Inflation Forecasting Humour

Gerard MacDonell has an amusing yet disturbing article at Beinn Bhiorach -- "Yeah, but if you control for the Jupiter effect." (Disclaimer: Gerard hired me in my first job in finance.) He discusses which prices indices economists should use to make the inflation rate higher than it is, in order to meet the bias that inflation is high, and going higher.

In addition to some useful technical insights, one might wonder -- aren't economists supposed to be forecasting what will happen in the economy, and not mining the data in order to find factoids to justify a predetermined conclusion? (I used to have such naive thoughts, and I imagine that I am not the only one.)

In any event, such behaviour helps explain how the economist consensus can do things like systematically over-predict future interest rates. (If you only read annual forecasts, and did not pay much attention to the levels of predicted yields, you would believe that 2016 is the seventh year of a bond bear market.) In addition to making a belief "rational expectations" even stupider than one would expect from theory, it calls into question any modelling technique that relies on consensus forecasts as an input. (You know that fancy affine term structure model that Fed researchers have been using to prove any number of things about the term premium? Guess what.*)

Footnote:

* "It is straightforward to include survey data within this Kalman filter framework. The model uses monthly data on the six-month and twelve-month-ahead forecasts of the three-month T-Bill yield from Blue Chip Financial Forecasts and semiannual data on the average expected three-month T-Bill yield from six to eleven years hence, also from Blue Chip. In weeks where these surveys are observed, they are incorporated in the measurement equation..." Kim and Wright https://www.federalreserve.gov/pubs/feds/2005/200533/200533pap.pdf

(c) Brian Romanchuk 2016