This is getting out of hand. A recent post received too many high quality comments. I discuss the discussion some here.

The general view expressed in the discussion is that it sure looks as if non residential fixed capital investment as a percent of GDP expressed (nrfinvgdp) and nominal interest rates are positively correlated, because high non residential fixed capital investment caused the Fed Opem Market Committee (FOMC) to raise target interest rates.

The bit of evidence supporting this story is that the correlation is high also with the Federal Funds interest rate (ffi) which is basically whatever the FOMC wants it to be. This does not explain why the short term safe ffi is highly correlated with the rate on long term moderately risky Baa rated corporate bonds. Nor does it explain why the correlation of nrfinv/gdp and the Baa rate is even higher than the correlation of nrfinv/gdp and ffi.

One thing which I think has to be added to the old comment thread is that the correlation of nrfinv/gdp and ffi is much too high. It is easy to tell a story about why it is positive, but the fact that it is so very high does not correspond to the standard assumption about FOMC policy in conventional macroeconomic models.

That assumption is called the Taylor rule and is very very standard in the literature. The models are closed with the assumption that ffi is a function of lagged inflation and unemployment. This is alleged to be an empirical observation. Now the Taylor rule is not derived from an optimization problem — the standard assumption that private economic agents optimize is not extended to policy makers. Since the assumption that ffi follows a Taylor rule is not an assumption about objectives and rational maximization, I have long considered it to be the only equation in standard macro models which is not problematic.

Now I'm down to zero. The analysis below is very crude, but I think is a whole new problem for standard macro models.