Vol : 42 July – December, 2019 No : 2...Vol : 42 July – December, 2019 No : 2 Content Research...

Vol : 42 July – December, 2019 No : 2Content

Research Papers

★ Telecommunications Infrastructure and Economic Growth in India

Mohina Saxena & Surajit Bhattacharyya ............................................ 1

★ Efficiency, Productivity and Returns to Scale of Indian General

Insurance Industry : Evidence Based on Panel Data

Ram Pratap Sinha .............................................................................. 23

★ Impact of Privatisation on the Performance of the Divested Firms :

Appraisal of Empirical Studies

Jugal Kishore Mohapatra .................................................................. 48

★ Corporate Governance Practices and its Impact on Non-Performing

Assets of Selected Commercial Banks in India

S.K.Chaudhury & Devi Prasad Misra ............................................... 78

★ Persistence in Performance of Indian Mutual Fund Schemes :

An Evaluation

Joyjit Dhar & Kumarjit Mandal ........................................................ 92

The Journal of Institute of Public Enterprise

The Journalof

Institute of Public Enterprise

Advertisement Tariff Details

Enquiries about corporate advertisements and appropriate art workshould be sent to :

The Editor,The Journal of Institute of Public Enterprise,

Institute of Public Enterprise, Hyderabad – 500 007.

Deliverablesa) A two page write-up about the sponsoring

organization in six issues of the journalb) Multi-coloured advertisement in the inner

cover page in six issues of the journalc) 20 copies of the journal for three years free

of cost.

Particulars

3.5 lakhs valid for3 years (6 Issues)

1.5 lakhs validfor 1 year

a) Advertisement (Back side cover page –Multi-colour in two issues )

b) A two page write-up about the sponsoringorganization in two issues of the journal

c) 10 copies of the journal for one year freeof cost

Issue-wise Details

75,000/- Backside Cover page in Multi-colour

65,000/- Inner Cover page in Multi-colour

Telecommunications Infrastructure and Economic Growth in India

1

Telecommunications Infrastructure andEconomic Growth in India

Mohina Saxena* & Surajit Bhattacharyya**

This paper attempts to analyze whether there has been any causal relationship betweentelecommunications infrastructure and economic growth in India over more than threedecades [1981-2018]; and identify whether such a causal relationship has beenunidirectional or bidirectional in nature. We use the Johansen and Juselius (1990) testfollowed by the Granger (1969) causality test between output and telecommunicationsinfrastructure. A multivariate VAR is also performed by considering other variables namely,domestic investment, trade volume and real effective exchange rate. There has been no longrun relationship between telecommunications infrastructure and economic growth in India;short run unidirectional causality emanating from output growth to telecommunicationsinfrastructure is observed. Interestingly, domestic investment does not Granger cause economicgrowth even in the short-run; however, there is unidirectional causality from REER todomestic investment and aggregate output. Short-run bidirectional causality between economicgrowth and volume of trade hints at a feedback effect.

Keywords : Telecom Infrastructure, Real GDP Growth, Granger Causality, Cointegration,Multivariate VAR.

1. IntroductionThere are at least three channels throughwhich development in telecom sectorfacilitates the economy; firstly, telecom-munications infrastructure has thepotential to increase efficiency in deli-vering the necessary social services alongwith increasing the level of productivity.This is because with the developmentin modern (communication) technologies,the cost of doing business drops thataccelerates the consequent economicoutput. Secondly, in the backdrop ofupsurge in number of call centers,BPOs as well as KPOs and software

The Journal of Institute of Public Enterprise, July-Dec, Vol. 42, No. 2ISSN : 0971-1864, © 2019.

* Mohina Saxena, Research Scholar, Departmentof Humanities & Social Sciences, IIT Bombay,Mumbai - 400 076.

** Surajit Bhattacharyya, Associate Professor ofEconomics, Department of Humanities & SocialSciences, IIT Bombay, Mumbai - 400 076.

companies particularly in the lastdecade and half, it is evident thatdevelopment in the telecom sector hasprovided significant employmentopportunities in India. And lastly, pene-tration of telecom services has facili-tated societies that were remotely loca-ted and women in particular, to reapeconomic benefits by increasing theirincomes significantly. A notable

The Journal of Institute of Public Enterprise, July-Dec, Vol. 42, No. 2© 2019, Institute of Public Enterprise

2

example is of the fishermen in the westcoast of India who regularly use cellphones to get information regardingmarket prices so as to decide the mostprofitable destination for selling offtheir stocks. This not only helps inreducing income volatility but on theother hand, contributes to the empower-ment aspect of Sustainable Develop-ment Goals (SDGs). Thus, in pursuitof achieving these goals, Indian tele-communications industry has not onlythrived to become one of the fastestgrowing in the world, it is in fact,ranked as the second largest in the worldas of March, 2018.

From another perspective, it has beenan undeniable argument that thegovernment’s ability to deliver varioussocial service schemes as well as copingwith the civil emergencies in an efficientway rests primarily on the efficacy ofthe telecommunications sector. Citizensnot only become more aware but canalso easily access as well as raise con-cerns regarding the implementation ofgovernment’s welfare programmes andactivities through a smooth well-func-tioning communication set up. Thus,the democratic essence of the nationstate is largely protected by develop-ments in the telecommunications sec-tor as it helps in building an informedsociety. Also, with regards to the impacton education, the development ofmodern communication technologies

has enabled a significant reach throughthe distance learning programmes thathave helped a larger section of the societywhich had been excluded earlier due toseveral supply-side bottlenecks. In fact,with the advent of internet, it is of nosurprise to witness the much-neededexistence of a full-fledged virtual com-munity, electronic market place andknowledge centres that have broughtwith itself rise in economic output. Inthis sense, an advanced telecommuni-cations infrastructure such as mobilephones and internet is pivotal not onlyfor generating economic gains but alsofor societal upliftment. In fact, in thebackdrop of India’s commitment toachieve the Sustainable DevelopmentGoals, it has been argued that deve-lopment of telecommunicationsinfrastructure leads India’s way tosucceed in achieving that target throughits multi-fold beneficial effects on theeconomy at large.1

The importance of infrastructure toeconomic growth was initially broughtout by the World Bank DevelopmentReport (1994) which highlighted thatgrowth as well as productivity is higherin countries that have an adequateand efficient supply of infrastructureservices. The India InfrastructureReport (1996) expressed a similar viewtowards the importance of infra-structure. Even, if we look back intothe last decade, in the 11th five year plan


3

Government of India (GoI) emphasizedthe urgent need for removing infrastruc-ture bottlenecks for sustained growth;[Economic Survey (2011-12)]. Subse-quently, in the 12th five year plan thegovernment had laid special emphasison the infrastructure sector and recog-nized that the availability of qualityinfrastructure is important not only forsustaining high growth but also toensure that the growth is inclusive;[Economic Survey (2012-13)]. Duringthe last quarter of the 20th century,infrastructure developments in Infor-mation and Communication Technolo-gies (ICT) were widely seen as havingheralded an information age in whicheconomic (and social) activity has beenmore productive, efficient, widenedand deepened to the grassroots level.Over the years ICT development hasbeen increasingly identified as the onethat has a strong association with over-all economic activities at a large scaleand it is the fast-paced advancement intelecom infrastructure that has theability to create spillover effectsthrough network externalities.

Our study explores the inter-linkagebetween available physical infrastructureand growth prospects of the Indianeconomy by especially focusing on thedevelopments in telecom sector infra-structure. Apart from other reasons thatmake the Indian telecom sector beingso important of late, it is by now well

argued in the policy-oriented literaturethat development in ICT is crucial toachieve the desired growth trajectoryfor the fast-growing developingeconomy of India.

According to the Network ReadinessIndex (2015) published by the WorldEconomic Forum, most of the deve-loping countries continue to lag in com-parison with the developed (industria-lized) countries in terms of physicalinfrastructure and preparation to parti-cipate and enjoy the benefits from ICTdevelopment. India has been relativelya late starter when it comes to identify-ing and developing state-of-the-art pub-lic infrastructure.Thus, telecommuni-cations infrastructure becomes increas-ingly relevant in today’s parlance whenthe policy makers and the academicsare debating on the plausible effects ofdigital divide in a vast (as well as fast)growing emerging economy like India.

The planning strategies and policyinitiatives induced by successive centralgovernments were matched with fastpaced positive response from the privatesector and that resulted in such a phe-nomenal progress which perhaps por-trays the biggest success story of India’sreforms; Panagariya (2013). In spite ofbeing plagued by so many challengessuch as infrastructure bottlenecks,relatively slower penetration in ruralareas and lack of skilled manpower in


4

telecom equipment manufacturingactivities, the growth story of Indiantelecommunications sector has remainedencouraging even in the scenario of fluc-tuations in other core sectors of theeconomy. The cost saving externalitiesgenerated due to the increasing returnsof communications foster the expan-sion of other markets and leads to lowertransaction costs. It is therefore, impera-tive to explore whether a fast-growingemerging economy like India had bene-fitted from the large scale FDI inflowsand massive investments made in thetelecommunications sector that alsohouse the highest level of employmentopportunities in the private sector ofthe economy.

