Historical Reliability and the Case of Basel III Credit-to-GDP Gaps

36 Pages Posted: 9 Aug 2022 Last revised: 28 May 2024

See all articles by Josefine Quast

Josefine Quast

affiliation not provided to SSRN

Date Written: May 26, 2024


Estimating trend-cycle decompositions to distinguish shorter- from longer-run developments is often of relevance to economic policymakers and for policy design. One case study for such decompositions are Basel III credit-to-GDP gaps which are used to assess whether aggregate credit is excessive or not. Basel III suggested estimates, however, remain historically unreliable beyond endpoint biases because the prescribed detrending method is prone to spurious regression issues. In consequence, official estimates are continuously reevaluated, do not converge to full sample estimates, are associated with ever-changing trend histories, and inconsistent ex post policy evaluation. Data revisions can further aggravate these issues. To illustrate that the historical variability in the derived gap and trend measures can get quite large, I introduce historical reliability bands as an intuitive concept which can also be applied to other indicators. These bands span the smallest to largest value estimated between initial and full sample assessment, accounting for all possible reevaluations in between. In the context of credit-to-GDP gaps, difference-filter-based approaches provide straightforward alternatives that resolve these inconsistencies while maintaining gap dynamics and amplitudes that the regulator seems to deem important.

Keywords: Detrending, reliability, stochastic trends, real-time analysis, robust procedures

JEL Classification: C10, C22, E32, E44, G01, G21, G23

Suggested Citation

Quast, Josefine, Historical Reliability and the Case of Basel III Credit-to-GDP Gaps (May 26, 2024). Available at SSRN: https://ssrn.com/abstract=4177293 or http://dx.doi.org/10.2139/ssrn.4177293

Josefine Quast (Contact Author)

affiliation not provided to SSRN

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Abstract Views
PlumX Metrics