Information sharing, credit booms, and financial stability Samuel Gu - PowerPoint PPT Presentation

Information sharing, credit booms, and financial stability Samuel Gu´ erineau CERDI, Universit´ e d’Auvergne Florian L´ eon CREA, University of Luxembourg ESRC-DFID Growth Research Program Financial Volatility, Macroprudential Regulation and Economic Growth in Low-Income Countries (Grant No. ES/L012022/1) S. Gu´ erineau and F. L´ eon Info sharing, credit booms, and financial stability ESRC-DFID Project 1 / 28

Outline Introduction 1 Data and variables 2 Empirical model 3 Results 4 Conclusion 5 S. Gu´ erineau and F. L´ eon Info sharing, credit booms, and financial stability ESRC-DFID Project 2 / 28

Introduction Introduction : Motivation This paper motivated by several strands of literatures related to the vulnerability of financial systems Credit boom and financial fragility 1 ◮ Credit boom is a main driver of financial crisis episodes ◮ Some questions remain • How can we limit the occurrence of credit booms ? • How can we alleviate their detrimental impact ? Recent development of information sharing 2 ◮ Development of IS around the World IS in the world ◮ IS have been created to favor credit access ◮ But they could affect financial fragility Few papers on financial fragility in developing countries 3 ◮ Episodes of financial fragility are less frequent in low-income countries ◮ But they may induce profound consequences S. Gu´ erineau and F. L´ eon Info sharing, credit booms, and financial stability ESRC-DFID Project 3 / 28

Introduction Introduction : Conceptual framework Existing literature focuses on the direct impact of information sharing (IS) Theory : 1 • IS ց fragility : Moral hazard, adverse selection, over-borrowing • IS ր fragility : Credit composition Empirical papers 2 • IS tend to strength financial stability (micro and macro evidence) • But how ? We study its indirect effect trough credit booms (1) Financial Information (3) fragility sharing (2) (3) Credit booms S. Gu´ erineau and F. L´ eon Info sharing, credit booms, and financial stability ESRC-DFID Project 4 / 28

Introduction Introduction : An overview We combine data from 159 countries over the period 2008-2014 divided in two groups 80 ”advanced” economies (High-income and upper-middle income countries) 1 79 ”developing” economies (Low-income and lower-middle income countries) 2 What we do ? Net effect of IS on financial fragility 1 Transmission channels 2 Attenuation effect of the detrimental effect of credit boom 1 Impact on the occurrence of credit booms 2 Main results IS reduces financial fragility ; no distinction between developing and other 1 countries Depth of IS limits the occurrence of credit booms (but coverage does not 2 matter) IS alleviates the detrimental effect of credit booms but only in advanced 3 economies S. Gu´ erineau and F. L´ eon Info sharing, credit booms, and financial stability ESRC-DFID Project 5 / 28

Data and variables Data and variables Datasets ◮ Bankscope (financial fragility) ◮ Doing Business (information sharing) ◮ WDI and IFS (other variables) Sample - 159 countries including : ◮ 79 developing countries (GNI per capita < US$ 4,125) ◮ 80 emerging and developed countries (GNI pc ≥ US$ 4,125) Period : 2008-2014 S. Gu´ erineau and F. L´ eon Info sharing, credit booms, and financial stability ESRC-DFID Project 6 / 28

Data and variables Data and variables Dependent variable : Financial fragility Measurement ∆( NPLs Loans ) ≥ 3 points ◮ Ratio of NPLs to loans is computed at the national level (weighted average) ◮ Authors’ calculation using Bankscope database Advantages ◮ Available for a large number of countries, including low income countries ◮ Identify episodes that were not transformed into financial crises ◮ Why do not we use financial crises dataset ? • Limited number of financial crises since 2005 in low-income countries (data before 2005 cannot be exploited due to the lack of data on information sharing mechanisms) S. Gu´ erineau and F. L´ eon Info sharing, credit booms, and financial stability ESRC-DFID Project 7 / 28

Data and variables Data and variables Independent variables Credit boom (CB) ◮ 2 criteria are used to define a credit boom An increase of the ratio of credit to GDP during at least three consecutive years 1 The average of increases is 5 percentage points by year 2 ◮ Data are extracted from WDI Information sharing (IS) ◮ Two alternative measures Depth of credit information 1 Coverage of credit registries and credit bureaus 2 ◮ Authors’ calculation using Doing Business data Control variables (X) Macroeconomic factors 1 • GDP per capita, growth, inflation, Exchange rate vol Financial factors 2 • PC/GDP, capital inflows, market concentration S. Gu´ erineau and F. L´ eon Info sharing, credit booms, and financial stability ESRC-DFID Project 8 / 28