Section-2 presents significance oftelecom sector in India’s growth his-tory. Section-3 discusses the empiricalliterature establishing the inter-linkagebetween telecommunications infra-structure and economic growth. InSection-4, we explain the empiricalframework followed by construction ofvariables in Section-4.1 and estimationresults in Section-4.2. Finally, Section-5concludes this paper along with someplausible policy implications.

2. The Indian Telecom SectorIndia has the second largest telecommu-nications network in the world withsubscriber base of over 1,200 millionas of March, 2018 [TRAI (2018)].

Since the last decade there has been anexponential growth in the Indian tele-communications sector especially in thecellular technology with the total num-ber of mobile phone subscribers grow-ing almost three hundred times in justten years from 3 million in 2001 to811.59 million in 2011 [TRAI (2011)].Apart from contributing around 2-3 percent of India’s GDP, in the last one anda half decade the telecommunicationssector has grown at the compoundedannual growth rate of 22 per centduring 2000-18 in terms of its totalsubscriber base. In fact, just on the eveof launching the New Telecom Policy2012 (NTP-2012), the Indian tele-communications sector had grown atthe highest compounded annualgrowth rate of 44 per cent during 2006-2007 to 2011-2012 [MoSPI, (2014)].

The reforms in telecommunicationssector infused competition and led tothe adoption of new technologieswhich proved to outweigh any benefitsthat scale economies may bestow;Panagariya (ibid.). A glimpse at themagnitude of success achieved throughthe reforms process is reflected throughthe following facts. As on March 31,1981, India had only 2.15 million tele-phone users. While alone in 2015,India added over 10 million telephonesin each quarter. The wireless densityhad increased exponentially from 0.3per cent in 1999-2000 to 91.09 per cent


5

in 2017-18; see TRAI, Annual Reports(various issues). Similarly, Figure-2 dis-plays the addition in number of tele-phone lines for each of the ten-yearperiod since 1981-82. The wirelesslocal call tariff has reduced from 16.40per minute in 1990 to 1 per minutein early 2000. The internet policyannounced in 1998 ended the monopolyof VSNL and allowed the entry of pri-vate operators to induce competitionwithout any significant licence fees. Theenunciation of this policy has led to asignificant increase in the number ofinternet subscribers from less than 1million in 2000-01 to 493.96 millionin 2017-18; (see Figure-3).

The telecommunications sector revenuehad gone up from 125.18 billionduring 1995-96 to 2556 billion in2016-17; see Table-1. The Indian IT

industry has evolved as the largestprivate sector employer with more than3.5 million employees as of March,2015. FDI in the telecom sector hasgrown at a compounded annual growthrate of 24.8 per cent between 2000-01and 2011-12. In the last decade andhalf, the cumulative FDI inflow in tele-communications sector constitutedaround 7 per cent of total FDI inflows;(Department of Industrial Policy andPromotion, 2017).

2.1 Challenges in the IndianTelecom Sector

The subscriber base for telecom servicesis highly skewed in favour of urban areas.The significant “digital divide” withfaster growth rate of urban subscribersas compared to the rural subscribersreflects that there are relatively lesser

Figure-1 : Teledensity in India (1981-2018)

100

80

60

40

20

0

-20

-40

Tele

dens

ity %

1981

1983

1985

1987

1989

1991

1993

1995

1997

1999

2001

2003

2005

2007

2009

2011

2013

2015

2017

Wireless Density Fixed Line Density Teledensity Linear (Teledensity)


6

Figure-2 : Number of Telephone Connections Added

Data Source : Telecom Regulatory Authority of India, Annual Report (various issues).

900

800

700

600

500

400

300

200

100

0

-1001980-81 1990-91 2000-01 2010-11 2014-15 2017-18

Num

ber o

f Tel

epho

ne C

onne

ctio

nsA

dded

(in

mill

ions

)

123123

123123 123

123

123123 123

123

123123123

12341234123412341234123412341234123412341234123412341234123412341234123412341234123412341234123412341234123412341234123412341234123412341234123412341234123412341234123412341234123412341234

123123

123123123123123123123123123123123

123123

123123123123123123123123123123123123123123

12341234

Number of Mobile Connections Added

Number of Fixed Line Connections Added

121212

123123

Figure-3 : Number of Internet Users in India

Data Source : The World Bank, World Development Indicators.

50045040035030025020015010050

0

No.

of I

nter

net U

sers

(in m

illio

ns)

1981

1983

1985

1987

1989

1991

1993

1995

1997

1999

2001

2003

2005

2007

2009

2011

2013

2015

2017


7

Table-1 : Contribution of Telecommunications Sector to GDP

Year

Telecom Sector Revenue Contribution to Total Government Levies(in Billions) GDP (%) (in Billions)

2004-05 720 2.2 104

2005-06 860 2.3 129

2006-07 1050 2.4 193

2007-08 1440 2.9 255

2008-09 1520 2.7 250

2009-10 1580 2.4 278

2010-11 1720 2.2 261

2011-12 1950 2.2 262

2012-13 2120 2.1 291

2013-14 2330 2.2 312

2014-15 2606 2.1 222

2015-16 2795 2 235

2016-17 2556 1.66 180

Data Source : TRAI, The Indian Telecom Services Performance Indicators, (various issues).

Year FDI Inflow (in Millions)

2001-02 39,384.61

2002-03 9,077.31

2003-04 3,978.40

2004-05 5,411.01

2005-06 27,514.50

2006-07 21,495.77

2007-08 50,995.61

2008-09 1,16,848.11

2009-10 1,22,696.62

Table-2 : FDI Inflow in the Indian Telecom Sector

Year FDI Inflow (in Millions)

2010-11 75,420.44

2011-12 90,115.26

2012-13 16,540.04

2013-14 79,872.83

2014-15 173,718.22

2015-16 86,370

2016-17 374,350

2017-18 397,480

Data Source : Department for Promotion of Industry and Internal Trade, Government of India.


8

beneficial effects of competition for therural population. The gap between urbanand rural teledensity was as high as 13times in 1996, it kept rising and reachedits peak in 2006 (20 times) after whichit started declining and currently theurban teledensity is around 3 timesmore than the rural teledensity. Perti-nent in today’s debate over reducing theso-called ‘digital divide’, it is argued infavour of identifying telecommunica-tions as ‘merit goods’ that universalaccess to telecommunication servicescan significantly reduce social andeconomic exclusion and offer increasedopportunities to the people at large.

The manufacturing of telecomequipment due to increased demand ofmobile phones is increasing steadily but

still lags behind the telecom services.In fact, the revenue of telecom manu-facturing sector is significantly smallerthan the services sector and has actuallydeclined since 2009-10. Also, the tele-com manufacturing sector is plaguedby poor research and development;NCAER (2012). The liberal tradepolicy has fostered imports of equip-ments but the lack of capacity buildinghas the potential to challenge thesuccess of this industry. This reducescost competitiveness of domesticindustry and exposes it to disadvantage.

3. Review of Select LiteratureThe expansion of telecommunicationssector has direct and indirect effects : itdirectly affects growth through newemployment opportunities, diffusion

Figure-4 : Post-Reforms Digital Divide in India

Data Source : TRAI, The Indian Telecom Services Performance Indicators, (various issues).

200180160140120100

80604020

0

Tele

dens

ity %

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

Rural Urban Overall


9

of information and knowledge,increased investment as well as demand;and has indirect effects through increasedproductivity, enhanced efficient func-tioning of the markets by reaping thebenefits from network externalities,among others; Thompson Jr. andGarbacz (2007).

One strand of literature using regressionanalysis depicts that telecommunica-tions infrastructure has a positive andsignificant impact on economic growth;see, Hardy (1980) and Norton (1992),among others. In the Indian context,Kathuria et al. (2009) find that Indianstates with higher mobile penetrationcan be expected to grow faster.

Another set of empirical studies exam-ine the relationship between telecom-munications infrastructure and growththrough Granger-causality test usingVAR framework and argue that tele-communications infrastructure is a pre-condition for economic growth; see,Dutta (2001), Cieslik and Kaniewsk(2004), among others. On the otherhand, Beil et al. (2005) and Chakrabortyand Nandi (2011) suggest that eco-nomic growth precedes telecommuni-cations infrastructure; i.e., telecommu-nications infrastructure is only a resultantoutcome of economic growth that hasalready taken place. This propositionasserts that a growing economydemands better telecommunications

infrastructure and hence, growth in tele-communications infrastructure takeplace. There is another notion thatclaims the existence of ‘feedback effect’between telecommunications infra-structure and economic growth;Cronin et al. (1991), Yoo andKwak (2004). This implies that thegrowth of any one of them fosters thedevelopment of other.

