What are the efficiency gains of bancassurance? A two-side analysis: the banking and the insurance standpoint

Emerging Scholars in Banking and FinanceContemporary Issues in Financial Markets and Institutions

Cass Business School, London 9th December 2009Franco Fiordelisi, University of Roma Tre, Bangor Business SchoolOrnella Ricci, University of Roma Tre oricci@uniroma3.it

Agenda•Introduction and motivations•Literature Review•Methodology•Data and variables

▫banking side▫insurance side

•Results▫banking side▫insurance side

•Conclusions

Introduction and motivations (1)

• The development of bancassurance is the result of many factors…▫Regulatory Framework

EU Second Banking Directive, 1989 USA Gramm-Leach-Bliley Act, 1999

▫Economic rationales Drop in the interest margin from traditional banking

activity Development of life products

o progressive ageing of population in all developed countries o decrease in welfare state protection offered by governments

Existence of some similarities and complementaries between the banking and the insurance activities

Introduction and motivations (2)• Most studies dealing with bancassurance have only been

descriptive in nature (Chen et al. 2008)• Empirical analyses focus on the risk diversification

hypothesis (Boyd et al. 1993, Genetay and Molyneux 1998, Laderman 2000, Estrella 2001, Fields et al. 2005, 2008, Chen et al. 2006, Nurullah and Staikouras 2008), while cost and revenue synergies resulting in efficiency gains are still a poorly investigated issue

• The aim of this paper is to assess if bancassurance results in efficiency gains from both the banking and the insurance sides, answering the following research questions:

1) Are banks engaged in the insurance business more cost and profit efficient than their competitors specialized in traditional and investment banking?

2) Are bancassurance companies more cost and profit efficient than independent companies operating in the life business?

3) Which model of bancassurance reaches the best performance?

Literature Review (1)• Efficiency hp “banking side”

• There are no studies focussing on bancassurance in an exclusive way!

ALLEN & RAI (1996)Universal banking countries more cost efficient than separated banking countries in 1988-1992

Universal banking as combination between traditional combination between traditional and investment bankingand investment banking

VANDER VANNET (2002)Financial conglomerates more revenue efficient than specialised banks in 1995-96

Financial conglomerates as combinations between combinations between traditional and investment traditional and investment banking/insurancebanking/insurance

CASU & GIRARDONE (2004) Profit efficiency increase for Italian financial conglomerates over 1996-99

Financial conglomerates as all all Italian banking groupsItalian banking groups (experimenting a trend towards conglomeration over the observed period)

Literature Review (2)• Efficiency hp “insurance side”

• Bancassurance as a distribution channel or as equity links in a mutually exclusive way

• In other studies investigating the relationship between efficiency and distribution, bancassurance is only marginally considered (Trigo-Gamarra 2007) or completely ignored (Klumpes 2004)

HWANG & GAO (2005) Higher cost efficiency for life insurance companies using bank branches for the distribution of their productsIrish market 1991-2000

Bancassurance as a distribution channeldistribution channel

BARROS et al. (2006) Higher cost efficiency for life companies controlled by banking institutions Portuguese market 1995-2003

Bancassurance as

equity link/ownershipequity link/ownership

Literature Review (3)• We can find only one study (Verweire 1999) analysing both

sides of the phenomenon, but it is based on balance sheet ratios

• The comparison of different structures of financial alliances between banks and insurance companies has been discussed only in studies adopting a managerial point of view (Voutilainen 2005, Staikouras 2006, Lin et al. 2009)

• Overall, our paper aims to advance the existing literature by: 1) measuring potential cost and profit efficiency gains of

bancassurance from both the banking and the insurance points of view

2) dealing with bancassurance from both an ownership and a distributional perspectives

3) providing quantitative findings to compare the performance attained by different organisational models

