Revisiting Bank Mergers Does Size Matter?

SPECIAL ARTICLE

Subhanil Banerjee

The central government’s policy favouring bank mergers he drive towards bank merger in India began with the assumes enhanced efficiency of the merged banks recommendations of the Narasimham Committee on the Financial System in 1991. The committee’s recom- through economies of scale and scope. An econometric T mendations on merger were ignored at the time. They made a analysis of India’s scheduled commercial banks, stronger comeback in 1997–98 as part of the Narasimham however, establishes that in the Indian banking sector, Committee on Banking Sector Reforms’ recommendations on mergers may actually be detrimental to efficiency. This structural reforms. This time the government embraced the recommendations and paved the way for bank mergers (Bagchi paper argues that public sector banks were set up to and Banerjee 2005). serve the welfare needs of the underprivileged and to Apart from mobilising funds from savers to investors, in a promote financial inclusion, not to make profits. In the developing country like India, public sector banks (PSBs) are case of bank mergers, economies of scale and scope are expected to facilitate ﬁ nancial inclusion and provide cheap loans to the poor, marginal farmers, small enterprises run by being used to veil the promotion of economies self-employed artisans and transport operators. The poor per- of exclusion. formance of the Indian banking sector in terms of ﬁ nancial inclusion is evident in the speech of Usha Thorat, former deputy governor of the Reserve Bank of India (RBI):

on an all-India basis ... 41% of the population is unbanked. In rural areas the coverage is 39% against 60% in urban areas … [The] number of loan accounts constituted only 14% of [the] adult population … In rural areas, the coverage is 9.5% ... Regional differences are signifi - cant, with credit coverage at 25% for the southern region and as low as 7%, 8% and 9% respectively in [the] north-eastern, eastern and central regions … 89 million are farmer households. 51.4% of farm households have no access to formal or informal sources of credit, while 73% have no access to formal sources of credit … The share of moneylenders in the debt of rural households increased from 17.5% in 1991 to 29.6% in 2002. (Thorat 2007: 2–3) Iyer (2015: 20) also confi rms that “small and marginal farmers who own more than 80% of the agricultural holdings are ‘disturbingly’ becoming more indebted to the moneylenders.” It is apparent that the Indian fi nancial system has not only fall- en short of attaining meaningful fi nancial inclusion but also displays regional and urban–rural disparity and increased dependence on traditional moneylenders. Instead of considering these lacunae in the Indian fi nancial sector, the Indian government has been promoting bank mergers as a possible solution, on grounds of enhanced effi ciency through economies of scale I am indebted to Amiya Kumar Bagchi for his valuable comments and and scope. suggestions during the development of this paper. I am also grateful to Zakir Husain for his technical insights, and to Sailabala Debi, Ashok 1 Declining Welfare Role of Indian Public Sector Banks Kumar Sar, L K Vaswani, Rabi Narayan Subudhi, Prashant Parida, After the so-called fi nancial liberalisation, Indian PSBs have Jogendra Behera and Sankalpa Bhattacharjee for their encouragement skirted their welfare responsibilities. From 1969–70 to 2014–15, and belief in me. the total bank credit outstanding for all scheduled commercial Subhanil Banerjee ([email protected]) is a research scholar at the banks (SCBs) grew from `3,971 crore to `65,36,420 crore, and KIIT School of Management, KIIT University, Bhubaneswar, Odisha. aggregate deposits grew from `5,028 crore to `85,33,285 crore