There are only a very few studies thathave actually considered Indian data andexplored the inter-linkage betweentelecommunications infrastructure, inparticular and economic growth.Narayana (2011) estimated the impactof telecom services on economicgrowth and argued that the contribu-tion of telecom services to GDP growthhas increased phenomenally in the postmobile phone era. The study also ana-lyzed the determinants of demand fortelecommunications infrastructure andfound that all the explanatory variables(access price, usage price, monthlyincome, education, caste, and location)were statistically significant at 1 per centlevel of statistical significance.

Ghosh and Prasad (2012) examined thenexus between telephone connectionsand economic activity in India duringthe period 1980-2007. The studyincluded real gross fixed capital forma-tion as a measure of investment and realGDP as a proxy for economic growth.


10

The ARDL bound test and Johansen andJuselius (1990) Maximum LikelihoodTest indicated that there was nocointegration among telephone connec-tions, real investment and real GDP.However, the Granger causality testfound the presence of short rununidirectional causality running fromtelecom infrastructure to economicgrowth and to investment.

Therefore, there exists inconsistencyacross literature regarding the directionof causality between telecommunica-tions infrastructure and economicgrowth. This can be attributed to diffe-rent data periods, varied econometricmodelling techniques and state of theeconomy considered in those studies.

4. Empirical AnalysisGiven the extant inconclusive empiricalevidences available, we aim to explorewhether the development in telecom-munications infrastructure has been astimulus to India’s growth story or hasit been merely a consequence ofgrowth, if not both.

Our study differs from Ghosh andPrasad (2012) as well as Mehta (2017)on the count that, this study is perhapsthe first attempt that encompasses alonger data period [1981-2018] thanany other research work with specificreference to the case of India. Ashighlighted earlier, in the decade of

1980s India’s teledensity was abysmallylow; slowly it started accelerating sincethe eve of liberalization and had risensteeply from 2005 onwards. Figure-5shows that during early 1980s to early1990s although there were instances ofhikes in real GDP growth rate but thosesharp hikes were not accompanied byany noteworthy increase in telecommu-nications infrastructure. Given theabsolutely low level of telecom userson the eve of economic reforms to evenlate 1990s, it seems that the occasionalhikes in the growth rate (e.g., in theyears 1983-84, 1988-89, and 1995-96)perhaps cannot be accrued to the deve-lopment in telecommunications infra-structure in particular. In fact, it is arguedthat the impact of telecom infrastruc-ture on a country’s growth rate essen-tially requires a critical mass of usersbefore any significant impact is felt;see, Röller and Waverman (2001).

The data period of Ghosh and Prasad(ibid.) ended at 2007, but the latestphase of telecom sector reforms beganin 2007 itself. This latest phase of reformsbrought in the removal of bindingrestrictions on the number of playerswithin a mobile circle, and 122 new2G licenses were given to various telecomcompanies on a first-come first-servedbasis in January 2008. The auction of 3Gspectrum was held in 2010 and mobilenumber portability was introduced in2011. Certainly, all of these initiatives


11

enabled the subscribers to execute awider choice and the facility to switchbetween different service providers. Inessence, such policy reforms enhancedthe level of competition among theoperators to retain their customers.Because increased demand for telecomusage induced the government as wellas the private players to boost up infra-structure facilities. Mehta (2017) on theother hand have focused only on thedata period 2007-2015. But we feelthat given the irreversible lumpy invest-ment required for building up some ofthe physical infrastructures, there has tobe long run dynamism in such infra-structure building activities. Therefore,

it is of significant importance to haveconsidered the available data till 2018since it becomes pertinent to examinethe impact (if any) of increasedeconomic growth on the developmentof telecommunications infrastructurein India.

Secondly, our study incorporates othermacroeconomic variables namely, totaltrade volume and real effective exchangerate that are either not included ordifferent from other studies on Indiantelecommunications sector.

We conceptualize the basic empiricalobjective of our study as per thefollowing diagrammatic representation.

Figure-5 : Telecom Infrastructure and Growth Rate

Data Source : TRAI Annual Reports and Handbook of Statistics on Indian Economy, RBI.

12

10

8

6

4

2

0

1981

1983

1985

1987

1989

1991

1993

1995

1997

1999

2001

2003

2005

2007

2009

2011

2013

2015

2017

1800

1600

1400

1200

1000

800

600

400

200

0

Gro

wth

Rat

e (%

)

Tota

l Tel

ecom

Use

rs (i

n m

illio

ns)

Total Telecom Users GDP Growth Rate12345671234567

121212121212 1 111

11

11

111

121212

1111

121212121212

1212121212121212

1212121212121212121212

121212121212121212121212121212

121212121212121212121212121212121212121212

121212121212121212121212121212121212121212121212

1212121212121212121212121212121212121212121212

12121212121212121212121212121212121212121212121212

1111111111111111111111111111

111111111111111111111111111111

1111111111111111111111111111111111

111111111111111111111111111111111111


12

4.1 Data and Variables Description

We consider time series data from varioussources; i.e., TRAI Annual Reports,Handbook of Statistics on Indian Economyprovided by the RBI and World Bank.Annual data is considered for the Indianeconomy as a whole and we go back to1981 till recent past (2018) to assessthe possible causality between telecom-munications infrastructure built up andthe apparently successful growthstory of India. Given the (secondary)data availability, our data period there-fore, captures arguably the earliest phaseof conceptualization of reforms thatwere planned during those early yearsof 1980s. Economic growth is proxiedby considering the level of real GDPin a particular year. On the other hand,we measure the telecom infrastructureby adding up the total number of fixedline users, number of mobile phone usersand number of internet users. Along the

lines, this study extends from testing onlythe bivariate causality to incorporate threeother macroeconomic variables, namely,gross fixed capital formation (GFCF),total trade volume (TRD), and real effec-tive exchange rate (REER). Logarithmictransformations of all the variables areconsidered in order to easily convertnon-stationarity series to stationaryseries after taking the first-difference.

4.2 Econometric Methods, Resultsand Discussion

We begin by examining the bivariatecausality between real GDP andtelecom infrastructure; and in the laterpart of the analysis we incorporate someother macroeconomic variables that weconceptualize to have affected the eco-nomic activity of Indian economy overthe past three decades. To begin with,the ‘unit root tests for stationarity’ of thedata series have been conducted.

Increased Capacity to Invest in TelecommunicationsMore Demand for Advanced Telecommunications

Increased and Speedy Flow of Information in the MarketLower Transaction Costs

Higher Level of EconomicActivity

State-of-the-ArtTelecommunications

Infrastructure

→

→

→ →

→

→


13

Unit Root Tests

It is by now oft-told and empiricallywell argued in the extant macro econo-metrics literature that many macroeco-nomic time series are found to containunit root and hence, are non-stationary;see, Nelson and Plosser (1982). There-fore, conventional hypothesis tests can-not be conducted based on the esti-mated coefficients of non-stationaryvariables using the traditional t-test and/or F-test. In fact, the statistical signifi-cance of the concerned test statistic isthen overstated and the results obtained

are a priori spurious; see Granger andNewbold (1974) for details. Thus, itbecomes imperative to first assess thestationarity of the relevant empiricalvariables in order to obtain statisticallymeaningful (and robust) results.

We conduct the two most commonlyused tests for assessing the presence ofunit roots following the seminal con-tributions of Dickey and Fuller (1979,1981) and Phillips and Perron (1988).Dickey and Fuller (ibid.) also extendedthe procedure to an augmented versionof the test which includes extra lagged

Table-3 : Variables and Sources of Data

Gross Domestic Product (GDP) RBI

Gross Fixed Capital Formation (GFCF) RBI

Telecom Infrastructure (TI) TRAI, World Bank

Trade Volume (TRD) RBI

Real Effective Exchange Rate (REER) Bruegel Datasets

Note : All are real variables constructed considering the base year as 2004-05.

Table-4 : Descriptive Statistics

LGDP LTI LTRD LGFCF LREER

Mean 16.94 4.82 10.80 15.66 4.67

Median 16.89 3.67 10.71 15.56 4.63

Max. 18.18 17.62 12.53 17.12 5.14

Min. 15.89 0.77 9.18 14.39 4.33

Std. Dev. 0.69 4.31 1.18 0.87 0.23

Skewness 0.20 1.82 0.08 0.21 0.50

Kurtosis 1.81 5.81 1.59 1.68 2.20


14

terms of the dependent variable inorder to eliminate autocorrelation. Thelag length is determined either by theAkaike Information Criterion [AIC](1974) or Schwartz Bayesian Criterion[SBC] (1978). We use the lag lengthwhich minimizes the value of the respec-tive criteria used. On the basis ofobtained Augmented Dickey-Fuller(ADF) statistic, the null hypothesis (thatthe series has a unit root) is accepted forall the data series at their respective levels.However, after first differencing the vari-ables, the respective series become sta-tionary. In both the tests (ADF as wellas Phillips Perron), the unit root testresults are very similar and do not

change qualitatively. Also, there are twoother possible forms of the ADF test :one depicts the presence of a drift andthe second captures the presence of botha drift and non-stochastic trend. We havetested stationarity for both the cases; i.e.,with and without trend in the presenceof an intercept term; see Table-5.