4) analysing the unexplored case of the Italian financial industry

Methodology: base SFACost efficiency

Profit efficiencyitititititit uvxTC ,ln

- xit ln output/netput quant

and input prices

- vit ~ N(0; σv2) random error

- ui ~ N+(0; σu2) inefficiency

Comparison based on common frontiers require that all firms share the same technology and environmental conditions. In the case of sample heterogeneity, efficiency estimates can be strongly biased (Bos et al. 2005, 2008)

i jjiij

i jjiij

i jjiij

i jjj

pygppdyyc

pbyaaTPTC

lnlnlnlnlnln2

1

lnln)(ln 110

itititititit uvxTP ,ln

• It assumes that all sample firms share the same production technology, but some firm specific factors - out of the management control - influence the distance from the best practice

• The deterministic kernel of the frontier is the same as in the base model, while the inefficiency component is now assumed distributed as uit ~ N(it; σu

2), where

• Using this model we can account for the heterogeneity problem and still benchmark all firms against a common frontier. We also measure the impact of exogenous factors on efficiency without incurring in the problems of a two step procedure

Methodology: TE Model (1)

Firm specific factors

• The best predictor for single firm efficiency is the conditional expectation of uit given the observed it

• With the TE Model, we obtain gross measure of efficiency. If we want net measure, we have to substitute in the above expression *****

2** 5.0expexp itituE

it

M

jitjj z

1,0* 1

22* 1 s

222vus 22

su

M

jitjj z

1,min

…assuming that all sample firms face the same optimal exogenous conditions (minimising inefficiency)

Methodology: TE Model (2)

Data and variables “banking side” (1)• We start from the operational definition of Vander Vennet

(2002); a bank is a financial conglomerate when:▫ It is engaged in investment banking and/or insurance trough

an in-house department or a consolidated subsidiary▫ The ratio of not interest income to total revenues exceeds 20%

• We restrict these criteria in order to focus only bancassurance combinations, BC. A bank is a BC if:▫ It consolidates at least one insurance company (captive or joint

venture)▫ It has a ratio of not interest income to total revenues

exceeding 20%• Banks not satisfying these criteria are defined as a

residual cluster (Not Bancassurance-oriented Institutions, NBI) Type 2005 2006 Total

BC 23 24 47

NBI 18 15 33

Total 41 39 80

Source of financial data:ABI BANKING DATASource of data on the group structure Companies' web sites

Data and variables “banking side” (2)

• 2 different input/output specifications

DEP VARTC Total cost = interest expenses + operating expenses

TP Total profit = pre-tax profit

OUTPUT QUANT

Value added approach Value added approach Resti 1997, Fiordelisi & Molyneux 2006, Fiordelisi 2007

Y1 Total loans Y2 Other earning assets

Y3 Customer deposits

Traditional and non traditional Traditional and non traditional banking outputbanking outputRogers 1998, Vander Vennet 2002

Y1’ Interest income

Y2’ Not interest income

INPUT PRICES

P1 Price of labour = personnel expenses / n° employees

P2 Price of fixed assets = depr&amort./fixed assets

P3 Price of capital = interest expenses/(deposits + other funding)

NETPUT Z Equity

Data and variables “banking side” (3)• We choose the firm specific factors on the basis of two criteria:

▫ Influence on efficiency on the basis of theory and empirical literature▫ Significant differences in means between the two subgroups

  LN_TA TCR% %INS LN_INTG

COMM LISTED

NBI_mean 15.2271 0.12034 0.0000 9.6064 0.4848 0.2121BC_mean 17.5177 0.09874 0.0297 13.1348 0.6383 0.7021diff NBI-BC -2.2905 0.0216 -0.0297 -3.5283 -0.1534 -0.4900t-stat -10.8054 2.6295 -7.3075 -9.0847 -1.3550 -4.9576p-value 0.0000 0.0126 0.0000 0.0000 0.1800 0.0000

• SIZE ln total asset (LN_TA)• RISK Total capital ratio (TCR%)• DIVERSIFICATION ratio between the book value of the insurance

companies and the consolidated equity (%INS)• INTANGIBLES ln total investment in intangibles assets (LN_INTG)• INSTITUTIONAL TYPE commercial or cooperative (COMM)• LISTING listed or unlisted institution (LISTED)