Economic & Political Weekly EPW FEBRuary 25, 2017 vol liI no 8 41 SPECIAL ARTICLE (RBI 2015: Table 48). Compare this staggering growth in credit rates, distancing themselves from the needs of common per- and deposits to the fact that only 53% of the adult Indian popu- sons and the underprivileged. Loans on consumer durables are lation had bank accounts in 2014 (Demirgüç-Kunt et al 2015)— at times interest-free, but such loans require a one-time pro- lower than what Thorat (2007) claimed—and defi nitely far cessing fee and are backed by high-value collaterals. Since fewer loan accounts than the 14% stated by Thorat (2007), and many types of personal loans are collateral-free, the higher it becomes obvious that the benefi ts of the Indian banking sys- percentage of personal loans signifi es the growing risk expo- tem have been skewed in the distribution. Indians bereft of sure of Indian PSBs. banking facilities comprise “marginal farmers, landless labour- The welfare role of PSBs has not been suffi ciently recognised ers, oral lessees, self-employed …” (Thorat 2007: 3). The Indi- by successive governments which, behind the veil of effi ciency, an Express, reporting National Sample Survey Offi ce (NSSO have “forced them to downsize their workforce and recapita- 2011) data stated that “51% of the country’s workforces are self- lised those PSBs whose non-performing assets were large” employed, while 33.5% are engaged as casual labour …” (Indian (Bagchi and Banerjee 2005: 1181). For example, in 1997–98 Express 2011). Hence, the majority of the working population— (when the Narasimham Committee on Banking Sector Reforms the self-employed and casual labourers—are not able to access was initiated), employees per branch in SBI and its associates the Indian banking system. Again, PSBs were set up not as profi t- (including administrative offi ces) numbered 23.16, which de- maximising bodies but to extend credit to small and marginal clined to 14.56 in 2012–13 (excluding administrative offi ces). farmers, small business holders and venture capitalists. In the Over the same time frame, employees per branch for other pre-liberalisation period (1975–76 to 1990–91), cumulative nationalised banks declined from 17.19 to 9.67 (RBI 1999, Tables 8 total credit (direct and indirect institutional credit, short- and and 33; RBI 2013, Tables B13 and B15). long-term) to agriculture and allied activities as a percentage of cumulative total credit by the SCBs was 17.06%. This decli- 2 Reasons for Financial Exclusion and Failure of ned in the post-liberalisation period (1991–92 to 2013–14) to Microfinance Institutions in India 13.07% (computed from RBI 2015, Tables 48 and 63). There are several reasons why microfi nance institutions (MFIs) Moreover, credit to agriculture and allied activities as a per- have failed to become a meaningful substitute to formal fi nan- centage of total credit of the State Bank of India (SBI) and its cial institutions for impoverished Indians. Ananth and Öncü associates, had increased from 8.72% (computed from RBI (2013) have illustrated that the rules imposed by the RBI and 1973, Table 4.1) in June 1973 to 17.09% (computed from RBI Government of India on the banking sector to extend formal 1991, Table 4.4) in March 1991, and declined to 11.73% (com- banking facilities to the underprivileged were mostly consid- puted from RBI 2014, Table 5.5) in March 2014. For small-scale ered a regulatory burden and the banks failed to anticipate industries, credit declined from 16.05% (computed from RBI prospective customers at the bottom of the pyramid. Norms 1973, Table 4.1) in June 1973 to 13.72% (computed from RBI such as “know your customer” jeopardised the situation fur- 1991, Table 4.4) in March 1991, and further to 4.99% (computed ther as impoverished Indians belonging mostly to the informal from RBI 2007, Table 5.5) in March 2007. For later years, the sector lacked the basic requirements to even get a formal in- data format has been changed and might not be comparable. troduction to the bank. Oblivious of the potential of these re- On the other hand, consumer durable credit and personal credit source-poor but high-potential customers, banks paid no at- (without housing credit) has increased from 2.22% (computed tention to designing customised savings products for them. from RBI 1973, Table 4.1) in June 1973 to 6.03% (computed from This led to the proliferation of informal fi nancial organisations RBI 1991, Table 4.4) in March 1991 and has further increased to through Ponzi schemes with false promises of abnormally high 9.98% (computed from RBI 2014, Table 5.5) in March 2014. returns that robbed customers of their last fi nancial resources. Outstanding credit to agriculture and allied activities as a Poor villagers often needed home loans, but because their percentage of total credit for the other nationalised banks had property rights are undefi ned, their properties cannot serve as increased from 9.45% (computed from RBI 1973, Table 4.1) in collateral and thus have no mortgage value. Informal money- June 1973 to 14.39% (computed from RBI 1991, Table 4.4) in lenders took advantage of this situation, charging higher inter- March 1991 and then declined to 13.62% (computed from RBI est on their loans and placing a huge interest burden on these 2014, Table 5.5) in March 2014. For small-scale industries, villagers. Landownership is also important in getting formal credit increased from 10.94% (computed from RBI 1973, Table credit (Kamath et al 2010). This keeps small and marginal 4.1) in June 1973 to 14.84% (computed from RBI 1991, Table farmers, share and hired tenants who have little or no land, 4.4) in March 1991 and then declined to 4.60% (computed beyond the consideration of formal credit. Earlier, self-help from RBI 2007, Table 5.5) in March 2007. However, the sum of groups (SHGs) introduced by the National Bank for Agriculture consumer durable credit and personal credit (without housing and Rural Development (NABARD), together with the business credit) increased from 2.31% (computed from RBI 1973, Table correspondent strategy, could have helped, as “the amount of 4.1) in June 1973 to 3.92% (computed from RBI 1991, Table 4.4) revenues that the business correspondents and business facili- in March 1991, and then to 4.84% (computed from RBI 2014, tators are likely to generate [are] at `66,000 crore and depos- Table 5.5) in March 2014. its they are likely to collect [are] at `30,000 crore annually” This portrays that PSBs in India are opting for a credit portfo- (Ananth and Öncü 2013: 82). However, this might not be as lio that will provide them quick revenues from higher interest simple as it seems. Only 11% of bank account holders save, and

