Identifying Credit Supply and Demand Factors in the Euro Area (very preliminary, don t quote or circulate) Sandra Eickmeier (BBk, CAMA), Giulio Nicoletti (ECB), Esteban Prieto (BBk, IWH)
Motivation and contributions Central banks need to assess credit supply and demand in real time. Furthermore, understanding the comovement of (or heterogeneity in) credit supply and demand in the euro area important for the ECB. Credit supply and demand are unobserved. For the euro area, central banks rely on survey measures, which are based on subjective assessments by bankers (BLS) or non-financial corporations (SAFE) and which are available on a quarterly/bi-annual base. Our contributions: - We propose bank credit supply and demand measures for the euro area based on actual monthly data. - We assess heterogeneity in credit supply and demand and explore to what extent it depends on individual bank and loan characteristics. Seite 2
Data Bank-level data on (real) new loans and loan rates to households (hh) and nonfinancial corporations (nfc). 84 banks from 11 countries July 2007 - September 2015 Source: ibsi/imir Loan spreads defined as loan rates over rates of Bund of same maturity Seite 3
Methodology Extract (3-4) factors with principal components from loans and spreads. Loadings are estimated for loan rates as well. Impose sign restrictions on factor loadings, following Eickmeier, Gambacorta and Hofmann (2015, EER): - Credit supply: loans (+), loan rate (-), loan spread (-) - Credit demand: loans (+), loan rate (-) - Restrictions hold for (unweighted) means of loadings across banks and at least 50-70% of the banks. Related to - empirical work using factor models with zero restrictions on loadings - SVARs using sign restrictions Discussion: - Factors incorporate any shifter of credit supply and demand - Dual meaning of loadings: weights / effects (here: changes in factors have specific effects on loans, loan rates and spreads) - Factors orthogonal by construction (but can influence each other with lags) Seite 4
Roadmap Commonality in banking data in the euro area Baseline: euro-area credit supply and demand factors Robustness analysis Comparison with Bank Lending Survey (BLS) and Survey on the Access to Finance of Enterprises (SAFE) data Heterogeneity Conclusion and outlook Seite 5
Commonality in banking data in the euro area Notable commonality in euro-area credit markets Higher for households than for non-financial corporations, for south than for north, and for large banks than for small banks. Seite 6
Credit demand and supply factors non-financial corporations (point estimates (black) and Median Target (red)) Seite 7
Credit demand and supply factors households (point estimates (black) and Median Target (red)) Seite 8
Results so far Sharp worsening of credit supply conditions during the GFC and the sovereign debt crisis. Credit supply boom between 2009 and 2011 (perhaps as a consequence of low interest rates) Credit demand elevated at beginning of GFC, possibly due to increased financing needs. Downward movement thereafter. At the end of the sample, positive credit supply conditions and weak credit demand conditions. Exception: (weak and uncertain) recovery of household credit demand. Moreover: - Credit supply factors more precisely estimated - Easier to find valid rotations for households than for non-financial corporations greater homogeneity among household credit markets Seite 9
Robustness - Credit demand and supply factors non-financial corporations (point estimates (black) and Median Target (red)) Seite 10
Robustness - Credit demand and supply factors households (point estimates (black) and Median Target (red)) Seite 11
Robustness (end of sample) - Credit demand and supply factors non-financial corporations (point estimates (black) and Median Target (red)) Seite 12
Robustness (end of sample) - Credit demand and supply factors households (point estimates (black) and Median Target (red)) Seite 13
Summary of robustness analysis Results quite robust. - More model uncertainty with more factors No major revision with incoming data good indicators Seite 14
How do results compare with survey data from the BLS and the SAFE? Conceptional differences - Credit supply and demand correlated in survey data, whereas orthogonality is an identifying assumption in our approach. - Survey data available at lower frequency (quarterly: BLS, bi-annual: SAFE). - Our factors based on actual data, survey data based on subjective assessments. Seite 15
Comparison with survey measures - Credit supply and demand non-financial corporations Seite 16
Comparison with survey measures - Credit supply and demand households Notes: Right panels: Solid BLS house purchases, dotted BLS consumer loans Seite 17
How do results compare with survey data from the BLS and the SAFE? Conceptional differences - Credit supply and demand correlated in survey data, whereas orthogonality is an identifying assumption in our approach. - Survey data available at lower frequency (quarterly: BLS, bi-annual: SAFE). - Our factors based on actual data, survey data based on subjective assessments. Findings - Supply factors broadly similar to survey measures, but our approach indicates deeper trough of credit supply during the sovereign debt crisis, whereas surveys measures indicate that the GFC has been more severe. - Credit demand factors consistent with SAFE. BLS credit demand measures very different! Seite 18
Heterogeneity - Credit demand and supply factors non-financial corporations (point estimates (black) and Median Target (red)) Seite 19
Heterogeneity - Credit demand and supply factors households (point estimates (black) and Median Target (red)) Seite 20
Heterogeneity - results Regional differences - Deeper drop in credit supply in south (where economies and banks were hit more strongly) than in north during the sovereign debt crisis. - Stronger recovery of business credit supply in the south than in the north at end of the sample, whereas credit demand above average in north, but not in south (where labor market and income situation is still weak). Banks with large government bond exposure have increased their credit supply after sovereign debt crisis, which then, however, declined and did not recover at the end of the sample period. Size of banks does not seems to matter. First results for loan maturities suggest, among others, high supply and demand of only longer-term loans at the end of the sample (risk taking?, cheap funding of longer-term investments?) [graphs not yet included in the presentation] Results also help understanding the survey measures. Remark: Characteristics barely correlated. Seite 21
Conclusion and outlook We identify credit supply and demand factors in the euro area by exploring actual, monthly individual bank data, which allows us to - assess the temporal evolution of credit supply and demand conditions in the euro area, - test to what extent the evolution of credit supply/demand depends on different bank/loan characteristics. Evolution of credit supply in line with surveys, credit demand in line with the SAFE, but not with the BLS. Recently, credit supply has picked up in most countries. Credit demand is still weak. But there are differences across loan maturities and regions. Outlook - Interaction of credit supply and demand and monetary policy Seite 22
Appendix Seite 23
(Selected) individual countries - Credit demand and supply factors non-financial corporations (point estimates (black) and Median Target (red)) Seite 24
(Selected) individual countries - Credit demand and supply factors households (point estimates (black) and Median Target (red)) Seite 25