In the blog post, the key point is that simple, static, savings/investment logic helps us think about what is going on. To see what they're getting at, here's their Figure 4: Rachel and Smith's argument is that world investment rates have not changed much from the 1980s, but real interest rates have fallen substantially. Therefore, according to them, this seems consistent with a Bernanke "savings glut" story (a shift to the right in the saving schedule in the figure)

*and*a Summers "secular stagnation" story (a shift to the left in the investment schedule in the figure). In fact, you can't explain their quantity/price data with only a savings glut story or only a secular stagnation story.

In the United States, we can tell a story similar to the above figure, in that real interest rates have fallen, and if we look at investment rates, we get:And, just for good measure, we can look at only nonresidential investment: So, in the United States, investment rates are somewhat high relative to the 1980s (the disaggregated data on nonresidential investment is available only post-1999) and, as is well-known (see for example this paper by Hamilton et al.) the real interest rate on government debt has fallen substantially since the 1980s.

But, we could argue that the relevant real rate of return that applies to the first chart is the real rate of return on capital. Conveniently, as it turns out, Gomme, Ravikumar, and Rupert have measured rates of return on capital from the NIPA (national income and product accounts) data. Here's one of their charts:So, for the United States, the rate of return on capital is higher than in the 1980s, and investment rates are higher. That doesn't square with savings gluts and secular stagnation.

As I argue in the slides I link to in this post, I think it's more useful to think about asset markets in explaining low real interest rates. For the U.S., it's important that we consider the market for U.S. Treasury debt, and what makes demand high and supply low in that market. That then has implications for monetary policy and inflation that I think are interesting. See the slides for the details.

Dear Steve,

ReplyDeleteJust a quick note as I have seen you use the graph of the ratio of real investment to real GDP on a couple of places. NIPA aggregates are chain-linked using quality-adjusted price indices, meaning that there is in fact non-additivity between components of real GDP. This explicitly invalidates the interpretation of the ratio of real investment to aggregate real output as being a "share" or an "investment rate" in any meaningful way. It is standard practice to look at nominal ratios instead.

Two useful references:

http://www.federalreserve.gov/pubs/feds/2000/200035/200035pap.pdf

http://www.karlwhelan.com/Papers/twosec_revision.pdf

Thanks. See the next post.

Deletein naomi robbins wonderful book on graphs, she explains how a trellis plot helped people see something important and new in an old, very, very well analyzed data set

ReplyDeletei'm also amused to read this just after reading, via thoma, this

https://theoryclass.wordpress.com/2016/01/16/reproducibility-and-publication/#comment-3725