MICROECONOMIC MODIFICATION OF AN INDUSTRY …€¦ · microeconomic modification of an industry-wide boone indicator: new estimates of the market power of russian banks april 8, 2015;

MICROECONOMIC MODIFICATION OF AN INDUSTRY-WIDE

BOONE INDICATOR: NEW ESTIMATES OF THE MARKET

POWER OF RUSSIAN BANKS

April 8, 2015; Moscow, Russia

XVI April International Academic Conference on Economic and Social Development

Mikhail Mamonov,

Center for Macroeconomic Analysis and Short-term Forecasting (CMASF) at the Russian Academy of Sciences – Institute for Economic Forecasting;

National Research University “Higher School of Economics”

Motivation

The recent multi-country research on non-linear relationship between the bank market power and risk exposure provides conflicting results:

(Berger et al., 2009; Beck et al., 2013): the inverse U-shaped form if Z-score=f(Lerner index; Lerner index2) whereas

(Tabak et al., 2012): the U-shaped form if Z-score=f(Boone; Boone2);

This could be a problem of measuring market power on the bank level rather than a problem of samples’ differences

The literature suggests only the Lerner index for the bank-level estimation of banks’ market power so far

Problems with the Lerner index (price mark-up over marginal costs): (1) based on averaged price rather than on actual price; (2) catches price competition, ignores quality competition; hence, it cannot describe competition completely

If so, at least one more competition indicator on the bank level needed

Introducing a new indicator: creating smth. new or modifying smth. old?

2

Literature review: just modification or smth. “new”?

Boone (2008), EJ: the “efficiency-structure” hypothesis of Demsetz (1973) Market share = f(Marginal costs…) a new competition proxy for industry

Recent contributed studies on the evaluation of bank-level market power:

1. Carbo, Humphrey , Maudos, Molyneux (2009), JIMF: translog revenue function with the interactions of input prices H-statistic

Shortcoming: other BSF (risk preferences, assets/liab. composition, outputs) are assumed to have no effect on the heterogeneity of H-statistic

2. Bolt, Humphrey (2010), JBF: estimation of competition frontier via SFA

3. Delis, Tsionas (2009), JBF: local optimization of the conjectural parameter (CP) from the Bresnahan model the first bank-level modification of CP

4. Brissmis, Delis (2011), JOR : local optimization a more broad alternative for H-statistic as compared to the CHMM(2009)

5. Delis (2012), JDE: local optimization of the Boone equation the first bank-level modification of the Boone indicator

Shortcomings of 3-5: bank-level heterogeneities in each case «fall from sky»

Beck, De Jonghe, Schepens (2013), JFI: Z-score=f(Lerner, Lerner*Institutions…)

3

Contribution

1. Propose a new version of bank-level Boone indicator (BI)

The external variables approach in a Beck et al. (2013)’s manner: BI is a linear function of bank-specific factors (business models – retail vs. corp. lending, loans vs. non-lending activities, risk profiles; and more);

Define and estimate this function within static panel framework using 2-step GMM;

It is an alternative to the Delis (2012)’s version of bank-level BI obtained through local optimization technique;

Use quarterly data bank-level for Russia (single country study) whereas Delis (2012) exploits Bankskope yearly data on 84 banking systems worldwide;

Contrasts to the bank-level H-statistics obtained through the internal variables approach by Carbo et al. (2009);

2. Divide the credit market into 4 niches by intersecting loan quality indicator with the share of retail loans in total loans. Analyze the BI values of:

largest banks both within and across these niches;

the other banks only within these niches, not across.