42 FEBRuary 25, 2017 vol liI no 8 EPW Economic & Political Weekly SPECIAL ARTICLE

8% take loans, with around 75% of savings accounts lying dor- MFIs in return for an agreement for collection of the loans. mant. This fi gure rises to 80% to 96% for accounts opened by Every time a portfolio was bought out, the MFI would get the business correspondents. Only 3% of households trust business ability to lend and borrow more and therefore expand” (Ghosh correspondents, as 47% of business correspondents become 2013: 1210). To remain profi table, profi t-seeking MFIs began untraceable after opening the account (Iyer 2015). charging rates of interest as high as 30% to 60% against their The recent Pradhan Mantri Jan Dhan Yojana sounds over- loans, often resorting to coercive measures to ensure repay- ambitious, impractical and inadequate. Moreover, it lacks any ment. There was no longer any difference between profi t-seek- offi cial document. By 2018, banks are required to open 150 ing MFIs and traditional moneylenders. Ghosh (2013) also million bank accounts for the estimated 75 million households mentions a legal innovation—“mutual benefi t trusts (MBTs) outside banking facilities. Overdraft facilities of `5,000 to that aggregated the member–borrowers of SHGs as members. each account as promised would burden the banks by `75,000 The grant money received for the purposes of ‘developmental’ crore. In addition, there are 4,90,000 unbanked villages in microcredit could then be placed in the MBT, which would in India which, at a ratio of 1:6, would require the setting up of turn contribute to the newly created for-profi t MFI” (Ghosh more than 80,000 rural branches of banks before 2016, where- 2013: 1211). Again, profi t-oriented MFIs always opted to operate as in the past three years only 7,459 bank branches have been at densely populated places, leading to the coexistence of mul- opened. The business correspondents, who use information tiple MFIs at a single place, giving multiple loans to the same and communication technologies (ICT), need to extend their borrower households, with a higher risk of default. Further, coverage by almost two lakh villages to provide absolute inclu- MFIs sometimes openly linked consumer goods with their sion, whereas ICT coverage in rural India is still low. Aadhaar- credit programme, aggravating the indebtedness of the bor- enabled direct benefi t transfers to the accounts of benefi ciaries rowers and leading to loan default. In such circumstances, the are also doubtful as biometric machines often fail to identify borrowers often resorted to traditional moneylenders to repay the biometrics of labourers and the aged. Moreover, for Aadhaar- these loans, being trapped in a vicious circle of debt that ulti- enabled RuPay cards to function properly, a large number of mately led to suicide (Ghosh 2013). ATMs are needed in rural and semi-urban areas (Shetty and Deokar 2014). While a bank account is a necessary condition 2.1 Inorganic Development Dreams for fi nancial inclusion, it is not suffi cient in itself, as It is apparent from the discussion so far that the Indian fi n- with no fi nancial capacity to save and invest, a dismal record of use ancial system has three main problems: the dismal condition of bank accounts and the severe lack of trust in the current ‘rock star’ of fi nancial inclusion, failure of MFIs, and failure of PSBs to model of using business correspondents … [it is diffi cult to see] how serve the welfare needs of the rural masses. How bank merg- opening of bank accounts will … inculcate the habit of saving among ers will address these problems is a mystery. Mega banks and the poor. (Iyer 2015: 20) internationally competitive fi nancial institutions hog the The much-celebrated MFIs, though set up as a tool for pov- newspaper headlines while the essence of the Indian banking erty alleviation, eventually became mere profi t-seeking enti- sector is lost amidst all these inorganic development dreams. ties, charging exorbitantly high rates of interest on their loans, The disinvestment spree within the Indian economy has shift- with very short repayment time. Most of them became non- ed the ownership of a considerable volume of public capital. bank fi nancial institutions after 2005 to attract higher capital PSBs are repeatedly encouraged to enter the stock market to from donors and international lenders. Their fl awed business raise resources and to introduce new investment options for models were subject to minimum prudential norms and regu- retail investors. Successive governments have carefully kept lations, and most had no record of the end use of their loans. the idea of more autonomy to the PSBs alive. The recent Coercive measures were used for loan recovery. The prolifera- P J Nayak Committee constituted by the RBI has suggested tion of MFIs did not augment the income of borrowers as only a reduction of the government’s stake in PSBs to less than 51% small portion of the loans was channelised to productive (Nayak 2014), which might eventually provide more autonomy activities. Moreover, strategic defaults and high interest on for state-owned banks in investment decisions, tenure and with- loans led to moral hazards (Ananth and Öncü 2013). Ghosh (2013) drawal of invested funds without prior government approval, illustrates that MFIs in India started with a development goal assuming a probable loss and dissatisfactory business return. and as non-profi t bodies. They began group lending for SHGs These steps would ultimately boost the stock prices of the PSBs and connected with commercial banks through the SHG–Bank in the Indian stock markets, benefi ting a minuscule percent- Linkage Programme (SBLP). Unlike other MFIs around the world, age of the Indian population. Indian MFIs were prohibited from keeping deposits by law and The present paper, however, does not digress into justifi ca- could not extend credit in case donor funds were not available. tions for turning PSBs into profi t-making bodies at the cost of That put them into capital scarcity for expansion and turned welfare, but discusses some of the recent proposed mergers on them into profi t-making entities (Ghosh 2013). Such a transfor- the basis of available data. This paper also makes no addition- mation was supported by the Small Industries Development al comment on the role of PSBs as stock market brokers instead Bank of India (SIDBI) and Industrial Credit and Investment of priority sector lenders and promoters of fi nancial inclusion. Corporation of India (ICICI). ICICI “launched a securitisation It attempts to empirically check whether there is any basis to product in 2003, wherein it would buy out the portfolio of the the promised fruits of bank merger in the Indian context or

Economic & Political Weekly EPW FEBRuary 25, 2017 vol liI no 8 43 SPECIAL ARTICLE whether there are some other vested interests on behalf of the Credit to small business and insignifi cant borrowers does government and other stakeholders that propel the merger wave. indeed decline after a merger of banks, irrespective of the country of origin (Berger et al 1998). However, the negative 3 Efficiency through Bank Mergers: Some Ambiguities static effect of bank mergers would eventually be offset by Proponents of bank merger usually forward economies of scale, positive dynamic effects, where other banks would come for- economies of scope and X-effi ciency gain as the rationale for ward to fi ll the vacuum of lending to small and insignifi cant the move. However, certain theoretical and technical bottle- borrowers (Berger et al 1998). The Indian government, how- necks need to be taken into account. With an increase in the ever, does not wish to leave enough players who might guide scale of operations, economies and diseconomies of scale are the course of actions reasonably. Mergers and acquisitions and both possible. Hence, the optimum scale of operation needs to benefi cial banking are usually inversely related, as evidenced be determined. Existing literature on the scale effi ciency in in Dymski (1999) who stated that fi nancial market consolida- banking fails to determine any common and defi nite volume tion leads to monopolistic pricing refl ected through higher of assets for banks that may be considered a scale effi cient bank fees and lower rates of interest on deposits for customers point. Measuring scope economies poses even more serious as well as declining rates of loan approval. problems, and existing literature has showed that measured Moreover, Ray (2003) had indicated that the credit–deposit scope economies at times far exceeded the credible level. The ratio in south India was higher than the eastern region pre- and X-effi ciency gain measurement, on the other hand, suffers post-nationalisation of banks. In fact, a similar trend is visible from the dilemma of how to measure the differences in X-effi - in the post-liberalisation period. In 2011–12, the credit–deposit ratio ciency arising out of random errors. The different econometric of all SCBs (outstanding credit/total deposit) for south, central methods applied to measure the same have yielded different and east India was 0.96, 0.47 and 0.51, respectively (RBI 2013, results and therefore cannot be trusted (Berger et al 1993). Table 2.2). Again, outstanding credit to small borrowers as a Surprisingly, the government has not made any effort to deter- percentage of total outstanding credit for all SCBs was 0.25 mine the scale effi cient point for banking, nor has it tried to and 0.04 for the north-eastern and western region (RBI 2013, measure scope or X-effi ciencies, if any, before favouring bank Table 2.2). This signifi es that not just the intensity but also the merger. There is also the basic question of whether banking nature of credit extension is different across regions in India. should be seen through the same lens as any other form of pro- Hence, banking is a highly diversifi ed industry that deals with duction (Bagchi and Banerjee 2005). The bank merger spree customers with different needs and activities and different worldwide followed the success of bank mergers in the United levels and types of risk exposure. For similar activity and States (US) that kept 20% of the US population out of banking related products, the two basic concepts of economies of scale services (Dymski 1999, 2004). and scope respectively do not apply to banking. The consumer debt-fuelled growth of US banks leading to un- sustainable levels of current account defi cit was also explained 4 The Reality of Proposed Bank Mergers in India in Bagchi and Banerjee (2005). Bank mergers in the US are In a recent paper, Sanyal and Das (2013) showed the tendency usually justifi ed as having added to the shareholders’ value, of the central government to infl ate micro data in macro sum- but in reality there is ample evidence to the contrary. Epstein mation to present a positive picture. Therefore, researchers (2005) had considered the merger of J P Morgan and Chase should be extra careful when accepting the central govern- Manhattan banks successful. In 2001, one year after the merg- ment’s postulated sequence of events. er, the annual average adjusted lowest share price for the con- Without any empirical and theoretical validation, the Indian cerned bank was $28.77 per share, which managed to rise to government indicated a push towards bank mergers. The pos- $29.39 for the fi rst time in 2004, before rising further to $40.12 sible mergers of SBI and State Bank of Patiala (SBP), United in 2007. After that, even until 2012 ($36.74), the annual aver- Bank of India (UBI) and IDBI, Dena Bank and Punjab National age adjusted lowest share price for the concerned bank never Bank (PNB), Oriental Bank of Commerce (OBC) and Bank of managed to cross the level attained in 2007.1 If the sharehold- India (BOI) are in the news. The government wants to put one ers had actually gained from such share price behaviour then weak and one strong bank together, hoping that the stronger it seems to have been a gain by chance rather than gain from bank will eventually transmit its effi ciency to the weaker bank sustained growth of the share price. Thus, it is apparent that and the merged entity would become effi cient. The govern- justifi cation of bank mergers on the basis of share price appre- ment has stated that UBI has been struck by a succession of ciation is fl awed as well. loan defaults whereas IDBI has a minor presence in eastern In- Bagchi (1989, 1997) had also elaborated on the insignifi cant dia and a weak retail network. On the other hand, SBP has relationship between quality of banking and size, by providing been accused of continuously high levels of non-performing historical examples in the Indian context. The Bank of Bengal assets (NPA). The government has also stated that the OBC possessed over three times the capital base of the Bank of might be better off if it got a helping hand from BoI (Gupta and Madras, yet the rate of growth of deposits and advances was Sidhartha 2014). much slower for the Bank of Bengal than the Bank of Madras. As of 30 July 2014, the gross NPA as a percentage of gross Again, the Bank of Bengal was more prone to serve the Euro- advances (henceforth NPA percentage) for SBP was 4.8% for pean and Indian elites, ignoring the cooperatives altogether. the preceding quarter (Sinha 2014). However, according to the

