Both Nick and I agree that investors appear to be substituting out of private securities and into government securities (in particular, USD and US treasuries). We appear to differ on the underlying cause of this behavior.
Nick takes the bubble and its behavior as exogenous. There is a growing bubble in money/bonds, which is drawing resources away from private capital investment. In a comment on Nick's post, I suggested an alternative interpretation. I take expectations over the future return to capital as exogenous and interpret the bubble asset as a socially beneficial alternative store of value (in theory, such bubbles can mitigate the adverse consequences of a dynamic inefficiency).
The type of model I have in mind can be found here. In that model, money (bonds) and capital compete as a store of value in the wealth portfolios of individuals. Assuming risk neutrality and diminishing returns to capital investment, a no-arbitrage-condition equates the expected marginal product of capital to the expected real rate of return on money. I then define "good news" as information that leads to an upward revision in the forecast of capital return; "bad news" is defined conversely. I argued that a bad news event (or a series of bad news events) would cause rational downward revisions in the forecasted return to future capital, thereby depressing capital investment and stock prices, and causing a flow of resources into government money/bonds. There would be "surprise" declines in the price-level (reflecting the increase in demand for money). The resulting behavior looks like an exogenous increase in the demand for government securities, but this is the wrong interpretation.
In a subsequent discussion, Simon van Norden asked me to explain what evidence I had to support the notion that shocks to expectations (over future capital return) varied significantly and at high frequency. These are the type of impolite questions that I think need to be asked. I replied, rather lamely I think, that these expectations were difficult to measure (except perhaps, indirectly via asset prices). I think I even stooped to the time-honored tradition of citing the importance that our predecessors attached to the hypothesis (think Keynes, Pigou, Marshall, and so on, back in time). Simon suggested that I look at the Federal Reserve Bank of Philadelphia's Survey of Professional Forecasters. Thanks, Simon!
I am kind of leery of this data, as the forecasts of professional forecasters are not necessarily the same thing as the forecasts of businessmen in charge of making multimillion dollar investments. Moreover, the survey does not ask about forecasts over the return to capital...the closest measure is future real GDP growth. But what the heck...some data is better than no data.
And the data looks kind of interesting. In Figure 1, I plot the average (over all forecasters) expected growth rate over five periods: the current quarter, one quarter ahead, ..., up to four quarters ahead. (Click on the figure to enlarge). I do this for each quarter, beginning with 2007:4 and ending with 2010:2. The dashed line measures actual (revised) real GDP growth. Let me summarize some facts about this data that I find interesting.
FACT 1: Short-run forecasts are much more volatile than long-run forecasts.
Personally, I find the resiliency in the year-ahead forecasts rather remarkable. Even in the depths of the financial crisis, these professional forecasters were forecasting a return to normal rates of growth within a year or so. Do they run the same canned VAR models to come up with these forecasts? I wonder whether businessmen in charge of capital budgets were as optimistic as these professional forecasters in the fourth quarter of 2008?
FACT 2: The recent recession appears to be associated with a persistent (possibly permanent) decline in potential GDP.
An expected return to "potential" GDP would require year-ahead forecasts to exceed 2% per annum coming out of the recession. This does not appear to be the case; at least, not so far.
I wonder how Pigou and Keynes would have reacted to this data. If I understand their views correctly, it is precisely the long-horizon that is subject to the psychology of animal spirits. And because capital investment today runs largely off of long-horizon forecasts, this is what makes investment so volatile. This view (which is related to my own) does not seem to be supported by this data.
On the other hand, even changes in short-horizon forecasts can induce changes in asset prices. And to the extent that assets are used as collateral in lending, a decline in asset prices may depress investment spending through the familiar accelerator. (I have written a model, with my SFU colleague Fernando Martin, that is being extended in this manner).
It is also of some interest to explore some higher moments associated with these growth forecasts. In Figure 2, I plot the standard deviation of forecasts (across forecasters) in each quarter and at each horizon. I interpret this standard deviation as a measure of "uncertainty."
Figure 2 suggests that the uncertainty over long-horizon growth increased substantially since the beginning of the recession and continues to remain high. In a nutshell, these forecasters are expecting a return to 2% growth in the long-horizon, but there appears to be much more uncertainty among these forecasters about the prospect of such an event.
It is also curious to see how "uncertainty" over the short-horizon increases dramatically. I view this as some evidence against Nick's hypothesis and in favor of my own (extended to include higher moments, of course). The emerging bond bubble is the consequence of increasing uncertainty over short and long term capital returns; this same uncertainty is contributing to depressed capital expenditure.
In short, I think that Nick might have the direction of causality ass-backwards.
* I thank Constanza Liborio for gathering and plotting this data.