3. Compare the BI values with the Lerner indices estimated for Russian banks and find no conflict between them as opposed to the recent research

4

Methodology 1/3

Specify the following translog operating cost function (similar to Turk Ariss, 2010; Fiordelisi et al., 2011):

where for bank i at quarter t OC – operating costs; Yh – h-th output (j=1…3 – commercial loans; deposits; fee and commission); Pm – m-th factor input price (m=1…3 – funds; personnel; physical capital); EQ – equity capital as a netput to control for risk preferences

Estimate it using ML (within SFA), OLS and GMM with currency and securities revaluations being dropped from or kept in OC – obtain 6 alternatives for marginal costs (j=1…6):

5

itq

r q

itrrq

m

itmmitl

k j

itkkl

h

ithhit PPPYYYOC ,

3

1

3

1

,

3

1

,,

3

1

3

1

,

3

1

,0 lnln2

1lnlnln

2

1lnln

2

21

2

21

3

1

,

3

1

,,

3

1

3

1

, lnlnlnlnln itit

m

itpm

h

ithhitu

s u

itssu EQEQTTTPTYPY

itit

m

ititmm

h

itithh EQTEQPEQY

lnlnlnlnln3

1

,

3

1

,

LNS

it

it

j

LNS

it

it

j

LNS

it

itLNS

itjY

OC

Y

OC

Y

OCMC

ln

ln,

Methodology 2/3

A set of multivariate modifications of the market shares equation:

where for bank i at quarter t MS – market share; MCj – j-th proxy for marginal cost (j=1…6); Xk – k-th bank-specific factor; DUMq – a dummy for bank group (q=1…2); Cycle – macro control: a deviation of loans-to-GDP ratio from its HP-filtered values

2-step GMM estimator is employed

The underlying bank-level Boone indicator is:

The more negative values – the stronger a competition on quantity

The more positive values – the stronger a competition on quality

6

K

k

itkkj

K

k

itk

LNS

itjkj

LNS

itjjtjijit XXMCMCMS1

,,

1

,,,,,, lnlnlnlnln

itjtjt

LNS

itjj

q

qqj

q

q

LNS

itjqj CycleCycleMCDUMDUMMC ,,

2

1

,

2

1

,, lnln

tj

q

qqj

K

k

itkkjjLNS

itj

ititj CycleDUMX

MC

MSBOONE

2

1

,

1

,,

,

, lnln

ln

Methodology 3/3

Compare the Boone indicator (competition on quantity or in quality) with the Lerner index (in Koetter et al. (2012) modification):

vs.

where for bank i at quarter t r – Interest income from Loans / Total loans; AFR – Interest expenses / Total Funds; MCj – j-th proxy for marginal cost (j=1…6)

7

LNS

it

LNS

itjit

LNS

it

itjr

MCAFRrLerner

,

,

tj

q

qqj

K

k

itkkjjLNS

itj

ititj CycleDUMX

MC

RBOONE

2

1

,

1

,,

,

, lnln

ln

Data

1. Bank-specific factors (BSF): The Bank of Russia web-site (www.cbr.ru) – monthly balance sheets of banks (Form 101); – quarterly profit and loss accounts (Form 102).

2. Macroeconomic controls (MACRO): The Federal State Statistics Service web-site (www.gks.ru)

3. Time period: Q1 2005 – Q4 2013 (40 quarters)

4. Number of banks (depending on the quarter): – in original sample: up to 1254;

– in adjusted sample: up to 971 after excluding the observations below 1st and above 99th percentiles on relative indicators. It provides up to 21446 observations

8

Preliminary estimation results: univariate Boone (Marginal costs obtained from SFA; revaluations were dropped from operating costs)

9

Outcome: omitted variables do matter ⇒ multivariate Boone indicator needed

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.45

0.50

-0.2

7

-0.2

5

-0.2

2

-0.1

9

-0.1

7

-0.1

4

-0.1

2

-0.0

9

-0.0

6

-0.0

4

-0.0

1

0.0

2

0.0

4

0.0

7

0.0

9

0.1

2

0.1

5

0.1

7

0.2

0

1. The ratio of retail loans in total loans, %

2. Loans-to-assets ratio, %

3. The share of overdue loans in total loans, %

4. Loans-to-deposits ratio (LTD), %

5. The share of reserve assets in total assets, %

6. Loan loss reserves to interest income ratio, %

7. Net deposits to total assets ratio, %

Bank-level Boone indicator (market for loans)

% o

f to

tal nu

mb

er

of

ban

ks

1.

