Research note: A model to predict RBA decisions

Research note: A model to predict RBA policy decisions

See... The RBA Framework

We have developed a model to aid the process of predicting RBA policy decisions. The model estimates the probability, measured in basis points relative to a 25bp tightening or 25bp ease, of a move in any given month. The model's fit is good, with a R2 of 0.76.

The model uses five independent variables, three for the domestic economy (capacity utilization, core inflation and credit growth) and two for the global economy (a global policy rate and JPMorgan's global PMI measure). Interestingly, the fit of the model was inferior when only domestic variables were used.

Back-testing shows that the model correctly predicts RBA decisions on 73 of the 75 occasions covered; that is, 97% of the time. This is vastly superior to the interbank (IB) futures market, which predicted two rate rises that didn't materialize, and only three of the seven hikes since 2003, when the IB contract was launched.

The model accurately predicted eight of the nine rate hikes since 2001 and two of the three eases. The two moves the model failed to predict were the 25bp cut in December 2001 and the 25bp hike in March 2005. On both occasions, though, the model was accumulating basis point probabilities in favour of the move.

The model correctly predicted all 63 times the RBA left policy unchanged, including in April this year when the futures market wrongly priced a high chance of a hike. The model predicts a 25bp rate hike for November 2007, but forecasts that the RBA will cut the cash rate at the end of 2008.

The most powerful driver for predicting RBA behaviour is the global policy rate, followed by the core CPI, domestic capacity utilization, credit growth, and the global manufacturing PMI. The model implies that, in the past, RBA decisions were more influenced by the actions of other central banks than the direction of domestic core inflation.

ENDS

© Scoop Media