44 FEBRuary 25, 2017 vol liI no 8 EPW Economic & Political Weekly SPECIAL ARTICLE latest annual data available, the gross NPA was 3.25% for the sector advances of OBC can explain its slightly lower NPA SBP and 4.75% for the SBI (RBI 2013, Table B7). Hence, justify- status, but the concerned bank has performed much better ing the proposed bank merger on grounds of improving the than its proposed acquirer in terms of percentage of total NPA of the SBP raises considerable ambiguity. Again, though income to total assets (9.65% to 7.88%) (RBI 2013, Tables B1 SBI registered a superior percentage of advances to deposits and B2). It is surprising how the proponents of bank merger, (86.94%) over SBP (83.23%) in 2012–13 (RBI 2013, Table B1), if who are committed to the strong bank–weak bank merger the percentage of domestic deposits to domestic advances is theory, can even think of merging these two banks. considered, the SBI actually becomes the weaker of the two As discussed above, the government lacks any defi nite merger (77.71% to 83.23%) (RBI 2013, Table B1). Moreover, SBP accou- policy based on strong theory and empirical analysis. The gov- nted for a higher percentage of total income to total assets ernment appears to be safeguarding its own interests and the (9.51%) than SBI (8.66%) in 2012–13 (RBI 2013, Tables B1 and interests of the few at the cost of the welfare of the masses. B2). Surprisingly, the percentage of priority sector advances to Economies of scale and scope are being used to veil the promo- total advances for SBP (31.33%) was much higher than that of tion of economies of exclusion. Most members of the govern- the SBI (25.28%) in 2012–13 (RBI 2013, Table B1). If economies ing bodies of banks live in the hope of more power, more of exclusion are the only way out for Indian banks then this remu neration and more perquisites, and are therefore com- would have been diametrically opposite. As SBI does not enjoy mitted to bank mergers as well. However, even if the above any meaningful superiority over the SBP, the sought merger is data considerations were discarded as casual observations, the not justifi ed by the assumption of better post-merger effi ciency existing academic literature on the possible costs and benefi ts gains for the merged banks. On the contrary, there is a possi- of bank mergers would also not help in justifying bank merg- bility that SBI, with the advantage of sheer size, might force the ers on grounds of effi ciency. merged entity to follow its own path of economies of exclusion. Although IDBI had a better NPA position than UBI in 2012–13 5 Bank Mergers: A Glance through Existing Literature (3.22% to 4.25%) (RBI 2013, Table B7), its percentage of total Among the advocates of bank mergers, Cornett and Tehranian income to total assets was lower (8.76% to 9%) (RBI 2013, (1992) considered the post-acquisition performance of large Tables B1 and B2). However, the percentage of priority sector bank mergers in the US and concluded that the merged banks advances to total advances was 17.53% for IDBI and 36.49% for outclassed the rest of the banking industry owing to their UBI (RBI 2013, Table B1). As a development fi nancial institu- improved ability to attract loans and deposits, better employ- tion, the higher percentage of advances to deposits for IDBI ment productivity and asset growth. Shaffer (1993) had per- (86.43%) over UBI (68.46%) (RBI 2013, Table B1) is not surpris- formed a merger simulation on US commercial banks to con- ing, but the dwindling of its priority sector lending raises ques- clude that if after the merger the more cost effi cient bank tions about its awareness of the domestic commercial and in- successfully transmits its X-effi ciency to the other then all the dustrial scenario. It seems IDBI’s better NPA has come at the mergers would be cost effi cient. Houston and Ryngaert (1994) cost of the deprivation of the priority sector. IDBI is now ex- lent their support to insignifi cant cost effi ciency realised pected to further de-risk its loan portfolio and focus on retail through bank mergers in the context of the US. Vennet (1996) loans (Narasimhan 2014). As the stronger bank is expected to wrote that in the European Union domestic mergers between become the lead bank after the merger and guide the course of banks of equal size signifi cantly augment the performance of action according to its will, would IDBI compel UBI to deprive the merged banks. the priority sector for better NPA? Resti (1998), using data envelope analysis (DEA), found that The proposed PNB and Dena Bank merger questions whether cost effi ciency of the merged banks improves gradually over the government has a uniform policy on bank merger. PNB had the years following merger. Rhoades (1998) considered nine posted a better percentage of total income to total assets case studies related with horizontal bank mergers and found (9.63% to 8.42%) (RBI 2013, Tables B1 and B2), total advances that all nine were associated with signifi cant post-merger cost- to total deposits (78.84% to 67.67%) (RBI 2013, Table B1) and cutting. Mishkin (1999) opined that fi nancial consolidation in priority sector advances to total advances (30.04% to 28.15%) the US would eventually augment the effi ciency of the Ameri- (RBI 2013, Table B1) in 2012–13 than Dena Bank. However, can fi nancial system but might expose the US economy to Dena Bank enjoyed a far better NPA position (2.19% to 4.27%) greater risk. He added that vigilance along with supervision by (RBI 2013, Table B7) than PNB. Hence, if SBP can be accu sed of the government would be necessary. Houston et al (2001) have high NPA and scheduled for merger, then how does Dena Bank mentioned that the operational effi ciency of merged banks emerge as a candidate for merger? improves as a result of cost savings following the elimination The proposed OBC and BOI merger is also irrational. The NPA of overlapping operations and consolidation of back-offi ce status of BOI (2.99%) is marginally better than OBC (3.22%) operations. Cuesta and Orea (2002) observed that in Spain, (RBI 2013, Table B7). The percentage of total advances to total merged fi rms exhibit better technical effi ciency than deposits is also slightly better for BOI (75.78% to 73.31%) (RBI non-merged ones. Penas and Unal (2004) stated that one of 2013, Table B1). However, the percentage of priority sector the most signifi cant benefi ts of bank merger is that after the lending for OBC (35.86%) is much higher than BoI (22.45%) merger the acquirer is subject to lower cost of funds regarding (RBI 2013, Table B1). The much higher percentage of priority its debt obligations. Sherman and Rupert (2006), using DEA,

