Vous êtes sur la page 1sur 14

Deregulation and the

Limits to Banking Market


Competition: Some
Insights From India
Iayati Sarkar
"Stimon Kumar Bhaumik

Poliq-mnkus around the warld haae ernphasizedthe airtues of dnegulati,on since


the 1980s. In f,nancial markets, deregulation has taken the form of remoual of
bh,nius to mtry and licmsing pol;icies, as well as the dismantkng of regulated
interest rate regimes.Theserneasureshaae beenwidely acclaimed as the harbingers
of competition, competition being the stepping stone to highrr leuek of fficimq.
Using theareticalcaaeats.fro* the industrial organization literature and empi,rical
nidencefrom theIndian banking industry, this papu a,rguesthat mtry and branch-
ing deregulationand interestrate libnalization might not bethepanacea that many
pokq-makers make them cut to be. Specif,cally, the paper argues that competition
might not naturally follow such dnegulation if the incurnbmts are well entrenched
in their markct segmcntsby uirtue of sheersizeand greatu accessibilityto customers.

I the goal of growth with distributive justice


INTRODUCTION through the nationalizattonof banks, the rapid
expansion of the banking system through ex-
Enu-y-andinterestratederegulationhavebeen tensive branching and the protection of banks
key elemens of banking sector liberalization from private sector competition through entry
programmes in a number of developing and and branch licensing restrictions and through
transitional economiessince the early 1980s.The administered interest rates. Subsequently, bank-
primary motivation behind such deregulation ing sector liberalization in these countries has
has been to generate increased competition in typically emerged in the form of entry and
the banking system so as to increase its opera- interest rate deregulation.
tional efiiciency. Such a prescription finds its The present case-study,focusing on the im-
theoretical roots in the industrial organization pact of deregulation on competition in the
Ii tera ture, specifically in the economics of imper- Indian banking sector, highlights the point
fect information, imperfect competition and that while policy-makers often seem to be con-
drn anric processes,that has established a positive fident that such deregulation would generate
re lation ship benrreencompetition, incentives and the much needed competition for incumbent
drnamic efficiency. banks, in practice, given the aforementioned
The objective of this paper is to analyse the initial conditions, the impact of deregulation
impact of deregulation on competition in deve- could be way below the critical level needed to
Ioping countn' banking systemswhere incum- generate elficiency gains for these banks. In
bent banks have been 'ruling the roost' for years India, private entry into the banking sector,
throtrgh an extensive branch network. In these hitherto dominated by public sector banks, was
countries. enqv and interest rate deregulation
have npicallv come after the banking systems
have been used for long by the government, as Jayati Sarhar is.with the Indira Gandhi Institute of Dnelopmenl
Rcsearch, Mumba'i, Indi.a, whib conesponding authm, Sumon
instrtrments for serving national priorities and I{u,mar Bhaumik is with the Inuestmmt Infarmation and Credit
objectives.Indeed, attemptswere made to attain Rnti.ng Agmq of Ind,ia, Calattta, India.
5U Inte'ru?,atxon&l
Journq,l oJ ueuelapTnent DaTLKt,ng

deregulated in January 1993. Between.fanuary Table I


1993 and March 1998, 24 new private sector Efficiency Indicators of Bank Groups:
banks - nine domestic and 15 foreign - en- l99l-92 and 199&97
tered the market, thereby increasing the total
number of scheduled commercial banks, eX- l99i-92 | 199G97
cluding specialized banks such as the regional PB DP\ts FB PB DP\B FB NDPVB
rural banks,from TSin 1991-92to 99 in 1997-98. Net interest margin 3.4 4.0 4.9 3 . 2 3 . 0 4 . 1 2.9
The policy of entry deregulation was accompa- as Voof working funds
nied by progressivederegulation ofinterest rates Operating cost as 2.6 3.0 2.2 2 . 9 2.5 3.0 1.9
on deposits and advances. Prior to 1994, the % of working funds
Staffexpenditure as 1.9 2.1 0.7 2 . r 1.5 1.0 0.3
entire term structure of term deposits as also
% of working funds
intere st on advanceswasfixed and administered Operating profit/ 1.0 1.0 6.8 0 . 8 t.2 3.4 10.0
by the Reserve Bank of India (RBI). Starting Staff expenses
October 1994, interest rateswere deregulated in
a phased manner and by October 1997, banks Note: PB: Public sector banks; DPVB: Old domestrc private
were allowed to set interest rates on all term sector banks; NDPVB : New domestic private sector banks; FB :
deposits of maturityof more than 30 daysand on Foreign banks.
all advancesexceeding Rs. 200,000.These steps Source: Pnfrnmanu Highlights of Banks,IBA, several issues.

