Univariate and multivariate filters to measure the credit gap

Within the framework of the Basel III capital regulation, macroprudential authorities may order the accumulation of countercyclical capital buffers in the period when systemic risks are building up. According to recommendations, it is worth setting the size of the capital buffer on the basis of the magnitude of the credit-to-GDP ratio gap. Therefore, the time series of Hungary's credit-to-GDP ratio is decomposed to trend and cyclical components (credit gap) using four trend filtering methods: univariate Hodrick-Prescott filter, univariate Christiano-Fitzgerald filter, univariate Beveridge-Nelson filter and multivariate Hodrick-Prescott filter. The decomposition was carried out separately for the household and corporate segments. Of the four methods, it is the results of the multivariate Hodrick-Prescott filter, which also uses the information content of other variables, that reflect experts' assessment relating to developments in lending in Hungary the most. In addition, endpoint uncertainty was also the smallest in this case, i.e. the receipt of new data caused the smallest changes in the values estimated for previous periods here.