Data and variables “insurance side” (1)

Type 2005 2006 Total

CB 13 12 25

IC 49 51 100

JV 21 22 43

Total 83 85 168

Source of financial data:INFOBILASource of data on the distribution system ANIASource of data on the ownership structure Companies' web sites

• Analysing the case of Italy (4° life insurance market in Europe and 6° in the world) we dispose of detailed and reliable data provided by ANIA

• Sample composition (ownership criterion)▫ Insurance companies controlled by banks (CB)▫ Joint ventures between banking and insurance partners (JV)▫ Independent companies (IC)

Data and variables “insurance side” (2)

• In order to define input and output variables, we choose the value added approach adopted in the most recent insurance studies (e.g. Fenn et al. 2008)

DEP VAR

TC Total operating cost from the life technical account

TP Net earned premiums + investment income – total cost

OUTPUTQUANT

Y1 Claims net of reinsurance, plus bonus and rebates, plus additions to reserves

Fenn et al. 2008Trigo-Gamarra, 2007

INPUT PRICES

P1 Net operating expenses plus other technical charges/Total assets P2 Investment charges/Total assets

Bikker and Van Leuvensteijn, 2008

NETPUTZ1 Total equity capital in t-1 Fenn et al. 2008

Z2 Total technical provisions in t-1 Fenn et al. 2008

Data and variables “insurance side”(3)

• In order to overcome the heterogeneity problem, we include in the model the following firm specific factors:▫ Z1_%banc, the percentage of premiums collected by bank branches▫ Z2_MS, the market share with reference to the Italian life industry▫ Z3_%fin, the weight of policies with an high financial content (falling

into the III or the V class under the 96/1992 Directive - respectively unit/index linked and capital redemption)

▫ Z4_comp, a dummy indicating if the firm is a life specialist or a composite company

Variable

Mean Difference in means

Total CB IC JV IC-CB JV-CB JV-IC

Z1_%banc

0.4406

0.7956

0.1648

0.8758

-0.6308**

* 0.0802 0.7110*

**

Z2_MS0.011

7 0.017

7 0.007

5 0.018

2 -

0.0103** 0.0005 0.0108*

**

Z3_%fin0.494

7 0.407

3 0.433

8 0.687

1 0.0265 0.2798*

**0.2533*

**

Z4_comp

0.3452

0.0400

0.4600

0.2558 0.42*** 0.2158

-0.2042*

*

*** p value<0.01; ** p value<0.05; * p-value<0.1

Results “banking side” (1)

COST EFFICIENCY INPUT/OUPUT SPECIFICATION NBI

MEANBC

MEANT-

STATa

P-value

Value added approach 0.8134 0.8197 -0.1719 0.86Traditional and non traditional banking

output0.8951 0.893 0.0850 0.93

ALTERNATIVE PROFIT EFFICIENCYINPUT/OUPUT SPECIFICATION NBI

MEANBC

MEANT-

STATa

P-value

Traditional and non traditional banking output

0.7858 0.7356 1.1694 0.25

a Two sample t-test of differences in mean H0: mean(NBI)-mean(BC)=0 H1: mean(NBI)-mean(BC)0

COST EFFICIENCY PROFIT EFFICIENCY(value added

approach)(traditional and non traditional output)

(traditional and non traditional output)

VARIABLE COEFFICIENT COEFFICIENT COEFFICIENTLN_TA -0.483*** -0.316*** -0.217**TCR% -2.794*** -4.193* -7.908***INS% 2.432** -2.429 -5.460***

