Photographing Phantom Invisible Bond Vigilantes

Robert Waldmann

A specter is haunting Paul Krugman — it is the specter of apparently sophisticated forecasters who predict a huge spike in US long term bond rates in the near future. He notes that most investors can’t believe this or rates would already be high. He also notes that some of those who are predicting a spike seem to be top advisors somewhere in the Obama administration.

He is puzzled by this phenomenon. I just note that it is not new at all. Basically such apparently sophisticated people predicting a spike in long term US interest rates are (almost) always with us. So far they have always been wrong. This issue is extensively researched by Nazaria Solferino et al in an article forthcoming in the journal of forecasting.

The phantom invisible bond vigilantes are not a new phenomenon. Consider Blue Chip Forecasts TM an organization which aggregates forecasts of, among other things, rates on Treasury bonds.

Usually there are some forecasters who predict that long term US interest rates will spike in the near future. There isn’t one such chicken little in the blue chip TM sample every month, but there is most months.

The interesting thing is that, so far, they have always been wrong. In an article forthcoming in The Journal of Forecasting Nazaria Solferino and I show that is it is possible to use a very simple formula to identify 1082 forecasts of US 30 year interest rates which are too high without one single solitary error so far. That’s a record of 1082 to zero (so far). The vast majority of the forecasters in the sample made such an incorrect prediction of a long term interest rate spike at least once. See
http://ideas.repec.org/p/rtv/ceisrp/135.html

There is no such fat tail of always wrong bond doves and almost no such 100% (so far) identifiable errors in forecasts of short and medium term interest rates.

Basically, phantom bond vigilantes are always with us. Almost always someone is convinced that the US is heading to high sustained inflation and that everyone else will agree a few months from now. It is an astounding feature of forecast data.