The effective federal fund rate plays a key role in the U.S. economy. It is not only the benchmark interest rate underlying financial instruments in the financial market, but also the pivotal factor in pricing of assets in other industry. Therefore, effective Federal Fund Rate has always been the focus of forecasting activities performed by academy and businesses. However, since December 16th, 2008, federal fund rates have been targeted within a range of 0%-0. 25%. Effective Federal Fund rates, which center on the target rate, become less unpredictable. However, as the economy recovering, effective federal fund rate will very likely resume its current role. Given this, it is still of profound significance to forecast fed fund rates.
Effective federal fund rate is determined by the federal fund market, where Financial institutions trade federal fund with each other overnight. Financial institutions and the Federal Open Market Committee (FOMC) are the major market forces that ultimately decide the effective fed fund rate.
There are a lot of forecasting models for effective federal fund rate. Hauwe, Paap and Dijk（2011）used a Bayesian forecasting model for federal funds target decisions using large set of macro economic predictors . Although this model initially is used for federal fund target rate prediction,it can be adapt to predict the effective federal fund rate. This is because effective federal fund rate is forced to stay close to federal fund target rate by the Trading Desk at Federal Reserve Bank of New York, see Taylor (2001). The Trading Desk performs Open Market Operation (OMO) and counteracts large deviations of effective federal fund rate from its target. In the meantime, financial institutions will also react to federal fund target changes immediately, given that they anticipated the OMO will bring the effective federal fund rate to new target rate very soon. Therefore, predicting effective federal fund rate can be roughly substitute for predicting target rate, which is determined by FOMC. Hauwe, Paap and Dijk examined the minutes of FOMC meetings and discovered that a large set of macro-economic indicators is used in determining the target rates.
Thorton (1998) proposed a model using the 30 days federal fund future rate as predictor. Federal fund futures rate can help predicting the target rates because futures market participants make commitments that are contingent on what they believe the federal funds rate will be and look to factors they believe will influence its course. The Fed targets the funds rate, and the overnight federal funds rate stays close, on average, to the Fed’s target. Hence, the federal funds futures rate naturally embodies the market’s expectation of what the Fed will do.
Theory suggests that model averaging can improve predictability. Ravazzolo, Dijk, and Verbeek (2007) discovered that averaging yield higher predictive gains than selecting the best model, and time varying model weights have higher statistical and economic values than other averaging schemes. Therefore, we can take the average of the above two models.