were taken to put increased pressure on banks to


adopt more competitive practices in the pricing banks. Finally, one cannotbuttake cognizance of
of loans and deposits and make these institu- the large differentials between the efficiency of
tions more efficient in the business of financial incumbent public and old private sector banks
intermediation. and the 'best-practice' levels as reflected in the
The expectation of increased efficiency of efficiency levels of the new domestic private sec-
banks notwithstanding, postderegulation trends tor banks.
ih different banking efficiency indicators of the The absence of any indication of quantitative
major bank groups do not reveal any marked improvements in the efficiency parameters of
changes in the efficiency levels of either the incumbent public sector banks may essentially
public or the private sector banks. Moreover, be due to two reasons.First, it can be argued that
efficiency differentials between the different increased competition in the banking market, if
bank groups have continued to persist despite a any, isyet to manifestitself in efficiencyimprove-
more deregulated environment. Indeed, as re- ments, given that the impact of cost-cutting exer-
vealed by Table 1, the efficiency of public sector cisesand restructuring of banks on the efficiency
banks, which control around 85 per cent of levels usually appear with a lag. Second, it is
banking business,has shown little improvement unlikely that there would be much meaningful
over the I 99 1-92levelswith respect to key indica- competition in the early years because entrants
tors. Worse, if we filte.r out the impact of large- need time to build up sufficient capacity to com-
scale technological upgradation on the expen- pete on a level plapng field with the more estab-
ditures of the banks, the data indicate that there lished incumbents.
has actually been a decline in efficiency levels of Nonvithstanding the above-mentioned line of
the public sector banks. As shown in the Table, reasoning, we present an alternative hypothesis
the proportion of staff expenditure in working about the absenceof efficiencv improvements in
funds, which is an indicator ofvariable costs,has incumbent banks subsequent to deregulation.
somewhat increased for public sector banks. We argue that, given the inherent structure of
Among the private sector banks, this proportion the Indian banking industn', there is very little
has increased for the foreign banks, but has potential for a'critical level' of competitive pres-
significantly declined for the old domestic pri- sure to be generated in the aftermath of deregu-
vate sector banks. A similar picture emerges with lation that should, in principle, motivate incum-
respect to the trends in operating profitper staff bentbanks to improve their efficiencylevels.We
expenses,a measure of labour productivi ty. Data present the theoretical arguments and empirical
indicate that labour productivity has declined evidence in support of this alternative hypothesis
for the public sector and foreign banks, but has at three levels.
increased significantly for the domestic private First, we offer theoretical justifications from
the industrial organization literature as to why study argues for supplementary policy mea-
deregulation may no t generate compe titive pre s- sures to increase the efficiency of the Turkish
sures on incumbents with characteristics similar banking sector.
to those in the Indian banking industry. In this The rest of the paper is organized as follows.
context, we examine specific characteristics of Section II provides the theoreticaljustification
the market structure of the banking industry that as to why competition may not emerge despite
might act as 'non-regulatory' barriers to compe- deregulation. Section III highlights the fea-
tition and prevent the emergence of a level play- tures of the Indian banking industry that are
ing field between entrants and incumbents. Sec- most likely to be barriers to competition. This
ondly, we analysetrends in key structural charac- section also summarizes the findings of our
teristics like concentration ratio and inter-firm econometric exercise. Finally, some policy im-
mobility to find out whether signs of increased plications are discussedin Section fV.
competition are emergingin the Indian banking
sector. Finaily, we econometrically test for the II
possible onset of competition, against the alter- STRUCTTIRAL BARRIERS TO
native hypothesis of insignificant competition, COMPETITION
by evaluating whether there have been statisti-
cally signifi cant reallocations in the marke t shares Our hypothesis involving the possible ab-
awayfrom incumbents in favor of entrants since sence of competition among banks in the after-
the beginning of deregulation. math of the deregulation in India has its theo-
We focus our analysis of competition on the retical underpinning in the extensive industrial
competition for bank's resources,viz. deposits.l organrzation literature on entry barriers and
Several differentiating factors relevant to the competition. One of the crucial assumptions
banking sector in a developing country come behind the expectation that free entry and
into play in the deposit market. Primary among deregulated prices would foster competition
this is accessibility,which is important for meet- and thereby improve dynamic efficiency is that
ing the liquidity needs of the depositors, particu- the product market is homogeneous, i.e., the
larlyin the savingsdepositmarket. The underde- incumbent and the entrant produce identical
veloped branch network of entrants could effec- goods. In this case, if the entrant is more effi-
tively discourage them from tapping such a nlar- cient than the incumbent, then the former can
ket and thereby encourage them to focus in offer a lower price or a better quality and cap-
resource generating activities in niche markets ture the latter's market share. However. such a
where accessibility is not a primary consider- scenario may not hold if the incumbents enjoy
ation. "product differentiation" advanages over the
Although it is.difficult to clearly predict the entrants in the sense of incurnbents already
long-term implications of dereguiation based on cornering the 'right' marketniches in the prod-
the experience ofjust the first sixyears of deregu- uc t space,enj oying customer loyalty, locational
lation, we feel thatl our diagnostic exercise of advantages or th e advantage accruing fro m large
whether competitive pressures are building up production capacities.z All these can make it
within the industry can give timely pointers to difficult for entrants to penetrate into the mar-
whether alternative or supplementary policy in- kets of incumbents and produce products iden-
tervention measures like bank prlatization and tical to those of the incumbents. As Klemperer
bank consolidation are needed to boost the effi- ( 1987) has noted, even if two firms sell function-
ciencyofpublic sectorbanks over the medium to ally identical pro ducts, produc t differen tiation
long term. Our con tention is particularly strength- can appear ex post after consumers have pur-
ened by u recent case-study of the impact of chaseda firm's product and increasethe cost of
financial liberalization and new bank entry in switching from one firm to another. In the
Turkey (Denizer, 1997), which finds that the new context of banks, Klemperer argues that two
entry of 31 banks over a 10-yearperiod, 1980- banks ntay, for example, offer identical check-
1990, has failed to generate sufficient competi- ing accounts,but there could be high (actual or
tion to alter inter-firm relationships and chal- perceived) transaction costs in closing an ac-
lenge the incumbents in the retail banking mar- count with one bank and operating with a
ket. Based on such a finding, the author of the competitor.
.luat,r ,0(tb uJ rJvuvtrty