Cointegration Test

Cointegration refers to a stationary linearcombination of non-stationary variables,implying an existence of long runequilibrium relationship among theconcerned economic variables; see forinstance, Enders (2004) and Asteriouand Hall (2007).

Table-5 : Augmented Dickey-Fuller Test Results

Variables Model with Constant Only Model with Constant and Trend

LTI (2) -2.14 -2.87

∆LTI (1) -5.94* -5.82#

LGDP (0) 2.24 -1.45

∆LGDP (0) -4.35* -4.92#

LGFCF (0) 0.68 -1.82

∆ LGFCF(0) -5.34* -5.97#

LTRD (1) 0.79 -2.93

∆ LTRD (1) -4.82* -4.95#

LREER (1) -2.15 -0.54

∆LREER (0) -3.36* -3.82#

Note : ∆ denotes the first-difference. Numbers in parentheses denote lag length using AIC or SBC.* : denotes significant at 5% level when compared with the critical value -2.97 in the first case;

similarly, in the second case# : denotes significant at 5% when compared with the critical value -3.56.


15

For assessing the causality in case of non-stationary variables the unit rootsshould be removed after differencingappropriately. Therefore, in case of twoI(1) variables the causality test shouldbe done only on the first difference ofboth the variables. However, if thevariables are stationary at first difference,then there is a possibility of cointegration.

Based on the result of the test forcointegration, it is decided which timeseries model is appropriate for estimatingthe causal relationship. If the series arenot cointegrated then an unrestrictedVector Autoregressive (VAR) model isused to assess causality, whereas, in thepresence of cointegration a Vector ErrorCorrection Model (VECM) is usedbecause VAR in first differences is miss-specified since the system omits the‘error correction term’ which is thelagged residual from the estimated longrun relationship; Engle and Granger(1987).

Therefore, in case of non-stationaryvariables it is imperative to assess thepresence of cointegration before estima-ting Granger Causality. In this paper,cointegration is tested using the Johansen

and Juselius (1990) method. Thecointegration test is done after findingout an appropriate lag structure on thebasis of either SBC or AIC. The teststatistics reveal that the null hypothesisof no cointegration cannot be rejectedat 5 per cent level of statistical significance.Therefore, there is no long run relation-ship among the variables; see Table-6.

Granger Causality TestA variable Yt is said to Granger causeXt, if Xt can be predicted with greateraccuracy by using the past history ofthe other variable (Y) in addition to thepast values of X rather than not usingsuch past values; Granger (1969).

Since it is by now established that thereis no long run relationship (cointegration)between the variables hence, the causa-lity is tested through the unrestrictedvector auto-regression framework bytaking the first differences of the varia-bles. Therefore, the unrestricted VARmodel will be as follows :

∆ ∆LTI LTIt i t ii

m

= + +−=∑α α0 11

∆LGDPi t ii

m

t+ +−=∑α ε21

1 ...(1)

Table-6 : Johansen Cointegration Test BetweenTelecom Infrastructure and Real GDP

Test Statistic p-value No. of Cointegrating Equations

Trace 11.13 0.06 0

Maximum Eigenvalue 9.60 0.17 0


16

∆ ∆LGDP LGDPt i t ii

m

= + +−=∑ ∑β β0 11

∆LTIi t ii

m

t+ +−=∑β ε21

2 ...(2)

We find that the optimum lag lengthof the bivariate VAR is one accordingto AIC and SBC. The bivariate Granger-causality test results are reported inTable-7.

It is observed that when ∆LGDP is thedependent variable, ∆LTI turns out tobe statistically insignificant at 5 per centlevel indicating absence of any causa-lity from ∆LTI to ∆LGDP. However,when ∆LTI is dependent variable then,∆LGDP is statistically significant indi-cating a short-run unidirectional causa-lity running from real GDP to telecom-munications infrastructure. This con-firms that economic growth causestelecommunications infrastructure and

is not caused due to it; see, Beil et al.(2005), Chakraborty and Nandi(2011) and Pradhan et al. (2014) forsimilar results.

Table-8 shows that the estimatedunrestricted VAR model satisfies the‘residual test’ with the null hypothesisof normal residuals, no autocorrelationand no heteroskedasticity respectively, atone percent level of significance.

The bivariate analysis might involvespecification bias due to the omission ofsome other relevant variables; therefore,we extend our econometric analysis toa multivariate framework. The follow-ing equation depicts the proposed func-tional relationship between telecom-munications infrastructure, economicgrowth and other macroeconomic vari-ables, such as real gross domestic invest-ment, total volume of trade and realeffective exchange rate.

Table-7 : Granger-Causality Test Results

Null Hypothesis F-Statistic Prob. Decision

∆LTI does not Granger Cause ∆LGDP 0.021 0.87 Accept

∆LGDP does not Granger Cause LTI 4.121 0.05 Reject

Table-8 : Diagnostic Tests Results

Test Statistic p-value Conclusion

JB 10.62 0.04Normality Kurtosis 3.12 0.25 Normal Residuals

Skewness 7.65 0.03

Autocorrelation LM(4) 4.55 0.35 No Autocorrelation

Heteroskedasticity χ2 (15) 25.10 0.06 No Heteroskedasticity


17

i.e., LTI = f (LGDP, LGFCF, LTRD,LREER)

There might also be reverse causality orbidirectional causality in this empiricalmodel. Since we know that all the varia-bles are I(1), we now verify if there existsany long run relationship or cointegra-tion among them by employing theJohansen and Juselius (ibid.) MaximumLikelihood procedure; see Table-9. Theoptimal order of lags for the VARmodel is again found to be 1 based onboth the AIC and SBC.

Again, we accept the null hypothesis ofno cointegration at 1 per cent level ofsignificance and conclude that there isno long run association among the variables.As earlier, the short run causality istested through the unrestricted vectorauto-regression framework by taking thefirst differences of all the variables (sincethe variables are I(1)); see Table-10.We find that the optimum lag length ofthe multivariate unrestricted VAR is 1according to AIC and SBC.

The unrestricted multivariate VARmodel will be as follows :

∆ ∆LTI LTIt i t ii

m

= + +−=∑α α10 111

∆ ∆LGDP LGFCFi t ii

m

i t i+ +−=

−∑α α121

13 ++=∑i

m

1

++ + +−=

−=

∑ ∑i t ii

m

i t ii

m

tLREER LTRD141

151

1α α ε∆ ∆ ...(3)

∆ ∆LGDP LGDPt i t ii

m

= + +−=∑α α20 211

∆ ∆LTI LGFCFi t ii

m

i t+ +−=

−∑α α221

23 iii

m

+=∑ ∑1

i t ii

m

i t ii

m

tLREER LTRD+ + +−=

−=

∑ ∑241

251

2α α ε∆ ∆ ...(4)

∆ ∆LGFCF LGFCFt i t ii

m

= + +−=∑α α30 311

∆LGDP LTIi t ii

m

i t+ +−=

−∑α α321

33 iii

m

=∑1

i t ii

m

i t ii

m

tLREER LTRD+ + +−=

−=

∑ ∑341

351

3α α ε∆ ∆ ...(5)

∆ ∆LTRD LTRDt i t ii

m

= + +−=∑α α40 411

∆ ∆LGDP LGFCFi t ii

m

i t+ +−=∑α α421

43 −−=

+∑ ii

m

1

−=

−=

+ + +∑ ∑i t ii

m

i t ii

m

tLREER LTI441

451

4α α ε∆ ∆ ...(6)

Table-9 : Johansen Cointegration Test Results

Test Statistic p-value No. of Cointegrating Equations

Trace 76.21 0.02 0

Maximum Eigen value 30.98 0.07 0


18

∆ ∆LREER LREERt i t ii

m

= + +−=∑α α50 511

∆ ∆LGDP LGFCi t ii

m

i+ +−=∑α α521

53 FFt ii

m

−=∑1

LTRD LTIi t ii

m

i t ii

m

t−=

−=

+ + +∑ ∑541

551

5α α ε∆ ∆ ...(7)

It can be observed from Table-10 thatwhen ∆LTI is the dependent variablethen ∆LGFCF, ∆LTRD and ∆LREERare statistically insignificant, whereas∆LGDP appears to be statistically signi-ficant at 5 per cent level of significance.In equation (4), both ∆LTI and∆LGFCF are statistically insignificantwhere as ∆LTRD and ∆LREER are sta-tistically significant. The results indicatethat there is no unidirectional causalityrunning from either telecommunicationsinfrastructure or real GFCF to real GDP.In case of real GFCF our results comply,whereas, for the case of telecommuni-cations infrastructure our results contra-dict the findings of Ghosh and Prasad(2012) that concluded the presence ofunidirectional causality running fromtelecommunications infrastructure toeconomic growth.