2.

3.

4.

5.

6.

7.

Market shares equations with one of the following bank-specific factor (X):

Final estimation results: Market shares equations (GMM) Explanatory variables (in logs)

Dependent variable – market share (in logs) M1 Model M2 Model M3 Model

MC1 (via SFA; Revals are dropped from OC) 0.010 (0.013) 0.035*** (0.014) 0.135*** (0.013)

MC1 × The share of retail loans in total loans 0.105*** (0.011) 0.103*** (0.011) 0.101*** (0.010)

The share of retail loans in total loans –0.112*** (0.010) –0.108*** (0.010) –0.112*** (0.009)

MC1 × Loans-to-assets ratio 0.001 (0.028)

Loans-to-assets ratio 1.026*** (0.023)

MC1 × Equity-to-assets ratio –0.007 (0.021) 0.002 (0.021) 0.021 (0.019)

Equity-to-assets ratio –0.869*** (0.017) –0.842*** (0.017) –0.662*** (0.015)

MC1 × The share of overdue loans in total loans 0.028*** (0.006) 0.029*** (0.006) 0.036*** (0.006)

The share of overdue loans in total loans –0.060*** (0.004) –0.059*** (0.004) –0.044*** (0.004)

MC1 × The share of non-interest income in total income –0.001 (0.014) 0.009 (0.014) –0.004 (0.014)

The share of non-interest income in total income –0.024** (0.011) –0.020* (0.011) 0.007 (0.010)

MC1 × Loans-to-deposits ratio 0.117*** (0.031)

Loans-to-deposits ratio 0.685*** (0.024)

MC1× The share of reserve assets in total assets 0.079*** (0.018) 0.076*** (0.017) 0.019 (0.015)

The share of reserve assets in total assets –0.100*** (0.015) –0.112*** (0.015) –0.159*** (0.013)

MC1 × Loan loss reserves to interest income ratio –0.0043*** (0.0006)

Loan loss reserves to interest income ratio –0.014*** (0.000)

MC × Federal state banks 0.281*** (0.108) 0.246*** (0.109) 0.294*** (0.103)

MC × Foreign-owned banks –0.107 (0.080) –0.116 (0.079) –0.088 (0.072)

Foreign-owned banks 0.113* (0.066) 0.124* (0.065) 0.191*** (0.063)

MC × Credit cycle (HP-filter) 0.010*** (0.002) 0.009*** (0.002) 0.007*** (0.002)

Credit cycle (HP-filter) –0.162 (0.124) –0.103 (0.122) –0.273** (0.118)

Obs. (banks) 21446 (971) 21446 (971) 21446 (971)

Exact identification Yes Yes Yes

10

No

tes:

***

, **

и *

- a

n e

stim

ate

is s

ign

ific

ant

at 1

%, 5

%, a

nd

10

%,

resp

ecti

vely

. R

ob

ust

sta

nd

ard

err

ors

are

in

par

enth

eses

Final estimation results: the distributions of alternative versions of Boone indicator

11

The final versions of bank-level Boone indicator: 1. MC is estimated within SFA; Revaluations of currency and securities were dropped from costs 2. Russian banks are mostly competing on quality, not quantity 3. Sberbank dominated in terms of bank-level Boone in all regressions 4. The M1 Model is preferred (provides the largest – 30% – part of banks competing on quantity)

0

5

10

15

20

25

30

35

-0.3

0-0

.26

-0.2

1-0

.16

-0.1

1-0

.07

-0.0

20

.03

0.0

80

.12

0.1

70

.22

0.2

70

.31

0.3

60

.41

0.4

60

.50

0.5

5

1. The M1 Market share Model




% o

f to

tal sa

mple

ass

ets

Competition

on quantity

(M1) 30% of

total sample

assets

Competition

on quality

(M1) 70% of

total sample

assets

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

0.18-0

.31

-0.2

2

-0.1

4

-0.0

6

0.0

3

0.1

1

0.2

0

0.2

8

0.3

7

0.4

5





% o

f to

tal num

ber

of

banks

1.