Ex-postproduct differentiation would be es- tive challenge to incumbent banks from the new
pecially strong between incumbent and entrant entrants could at best be effective in urban-
banks, as the former will be already operating metropolitan areaswhere new entrants are most
with. an established depositor base while en- likely to establish their offices.
trants will have to build one from scratch. Un- At the time of entry deregulation in 1993,
der such circumstances, entrants maybe forced incumbent public sector banks in India had al-
to develop market niches of their own that are ready established an extensive branch network
distinct from that of incumbents. In fact, evi- throughout the country, conrolling around 91
dence to this effect is borne in an analysis of per cent of bank branches and accounting for 89
post-entry competition in banking in Portugal per cent of the deposits mobilized (Table 2).
(Barros, 1995) which finds that entrants have Such predominance of public sector banks has
expanded their branches more in markets other been the direct result of widespread, policydi-
than the ones where incumbents have increased rected branch expansion thatwas undertaken to
branch networks the most. The implication of spread banking among the masses.Branch ex-
such tendencies is thatincumbents are unlikely pansion by private sector banks, both domestic
to face any effective competition from entrants and foreign, was on the other hand stringently
'monopoly'
upon deregulation and the power regulated. Hence, even though five years have
of incumbents would persist even in the pres- elapsed since the deregulation of entry and
ence of entry. branching, private sectorbanks continue to have
The persistence of monopoly arising out of a marginal foothold in the market, with new
product differentiation advantages, notwith- entrants among them controlling less than one
standing deregulation, is a strong possibility in per cent of total branches and lessthan three per
the Indian banking industry as well as in many
other developing countries. As noted earlier, Table 2
dere gulation in the secoun trie shave often come Structurd Characteristics of Indian Banks:
at a time when incumbent banks have already 1992 and 1997
attained an entrenched position through an
extensive branch network and a deep customer Incumbents Entrants
base developed in the course of several years of 1991-92 199&97199G97
virtual monopoly. 3 Under such a scenario, it is
to be expected that ex-posfdifferentiating fac- 1. No. of Scheduled 74 t5 24
tors such as accessibility, customer loyalty etc. Commercial Banks
(excluding RRBs)
would actas implicitstructural barriers and that
these barriers would dissuade new private sec- 2. Ownership
- Public 28 27
tor banks from direct competition by locating
- Indian private 23 25* 9
themselves in the market niches of the incum- - Foreign private
bent banks. In fact; there is evidence in the
23 23 15
literature suggesting that the size of a bank (in 3. Share in Total Branches 100 99.7 0.3
- Public sector 91.5 90.6
terms of its branch network) becomes a signifi- - Indian private 8.2 8.8 0.3
cant mobility barrier for new entrants as prox- - Foreign private 0.3 0.3 0.03
imity to bank offices is an important factor in
consumers' choice (Dermine, 1993). This, in 4. Share in Total Deposits 100 97.4 2.6
- Public sector banks 89.0 82.9
turn, leads us to hypothesize that the impact of - Indian private 4.6 7. 3 2. 5
deregulation on banking market competition - Foreign private 6.4 7.2 0.1
would be relatively insignificant in the short to
5. Share in total assets 100 97.3 2.7
medium run, especially in developing coun- - Public banks 88.5 82. 6
tries like India, since a sufficiently long time will - Indian private 4.2 6. 5 2. 4
have to necessarily elapse before which the - Foreign private 7.3 8. 2 0. 3
entrants can build up a branch network and
customer base that could closely match that of Note: * Two non-scheduled old private banks gaining
the existing incumbent banks, thereby neutral- scheduled status.
izin g the latter' scomparative advantage in terms Sorrrce: Computed from PerformanceHighlights ofBanks,
of accessibility and customer loyalty. Competi- IBA, several issues.
L'eregul,anon 53

cent of deposits and bank assets.Among the size of public sector banks is on the average, at
incumbents, the domestic private sector banks least l0 times that of private sector banks. The
have increased their share in total branches, average branch size of the former is, on the
while the share of the foreign banks has re- other hand, a fraction of that of the foreign and
mained the same. Hence, the share of incum- new private sector banks.
bent public sector banks in total branches has The "unparalleled" branch network and as-
only marginally declined from 91.5 per cent to setbase of public sector incumbents alongwith
90.6 per cent. their "unmatched stock of information capi-
A corollary of continuous expansion of the tal," the latter acquired from established rela-
public sector banks has been the extreme skew- tionships with their existing customer base,
nessin the assetdistribution between public and have been deemed as "near impossible" for
private sector banks. As Table 3 reveals,the asset new entrants to replicate (MOF, 1992)4.It has,

Table 3
Characteristics of Asset Distribution byBank Groups: March 1995-96

Size (Rs. crore) S B I& A NB DPVB NDPVB FB


Minimum Size 3428 7289 44 128 32
Maximum Size r50266 37593 6100 22Br 8712
Mean Size 24367 r7776 1572 1063 1735
Average Size of Branch l5 r0.8 9.29 116.7 314.4

Note: SBI & A: State Bank of India and Associate banks; NB: Nationalized banks;
DPVB: Old domestic private sector banks; NDPVB: New domestic private sector banks
FB: Foreign banks
Source: Computed from PerformanceHighlights of Banks: 1995-95,18A.

Table 4
Characteristics of Deposits Mobilized,by Major Bank Groups

SBI&A NB PVT FB
I. Average size of account, 1995 (Rs.)
- Term deposits 18,430 22,179 28,225 2,35,040
- Savingsdeposits 3,775 3,547 3,295 20,793
- Current deposits 52,295 32,083 21,065 l,ll,l42
II. Composition of deposits, 1997 (% share)
- Term deposits 57.9 62.3 78.3 72.7
- Savingsdeposits 24.5 24.4 10.7 8.4
Current deposits 17.6 13.3 I 1.1 19.0
III. Deposits mobilized from urban areas
(% of total deposits) (June, 1997) 58.2 67.6 73.0 99.8
lV. Ownership of deposits (1997)
l. Government Sector l2.B 8.0 6.6 0.2
2. Private Corporate Sector (Non-financial) 3.0 2.5 6.6 15.9
3. Financial Sector 11.8 4.6 16.1 6.5
4. Household Sector 62.9 74.8 58.0 33.3
5. Foreign Sector 9.6 10:I 12.7 44.1

Note: PW.zTiivate Sector Banks


Source: ResmteBank of India Bull,etin, several issues, RBI.
34 Intemational.foum,al of Dnelopment Banking

therefore, defacto eliminated the possibility of a better position to afford. Consequently, the
private banks, both incumbents and entrants, average size of savings and fixed deposit account
to compete with incumbent public sector banks are much larger for foreign banks than for do-
by way of branch expansion. A direct conse- mestic banks.
quence of this has been the use of price and Like foreign banks, co{porate sector deposits
product differences as strategic variables by appear to be a niche for SBI&A. However, the
these banks during the deregulated period. larger average size of fixed and current deposit
The four major bank groups operating in the accounts offoreign banks ascompared to SBI&A
retail banking market are State Bank of India suggests that foreign banks have tended to spe-
and its seven associatebanks (SBI&A), the na- cialize in handling wholesale rather than retail
tionalized banks (NBs), the domestic private accounts. Finally, private domestic banks appear
sector banks (DPVBs) and the foreign,banks to target financial sector deposits significantly
(FBs). The DPVBs, in turn, can be separated more than other bank groups.
into old banks and new entrants (NDPVBs). With regard to the new private sector entrants,
Product differentiation in Indian banking has these predominantly urban-centric banks have,
traditionally manifested through the emergence like their foreign coun terparts, focused on whole-
of 'niche clienteles' for the different bank salecorporate deposits, foreign exchange opera-
groups. As Table 4 reveals,there is (expectedly) tions, NRI services and international banking,
greater difference between domestic and for- rather than on mobilizing retail deposits. Ac-
eign banks than between public and domestic cordingly, their presence, although minuscule
private sector banks. While individual accounts compared to their public sector counterparts, is
constitute a major portion of bank deposits for relatively the maximum in the fixed and current
all bank groups, the ownership composition of deposit markets.
such accounts are biased in favour of non-
resident accounts in the case of foreign banks. III
Further, these banks target the private corpo- DFREGUIATION AND IVIARKE'T
rate sector relatively more for deposit mobiliza- STRUCTURE
tion compared to other bank groups. That
foreign banks target the premium segment(s) This section will analysethe impact of deregu-
is also evident from the high-priced, high qua- lation on three aspectsof banking market struc-
lity service that these banks of,&!Lq the deposi- ture in India. These are: (a) concentration ratio,
'big savers'are in
tors and which obviously the (b) interbank mobility and (c) changes in mar-

