Systematic Relative Strength

The topic of interest rates is of concern to investors for a couple of reasons. Savers are interested in finding out when interest rates might rise and they might earn more than 0% on their accumulated capital, and bond investors would like some kind of early warning if there is trouble ahead. I’ve seen lots of opinions on this, and they’ve mostly all been wrong. My personal answer to the question about when interest rates would rise would have been something like “2009,” which explains why 1) I am not a prominent interest rate forecaster and why 2) everyone should use a systematic investment process (as we do)!

To attempt to answer the interest rate question, Eddy Elfenbein of Crossing Wall Street stepped in in a way I particularly admire—with actual data, not just opinions. He showed two competing interest rate models developed by Greg Mankiw at Harvard and Paul Krugman at Princeton. Although the coefficients are slightly different, it turns out that their models are pretty similar. I’ve shown the two graphs below.

Mankiw Interest Rate Model

Krugman Interest Rate Model

Source: Crossing Wall Street (click on images to enlarge)

Up until the recent financial crisis, the forecast fit the data rather well for both models. That is to be expected, since the model is derived from the data and each modeler is searching for the best fit equation. Both models show that, given the past behavior of interest rates in relation to the variables they use (core inflation and unemployment), current interest rates should be negative! The Fed seems to be coping with this situation by holding rates at zero and using quantitative easing to simulate negative rates.

What will make these models suggest that interest rates should start to move higher? If core inflation increases and the unemployment rate begins to decline, both of these models would call for higher rates. For Krugman’s model, for example, core inflation would have to rise to 2.5% (from the current 1.8% level; I used PCE excluding food and energy), while unemployment would need to decline to 7.5% from 7.8%. (Or it could be a different combination that was mathematically equivalent.) For Mankiw’s model to call for higher rates, only a slight increase in inflation or a drop in unemployment would be needed.

If the economy continues to plug along with slow growth, low inflation, and relatively high unemployment, both of these models would continue to suggest that negative rates are needed to revive the economy.

So much for theory. In reality, many considerations go into setting the Fed Funds rate. Watching the behavior of inflation and unemployment probably enters into it, but I’m guessing the Fed is examining other data as well. From the outside, perhaps the best thing we can do is monitor the relative strength of bonds versus other asset classes to get a handle on the expectation for interest rates.