LN_INTG 0.192*** -0.0154* 0.406***COMM -0.302** -0.1755 -0.457LISTED 0.480** -0.3188** -0.077

• Results from the model including firm specific factors

Results “insurance side” (1)• Gross cost efficiency from the TE model

• Impact of firm specific factors

• Net cost efficiency from the TE model

Type Mean

Std. Dev Min Max

CB 0.939 0.033 0.876 0.974IC 0.902 0.085 0.620 0.988JV 0.953 0.034 0.830 0.985

 Groups Diff. Eff(IC)-Eff(CB) -0.0368*Eff(JV)-Eff(CB) 0.0142

Eff(JV)-Eff(IC)0.051**

*Variable Coeff.Cost 0.191**Z1_

%banc -0.465**Z2_MS -6.334**Z3_%fin -0.713*Z4_comp -0.308**

Type Mean

Std. Dev Min Max

CB 0.967 0.014 0.939 0.984IC 0.946 0.066 0.698 0.990JV 0.972 0.018 0.887 0.988

Groups Diff.Eff(IC)-Eff(CB) -0.0217Eff(JV)-Eff(CB) 0.0042

Eff(JV)-Eff(IC)0.0259*

*

Results “insurance side” (2)• Gross profit efficiency from the TE model

• Impact of firm specific factors

• Net profit efficiency from the TE model

Type Mean

Std. Dev Min Max

CB 0.916 0.107 0.579 0.984IC 0.911 0.118 0.498 0.988JV 0.830 0.113 0.620 0.972

 Groups Diff. Eff(IC)-Eff(CB) -0.0052Eff(JV)-Eff(CB) -0.0855**

Eff(JV)-Eff(IC)-

0.0803***Variable Coeff.

Cost-0.571***

Z1_%banc -0.136Z2_MS 3.898Z3_%fin 1.19***Z4_comp -0.298**

Type Mean

Std. Dev Min Max

CB0.9875

3 0.000810.9850

70.9887

9

IC0.9873

9 0.000770.9848

70.9889

2

JV0.9872

1 0.000760.9854

50.9889

4

Groups Diff.Eff(IC)-Eff(CB) -0.0001Eff(JV)-Eff(CB) -0.0003

Eff(JV)-Eff(IC) -0.0002

Conclusions (1)1) Are banks engaged in the insurance business more cost and profit efficient than their competitors specialized in traditional and investment banking? Bancassurance combinations (BCs) and Not Bancassurance-oriented institutions (NBIs) show very similar levels of cost and profit efficiency, under both the value added and the alternative approach used to define input and output variablesThe firm specific factor measuring diversification into the insurance business shows results very sensitive to the input and output specification. The relationship with cost efficiency is negative under the valued added approach and positive (but not significant) under the specification with traditional and non traditional banking output. Moving to the profit side we also find a positive and significant relationship between diversification in insurance and the attained performance The supposed cost and revenue synergies resulting from the alternative approach, however, are probably of a scarce magnitude or compensated by other diseconomies, given that the medium level of cost and profit efficiency is very similar for BCs and NBIs

Conclusions (2)2) Are bancassurance companies more cost and profit efficient than independent companies operating in the life business?

Considering gross efficiency scores, we find a significant cost advantage for bancassurance companies. When we observe net scores the advantage of CBs over ICs disappears while that in favor of JVs is strongly reduced, revealing that the bancassurance overperformance is mainly explained by some positive features, such as the share of premiums collected by bank branches and the proportion of high financial content policies in the business mix, that are not exclusive of insurance companies wholly or jointly owned by banks

On the profit side we do not find any strong evidence in favor of bancassurance: the relationship between the performance attained and the share of premiums collected by bank branches is still positive, but not statistically significant. The percentage of product with an high financial content is negatively related to profit efficiency in a significant way, revealing that these policies are less costly to manage than protection insurance but less profitable

Conclusions (3)3) Which model of bancassurance reaches the best

performance?Results seem to be in favor of JVs for the cost side and

present a substantial parity among sample firms on the profit side

The convenience of bancassurance as a distribution channel is relevant and consolidated while the success of insurance products with an high financial content is more volatile and strictly dependent on current market trends, requiring a continuous revision of the business mix

So the existence of ownership links is not necessarily the best strategy for the realization of cost and revenue synergies between the banking and the insurance activities and the subjects involved should consider also the alternative of more flexible and reversible forms of cooperation, such as cross selling agreements and non equity strategic alliances

The identification of the best bancassurance model probably depends also on the characteristics of the subjects involved and undoubtedly deserves further investigation