Table 5
Trends in Concentration Ratios in Deposit Markets: 1985-1996-97
Aggregate Deposit Market SavingsDeposit Market
Year 1-bank 3-bank 5-bank 10-bank l-bank 3-bank 5-bank lO-bank
1985 23.6 37.5 49.8 69.8 2L.7 35.9 47.9 68.2
1986 22.6 37.4 48.9 69.9 21.5 35.9 47.8 67.8
1987 2r.2 36.7 47.9 67.9 21.9 36.7 48.4 67.8
19BB-89 999 35.7 48.5 69.8 22.7 37.3 48.9 68.0
1989-90 2 1 . 8 36.0 48.0 68.1 22.6 37.r 48.8 68.0
1990-91 2r.9 35.9 48.3 67.2 22.5 37.0 48.5 67.7
1991-92 99q 37.4 49.0 67.0 2r.4 35.8 47.4 66.9
1992-93 22.r 35.9 47.2 65.7 21.6 35.7 +7.3 66.7
1993-94 21.8 34.8 45.2 63.9 27.7 36.5 48.0 66.8
199495 2L.0 33.7 44.0 62.8 2r.6 36.3 48.0 66.7
1995-96 rs.7 33.8 4L.3 63.0 21.8 36.2 47.8 66.8
199G97 20.2 32.2 44.r 60.6 2r.5 36.1 47.8 66.6

Source: Computed from Fznancial Analysis of Banks and PerformanceHighkghts of Banks,several issues,
IBA.
Lteregumnon 55

ket share. number of accounts, with the share remaining


Concentration Ratio: Trends in m-bank con- steady around 26 per cent. Overall, the esti-
centration ratio for Indian commercial banks for mates of different concentration ratios in the
different typesof deposits for selectedyearssince savingsdeposit market tend to suggestthat the
1985 are presented in Table 5. In India, 19 out of leading public sector banks continue to domi-
t}:.e27 public sector banks are among the top 20 nate the savingsdepositmarketboth in terms of
banks. Table 5 reveals that although the l-bank, number of accounts as well as in aggregate
3-bank, 5-bank and lO-bank concentration ratios value. They continue to attract bulk of . the
have each declined a few percentage points be- savings depositors, while foreign and private
tween 1993, the year prior to deregu'lation and domestic banks continue to remain fringe play-
1997, some of the ratios do not show a clear ers.
declining trend in the post-entry years. This, in Interbank Mobility: Another meaningful mea-
spite of the fact that the new banks have been sure of competition that has been used in the
increasingly attempting to consolidate their posi- industrial organization literature is the mobility
tion over the years. In fact, in the total deposit among firms in an industry. This proxy measure
market, the l-bank concentration ratio (that is, of competition takes into account both the
the market share of the largest bank, the State price and non-price dimensions of competi-
Bank oflndia) has even increased, albeitmargin- tion. It hypothesizes that an increase in compe-
ally, between 1995-96and 1997-98,from 19.7per tition in the market is likely to be reflected in
cent to 20.2 per cent. Trends in 5-bank ratio changes in the relative rankings of the top firms
showsthatithas remained the same, ataround44 in a market. As Heggestad and Rhoades (1976)
percentbetween 199+95 and 199&97, declining have noted, a rank change among industry
temporarily in between. A more pronounced leaders may reflect some significant event that
declining trend in the l0-bank ratio since de- influence structural inter-firm relationships
regulation, however, indicates increased compe- while rankchanges among smaller, fringe firms
tition among the relatively smaller banks in the maybe due to chance orotherfactorsunrelated
market. to the structure of the market.
Any analysisof competition in retail banking With respect to the Indian banking industry,
should particularly focus on the dynamics in the we propose to study interbank mobility among
savingsdepositmarketas savingsdepositis one of
the main outputs of retail banks and is the most Table 6
basicfinancial assetpeoplehold (Denizer, 1997). Trends in Interbank wtobiliff: 1985 - 1996-97
In 1995,savingsaccountsconstituted 70 per cent
of the total of around 390 million deposit ac- Top 10 Banks All Banks
counts and accounted for 25 per cent of total Sumof Average Sumof Average
value of deposits. Post-deregulation trends in the rank rank rank rank
l-bank, 3-bank, 5-bank and lO-bank concentra- changes change changes change
tion ratios in the savings deposits market and
1985 0 0
these ratios continue to hover around 21 per
1986 6 1.0 4T 1.8
cent, 36 per cent, 48 per cent and 67 per cent E
1987 I l.l 43 1.8
respectively (Table 5). In fact, as Table 5 shows,
1988-89 6 2.0 48 l.B
concentration ratios in the savingsdeposit mar-
1989-90 5 2.4 44 2.4
ket have remarkably remained constant over the
1990-91 9 1.0 43 2.3
last 12years.One implication ofthe nearabsence
1991-92 6 1.3 5t) 2.6
of a decline in concentration ratios in the savings
1992-93 5 1.0 44 2.1
deposit market following deregulation is that a
1993-94 5 1.6 5l 2.7
critical number of branches is required for any
199495 I 1.0 44 2.3
bank to be an effective competitor in the retail
1995-96 3 1.0 3t 5.f)
banking market.
1996-97 6_ 1.6 79 4.0
Arelated and interesting indicator of possible
changes is the number of savings deposit ac- Source: Computed from Financial Analysis of
counts. Data for SBI&A till 1995 show no discern- Banks and Prfonnance Highlights of Banks,lBA,
ible change in this bank group's share in the several issues,
two groups of banks: first, the rank changes not really expand in any significantway) particu-
among the top l0 banks in the market, and lar specialized niches distinct from those of the
second, rank changes among all the banks in incumbents. This is in line with existing empiri-
the market. The latter measure will seek to cal work on entry and market dy-namics(Geroski,
capture the aggressive behaviour of the fringe 1991; Geroski and Schwalbach,1991). Note that
banks in the market. As stated earlier, the top l0 while it can be argued that any absence of a
banks in India have consistently been public significant evidence of reallocation of market
sector banks with private sector domestic and shares is but a consequence of enhanced effi-
foreign banks limited to operating as fringe ciency of the (incumbent) public sector banks,
players. As of 199G97, only one foreign bank data from Table I point to the contrary.
appears among the top 20 banks and three An examination of the trends in share of
foreign and two old private sector banks appear public sector banks since 1980 reveals that the
among the top 30. share of privately owned banks in deposit mobi-
Measures of interbank mobility fbr the top lization has shown a noticeable improvement at
10 banks and for all banks for the period 1985- the expense of .the share of its public sector
'Ihis
97 are presented in Table 6. We present two counterparts. is particularly true since 1991-
measuresof mobilityfor each set,an "extensity" 92 with regard to the share of domestic private
measurewhich is defined asthe number ofrank banks (including the nervbanks) in total deposits
changes among the banks in each year and an which has increased steadily from 4.6 per cent in
"intensity" measure which is defined as the 1991-92to 10.2 per cent in 1997-98.In this sub'
average "size" of rank changes in a particular section, we will econometrically testwhether the
year. Both measures a-recomputed taking 1985 decline in public sector market share is part of
as the base year, i.e., changes in rank are mea- the historical time trend or is it a statisticalll
sured relative to the ranks in 1985. It is evident deviation frorn the normal trend.
si,gnifi,canf
from the Table thatwhile there is no discernible Methodology: Our empirical exercise involves
trend in interbank mobility among the top l0 testing for a structural break.between pre-de-
banks since deregulation, there has been a regulation and post-deregulation rnarket shares
significant increase in the mobility among all of incumbent public sector banks. \AIe accord-
banks, especially during the last two years. This ingly develop a simple analvsis of covariance
evidence, together with the fact that the nine model with time effects in which we apply the
new domestic private sector banks have signifi- Chow test or analysisof variance test to look for
cantly improved their ranking between 1995-96 structural break in the post-1992 period.
'residual method' under which
and 199G97, suggests that the new entrants We adopt the
have had a significant irrrpact on existing mar- we test if each observation in the post-policy
ket relationships but only at the fringe. How- period could have been generated by the pre-
ever,.since the scale of operations of the en- policy structure. This method is based on post.
trants and other private sector incumbents is sample predictive testing. Specifically, one pre-
significantly smaller than that of the top ranked dicts the market shares for the post-reforms pe-
public sector incumbents, it is unlikely that riod by apph{ng the structure prevailing in the
entry and expansion at the lower endwill change pre-reforms period and then tests whether the
the nature of established relationships among prediction errors are statistically significant. Sig-
the top banks. nifi cant prediction errors imply a structural break
Competition in Market Shares: Reallocation provided the division of the sample based on a
of market shares among competitors is one of priori knowledge is correct (Harvey, 19Bl). An
the early manifestations of competition in any advantage of, such a methodology is that it en-
market consequent to deregulation. A signifi- ables us to circumvent the problem of having
cant loss of market share of incumbent banks only few observations in the post-policy period
would signal the fact that entrants have been and helps to ident-i$'the exact break-point in the
successful in competing av{ayshares from the post-policy period.s
incumbents by offering lower prices/better ser- In order to testfor structural change in market
vices. On the other hand, the absence of such shares of different bank groups for any given
changes subsequent to deregulation, would deposit market, we estimate simple regression
imply that entrants have sought to occupy (but equations of the form:
Dnegulah,on 37