However, there is presence of short rununidirectional causality emanating fromreal GDP to telecommunications infra-structure. Thus, economic growthcauses the development of telecominfrastructure in India and not the otherway round. Also, in equation (6)

∆LGDP turns out to be statistically sig-nificant indicating presence of bidirec-tional causality between real GDP andtotal trade; see Pradhan et al. (2014)for similar results. In equation (7),∆LTI, ∆LGDP, ∆LTRD and ∆LGFCFare statistically insignificant; indicatingpresence of short run unidirectional cau-sality running from REER to real GDP.Short run unidirectional causality isfound to be running from real GFCFto total trade and from REER to GFCF.The diagnostic test results of estimatedunrestricted VAR model are reported inTable-11.

5. Summing UpA significant section of rural India stillstruggles with social exclusion fromquality health services, secondary andhigher education, housing, water supply,sanitation and overall social security. Itis indeed true that growth without socialjustice is inhuman and social justicewithout adequate growth is inconceivable.The desired benefits of telecom revolu-tion in terms of attaining the SDGs canonly be fruitfully reaped once Indiaprogressively steps toward addressingsuch issues of social exclusions.

Our study addressed the intriguing ques-tion—whether the phenomenal deve-lopment in the Indian telecom infrastruc-ture caused a spur in economic growth;or, is it due to the steady economic growthin India that induced the much-needed


19

Table-10 : Granger-Causality Test Results

Null Hypothesis F-Statistic Prob. Decision

∆LTI does not Granger Cause ∆LGFCF 0.002 0.95 Accept

∆LGFCF does not Granger Cause ∆LTI 2.082 0.19 Accept

∆LTI does not Granger Cause ∆LGDP 0.019 0.93 Accept

∆LGDP does not Granger Cause ∆LTI 4.095 0.03** Reject

∆LTRD does not Granger Cause ∆LTI 0.172 0.61 Accept

∆LTI does not Granger Cause ∆LTRD 1.875 0.13 Accept

∆LGFCF does not Granger Cause ∆LTRD 5.924 0.02** Reject

∆LTRD does not Granger Cause ∆LGFCF 1.028 0.33 Accept

∆LGDP does not Granger Cause ∆LTRD 4.896 0.02** Reject

∆LTRD does not Granger Cause ∆LGDP 2.934 0.07* Reject

∆LGDP does not Granger Cause ∆LGFCF 1.431 0.20 Accept

∆LGFCF does not Granger Cause ∆LGDP 0.483 0.53 Accept

∆LTI does not Granger Cause ∆LREER 1.295 0.25 Accept

∆LREER does not Granger Cause ∆LTI 6.486 0.65 Accept

∆LGDP does not Granger Cause ∆LREER 2.643 0.15 Accept

∆LREER does not Granger Cause ∆LGDP 4.987 0.03** Reject

∆LGFCF does not Granger Cause ∆LREER 0.891 0.18 Accept

∆LREER does not Granger Cause ∆LGFCF 4.912 0.03** Reject

∆LTRD does not Granger Cause ∆LREER 2.671 0.20 Accept

∆LREER does not Granger Cause ∆LTRD 0.175 0.63 Accept

Note : **and * denote significant at 5% and 10% level respectively.

Table-11 : Diagnostic Test Results

Test Statistic p-value Conclusion

JB 18.83 0.06 Normality Kurtosis 7.86 0.17 Normal Residuals

Skewness 8.65 0.08

Autocorrelation LM (16) 12.89 0.81 No Autocorrelation

Heteroskedasticity χ2 (140) 165.94 0.09 No Heteroskedasticity


20

telecom infrastructure to maintainthe growth momentum. Or, is there a‘feedback effect’ ?

We first examined the long run relation-ship through the co-integrationbetween telecom infrastructure and eco-nomic growth in India using annualdata for the period 1981-2018. Ourresults fail to establish any long run rela-tionship between telecom infrastructureand output level. However, we findshort run unidirectional causal relation-ship emanating from economic growthto telecom infrastructure, indicatingthat advancement in telecommunica-tions infrastructure has been a conse-quence of economic growth. Thecointegration results remained un-changed even in the multivariate VARanalysis. Interestingly, the causality resultsremained valid as well. Apart fromthat, we also observe the presence ofshort-run unidirectional causality fromreal (gross) domestic investment totrade volume, from REER to realinvestment as well as to real GDP.

It is often argued that the post-1991structural reforms in the Indian economybrought in a phenomenal growth in theservices sector and the result suggeststhat the recent development of telecominfrastructure has been apparently dueto the growing demand of the flourishingservices sector. Thus, it can be argued

that services led increased economicgrowth had induced the advancementin telecom infrastructure of late.

The absence of long run relationshipcould arise due to heterogeneity of tele-communications penetration betweenrural and urban areas. Such a significant‘digital divide’impedes the developmentof telecom infrastructure in rural areasand hence, constraints the overall effectof telecommunications infrastructureon economic growth. Also, the telecomequipment manufacturing sector ishighly dependent on imported equip-ment and therefore, most of thedomestic demand is met through importsrather than home production. This alsoaffects competitiveness of Indiantelecom products in the internationalmarkets and contributes to decay theimpact of telecommunications infra-structure on economic growth.

Given the spatial unevenness in the ruraland urban telecom infrastructure, eitherthe government has to mitigate theproblem of relatively poor teledensityor adequately incentivize the privateplayers to build up reasonable telecominfrastructure to tap the ‘bottom of thepyramid’ which is large enough for apopulous country like India.

It would have been interesting toexamine the nexus at the state levelusing panel cointegration framework.


21

Unfortunately, it is difficult to extractdisaggregated data for all the concernedvariables, especially for the telecommu-nications infrastructure. For instance,given the data period of our study wecould not get the FDI data for theentire period and examine whether thereis any inter-linkage between past FDIinflow into the country and improvementin telecom infrastructure.

End Note1. The SDGs are a collection of 17 global

goals set by the UN General Assembly in2015 for the targets to be achieved by2030.

ReferencesAsteriou, D. & Hall, S.G. (2007) Applied

Econometrics : A Modern Approach UsingE-Views and Microfit, New York : PalgraveMacmillan.

Beil, R.O., Ford, G.S. & Jackson, J.D. (2005)On the Relationship between Telecommu-nications Investment and EconomicGrowth in the United States, InternationalEconomic Journal, 19(1), 3-9.

Chakraborty, C. & Nandi, B. (2011) MainlineTelecommunications Infrastructure, Levelsof Development and Economic Growth :Evidence from a Panel of DevelopingCountries, Telecommunications Policy, 35(5),441-49.

Cieslik, A. & Kaniewsk, M. (2004) Telecom-munications Infrastructure and RegionalEconomic Development : The Case ofPoland, Regional Studies, 38(6), 713-25.

Cronin,F.J., Parker, E.B., Colleran, E.K. & Gold,M.A. (1991) Telecommunications Infra-structure and Economic Growth : AnAnalysis of Causality, TelecommunicationsPolicy, 15(6), 529-35.

Dickey, D.A. & Fuller, W.A. (1979) Distribu-tion of the Estimators for AutoregressiveTime Series with a Unit Root, Journal of theAmerican Statistical Association, 74(366a),427-31.

Dickey, D.A. & Fuller, W.A. (1981) LikelihoodRatio Statistics for Autoregressive TimeSeries with a Unit Root, Econometrica,49(4), 1057-72.

Dutta, A. (2001)Telecommunication andEconomic Activity : An Analysis ofGranger Causality, Journal of ManagementInformation Systems, 17(4), 71-95.

Engle, R.F. & Granger, C.W. (1987) Co-Inte-gration and Error Correction : Representa-tion, Estimation and Testing, Econometrica,55(2), 251-76.

Ghosh, S. & Prasad, R. (2012) Telephone Pene-trations and Economic Growth : Evidencefrom India, NETNOMICS : EconomicResearch and Electronic Networking, 13(1),25-43.

Granger, C.W.(1969)Investigating CausalRelations by Econometric Models andCross-Spectral Methods, Econometrica,37(3), 424-38.

Granger, C.W. & Newbold, P. (1974) SpuriousRegressions in Econometrics, Journal ofEconometrics, 2(2), 111-20.

Hardy, A.P. (1974)The Role of the Telephonein Economic Development, Telecommuni-cations Policy, 4(4), 278-86.


22

Johansen, S. & Juselius, K. (1990) MaximumLikelihood Estimation and Inference onCointegration with Applications to theDemand for Money, Oxford Bulletin ofEconomics and Statistics, 52(2), 169-210.

Kathuria, R., Uppal, M. & Mamta. (2009) AnEconometric Analysis of the Impact ofMobile in India, ICRIER Policy Paper Series,(9), 1-68.

Mehta, B.S. (2017) Performance of MobilePhone Sector in India, Economic andPolitical Weekly, 52(14), 50-56.

Narayana, M.R. (2011) TelecommunicationsServices and Economic Growth : Evidencefrom India, Telecommunications Policy, 35(2),115-27.

Nelson, C. & Plosser, C.I. (1982) Trends andRandom Walk in Macroeconomic TimeSeries : Some Evidence and Implications,Journal of Monetary Economics, 10(2),139-62.