Economic & Political Weekly EPW FEBRuary 25, 2017 vol liI no 8 45 SPECIAL ARTICLE had shown that the cost reduction opportunity in terms of advances, net profi t, total assets and total employees. The data branch operating costs is 22% in merged banks, while it was are taken from the RBI’s Statistical Tables Relating to Banks in only 7% before the merger. Campa and Hernando (2006) India (“Table 2: Liabilities and Assets of Scheduled Commer- opined that in the European Union merger and acquisition cial Banks” and “Table 13: Bank-wise and Category-wise Em- leads to improved operational effi ciency. Altunbas and ployees of Scheduled Commercial Banks”). Bagchi and Baner- Marques-Ibanez (2008) declared that on average, bank merg- jee (2005) considered data of the SCBs from 1991 to 2004, thus ers led to better performance for banks in the European Union. leaving out IDBI, one of the major players of the Indian bank- Egger and Hahn (2010) found signifi cant post-merger cost effi - ing sector, which was made public in 2005. The analysis intends ciency gain for the merged banks. to estimate three regression equations as specifi ed below Table 1: Econometric Tests Results through panel data analysis. Dependent Variable Advance-deposit ratio (ar ), ar ir pre it it it it inve stment-deposit ratio (ir ), tasit (Independent Variable) Stat Prob Stat Prob Stat Prob it Wooldridge test for 2.028 0.1587 2.627 0.1095 3.863 0.0532 profi t rate (preit) measured as autocorrelation net profi t divided by number Modified Wald test for 3.6e+06 0.00 4.7e+06 0.00 1.5e+08 0.0000 of employees have been con- heteroscedasticity sidered as dependent varia- Pesaran test for Fixed effect Random contemporaneous model effect model FE RE FE RE ble (one at a time). Repre- correlation 35.082 37.813 0.0000 26.193 25.677 0.0000 42.353 67.177 0.0000 senting effi ciency and total Average correlation assets (tas ) has been taken across units 0.459 0.466 0.421 0.415 0.439 0.503 it as the independent variable On the other hand, those opposing bank mergers such as representing size (scale). Merger necessarily increases the to- Rhoades (1993) concluded that there were no effi ciency gains in tal assets of banks, hence considering total assets representa- horizontal mergers of banks. Berger and Humphrey (1994) stated tive of merger is justifi ed. If any positive and statistically sig- that there was no improvement in post-merger operational effi - nifi cant association is found between the dependent and the ciency for the merged banks. Amel et al (2004) opined that there independent variables as specifi ed, it might be concluded that is no connection between merger and economies of scale, scope merger indeed has a positive impact on effi cient banking. and managerial effi ciency. Rezitis (2008) observed that the tech- The three regression equations to be estimated are: nical effi ciency of merged banks in Greece had decreased after arit = a + b. tasit+ uit …1 i =1, 2, 3, ….. 73 and t=2005, the merger, whereas that of non-merged banks increased. Bernad 2006, …. 2014 et al (2010) found that only half of the Spanish banks posted im- irit= c + d. tasit + vit …2 provements of productivity after a merger. Halkos and Tzeremes preit = e + f. tasit + wit …3 (2013) showed that in the midst of the Greek fi scal crisis the ma- In the above three equations, a, c, and e are the constant jority of the potential mergers would not be posting any short-run terms, b, d, and f are the coeffi cients of tasit and uit, vit, and wit operating effi ciency gain. Further, any potential merger between are the stochastic error terms. Considering the panel structure two cost-effi cient banks would not directly translate to better ope- of the data under consideration and considerable heterogenei- rational effi ciency for the merged entity. ty among units, initially the heteroscedasticity, autocorrela- It is evident from the above discussion that the literature in tion and cross sectional dependence in the residuals have been favour of or against bank mergers is plentiful and mostly con- checked. cerned with the developed countries and post-merger perfor- As illustrated in Table 1, according to the Wooldridge (2002) mances of banks. The acceptability of the results of such anal- test for fi rst order autocorrelation, no fi rst order autocorrela- ysis in the Indian context is doubtful as it is still a developing tion has been found in error structure of the data under con- country and not many bank mergers have taken place in India sideration. The Greene (2000) test for group-wise heterosce- so far. Adopting or refuting bank mergers on the basis of the dasticity declares the presence of heteroscedasticity in the er- experiences of developed countries or based upon individual ror structure of all three panel data models. The Pesaran test successes or failures would be another fl awed decision. (2004) for cross sectional dependence fi nds the presence of At this juncture, this paper opts for an econometric analysis cross sectional dependence in the error structure of all three of Indian SCBs—with the exception of regional rural banks models, with high-level average correlation across units. Cam- and the Bharatiya Mahila Bank—to fi nd the economies of scale, eron and Trivedi (2005) maintain that ignoring heterogeneity if any, in banking. The data analysis does not venture into and correlation across units and over time might lead to biased economies of scope and X-effi ciency, and bank merger enthusi- statistical inference. Moreover, Chudik et al (2011) have said asts might consider this the limitation of this paper. that if the number of cross sectional units (N) is much larger than the time units (T) then cross sectional dependence should 6 Economies of Scale in Indian Banking: always be taken into account. Most recent studies with panel An Econometric Analysis data address heteroscedastic and autocorrelation problems, The present econometric analysis considers balanced panel but cross sectional correlation has been largely ignored (Ardiz- data of 73 Indian SCBs over 10 years (2005–14) on investments, zi et al 2014). With growing fi nancial and economic integration