I r,= ar+ B, time * Er, (1) is particularly applicable in the context of our
analysis since deregulation in the banking in-
where )* is the observed market share of the gth dustry in India involved not only dismantling
bank grbup (g = SBI&A, NBs, DPVBs, FBs) in barriers to new entry, but also barriers to mobil-
period t ( t= 1980,......,1997-98). The rnarker share ity through liberalization of branch licensing
is assumed to be a function of an intercept term and expansion policy.
and a trend variable, tirne. This simple specifica- For our analysis,we use time-series data on
tion of the regression equation fits the data quite total deposits of the four major bank groups,
well asjudged by the R-square and the F-values disaggregated by three primary regions. Ide-
reported later in the paper. ally, an analpis ofcompetition in marketshares
The period of our study lies between 1980 and should focus on the local geographical market
1997-98, subdivided into two groups, January where branches of different bank groups com-
1980 to March 1994 and April 1994 to March pete head on. Flowever, we do not have such
1998. To test for structural break, we estimate branch level data disaggregated by local mar-
equation (1) using the ordinary least square kets. Instead, we use bank group ievel data
(OLS) techniqueo, for each bank group covering disaggregated by major states in which these
the / pre-policy years, i.e., l9B0 to 1993-94.N- groups operate. The deposit data has been
though, officially, the initiation of banking sec- drawn from 15 major states, accounting for
tor deregulation can be traced back to 1991-92, more than 90 per cent of total deposits for each
we take 1993-94 as the cut-off point for the pre- bank group and these stateshave been divided
policy period because the first set of entrants into three geographical regions according to
established themselves in the market during the the following criteria.
course of 1993-94 and their first full year of
operation was 199495. The post-deregulation
Table 7
sub.period is 1993-94to1997-98.We then test rhe
Share of Deposits and Share of Brandres of Bank
null hypothesis that the observed market share in
Groups Across Selected Regions
the /+lthyear (i.e., the firstyearin the post-policy
period) is not statisticallydifferent from the mean Share of Share of
market share estimated for the pre-policy period. Deposits in Branches in
If the null hypothesis of "no structural break" is Region Region
rejected, i.e., there is a structural break in the
1991941997-981993941997-98
t+lth year, we repeat the same test using the
pooled three-year data of the post-policy period.
I. Region1
This is to test whether all the three observations
in the post-policv period, taken together, could
SBI&A 2z.s 18.7 2r.0 20.6
NB 54.8 54.9 7 l . B 7 0 . 7
have been generated by the pre-policy structure. DP\ts 3 . 7 9 . 7 6.0 7.4
On the other hand, if the null hypothesis is not FB 19.0 16.6 1 . 1 1 . 3
rejected, thenwe pool the t+1th observationwith
the pre-policy period and repeat the test for the II. Region 2
remaining two years. SBI&A 31.6 29.2 27.2 26.7
An important point to be noted in the context NB 52.1 47.6 54.0 53.5
of our empirical analysisis thatwe use the terms DPVB r4.B 20.6 l B . 7 1 9 . 6
'entry' 'entrant' FB 1.5 2.6 0.1 0.2
and in a broad sense by consi-
dering not only nelv private sector banks that III. Region3
move from 'zero outputs to positive outputs' SBI&A 31.8 30.7 28.9 28.7
through entry (Bain, 1956), but also intergroup NB 66.5 66.2 69.3 69.2
shifts among incumbent banks (Caves and Por- DPVB 1.5 2.7 r.7 2.2
ter,L977). As Cavesand Porter argue, in the case FB 0.2 0.4 = Q = Q
where an industry contains subgroups of firms III. All India
with differing structural characteristics "barriers SBI&A 30.2 25.7 26.s 26.6
to mobiliLy betweengroupsrest on the same struc- NB 56.2 56.8 64.3 63.8
tural features as barriers to entry into any sroup DP\ts 5.6 10.2 8.5 9.4
from outside the industry. " This characterization FB 8.0 7.3 0.3 0.4
38 Intnnational loumal of Dnelopment Banking

