirmed Plosser's original conclusion that a zero effect of deficits cannot be rejected.

Our view is that this literature, like the literature regarding the effect of fiscal policy on consumption, is ultimately not very informative. Examined carefully, the results are simply too hard to swallow, for three reasons. First, the estimated effects of policy variables are often not robust to changes in sample period or specification. Second, the measures of expectations included in the regressions generally explain only a small part of the total variation in interest rates. For example, the average R-squared of Plosser's basic monthly regressions (1987, tables 6 and 7) is .06, and the corresponding value of Evans's basic quarterly regressions (1987b, tabled 1) is .09. This poor fit suggests some combination of measurement error in expectations and the omission of other relevant (and possibly correlated) variables. Under either explanation, the estimated coefficients on the policy variables must be viewed with skepticism. Third, Plosser (1987) and Evans (1987b) generally cannot reject the hypothesis that government spending, taxes, and monetary policy each have no effect on interest rates. Plosser (1987) also reports that expected inflation has no significant effect on nominal interest rates. These findings suggest this framework has little power to measure the true effects of policy.

Posted by DeLong at January 23, 2003 03:00 PM | Trackback