Norton, S. (1992)Transaction Costs, Telecom-munications and the Microeconomics ofMacroeconomic Growth, EconomicDevelopment and Cultural Change, 41(1),175-96.

Panagariya, A. (2013) India : The EmergingGiant, Oxford University Press.

Phillips, P.C. & Perron, P. (1988) Testing for aUnit Root in Time Series Regression,Biometrika, 75(2), 335-46.

Pradhan, R., Arvin, M.B., Bahmani, S. &Norman, N.R. (2014) TelecommunicationsInfrastructure and Economic Growth :Comparative Policy Analysis for the G-20Developed and Developing Countries,Journal of Comparative Policy Analysis :Research and Practice, 16(5), 401-23.

Roller, L. & Waverman, L. (2001) Telecommu-nications Infrastructure and EconomicDevelopment : A Simultaneous Approach,American Economic Review, 91(4), 909-23.

Thompson, H.G. Jr. & Garbacz, C. (2001)Mobile, Fixed Line and Internet ServiceEffects on Global Productive Efficiency,Information Economics and Policy, 19(2),189-214.

World Bank. (1994) World Development Report :Infrastructure for Development, OxfordUniversity Press.

Yoo, S.H. & Kwak, S.J. (2004) Information Tech-nology and Economic Development inKorea : A Causality Study, InternationalJournal of Technology Management, 27(1),57-67.

— x — x —

Efficiency, Productivity and Returns to Scale of Indian General Insurance Industry :Evidence Based on Panel Data

23

Efficiency, Productivity and Returns to Scaleof Indian General Insurance Industry :

Evidence Based on Panel DataRam Pratap Sinha*

The present study compares the efficiency-productivity-returns to scale performance of 15general insurance companies operating in India during the period 2009-10 to 2013-14.We have used a panel data approach towards estimation leading to inter-temporalcomparison of performance in respect of both efficiency and returns to scale. The estimationof efficiency indicates that under the assumption that technology is local, the efficiencyperformance of the insurer’s has improved during the period. However, returns to scaleestimation indicates that most insurance companies were not operating at the optimalscale. Further, estimation of Malmquist productivity index indicates that while efficiencyshifts have been positive during the period under consideration, technical change has beenon the negative side indicating a lowering of the frontier.

Keywords : General Insurance, Window Analysis, Total Factor Productivity, Bootstrap, DEA.


* Dr.Ram Pratap Sinha, Associate Professor ofEconomics , Government College of Engineeringand Leather Technology, LB Block,Sector III,Salt Lake, Kolkata-700106, India.

IntroductionThe Indian insurance sector experiencedderegulation of entry in the aftermathof the introduction of banking sectorreform in India. The insurance sectorchanges can be considered as an inte-gral part of the overall program offinancial sector reform. Infusion of com-petition through private sector entry inthe insurance sector was essential topromote scale and scope economies inthe sector. From the international pers-pective also, this was quite essential asIndia was committed to the interna-tional community to open up its financialservices sector which, for long,remained a monopoly of the public

sector. Thus, the deregulation of entryin the Indian insurance sector took placein 1999 along with the establishmentof the insurance market regulatornamely the Insurance Regulatory andDevelopment Authority. The changesin the regulatory architecture andcompetition scenario were followed byderegulation of tariffs and many otherimportant policy changes which signifi-cantly influenced the working of thegeneral insurance sector. Against thebackdrop, the present study seeks toestimate efficiency, total factor productivity


24

changes and returns to scale for fifteengeneral insurance companies for theperiod 2009-10 to 2013-14.

While adopting the non-parametricmethodology for the estimation ofinsurer performance, we have departedfrom the extant approach used in theIndian context in several respects. In theestimation of efficiency performance ofthe in-sample insurers, we have adoptedthe panel data approach instead of thecross-section approach used in the cur-rent literature. Thus the benchmarkused in our study is based on the entireobservation period. A similar procedurehas been adopted for the estimation ofreturns to scale.

Estimation of efficiency on the basis ofa panel data approach requires theassumption there is no technical changecausing the frontier to shift in between.This is a limitation of the efficiencyanalysis. Thus, in the next phase, wehave also estimated total factor produc-tivity change for the period under obser-vation. While we have followed thepopular non-parametric approach forthe estimation of total factor produc-tivity, we have made a major departurefrom the past. Since the sample size forthe present study is small, we haveresorted to bootstrap DEA estimationof productivity and its components.Apart from giving bias-correctedestimates, this methodology enables us

to provide interval estimates of produc-tivity thereby enabling us to getstatistical interpretation of the scoresincluding interval estimates.

The paper has five sections and proceedsas follows. Section-1 provides an over-view of the post-reform general insu-rance sector. Section-2 describes themethodology. Section-3 describes therelated research work. Section-3 providesa brief overview of the relatedresearch work. Section-4 discusses theresults.Section-5 concludes.

Section-1 : An Overview of theIndian General Insurance IndustryIndian general insurance industry isquite small when compared to theInternational standard. The sector isalso smaller than the life insurancecounterpart. Thus there is considerablescope for the expansion of the sector.

During the period under consideration,the total number of diversified generalinsurers increased from 19 to 21. Onthe other hand, the total number ofstandalone health insurers increasedfrom 1 to 4 during 2009-10 to 2013-14.The number of specialized insurer’s,however, remained unchanged duringthe period. See Table-1 for details.

Of particular interest is the standing ofthe Indian general insurance sectorvis-a-vis the global scenario. For the sake


25

of international comparison, we consi-der two performance indicators-insur-ance density and insurance penetration.Insurance density is calculated as theratio of insurance premium (in US $)to the total population of the country. Onthe other hand, insurance penetrationis computed as the ratio of insurancepremium to GDP (in % terms).

Section-2 : Efficiency, TotalFactor Productivity and Returnsto ScaleIn a market driven economy with com-petition coming from both domesticand overseas competing firms, each pro-ductive entity needs to remain con-cerned about efficiency, productivityand returns to scale. In plain language,

efficiency corresponds to the perfor-mance of a productive unit with respectto some observed/ virtual best practiceunit. Productivity corresponds to theratio of output(s) to input(s). Returnsto scale implies the change in outputin relation to changes in the scale ofoperation. Thus, depending on whetherthe return is more than proportionate,exactly proportionate or less thanproportionate, we can classify returnsto scale into three types : increasing,constant and decreasing.

In the computation of efficiency andproductivity, the concept of returns toscale plays an important role. If thetechnology exhibits constant returns toscale throughout, the technology isglobal implying that the same benchmark

Table-1 : Players in the General Insurance Market(1999-00 to 2013-14)

Year 1999-00 2009-10 2013-14

No of diversified 4 19 21general insurers

Standalone health insurer 0 1 4

Specialised general insurers 1 2 2

Source : IRDA (2014) : Handbook on Indian Insurance Statistics 2013-14.

Table-2 : General Insurance Density and Penetration :India and the World (2014)

General Insurance Insurance Density Insurance Penetration

India 11 0.70

World 294 2.70

Source : IRDA (2017) : Handbook on Indian Insurance Statistics 2016-17.


26

applies for all firms irrespective of theirscale of operations. However, if thereturns to scale is variable, then the refe-rence units can be different for differentunits under observation. In a similarvein, the returns to scale exhibited byan unit has important implications forthe average productivity of the firm.Depending on the nature of returns toscale, the observed firm observed scalesize will be less than, equal to or greaterthan the most productive scale size(Banker, 1984).

Efficiency, Productivity andReturns to Scale for a SingleInput-Single Output TechnologyIn order to provide a more formal pre-sentation of the relationship of returnsto scale with efficiency and producti-vity, let us consider a single input-single output technology (Ray, 2004)characterized by the productionpossibility set :

PS = [(x,y) : y<f(x)]

Here y and x stand for output and inputrespectively. The inequality sign allowsthe possibility of x-inefficiency : obser-ved output can be less than or equal tothe potential output suggested by thetechnology. Efficiency from the output

perspective is defined as : E yf x

o =( )

. Let

firs(x), fcrs(x) and fdrs(x) represent the bestpractice output under increasing,

constant and decreasing returns to scale.Since in the case of Most ProductiveScale Size (MPSS), the technologyexhibits constant returns to scale, wehave firs(x) < fcrs(x) and fdrs(x) < fcrs(x).Thus if we accommodate variable re-turns to scale (i.e. returns to scale canbe increasing, constant or decreasing)then E Evrs

ocrso≥ .

Now let us consider the linkage of pro-ductivity with the returns to scale. We

define average productivity AP= f xx( ) .

By using the previous logic,

APf xx

f xxcrs

crs vrs= ( ) ≥ ( ) . Thus average pro-

ductivity is highest when the firmexperiences MPSS.

Efficiency, Productivity andReturns to Scale in a Multi-Input Multi-Output SettingWe now consider a technology PT

which models the transformation ofinputs x (numbering 1,2, ..…, n) onto outputs y (numbering 1,2, ….., m).Thus x RN∈ + and y RM∈ + .

