Potential output in the CNB's forecasting system
In the CNB's forecasting system, potential (or equilibrium) output is taken to mean the level of real output at which real economic activity does not generate upward or downward pressures on inflation. Thus, the estimate of equilibrium output is not associated with any specific level of inflation, but is only linked with the absence of pressures for a change in inflation.
Like the equilibrium levels of other macroeconomic variables, the equilibrium level of output is not an observable and measurable variable. A whole range of different methods can be used to estimate equilibrium output. The CNB's main method of estimating potential output is closely linked with its definition. The estimate uses a multi-equation model, which has the advantage that it includes other information on the labour market and inflation in addition to data directly concerning output. In this model, potential output, the non-accelerating inflation rate of unemployment, the effect of monetary policy and several other parameters are determined simultaneously, which enables their interactions to be taken into account. For example, if inflation accelerates without the presence of a supply shock and a fiscal impulse, it is likely that monetary policy was easy in the past and, as a result, there was an excess of demand over production. A recursive algorithm known as the Kalman filter is used to generate the estimate in the model.
As an alternative approach to determining potential output, the CNB also uses an estimate based on the production function. This looks exclusively at the supply side of the economy, identifying factors of current and future growth in real economic activity on the supply side (technological progress, labour and capital). However, this method of estimating potential output is affected by the problem of low robustness of the estimate of the final values using the univariate Hodrick-P rescott filter.
Charts 1 and 2 compare the estimates of potential output obtained using these two methods. Compared to the Kalman filter, the production function approach leads to a slightly lower estimate of the level of potential output, but to roughly the same estimate of the current growth rate of potential output, at 4%. The October forecast projects the same rate of growth going forward.