46 FEBRuary 25, 2017 vol liI no 8 EPW Economic & Political Weekly SPECIAL ARTICLE among countries and fi nancial entities, cross sectional de- As depicted in Table 2, all the coeffi cients of total assets (tasit) pendencies among banks are becoming obvious and consider- are statistically signifi cant and negative. Hence, total assets im- ation of the same is becoming important. In case cross sec- part statistically signifi cant negative impact on advance-deposit tional dependence is for unobserved common factors that are ratio (arit), investment-deposit ratio (irit) and profi t rate (preit). uncorrelated with regressors, fi xed effects (FE) or random ef- This lends support to Bagchi and Banerjee (2005) and proves fects (RE) estimators might be considered and the standard er- that mergers in the context of the Indian banking sector may rors might be corrected following the Driscoll and Kraay (1998) even be detrimental to any sort of effi ciency gain, and economy approach. However, if the unobserved common factors re- of scale in the Indian banking sector is nothing but a web of lies. sponsible for cross sectional dependence are correlated with the regressors then the Pesaran (2006) method should be used Why Persist with Bank Mergers? for estimation of the coeffi cients as the Driscoll and Kraay The lacunae of fi nancial liberalisation in India stand exposed. (1998) method would not work and FE and RE estimators would Banerjee (2015) had shown the increased volatility and higher be biased and inconsistent (De Hoyos and Sarafi dis 2006). negative impact of exchange rate on trade balance after liber- However, if the relation (correlated or not correlated) among alisation against the pre-liberalisation period for India, and questioned the effi cacy of fl exible exchange rate as an Table 2: Regression Results Independent Value Z-Statistics Probability R2 adjR2 F Statistics Probability economic stabiliser facing an economic crisis. Never-

Variable tasit theless, some demonic force drives subsequent gov- Intercept 0.9509185 26.89 0.000 0.0044 0.003 25.21 0.000 ernments to persist with their decades-long mistakes. arit Coefficient -4.45e–08 -5.02 0.000 Apart from the already mentioned problems of the In- ir Intercept 0.588865 35.26 0.000 0.0215 0.020 80.10 0.000 it dian ﬁ nancial sector, Bhattacharjee and Chakrabarti Coefficient -5.44e–08 -8.95 0.000 (2013) had mentioned that one of the main causes of pre Intercept 1.984901 7.45 0.000 0.0152 0.013 27.28 0.000 it the dismal performance of the manufacturing sector is

variable Dependent Coefficient -2.86e–07 -5.22 0.000 the risk-averse behaviour of banks. These immediate unobserved common factors and regressors is not known then problems have nothing to do with bank mergers. Again, as em- the choice bet ween Driscoll and Kraay (1998) and Pesaran pirically proven, increased size of the PSBs may be detrimental (2006) is dubious. Beck and Katz (1995) have used Monte Car- to their performance and the weak bank–strong bank theory of lo Simulation to explain and favour the use of ordinary least merger as conceived by the government is based upon poor em- squares with panel corrected standard errors over more ad- pirics and theory. Perhaps, the government is looking towards vanced FE and RE models as well as generalised least squares the international market and expects PSBs to compete with (GLS) estimator for panel data sets exhibiting both heterosce- multinational banks. Given the greater autonomy regarding in- dasticity and cross sectional dependence. Ardizzi et al (2014), vestment decisions, however, such infl ated fi nancial entities are encountering a similar problem with the data set, has also fol- likely to fi nd the domestic investment scenario more lacklustre lowed the method proposed by Beck and Katz (1995) but with than the international market and may ignore micro-level do- Prais and Winsten (1954) regression to adjust for autocorrela- mestic needs altogether. RBI sometimes says that full capital action. Similar to Beck and Katz (1995), the present analysis con- count convertibility is its ultimate goal, and it may think that siders ordinary least square (OLS) regression with panel cor- bank mergers would move them towards that goal. The dan- rected standard errors (PCSE) to properly adjust for the stand- gers of full capital account convertibility have been evident in ard errors. Unlike Ardizzi et al (2014), Prais and Winsten the East Asian fi nancial crisis, and if the government wants to (1954) regression has not been considered as the data set is gamble with the Indian economy in similar ways then the future is bereft of autocorrelation problem. doomed. Bank mergers should be resisted at the idea stage.

