Some Reflections on the Interactions between the Bank of England's Forecasts and the MPC's Policy Decisions

This paper makes three main points. First, whereas the Monetary Policy Committee's forecasts of inflation and output growth in the UK are comparatively accurate, they cannot forecast deviations around trend, except at short horizons. Second, this is primarily because they adjust policy, the short-term interest rate, to drive inflation/output back to trend at their forecast horizon. This is not apparent when using a Taylor-rule using ex post forecasts, since these are published after taking account of policy changes. I use a rule of thumb to re-engineer estimates of the ex ante forecasts, upon which the policy decision was based. Also, because of the lengthy lags in the transmission mechanism, Central Bank decision-makers relate their interest decisions, not to current variables, but to forecast values for future inflation and output, with a forward-looking interest rate reaction function of the form: $$i_{t} =a b_{1} E_{t} {+AFw-left( {+AFw-pi _{{t j}} - +AFw-pi *} +AFw-right)} b_{2} E_{t} {+AFw-left( {y_{{t j}} } +AFw-right)}$$ Taking account of ex ante forecasts, with a forward-looking reaction function, gives very different results from the standard Taylor reaction function estimates. Third, the coefficient of reaction to inflation deviations at the forecast horizon has been almost exactly enough to return inflation to trend without need for any further change. So one might expect interest rates to follow (nearly) a random walk. Yet they are strongly auto-correlated. This latter remains a conundrum which requires further research. Copyright International Atlantic Economic Society 2005