Region 1 comprisesthree shtes/union terri- Table 8


tory, viz., New Delhi, West Bengal and Impact of Deregulation on the Deposit Market
Maharashtra, where the operations of foreign Shares of Bank Groups Across Different
banks are relatively the most concentrated com- Regions (F-statistics from Chow test)
pared to that in the rest of India, as is reflected
in their share in these regions being higher 1980-94- 1980-95-1980-9G1980-97-
than the national average. Foreign banks mobi- 199+95 1995-96199G971997-98
lize about 93 per cent of their total deposits
Region 1
from Region l. The presence of domestic pri-
SBI&A 3.20 0.03 1.29 0.50
vate sector banks, both old and new, is also
significant in this region. Region 2 comprises
NB 2.16 r.Lz 0.10 0.13
PVB 15.02*
six states,Tamilnadu, Kerala,.Andhra Pradesh,
FB 2.40 L.34 0.30 2.2r
Karnataka, Jammu & Kashmir and Rajasthan,
PB 00 t.t4 0.60
where the old domestic private sector banks are
mor'e concentrated as is reflected in their mar- Region 2
ket share in these regions being higher than the SBI&A 1.60 0.07 0.65 4.72*
national average. In Region 2, the presence of NB 1.51 0.26 2.04 0.59
foreign banks is marginal, accoun ting for around PVB l0.l7x
six to seven per cent of their total deposits. FB 1.65 1.60 0.40 15.00*
Region 3 comprises six states, Uttar Pradesh, PB 12.L7*
Madhya Pradesh, Bihar, Orissa, Gujarat and
Region 3
Punjab, where foreign banks are virtually ab-
SBI&A 0.34 0.05 0.05 0.05
sent (barring a recently opened branch in Uttar
NB 0.50 0.68 0.03 0.44
Pradesh) andwhere the domestic private banks
PVB 2.35' 0 3.60 3.08
also have a marginal presenqe. The average
FB 00 0 3.75
sharesof different bank groups in deposits and
PB t.54 0.10 3.70 4.r0
in total branches in the three regions are given
in Table 7. IV. All India
The rationale behind the regional classifica- SBI&A 4.L5 0.15 1.50 1.95
tion is to isolate the different tFpesof competi- NB 3.36 0.56 0.05 0.03
tion that public sector banks are facing in the PVB 16.4*
deregulated regime. Region l, for example, FB L.37 2.22 0.20 I.02
represents markets where public sector banks PB 2.80 0.r4 3.23 0.35
face the most competition from foreign banks
and private banks. Similarly, Region 2 repre- * Denotes that the F-valueis significant at 5 per cent
sentsmarketswhich capture the extent of com- leveland marksa structural break in market sharesfor
petition being faced by public sector banks that particular bank group.
from the old private sector banks, which are
now allowed to expand their operations in the share in branches, the presence of the foreign
deregulated regime. Region 3, where public banks is most felt in Region I and domestic
sector banks dominate, will capture the extent private banks are relatively the most concen-
to which public sector bank groups are compe- trated in Region 2. However, despite an in-
tingwith each other in the deregulated regime. crease in the shares of both these groups of
Data on state-wise deposits by four major banks in the branch network and the conse-
bank groups, namely SBI&A, NBs, DPVBs and quent improvement in the accessibilityof these
FBs, have been compiled from Basic Statistics banks to customers, the impact of greater post-
(BasicStatisticalRzturns), published annually by deregulation .activity of these banks on their
the RBI. These publications provide bank group market shares has at best been marginal.
wise data on afgregate deposits for all states. Regression results reveal an interesting pic-
Results:The mean sharesof each bankgroup ture of the dynamicsin the differentregions.In
for each region and for the country as a whole Region I (Table 8), only the domestic private
are shown in Table 7. As is evident from the sector banks as a group have experienced a
Table, both with respect to market share and structural break in their market shares upon
LrereguLanon 5v