2.

3.

Boone indicator in the 4 niches of the market for loans

12

0

10

20

30

40

50

60

70

80

90

100

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

The share of overdue loans in total loans, %

Th

e r

ati

o o

f re

tail

lo

an

s

in t

ota

l lo

an

s, %

SBER-

BANK

VTB24 RUSSIAN STANDARD HOME

CREDIT OTP

ROSBANK

URALSIB

MDM

TRUST

MTS-BANK

GAZPROMBANK

VTB

ROSSELKHOZBANK

BANK OF MOSCOW

20.0

CITY

ROSGOSSTRAKH

UNIASTRUM

PETROCOMMERZ

SOYZ

ALFA-BANK

VOZROZDENIE

Notes: The competition map as of the end of 2013 Circle area corresponds to the value of Boone indicator interacted with respective bank size (in terms of assets)

Boone vs Lerner: no conflict revealed 1/2

13

-0.02

-0.03

0.01 0.01

0.01

0.04

0.05

0.06

0.14 0.13

0

20

40

60

80

100

120

140

160

180

200

-0.04

-0.02

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

-53 -37 -20 -4 13 29 46 62 78 95

Boone indicator: averaged value in respective Lerner decile

Share of total sample assets (secondary axxis), %

Number of banks (secondary axxis)

Lerner index (market for loans), %

Ban

k-le

vel B

oo

ne

ind

icat

or

(mar

ket

for

loan

s)

Nu

mb

er

of

ba

nk

s a

nd

th

eir

siz

e

Pairwise correlation of Boone and Lerner is 93% at the decile level

Boone vs Lerner: no conflict revealed 2/2

14 All banks in the sample

Boone

Lerner, lag = 0 q 0.222



Lerner, lag = 3 q 0.211 Lerner, lag = 4 q 0.205

The sample’s median

Boone

Lerner, lag = 0 q -0.03



0.00

0.06

-0.02

0.03

30.6

11.3

19.2

0

5

10

15

20

25

30

35

-0.03

-0.02

-0.01

0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

20

05

q1

20

05

q3

20

06

q1

20

06

q3

20

07

q1

20

07

q3

20

08

q1

20

08

q3

20

09

q1

20

09

q3

20

10

q1

20

10

q3

20

11

q1

20

11

q3

20

12

q1

20

12

q3

20

13

q1

20

13

q3

Bank-level Boone indicator (market for loans), sample's median

Lerner index (market for loans), sample's median; secondary axxis

Conclusion

Within the panel framework, propose to use the external variables approach in order to disaggregate an industry-wide Boone indicator of banking competition to the bank level

For the sample of Russian banks, show that the key role in such bank-level modification belongs to (i) The share of retail loans in total loans, (ii) The share of overdue loans in total loans, (iii) The share of reserve assets in total assets, (iv) Loan loss reserves to interest income ratio

Having estimated the bank-level Boone indicator (BI) for Russian banks, show that

(1) Russian credit market is a monopoly with Sberbank being the dominant player and possessing highest value of BI;

(2) Russian banks mainly compete on quality rather than quantity in credit market. This effect becomes larger during the upward phase of credit cycle or when loan quality deteriorates;

Having divided Russian credit market into 4 niches using risk and loans compositions criteria and having analyzed the BI of banks in that niches, find that

(1) Competition is strongest within the “low risk – corporate lending” niche and

(2) Competition is weak within the “high risk – retail lending” niche

For Russian banks, bank-level Boone indicator and Lerner index have no conflict

15

THANK YOU!

MIKHAIL MAMONOV

EMAIL: [email protected]

April 8, 2015; Moscow

XVI April International Academic Conference on Economic and Social Development

MICROECONOMIC MODIFICATION OF AN INDUSTRY …€¦ · microeconomic modification of an industry-wide boone indicator: new estimates of the market power of russian banks april 8, 2015;

Documents