A short disquisition on Cowen’s Third Law

“All propositions about real interest rates are wrong,” and this one is in the form of a Bloomberg column.  Here is one part:

The law of demand is one of the sacrosanct principles of economics: If the price of apples goes up, the demand for apples will go down. Yet when real interest rates go up, it is not obvious that the demand to invest goes down, even if the numbers are adjusted for possible reasons why real interest rates might have changed in the first place. This raises a profound question: If the law of demand doesn’t apply, how well do we understand investment and real interest rates at all?

The puzzles deepen. There was a longstanding debate in economics about whether the U.S. Federal Reserve, using monetary policy, could affect real rates of interest. After extensive research, the conclusion was reached that the Fed can indeed have a marginal effect by supplying more liquidity to markets. Yet the effect is sufficiently small that there can be a plausible debate about whether, statistically speaking, it exists at all.

These days, however, the real federal funds rate measures as below -4%, based on measures of core inflation. No one doubts that the monetary expansions of the Fed, which have brought much higher inflation, are a major factor behind that shift. In other words, the Fed’s impact on real rates is far more potent now than in times past.

It gets worse yet. Most observers were not anticipating that expected short rates could fall and stay below -4%. Why hold those assets at all? I don’t have a good answer to that question.

Recently the term structure of interest rates has become inverted, by which it is meant that the short-term rates are higher than the long-term rates. Economists have debated for decades whether such a sign might be a good predictor of a recession. (The theory is that low long-term rates mean that future demand to invest will be low, a bearish sign.) Yet the current data are ambiguous. So not only are real interest rates often hard to predict, but they themselves are not typically clear predictive signals on their own.

There are further conundrums at the link.


Comments for this post are closed