Then the production technology canbe represented as :

PT = [(x,y) : x can produce y]

Invoking Shephard (1953,1970), theoutput distance function may beexpressed as :


27

D x y x y Tfo , min : ,( ) = ⎛

⎝⎜

⎞

⎠⎟∈

⎧⎨⎪

⎩⎪

⎫⎬⎪

⎭⎪=µ

µ

x y PTmax : ,= ( )∈{ }⎡⎣ ⎤⎦−

δ δ11

An observed firm is inefficient whenD x yfo ,( ) <1 and is efficient whenD x yfo ,( ) =1 .

For a firm with observed outputand input vectors being y0 and x0

respectively, the optimization problemis : Max δ

Subject to : δ λy Y0 ≤ and x X e0 1≥ =λ λ,

(for vrs). Thus = Eo =1δ

Note that, for period-wise data, wemake use of the period specific obser-vations on input and output as the refe-rence sets. However, for panel data, wecan use the entire panel as the referenceset (see Charnes et.al., 1985). Thus ifwe have a panel of observations relatingto the inputs and outputs (in our casen.t observations on inputs and m.tobservations on outputs assuming thatobservations are available for timeperiods 1,2, ....., t), then the modifiedmathematical program would be :

Max δp

Subject to : δ λy Yp0 ≤ and x X ep0 1≥ =λ λ,

(for vrs). Thus Ep

01=δ

where the sub-

script p implied that the reference set isactually a panel.

While estimating productivity, we areprimarily interested in knowing howproductivity has changed inter-tempo-rally. The non-parametric approach ofproductivity estimation assumes theexistence of a production/cost functionbut does not require a parametric rela-tionship between the outputs and inputs.The most popular non-parametricmeasure of total factor productivitychange is the the Malmquist Produc-tivity Index which was introduced by.Caves, Christensen and Diewert (1982).The computation of Malmquist Pro-ductivity Index, like the case of DEAbased efficiency estimation, is based onthe concept of input and output dis-tance functions originally introducedby Shephard (1953, 1970).

For understanding the methodology ofMalmquist productivity estimation, letus consider two consecutive time periods0 and 1. Using the technology of timeperiod 0, the Malmquist ProductivityIndex as per Caves, Christensen andDiewert (1982) can be defined as :

MD x y

D x y

xyxy

f

f0

0 1 1

0 0 0

1

1

0

0

=( )( ) ×,

, and

MD x y

D x y

xyxy

f

f1

1 1 1

1 0 0

1

1

0

0

=( )( ) ×,

,


28

Under the assumption of constantreturns to scale, the frontier is anupward rising straight line implying

that, xy

xy

1

1

0

0= . we have MD x y

D x yf

f0

0 1 1

0 0 0=

( )( ),

,

and MD x y

D x yf

f1

1 1 1

1 0 0=

( )( ),

,.

The Malmquist productivity change

index is computed as = M M0 1.

Under the operation of constant returnsto scale, Färe, Grosskopf, Lindgren andRoos(1989,1992) decomposed theoutput based Malmquist index into thefollowing two components: EfficiencyChange and Technical Change.

Efficiency Change = D x y

D x yf

f

0 1 1

0 0 0

,

,( )( )

Technical Change =

D x y

D x y

D x y

D x yf

f

f

f

0 1 1

1 1 1

0 0 0

1 0 0

12,

,

,

,( )( ) ×

( )( )

⎡

⎣⎢⎢

⎤

⎦⎥⎥

Under variable returns to scale, bothefficiency change and technical changecomponents can be decomposed further.Färe, Grosskopf, Norris and Zhang(1994) decomposed the efficiencychange component of Malmquist indexinto pure efficiency and scale components.

Pure Efficiency Change = D x y

D x yf VRS

f VRS

1 1 1

0 0 0( )

( )

( )( ),

,

Scale Efficiency Change =

=

D x y

D x y

D x y

D x y

f CRS

f VRS

f CRS

f VRS

1 1 1

1 1 1

0 0 0

0 0 0

( )

( )

( )

( )

( )( )( )

,

,

,

,(( )

Wheelock and Wilson (1999) decom-posed the Technical Change (CRS) into the two components : Pure TechnicalChange and Change in Scale of Techno-logy. Thus as per Wheelock andWilson (1999), Malmquist ProductivityIndex (VRS)= Pure Efficiency ChangeX Scale Efficiency Change X Pure Tech-nical Change X Change in Scale ofTechnology.

Change in Scale of Technology =

=

D x y D x y

D x y D x yf CRS f CRS

f VRS f VRS

0 1 1 0 0 0

0 1 1 0 0( ) ( )

( ) ( )

( )× ( )( ), ,

, , 00

1 0 0 1 0 0

1 0 0 1

( )( ) ( )( )

( ) ( )

( ) ( )

D x y D x y

D x y D xf CRS f CRS

f VRS f VRS

, ,

, 00 0, y( )

⎛

⎝

⎜⎜⎜⎜⎜⎜

⎞

⎠

⎟⎟⎟⎟⎟⎟

Bootstrap Estimation ofMalmquist ProductivityIn the present context, we have obser-vations on only 15 insurance compa-nies per year. In the estimation ofMalmquist productivity index, wemake a pair-wise comparison for twoyears. Thus the sample size is quitesmall. For a finite sample size, Bankeradmitted that for a finite sample size,the estimated production frontier would


29

lie below the theoretical frontier. Thus,DEA estimates of efficiency wouldoverestimate the true position. Further,Korostelev, et al (1995a, 1995b) havedemonstrated that while the DEA esti-mates are consistent (under very weakgeneral conditions) the rate of conver-gence ( to the true frontier) is very slow,Thirdly and finally, the DEA basedestimates by giving only point esti-mates, are not amenable to statisticalinterpretation.

In order to overcome the problem, wehave made use of smoothed bootstrapmethod suggested by Simar and Wilson(1998). The plain vanilla bootstrapmethod suggested by Efron (1979)implies procedure of drawing withreplacement from a sample and theresultant samples mimic the data gene-rating process of the underlying truemodel which can be used for statisticalinference. In the context of efficiencyanalysis, the conditional density of thedistance function has bounded supportover the interval (0,1) and is right dis-continuous at 1. In order to cope withthe problem, Simar and Wilson (1998)made bootstrap estimation of the dis-tance function using univariate kernelestimates of the marginal density of theoriginal estimates. Thus for a randomvariable X having values x1, x2, …., xn

with probability density function f andcumulative distribution function F, f isestimated by using the Parkan-Rosenblatt

kernel = f nh k xhn

ei

n= ( ) −⎛⎝⎜

⎞⎠⎟

−

=∑1 1

θ where the

kernel k is an univariate pdf and h is thesmoothing parameter. In case of efficiencyestimation, x <1. then a symmetric kernel

is used : g nh kxh

k xhn

ei

n=−( ) + − +⎧

⎨⎪

⎩⎪⎫⎬⎪

⎭⎪−

=∑1 1

2θ θ .

Appropriate values of the smoothingparameter is found by maximizing thelikelihood cross validation function(refer Silverman (1986)).

For incorporating the bootstrap processfor the estimation of efficiency, weassume a data generating process (DGP)in which the firms deviate in a randomfashion from the production frontierat time t. From the output perspective,this is measured by the output distancefunction Dt

0 . In the bootstrap approach,

this DGP is replicated multiple timesresulting in a large number (say B) ofpseudo samples of input and output

variables : T x ybitb

itb= { }, where i = 1,2,

...., n and t = 0,1, The original estimatorsare now applied to these pseudo samples.For each bootstrap sample b Bs ∈ , effi-ciency is computed by measuring thedistance of each observation belongingto the original sample from the frontierconstructed from the pseudodata.

In the context of computation of theMalmquist productivity index, it isessential to utilise the relationshipP x yT it it* * *,= { } for measuring the distance


30

of each observation belonging to theoriginal sample from the frontier createdout of the pseudo sample of inputs andoutputs.

D x yf y0

0 0

1, sup ,( ){ } =

−

θ θ

Subject to : θy Yb0 0≤ .....(5)

x Xbj j j0 0 0≥ ∑ ≥λ λ λ, ,

D x yf1

1 1

1, sup ,( ){ } =

−

θ λ θ

Subject to : θ λy Yb1 1≤ .....(6)

x Xbj j j1 1 0≥ ∑ ≥λ λ λ, ,

D x yf0

1 1

1, sup ,( ){ } =

−

θ λ θ


x Xbj j j1 0 0≥ ∑ ≥λ λ λ, ,

D x yf1

0 0

1, sup ,( ){ } =

−

θ λ θ


x Xbj j j0 1 0≥ ∑ ≥λ λ λ, ,

In case of estimation of Malmquist indexof productivity change, we deal with paneldata with the possibility of temporalcorrelation. To preserve this, the jointdensity of two period distance functionsis estimated. The bivariate kernel can

be written as : f nhk x

hbne

i

n= −⎛

⎝⎜

⎞

⎠⎟=∑1

2 1

* *θ

where x* has (1X2) dimension andθ* / /( , )= D D0 0 1 1 . However, when the supportof f is bounded (as is the case with

Malmquist Productivity Index) theestimated density from the bivariatekernel is inconsistent and asymptoti-cally biased. Thus we use the symmetrickernel using Silverman’s reflectionmethod. For further details on thisissue, see Simar and Wilson (1999).