Note Andhra Pradesh,” Economic & Political Weekly, Pre- and Post-reform India: Impact on Balance of Trade,” Utkal Economic Papers 17, pp 60–72. 1 http://fi nance.yahoo.com/q/hp?s=JPM+ Vol 48, No 7, pp 77–83. Beck, N and J N Katz (1995): “What to Do (and Not Historical+Prices Ardizzi, G, C Petraglia, M Piacenza, F Schneider and G Turati (2014): “Money Laundering as a to Do) with Time-Series Cross-Section Data,” Crime in the Financial Sector: A New Approach American Political Science Review, Vol 89, No 3, to Quantitative Assessment, with an Applica- pp 634–47. References tion to Italy,” Journal of Money, Credit and Berger, A N, W C Hunter and S G Timme (1993): “The Effi ciency of Financial Institutions: A Re- Altunbas, Y and D Marques-Ibanez (2008): “Merg- Banking, Vol 46, No 8, pp 1555–90. view and Preview of Research Past, Present ers and Acquisitions and Bank Performance in Bagchi, A K (1989): The Presidency Banks and the and Future,” Journal of Banking and Finance, Europe: The Role of Strategic Similarities,” Indian Economy 1876–1914, Kolkata: Oxford Vol 17, No 2–3, pp 221–49. Journal of Economics and Business, Vol 60, University Press. Berger, A N and D Humphrey (1994): “Bank Scale No 3, pp 204–22. — (1997): The Evolution of the State Bank of India, Economies, Mergers, Concentration, and Amel, D, C Barnes, F Panetta and C Salleo (2004): Vol 2, The Era of the Presidency Banks 1876– Effi ciency: The US Experience,” Working “Consolidation and Effi ciency in the Financial 1920, New Delhi: Sage. Paper 94–25, Wharton School, University of Sector: A Review of the International Evi- Bagchi, A K and S Banerjee (2005): “How Strong Pennsylvania, http://fi c.wharton.upenn.edu/ dence,” Journal of Banking & Finance, Vol 28, Are the Arguments for Bank Mergers?,” fi c/papers /94/ 9425.pdf. No 10, pp 2493–2519. Economic & Political Weekly, Vol 40, No 12, Berger, A N, A Saunders, J M Scalise and G F Udell Ananth, S and T S Öncü (2013): “Challenges to pp 1181–89. (1998): “The Effects of Bank Mergers and Financial Inclusion in India: The Case of Banerjee, S (2015): “Exchange Rate Volatility in Acquisitions on Small Business Lending,”

Economic & Political Weekly EPW FEBRuary 25, 2017 vol liI no 8 47 SPECIAL ARTICLE

