Objetive:

To quantify the uncertainty associated to the estimation of the current value of the business cycle.

Methodology:

The Baxter and King filter is applied to an time series expanded with forecasts, assuming a random walk, and confidence intervals are build for the cyclical component both analytically and using bootstrap.

Limitations:

The quality of the confidence 1 intervals deteriorates when there is a structural break in the time series.

Originality:

The original contribution of this article is a proposed methodology to jointly determined the current phase of the business cycle and the uncertainty associated with such measurement.

Conclusions:

For Costa Rica, the proposed method provides estimations of the business cycle that are less sensitive to new data points than the Hodrick and Prescott filter, and simulations show that the proposed confidence intervals provide the specified nominal confidence.