Measuring credit procyclicality database

• Accueil
• Measuring credit procyclicality database

The final database is an unbalanced database consisting of 163 countries over the period 1957Q1-2018Q4. A detailed presentation of data sources, data treatments and trend-cycles decompositions is provided in Bouvatier et al. (2021).

1. Credit series

Main credit series come from the IFS database provided by the IMF. These credit series are called IFS_22d and correspond to credit granted by banking institutions to the private sector. These credit series have been adjusted using alternative discontinued series also from the IFS database; level shifts, gaps and seasonality have been corrected.

The final database also includes credit series from the BIS database. These credit series are called bank_credit and private_credit for narrow and broad credit respectively. However, these credit series are available for 43 countries (Dembiermont et al., 2013) so they have been included only to allow comparison.

2. Measurement units

Credit series are expressed in domestic currency at current price (called XDC), in percentage of GDP (called 770) or in domestic currency at constant price (called RDC). In addition, all credit series include adjustments, indicated by A in the credit series names.

For instance, credit series called bank_credit_XDC_A correspond to narrow credit expressed in domestic currency at current price adjusted by the BIS; credit series called IFS_22d_770_A correspond to bank credit (from the IFS database) expressed in percentage of GDP that have been adjusted (for level shifts, gaps and seasonality).

3. Trend-cycle decompositions

The final database provides a trend-cycle decomposition for all credit series. We rely on 3 methodologies: Hodrick & Prescott (1997) filter with lambda parameter fixed to 400,000 (called HP4), modified Hodrick-Prescott (HP) filter following Kaiser & Maravall (1999, 2001) with lambda parameter fixed to 400,000 (called HPM), and basic Singular Spectrum Analysis (called SSA). All methodologies are implemented in one-sided approach (called os) and in two-sided approach (called ts). In one-sided approach, the first window used to generate the trend-cycle decomposition includes 40 observations (i.e. 10 years). Further, the trend component is called trend and the cyclical component is called gap.

For instance, credit series bank_credit_770_gap_HP4os correspond to the credit gap assessed by one-sided HP filter for the narrow credit-to-GDP ratio provided by the BIS database; credit series IFS_22d_770_gap_SSAts correspond to the credit gap assessed by two-sided basic SSA for adjusted credit-to-GDP ratio computed from the IFS database.

References:

• Bouvatier V., A.L. Delatte & P.N. Rehault, Measuring credit procyclicality: a new database, Discussion Paper DP16519, CEPR, 2021.
• Dembiermont, C., Drehmann, M., & Muksakunratana, S. (2013). How much does the private sector really borrow - a new database for total credit to the private non-financial sector. BIS Quarterly Review, March 2013, 65.81.