Transcript Optimal financial structures and development

OPTIMAL FINANCIAL
STRUCTURES AND DEVELOPMENT
The evolving importance of banks and markets
Asli Demirguc-Kunt, Erik Feyen, and Ross Levine
Question:
 Does financial structure—the mixture of
intermediaries and markets—matter for
economic development?
One answer is “no.”
 Banks provide services that promote development
 Markets provide services that promote development
 The mixture per se does not matter
 Demirguc-Kunt and Levine (2001)
 La Porta et al (1997, 1998, 1999)
Another answer is “yes.”
 Banks & markets provide different services
 Economies at different stages of development
 require different mixtures of these services
 require different financial structures
 Thus, economic growth alters
 the optimal mixture of financial services
 And hence the optimal financial structure
 Deviations from the optimum hurt the appropriate
mixture of financial services, curtailing economic activity
 Lin, et al. (2011)
 Allen and Gale (1995, 2000), Boyd and Smith (1998), Morck and
Nakamura (1999), Weinstein and Yafeh (1998)
 Sort of like electrolytes in the body
This paper reassesses
① the evolving importance of banks and markets
during the process of economic development.
② the association between financial structure and
economic development.
 Rather than focus on financial structure per se, as in
Demirguc-Kunt and Levine (2001)
 We focus on deviations from an estimated optimum
that evolves during the process of development.
What’s different from past
X-country studies?
 New:
Quantile regressions
2. Examine deviations from optimal financial
structure and the association with development
1.
 But:
 We do not nail down a causal mechanism
 We do not derive sharp policy recommendations,
though the work is policy relevant
Quantile regressions
EVOLVING IMPORTANCE OF BANKS &
MARKETS
Specification quantile regressions
 Yc,t = Xc,t + 1Pc,t + 2Sc,t + c,t






Yc,t = log real per capita GDP
Pc,t = private credit
Sc,t = securities capitalization
Xc,t = Yc,0 , Gc,t, Tradec,t, c,t, Schoolc,t, T-FEs.
72 countries (including all available OECD)
Five-year periods, 1980-2008, max 6 obs per country
 Estimation
 OLS … yields “” for the average country
 Quantiles …
 obtain “” for each percentile (quantile) of Y
 How does “” evolve with economic development?
150
.01
100
.008
.006
0
20
40
60
Quantiles of Log GDP per capita
80
100
50
.004
0
.002
0
.005
20
.01
40
.015
60
.02
80
100
.025
Quantile coefficients for Private credit &
Securities Market Cap.
0
20
40
60
Quantiles of Log GDP per capita
80
100
Priv. Cred. Coeff. Quant. Reg.(LHS)
OLS estimate (LHS)
Sec. Mkt. Coeff. Quant. Reg.(LHS)
OLS estimate (LHS)
Fitted values
Private Credit / GDP (RHS)
Fitted values
Securities Market / GDP (RHS)
0
20
40
60
Quantiles of Log GDP per capita
80
100
150
.0015
100
.001
.0005
0
-.0005
0
-.001
0
.0035
20
50
.004
40
60
.0045
80
.005
100
Quantile coefficients for Private credit &
Securities Market Cap. (standard controls)
0
20
40
60
Quantiles of Log GDP per capita
80
100
Priv. Cred. Coeff. Quant. Reg.(LHS)
OLS estimate
Sec. Mkt. Coeff. Quant. Reg.(LHS)
OLS estimate
Fitted values
Private Credit / GDP (RHS)
Fitted values
Securities Market / GDP (RHS)
Evolving importance of banks & markets
 Evidence consistent with Allen/Gale & Boyd/Smith
 Development increases demand for services provided
by securities markets more those provided by banks
 Specifically:
 “Quantities” of both increase
 Sensitivity of Y to banks falls
 Sensitivity of Y to markets increases
 Evidence is inconsistent with …
 Only supply of finance rising with development
 Roles of banks & markets following similar paths
Economic magnitude of banks and markets
at different levels of development
Percent change in GDP per capita of 1 SD increase
in Private credit or Securities market capitalization
(controls included)
%
%
25.0
20.4
20.0
15.9
15.4
15.0
Private Credit to GDP
Sec. Market Capitalization to GDP
10.0
6.9
5.0
3.4
0.7
0.0
25th pctile
50th pctile
75th pctile
Financial structure gap
OPTIMAL FINANCIAL STRUCTURE
AND DEVELOPMENT
Computing the financial structure gap
Financial structure gapc,t  Ln( Actual structurec,t – Optimal structurec(y) )
 Actual structurec,t = Bank credit / Securities market capitalization
Computing the financial structure gap
Financial structure gapc,t  Ln( Actual structurec,t – Optimal structurec(y) )
 To compute Optimal structurec(y)
1.
Select benchmark countries (high income OECD) with few
impediments to achieving the optimal financial structure
2.
For benchmark countries, estimate:
Actual structurec,t = a*Yc,t + B*Xc,t+ uc,t
Xc,t includes: LO, Equator, PopSize, PopDen, & Natural Res.


3.
For ALL countries, use these estimated parameters to construct
“Optimal structurec(y)”
Financial structure gap (FSG) and
economic development
 Yc,t = Xc,t + bBc,t + sSc,t + FSGc,t + c,t
 Sample excludes high income OECD countries
Financial structure gap (FSG) and
economic development
Financial Structure Gap
-0.04*** -0.02*
[0.00]
[0.10]
Private Credit to GDP
7.29*** 5.06*** 4.67*** 4.52***
[0.00]
[0.00]
[0.00]
[0.00]
Securities Market Capitalization to GDP
2.99***
[0.00]
0.38
[0.45]
-0.02*
[0.07]
0.11
[0.81]
Financial Structure Ratio
-0.03*
[0.08]
0.23
[0.65]
0.02
[0.45]
Standard controls
Country-fixed effects
Time-fixed effects
No
Yes
No
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Observations
Adjusted R-squared
Countries
263
0.56
72
263
0.75
72
238
0.76
64
238
0.76
64
Structure matters …
 The financial structure gap is negatively
associated with economic activity, controlling for
 Bank development
 Market development
 Financial structure
 Time FEs
 Country FEs
 & Time-varying country characteristics
Economic magnitude
 With only country FEs as controls:
 1 in FSG  6% in Y
 With all controls:
 1 in FSG  3% in Y
-.07
-.06
-.05
-.04
-.03
-.02
FSG is NOT less important at lower levels
of development
0
20
40
60
Quantiles of Log GDP per capita
Lgap Coeff. Quant. Reg.
Fitted values
80
OLS estimate
100
Empirical findings:
 Financial structure matters
 Deviations from the optimal structure are associated with less
economic activity
 The sign of deviation does not matter
 Earlier studies did not deviation from optimum
 As Y , countries require relatively more market services
 Countries become more bank and market based
 But
 The sensitivity of output to bank development falls
 The sensitivity of output to market development increases