Journal of Financial Economics, Vol 50, No 2, Outsiders,” Journal of Financial Economics, — (2007): Basic Statistical Returns of Scheduled pp 187–229. Vol 60, No 2–3, pp 285–331. Commercial Banks in India, Vol 36, Reserve Bernad, C, L Fuentelsaz and J Gómez (2010): “The Indian Express (2011): “Over Half of India’s Work- Bank of India, Mumbai. Effect of Mergers and Acquisitions on Produc- force Self-employed: NSSO Survey,” Indian Ex- — (2013): Statistical Tables Relating to Banks in tivity: An Empirical Application to Spanish press, 25 June 2011, http://archive.indianex- India 2012–2013, Reserve Bank of India, Mumbai. Banking,” Omega, Vol 38, No 5, pp 283–93. press.com/news/over-half-of-india-s-work- — (2014): Basic Statistical Returns of Scheduled Bhattacharjee, S and D Chakrabarti (2013): “Finan- force-selfemployed-nsso-survey/808442/. Commercial Banks in India, Vol 43, Reserve cial Liberalisation, Financing Constraint and Iyer, I (2015): “Financial Inclusion: Need for Differ- Bank of India, Mumbai. India’s Manufacturing Sector,” Economic & Po- entiation between Access and Use,” Economic & — (2015): Handbook of Statistics on the Indian litical Weekly, Vol 68, No 6, pp 61–67. Political Weekly, Vol 50, No 7, pp 19–20. Economy 2014–15, Reserve Bank of India, Cameron, A C and P K Trivedi (2005): Microecono- Kamath, R, A Mukherji and Maria Sandström Mumbai. (2010): “Accessing Institutional Finance: A De- metrics: Methods and Applications, New York: Resti, A (1998): “Regulation Can Foster Mergers, mand Side Story for Rural India,” Economic & Oxford University Press. Can Mergers Foster Effi ciency? The Italian Political Weekly, Vol 45, No 37, pp 56–62. Campa, J M and I Hernando (2006): “M&As Perfor- Case,” Journal of Economics and Business, Mishkin, F S (1999): “Financial Consolidation: mance in the European Financial Industry,” Vol 50, No 2, pp 157–69. Journal of Banking & Finance, Vol 30, No 12, Dangers and Opportunities,” Journal of Bank- ing & Finance, Vol 23, Nos 2–4, pp 675–91. Rezitis, A N (2008): “Effi ciency and Productivity pp 3367–92. Effects of Bank Mergers: Evidence from the Narasimhan, T E (2014): “IDBI Bank Puts Focus on Chudik, A, M H Pesaran and E Tosetti (2011): Greek Banking Industry,” Economic Modelling, Retail Loans,” Business Standard, 21 August, “Weak and Strong Cross Section Dependence Vol 25, No 2, pp 236–54. and Estimation of Large Panels,” the Economet- http://www.business-standard.com/article/fi - Rhoades, S A (1993): “Effi ciency Effects of Horizon- rics Journal, Vol 14, No 1, pp 45–90. nance/idbi-bank-hopes-to-turnaround-in-the- next-2-years-114081800526_1.html. tal (In-market) Bank Mergers,” Journal of Cornett, M M and H Tehranian (1992): “Changes in Banking & Finance, Vol 17, Nos 2–3, pp 411–22. Corporate Performance Associated with Bank Nayak, P J (2014): “Report of the Committee to Re- — (1998): “The Effi ciency Effects of Bank Merg- Acquisitions,” Journal of Financial Economics, view Governance of Boards of Banks in India,” ers: An Overview of Case Studies of Nine Merg- Vol 31, No 2, pp 211–34. Reserve Bank of India, Mumbai, http://rbidocs. rbi.org.in/rdocs/PublicationReport/Pdfs/BCF- ers,” Journal of Banking & Finance, Vol 22, Cuesta, R A and L Orea (2002): “Mergers and Tech- 090514FR.pdf. No 3, pp 273–91. nical Effi ciency in Spanish Savings Banks: A Stochastic Distance Function Approach,” NSSO (2011): “Employment and Unemployment Sanyal, M K and P Das (2013): “National Accounts Journal of Banking & Finance, Vol 26, No 12, Situation in India 2009–10,” 66th Round (July Statistics and State Estimates of Domestic pp 2231–47. 2009–June 2010), Report No 537, New Delhi: Products: Some Unresolved Issues,” Journal of National Sample Survey Organisation, Minis- Income and Wealth, Vol 35, No 1, pp 35–45. De Hoyos, R E and V Sarafi dis (2006): “Testing for try of Statistics and Programme Implementa- Cross Sectional Dependence in Panel Data- Shaffer, S (1993): “Can Megamergers Improve tion, Government of India, http://www.mospi. models,” Stata Journal, Vol 6, No 4, pp 482–96. Bank Effi ciency?,” Journal of Banking & Fin- gov.in/sites/default/fi les/publication_reports/ ance, Vol 17, Nos 2–3, pp 423–36. Demirgüç-Kunt, A et al (2015): “The Global Findex NSS_Report_No_537.pdf. Sherman, H D and T J Rupert (2006): “Do Bank Database 2014: Measuring Financial Inclusion Penas, M F and H Unal (2004): “Gains in Bank Around the World,” Policy Research Working Mergers Have Hidden or Foregone Value? Real- Mergers: Evidence from the Bond Markets,” ised and Unrealised Operating Synergies in Paper 7255, World Bank. Journal of Financial Economics, Vol 74, No 1, One Bank Merger,” European Journal of Opera- Driscoll, J and A C Kraay (1998): “Consistent Cova- pp 149–79. tional Research, Vol 168, No 1, pp 253–68. riance Matrix Estimation with Spatially Depen- Pesaran, M H (2004): “General Diagnostic Tests for Shetty, S L and B K Deokar (2014): “Financial Inclu- dent Data,” Review of Economics and Statistics, Cross Section Dependence in Panels,” Cam- sion: Differences between the Government and Vol 80, No 4, pp 549–60. bridge Working Papers in Economics No 0435, Dymski, G (1999): The Bank Merger Wave, Armonk, http://www.econ.cam.ac.uk/research/repec/ RBI?,” Economic & Political Weekly, Vol 49, NY: M E Sharpe. cam/pdf/cwpe0435.pdf. No 35, pp 12–15. — (2004): “Korea’s Post-crisis Banking Crisis and — (2006): “Estimation and Inference in Large Sinha, S (2014): “Merger with SBI Will Create a the Global Bank Merger Wave,” University of Heterogeneous Panels with a Multifactor Error Strong Bank, Says State Bank of Patiala MD,” California Centre, Sacramento, 28 November. Structure,” Econometrica, Vol 74, No 4, Hindu Business Line, 30 July, http://www.the- Egger, P and F R Hahn (2010): “Endogenous Bank pp 967–1012. hindubusinessline.com/industry-and-econo- Mergers and Their Impact on Banking Perfor- Prais, S J and C B Winsten (1954): “Trend Estima- my/banking/merger-with-sbi-will-create-a- mance: Some Evidence from Austria,” Interna- tors and Serial Correlation,” Cowles Commis- strong-bank-says-state-bank-of-patiala-md/ar- tional Journal of Industrial Organization, sion Discussion Paper No 383, Chicago. ticle6265152.ece. Vol 28, No 2, pp 155–66. Ray, M (2003): “Regional Financial Institutions: Thorat, U (2007): “Financial Inclusion: The Indian Epstein, M J (2005): “The Determinants and Evalu- Lending Behaviour and Path-Dependence, Ma- Experience,” speech at the Financial Inclusion ation of Merger Success,” Business Horizons, dras 1921–69,” PhD thesis, University of Calcutta. Conference, Whitehall Place, London, 19 June, Vol 48, No 1, pp 37–46. RBI (1973): Banking Statistics: Basic Statistical Re- http://www.bis.org/review/r070626f.pdf. Ghosh, J (2013): “Microfi nance and the Challenge turns, Vol 2, Reserve Bank of India, Mumbai. Vennet, R V (1996): “The Effect of Mergers and Ac- of Financial Inclusion for Development,” Cam- — (1991): Banking Statistics: Basic Statistical Re- quisitions on the Effi ciency and Profi tability of bridge Journal of Economics, Vol 37, No 6, turns, Vol 20, Reserve Bank of India, Mumbai. EC Credit Institutions,” Journal of Banking & pp 1203–19. — (1999): Statistical Tables Relating to Banks in Finance, Vol 20, No 9, pp 1531–58. Greene, W (2000): Econometric Analysis, Upper India 1998–99, Reserve Bank of India, Mumbai. Wooldridge, J M (2002): Econometric Analysis of Saddle River, NJ: Prentice Hall. — (2006): Statistical Tables Relating to Banks in Cross Section and Panel Data, Cambridge, MA: Gupta, S and Sidhartha (2014): “Govt Asks IDBI India 2005–06, Reserve Bank of India, Mumbai. MIT Press. Bank, UBI to Plan Merger,” Times of India, 26 July, http://timesofi ndia.indiatimes.com/ business/india-business/Govt-asks-IDBI-Bank- UBI-to-plan-merger/articleshow/39005996. EPW Index cms. Halkos, G E and N G Tzeremes (2013): “Estimating An author-title index for EPW has been prepared for the years from 1968 to 2012. The PDFs of the the Degree of Operating Effi ciency Gains from Index have been uploaded, year-wise, on the EPW website. Visitors can download the Index for a Potential Bank Merger and Acquisition: A DEA Bootstrapped Approach,” Journal of Bank- all the years from the site. (The Index for a few years is yet to be prepared and will be uploaded ing & Finance, Vol 37, No 5, pp 1658–68. when ready.) Houston, J F and M D Ryngaert (1994): “The Over- all Gains from Large Bank Mergers,” Journal of EPW would like to acknowledge the help of the staff of the library of the Indira Gandhi Institute Banking & Finance, Vol 18, No 6, pp 1155–76. of Development Research, Mumbai, in preparing the index under a project supported by the Houston, J F, C M James and M D Ryngaert (2001): “Where Do Merger Gains Come From? Bank RD Tata Trust. Mergers from the Perspective of Insiders and

48 FEBRuary 25, 2017 vol liI no 8 EPW Economic & Political Weekly