As promised below, I have written something about investment (warning google document)

In the note, I calculate some variables related to discounted future profits. Here is a snip from the paper in which I define them.

Standard micro founded models of investment are based on shareholder value maximization and assume that there are adjustment costs, so investment decisions must be forward looking. They imply high investment when the expected discounted stream of profits per unit of capital is high. If expectations are rational, expected discounted profits should be correlated with actually achieved discounted profits.

The interesting part of the note is just a graph of time series. For this graph I assume depreciation of 1.25% per quarter. I use Moody's BAA index for the interest rate and look at non residentical fixed capital investment.

invest  prret10 is prret defined in the equation divided by 10. prprfgdp20 is prprfgdp divided by 20.

The pattern is the opposite of that predicted by standard macroeconomic theory — the variables which should be positively correlated are negatively correlated.

In particular, graph shows two episodes of high non residential fixed capital investment and low discounted profits. First there was extremely high investment during the Carter years. Low profits and high interest rates during the Reagan years make this investment seem to have been a mistake. I note in passing that the events which are clearly there in the official data, contradict the generally accepted story about what happened in the US economy back then. Second, as is more widely recognized, there was a huge amount of investment in the late 90s followed by painful disappointment in the 00s.

In any case, the pattern in the data is very clear and is dramatically the opposite of the pattern predicted by standard macroeconomic models.