Empirical model Baseline model Empirical model 1 st step : Baseline model (net effect of IS) Pr ( BSD it ) = α + β IS it + Γ X it + ε it Where ◮ BSD it : dummy equals to 1 if a country i experienced an episode of financial fragilityin year t ◮ IS it : Indicator of credit information sharing (depth and coverage) ◮ X it : Control variables (including time dummies) Method ◮ Econometric method : Random-effect probit ◮ Binary nature of dependent variable ◮ Random effect : Control for unobserved heterogeneity Expected result : CIS reduces financial fragility ( β < 0) S. Gu´ erineau and F. L´ eon Info sharing, credit booms, and financial stability ESRC-DFID Project 9 / 28

Empirical model Baseline model Empirical model 1 st step : Baseline model (net effect of IS) Pr ( BSD it ) = α + β IS it + Γ X it + ε it Where ◮ BSD it : dummy equals to 1 if a country i experienced an episode of financial fragilityin year t ◮ IS it : Indicator of credit information sharing (depth and coverage) ◮ X it : Control variables (including time dummies) Method ◮ Econometric method : Random-effect probit ◮ Binary nature of dependent variable ◮ Random effect : Control for unobserved heterogeneity Expected result : IS reduces financial fragility ( β < 0) S. Gu´ erineau and F. L´ eon Info sharing, credit booms, and financial stability ESRC-DFID Project 10 / 28

Empirical model Transmission channels Empirical model 2 nd step : Transmission channels Inclusion of credit booms (CB) 1 Pr ( BSD it ) = α + β IS it + δ CB it + Γ X it + ε it ◮ Expected results : β < 0 and δ > 0 Interaction between IS and CB 2 Pr ( BSD it ) = α + β IS it + δ CB it + γ IS it ∗ CB it + Γ X it + ε it ◮ Expected results : β < 0, δ > 0 and gamma < 0 Determinants of CB 3 Pr ( CB it ) = α ′ + β ′ IS it + Γ ′ X it + ε it ◮ Expected result : Sign of β ′ can be positive or negative S. Gu´ erineau and F. L´ eon Info sharing, credit booms, and financial stability ESRC-DFID Project 11 / 28

Empirical model Transmission channels Empirical model 2 nd step : Transmission channels Inclusion of credit booms (CB) 1 Pr ( BSD it ) = α + β IS it + δ CB it + Γ X it + ε it ◮ Expected results : β < 0 and δ > 0 Interaction between IS and CB 2 Pr ( BSD it ) = α + β IS it + δ CB it + γ IS it ∗ CB it + Γ X it + ε it ◮ Expected results : β < 0, δ > 0 and gamma < 0 Determinants of CB 3 Pr ( CB it ) = α ′ + β ′ IS it + Γ ′ X it + ε it ◮ Expected result : Sign of β ′ can be positive or negative S. Gu´ erineau and F. L´ eon Info sharing, credit booms, and financial stability ESRC-DFID Project 12 / 28

Empirical model Transmission channels Empirical model 2 nd step : Transmission channels Inclusion of credit booms (CB) 1 Pr ( BSD it ) = α + β IS it + δ CB it + Γ X it + ε it ◮ Expected results : β < 0 and δ > 0 Interaction between IS and CB 2 Pr ( BSD it ) = α + β IS it + δ CB it + γ IS it ∗ CB it + Γ X it + ε it ◮ Expected results : β < 0, δ > 0 and gamma < 0 Determinants of CB 3 Pr ( CB it ) = α ′ + β ′ IS it + Γ ′ X it + ε it ◮ Expected result : Sign of β ′ can be positive or negative S. Gu´ erineau and F. L´ eon Info sharing, credit booms, and financial stability ESRC-DFID Project 13 / 28

Empirical model Transmission channels Empirical model 2 nd step : Transmission channels Inclusion of credit booms (CB) 1 Pr ( BSD it ) = α + β IS it + δ CB it + Γ X it + ε it ◮ Expected results : β < 0 and δ > 0 Interaction between IS and CB 2 Pr ( BSD it ) = α + β IS it + δ CB it + γ IS it ∗ CB it + Γ X it + ε it ◮ Expected results : β < 0, δ > 0 and gamma < 0 Determinants of CB 3 Pr ( CB it ) = α ′ + β ′ IS it + Γ ′ X it + ε it ◮ Expected result : Sign of β ′ can be positive or negative S. Gu´ erineau and F. L´ eon Info sharing, credit booms, and financial stability ESRC-DFID Project 14 / 28