entry and expansion. It is in this region that Finally, the aggregate picture of changes in
mostof the newentryofprivate sectorbanks has market share testifies to the fact that there has
occurred and, hence, the result is not surpris- been statistically no change in the public sector
ing. The overall share of these banks has in- banks' share in the deposit market during the
creasedfrom 3.7 per cent in 1993-94to 9.7 per last five years. While the share of these banks
cent in 1997-98.However, such a gain in market have come dorrm, our econometric analysis
share has not been sufficient to cause'any sig- seemsto suggestthat this decline till now is still
nificant decline in the market share of any of part of the time trend since 1980 and cannot be
the public sector bank groups either separately specifically attributed to deregulation.
or taken together. Instead, the increase in the The results of our regression analysis are
share of the domestic private banks has largely consistentwith our theoretical arguments and
been at the expense of the foreign banks whose the exploratory data on the structure of the
share declined from 19 per cent in 1993-94 to Indian banking industry which was presented
16.7 per cent in 1997-98, notwithstanding an earlier. Importantly, our analysisindicates that
increase in the latter's share in total branches. foreign banks have not been a competitive
This result is consistentwith the earlier observa- threat where its presence is the most, i.€.,
tion that the new private banks have their busi- Region l. It points to the fact that despite their
ness profile very similar to that of the foreign increased presence and superior quality of
banks. services they have not been able to match the
In Region 2, where private sector banks are public sector banks in terms of accessibility
relatively more concentrated than in the rest of and hence have continued to remain in their
India and hence are relatively more accessibleto pre-existing niche.
the depositor, the signs of competition are more With regard to the domestic private sector
apparent. Here too, the clomestic private sector banks which are more similar to public sector
banks have experienced a significant increase in banks in terms of depositor-profile than the
market share, with its share increasing from l4.B foreign banks, there are some signs of'limited
per cent in 1993-94to 20.6 per cent in 1997-98, competition especially where the former are
with the structural break in market shares ap- concentrated. While part of the increase in
pearing in 199495. The postderegulation gain market share may have followed from new entry
in the market share of domestic private banks in and liberalized b.ranch licensing policies, this
Region 2 has resulted in a significant loss in the increase is also consistent with the fact that the
market share of public sector banks asa group in old domestic private sector banks have become
199495 and a significant fall in the market share more efficient since deregulation, unlike the
of the State Bank group between 199G97 and other groups and the new private banks are the
1997-98. With regard to foreign banks, while most efficient among all groups (Table l). The
their presence is relatively minuscule in Region latter conjecture has its underpinning in the
2, there has been a significant change in their efficiency market hypothesis (Demselc-, 1973)
market share in 1997-98,pointing out to the fact which states that more efficient firms are in a
that these banks are now exploring new centres position to lower their prices and thereby in-
for resource mobilization awayfrom their tradi- crease their market shares.
tional markets. The number of branches of With regard to public sector banks, Region 3
foreign banks in this region increased from 22in gives us a benchmark of competition between
1993-94 to 34 in 1997-98, with market share these banks in the relative absence of private
increasing from 1.5 per cent to 2.6 per cent sector banks. Given that both the State Bank
between the two years. group and nationalized banks have an exten-
With respect to Region 3 where public sector sive branch network, accessibility is unlikely to
banks have predominated both during the regu- be a major determinant of market share. In-
lated and deregulated periods, the markets share stead,competition for market sharescould have
of none of the bank groups show any significant stemmed from competition in prices and ser-
change. The State Bank group and the national- vices. However, there are no indications of any
ized banks have continued to hold on to their marketreallocation in Region 3, pointing to the
respective market shares as well as the share in possibility that the public sector banks are still
the total branch network. perceived as close substitutes b,vdepositors.
40 Intem,ational loumal oJ Dnelopment Banhtng

ry and Gorton (1994), large-scale entry of banks


POLTCY IMPLICATIONS may reduce the franchise value of bank licenses
through the erosion of bank profits and increase
Our theoretical arguments and empirical the probability of bank insolvency. Profits de-
analysishighlight the limits of policies that rely crease on entry as entrant banks, in an effort to
prirparily on entry and interest rate deregula- capture market shares, underprice loans and
tion to generate increased competition in a overprice deposits. International evidence sug-
banking systemwhere the less efficient incum- geststhat the chances of bank insolvency may be
bents (i.e., publicly owned banks) are already further aggravated if interest rates are fully liber-
entrenched in the market through an extensive alized (Caprio eta1.,1994).In light of our results
branch network. Our regression analysis of ofincreased competition atthe fringe, such com-
changes in market shares reveals that some petition will inevitably affect the smaller public
pockets of competition (Region 2), have in- sector banks and increase the probability of their
deed appeared in the lastfive years,but that the insolvency. As such, an increase in the number of
oligopolistic dominance of public sector banks fragile banks is not inconsistent with the logic of
in the deposit market and branch network con- competition. But, in that case,exit of the fragile
tinues. In the presence of such size-based"cush- and lessviablebanks shouldform an integral part
ions," the critical minimum level of competi- of a policy that seeksto promote efficien cy by way
tion needed for generating efficiency gains of increased domestic and international corSrpe-
among public sector banks may be difficult to tition. In the Indian context, given the rigidities
attain. in the labour market and the virtual absence of
Our analysis of trends in concentration ra- an exit policy, dissolution as well as large-scale
tios in the total and savingsdeposit market and mergers and acquisitions of the non-competitive
trends in interbank mobility among the top l0 weaker banks are unlikely, atleast in the short to
banks and among all banks suggestthat atbest medium run. Herein lies the fallacy with the
competition has emerged at the fringe. This, in policy of continuous deregulation.
turn, is presumably because the entry of banks Given the potentially limited impact of de-
has been at the lower end, with the position of regulation in the long run, with the risk of pos-
the dominant public sector banks not under sible insolvency of the relatively weak banks with
any threat either from the private sector en- large-scale entry of private banks and interest
trants or the incumbents. Given non-regulatory rate deregulation, a logical alternative could be
barriers as the potentially non-bridgeable size the privatizationofpublic banks in order to both
'Jump-start" efficiency improvements in these
of public sector banks in India, our analysis
highlights the fact that the size of entrants is as banks and infuse fresh capital into the weaker
important as the policy of entry deregulation banks. Cross-countryexperiences have suggested
itself. A similar conclusion wasdrawn by Denizer that "banking is no exception to the general rule
( 1997) , after analysing l0 years of deregulation that autonomy, competition and hard-budget
in Turkish banking. An important factor that constraints cannot be achieved without
should be noted in this context is thatwhile the privatization" (foshi and Little, 1996) and
policy of deregulation has been introduced to privatization of banks has been an option exer-
remove distortions in the market and increase cised in countries such as Korea, Chile, Hungary
bank efficiency, the distortions caused by the and Poland. The other alternative is to increase
previous 25 yearsof regulated interest rates and competition for the banking sector as a whole
excessivebranch expansion, still continue to from non-banks and local banks, a policy move
persist in large measure. that has been adopted in India in recent years.
While, in the absence of tell-tale signs of However, here too, the retail banking sector is
competition, a case can be made in favour of much larger in size relative to the non-banks,
further liberalization, banking disastersaround although the importance of the larger is growing
theworld are grim reminders thatunmitigated ais-d,-aisthebanking sector; in 1995-96,the regu-
deregulation can have a high cost, a possible lated deposits of non-bank financial intermedi-
outcome being a decline in the chartervalue of aries were 7.6 per cent of total bank deposits.
banks and excessiverisk-taking in the banking As Caprio et al., (1993) have noted, the expe-
sector.As argued by Caprio and Summers ( 1993) riences of both the developed and developing
Davgul,ah,on 4l

countries which implemented financial deregu- estimate the mean marketsharesfor the struc-
lation and liberalizattonpolicies since the 1980s ture that actually existed in the pre-policy
suggest that a balance needs to be struck be- period (1970-92) and the mean market shares
tween a number of potentially conflicting objec- for the structure in the post-policy period
tives and the risk that deregulation entails. Our (1992-95) and then assessif the differences
analysis has attempted to highlight the poten- between the mean shares estimated for the
tially marginal impact of belated deregulation monde andfor the anti-monde are statisticallv
and the limitations of considering deregulation significant. In our case,the limitation ofapply-
as a sure-fire way of increasing the efficiency of ing the analytical method is thatwithjust three
the banking sector in the long run. The onus on observations in the postpolicy period, it is not
the policy-makers, therefore, is to supplement possible to estimate the mean market shares
deregulation with other policy measures, while accurately. The residual method circumvents
keepingin mind the trade offbetween efficiency this limitation.
gains and systemic problems. 6. Subsequent to the estimation of the model,
the Durbin-Watson statistic revealed the pres-
ACKNOWLEDGEMENTS ence of autocorrelation. The equations were
therefore re-estimated assuming that the er-
The authorswould like to thank Subrata Sarkar rors followed a first order: auto-regressive pro-
and Rajendra Vaidya for some valuable com- cess. The Durbin-Watson statistics of the re-
ments on earlier drafts of the paper. The usual estimated model suggested the absence of
disclaimer applies. higher order autocorrelation.

NOTES REFERENCES
1. While a consideration of competition in the
loans market is equally important in provid- Bain, J. S. , Ba,ffiers to Nal Competition, }{anrard
inguswith acomplete picture ofthe competi- University Press,Cambridge, 1956.
tive environment, its exclusion is dictated
mainly by the non-availability of recent data. Barros, P. P., "Post-entqyExpansion in Banking:
However, we feel that this exclusion does not The Case of Portugal", IntrnationalJaumal of
dilute the importance of our present exer- Ind,ustrial Organisation, Volume 13, No. 4, 1995,
cise. This is especially true, since in the ab pp.593-612.
sence of little overt link benveen capital mar-
Caprio, G., arrd Summers,L., Fi,nana and,ItsRefonn:
kets and banks in developing countries like
India, deposits are the life blood ofbanks and ryond l^aksa,Fai,re,World Bank Working Paper,
World Bank, Washington D.C., 1993.
competition for deposits is thus a key area of
competition. Caprio, G., I. Atiyas and J. Flanson, Financial
2. See Tirole (1988) fqr a discussion of the Rcforms: Theory and Expni.mce, Carnbridge Uni-
theoretical arguments. versity Press, 1994.
3. Typically, such banks belong to the public
sector, as in Chile (prior to privatizaion), Caves, R. E., and Porter, M. E. , "From Entry
Indonesia, Sri Lanka, Turkey, India and the Barriers to Mobility Barriers: Conjectural Deci-
east European countries. sions and Contrived Deterrence to New Compe-
4. The importance of the branch network of tition", QuartnlyJoumal ofEconomics,Volume 91,
incumbent banks is further highlighted by No. 2, 1977, pp.24l-262.
the observation that such a network would Demsetz, H., "Industry Structure, Market Rivalry
provide a "platform" for distribution of a and Public Policy",Jru,mal of Law and Economics,
wide range of newfinancial services (such as April lg7\,pp. i-tO.
insurance and equity related services), in
addition to its existing functions of deposit Denizer, C., 'The Effects of Financial Liberaliza-
mobilization and credit delivery (MOF, tion and New Bank Entrv on Market Structure
leez). and Competition in Turke/", Policy Research
5. Another method of applying the Chow test is Working Paper 1839, World Bank, Washington,
the "analytical method" by which one can D.C. 1997.
42 Intrnational.loumal of Dnelapmar,t Banhtng -

Dermine, Jean, European Bankirtg in the 1990s, Indian Banks'Association, Perfom,anceHighkght


Second edition, Blackruell Publishers, Oxford, ofFmeign Banhs,various issues.
1993'
r^o}.i \/iio.,
Vtjry o-l r il,r r\ Little,
and I.M.D. r i++lo Ind,ia's
) Economi
Joshi,
"Refurts
Geroski, P., Marhct Dynamia and,Entry, Blackwell iggl.ZOOl, Oxford University Press,Nev
Publishers,'Oxford, UK 1991. Delhi, 1996.
'Markets
Geroski, P. andJ. Schwalbach (ed.) , Entry and Klemperer, P. , with Consumer Switch
Marhet Contestabikty: An Intmmtional Compafi,son, irtgCostsiQua,rtnlyloumalof Economia,Volum<
Blackwell Publishers, Oxford, UK 1991. 102,LggT,pp. 375-394.
Harvey, A. C. , The EconometricAnalysi,s of Timc McKinnor, R., MonE and Capital in Econorni
Sriu, Philip Allan, Oxford, 1981. Danlapmmf, Brookings Institution, Washingtor
.D.C., 1973.
Heggestad, A. and S. A. Rhoades, 'Concentra-
tion and.Firm Stability in Commercial Banking", Ministry of Finance, Reportof the Committeeon th
Rniar ofEconomia and Statistia,Volume 58, I 976, Finan cialS)seera(Narasimham Comrnitee Report)
pp.443452. Ministry of Finance, Government of India, Nev
Delhi, 1973.
Indian Banks' Association, Ind,ian Banking Year-
book(1994, 1995), Mumbai. Ministry ofFinance, Public SectorCommerciatBank
andFinanci,al SectmRcfom,: Rzbu,ild,ingfm a Bette
Indian Banks' Associatio n , Financial Analysis of Future, Discussion paper, Ministry of Finance
Banks, various issues. Government of India, New Delhi, 1973.
Indian Banks' Associatio n, Prfonn ana Highkghts Reserve Barrk of India, Trmds and hogras o
of Public SectmBanhs, various issues. Banking in Indi,a,l99394, Mumbai, 1994.
Indian Banks' Associatio n, Performana Hi ghtights Tirole , J., Thd Tlreury of Industrial Organization
of hiaate SectmBanhs, various issues. The MIT Press,Cambridge, Massachussetts,1988