The next issue is that of bias correction.After the computation of the bootstrapvalues it is essential to correct for thefinite sample bias prevailing in the originalestimators of the distance function. Thebias is calculated as :

Bias MMB

Mb

B

001

0( ) = −=∑ *

Thus the bias

corrected estimate of Malmquist indexof productivity change is computed

as : M MMB

bc b

B

0 0012= − =∑ *

Finally, we need to generate intervalestimates of Malmquist index of producti-vity change. For this, we approximate

the unknown distribution of M M0 0−( )by M M0 0

* −( ) conditioned on the originalsample data. Since we do not know the

distribution of M M0 0−( ) , we can use

the bootstrap estimates to find out val-ues aα

* and bα* such that Prob

− ≤ −( ) ≤ −{ }b M M aa* *

0 0 α tends to 1-α. By

reversing the sign and rearranging theterms we get a confidence interval of1-α :

M a M M b0 0 0+ ≤ ≤ +α α* *


31

Related Research WorkOne of the earliest efficiency-produc-tivity study in the context of the gene-ral (non-life) insurance industry wasby Toivanen (1997) who researched forthe presence of economies of scale andscope in the Finnish non-life insurancesector using data for the period 1984-91. The study involved the estimationof a quadratic cost function. During theperiod of study Finnish non-life busi-ness was branch centric in nature andthe insurance firms tried to expand theirbranch network for gaining marketpower or informational advantages.The study revealed that diseconomiesof scale existed at the firm level whereaseconomies of scale was present at thebranch level. Economies of scope waspresent in the production process.

Fukuyama and Weber (2001) estimatedefficiency and productivity growth ofJapanese non-life insurance companiesfor the time period 1983-1994. Thestudy estimated Farrell, Russell andZieschang measures of output orientedtechnical efficiency and on the basis ofthese measures, constructed Malmquistindex of total factor productivitychange. Then the index is decomposedinto efficiency change and technologi-cal change indices. The outcomes of thestudy showed that between 1983-90productivity improved significantlyand it was mainly due to technological

change. In the next three years, thecollapsed bubble economy resulted inthe stagnation of technological change.However, by 1993-94, there was againan upturn in technological change.

Cummins and Xie (2008) assessed theefficiency and productivity effects ofmergers and acquisitions in the USproperty-liability insurance industryduring the period 1994-2003 usingdata envelopment analysis (DEA) andMalmquist productivity index. Thestudy examined efficiency and produc-tivity shifts for three types of insurers :acquirers, acquisition targets, and non-M&A firms. For examining characteris-tics of the in-sample firms, the studyemployed probit analysis. The resultsprovide evidence that mergers and acqui-sitions in property-liability insuranceindustry enhanced firm valuation. Thestudy further found that the acquiringfirms achieved more revenue efficiencygains than the non-acquiring firms, andtarget firms experienced greater costand allocative efficiency growth com-pared to non-targets.

Kasman and Turgutlu (2009) appliedthe Malmquist total factor productivityindex to examine productivity growthin Turkish insurance industry for theperiod 2000-2005. The overall produc-tivity growth is the decomposed in totechnological change and efficiencychange. Their study found that during


32

the period the non-life sector had apositive productivity growth (especiallyduring 2003-05) while the life insur-ance sector regressed. The growth in thenon-life sector was mainly due tofavourable efficiency shifts.

Barros, Nektarios and Assaf (2010)employed the two-stage double boot-strap approach to evaluate the perfor-mance of Greek life and non-lifeinsurance companies using the globaltechnology framework for the period1994-2003. The first stage efficiencyresults indicate a decline in efficiencyover the sample period. The secondstage truncated regression confirmedthat the competition for market sharesis a major influencing factor ofefficiency performance in the Greekinsurance industry.

Vencappa et.al. (2013) estimated anddecomposed productivity growth foran unbalanced panel of Europeaninsurance companies for the period1995-2008. The study estimatedproductivity growth using a parametricstochastic frontier method. Total factorproductivity change is then decomposedin to four components : (i) technicalchange, (ii) scale efficiency change,(iii) technical efficiency change and (iv) scaleefficiency change. In order to capturethe inherent variability encounteredin the insurance sector, the study usedthree output proxies : total premiums

collected, total claims incurred and thesum of claims paid and any changesmade to the insurer’s loss reserves. Thestudy found that total productivitygrowth in the European insurancesector was volatile in nature and drivenprimarily by changes in mean technicalefficiency.

Javaheri (2014) estimated total factorproductivity change in respect of allIranian insurance companies relative tothe period 2003-2009 using data enve-lopment analysis. In order to examinethe influence of environmental variableon productivity, he used to bit regres-sion was. The results indicated that thepolicy of liberalization had a positiveimpact on productivity growth.Further, the results also indicated thatdimension and the field of activityhad significant positive effect onproductivity changes.

Alhassan and Biekpe (2015) analysedeficiency, productivity and returns toscale economies in the non-life insu-rance market in South Africa for thetime span 2007-2012. Data envelop-ment analysis was employed to estimateefficiency and returns to scale whileproductivity growth was analysed byusing Malmquist productivity index.They applied Truncated bootstrappedand logistic regression techniques forfinding out the determinants of efficiencyand the probability of operating under


33

constant returns to scale. The resultsshowed that non-life insurers operatedwith about 50 per cent efficiency.Approximately 20 per cent of insurerswere scale efficient. The study alsofound productivity improvementsduring the period which was mainlydue to technological changes. The resultsof the regression analysis indicated anon-linear impact of size on efficiencyand constant returns to scale. Variableslike product line diversification, rein-surance and leverage also had a signifi-cant relationship with efficiency andconstant returns to scale.

Data, Results and DiscussionInputs, Outputs and DataEstimation of efficiency and produc-tivity performance of the non-lifeinsurance companies require the iden-tification of performance indicators(inputs and outputs of the productionprocess). However, this is a challengingtask in so far as the general insuranceindustry is concerned. There are long-standing disagreements among theresearchers regarding the appropriatechoice of inputs and outputs. The samevariable has been used as an input insome research studies and as an out-puts in other studies. For getting moreinformation on the extant approachesto the selection of inputs and outputs,see Eling and Luhnen (2010), Levertyand Grace (2019) and Jarraya and

Bouri (2012). On the whole, assuggested by Leverty and Grace (2010),there are two dominant competingapproaches for the identification andselection : the Flow Approach and theValue Added Approach.

The Flow Approach treats the insu-rance firms as financial intermediarieswhich acts as transforming institutionsof premiums in to claims payment.The important Flow Approach outputindicators are rate of return on invest-ments, the ratio of liquid assets to lia-bilities and the probability of solvencyof the insurance company. The inputsinclude the current policy holder’s sur-plus, the sum of the costs incurred forperforming the underwriting andinvestment functions and the policy-holder supplied debt capital (represen-ted by the sum of unpaid net losses,unpaid loss adjustment expenses andunearned premium reserves).

The alternative approach i.e., the ValueAdded Approach uses outputs relatedto the amount of financial services pro-vided by the insurance firms. Theimportant output indicators in theValue Added Approach include claimsexpected to be paid as a result of pro-viding insurance coverage during a par-ticular period and the average realinvested assets of a firm. The importantinput indicators include expenditure onlabour and physical capital, financial


34

equity capital and policy holdersupplied debt capital.

In the present context, we have used ahybrid approach with the inclusion oftwo inputs and two outputs. The twoinputs are operating expenses and netpremium income. Since item-wise ex-penses relating to labour and physicalcapital are not available, operating ex-penses relating to the insurance busi-ness serves as a broad indicator for theseexpenses. On the output side, there aretwo important indicators: benefits paidand asset under management. The firstone i.e. benefits to the policy holdersrepresents the claims paid to the policyholders i.e. it represents the real insur-ance services provided by the generalinsurers. The second one represents in-vested assets of the firm and it has beentaken as a free output link. The in-sampleinsurance companies are assumed tooperate under variable returns to scale.Productivity estimation is made forthe output-oriented approach. Thenominal data have been appropriatelydeflated to facilitate inter-temporalcomparison.

Data related to the Indian generalinsurance companies have been collectedfor the period 2009-10 to 2013-14from IRDA Annual Reports and theHandbooks of Indian Insurance Statis-tics for the financial years 2011-12 and2013-14. Fifteen general insurancecompanies have been included in thestudy. Other companies could not beenincluded because it was essential to forma balanced panel of observations.

Results and DiscussionIn the present section, we provide esti-mates of panel (window) based esti-mates of efficiency, bias corrected boot-strap estimates of Malmquist produc-tivity estimate and it components andreturns to scale exhibited by the insample general insurers.

Window Analysis of EfficiencyTable-3 provides the descriptivestatistics of output oriented efficiencyestimates for the in-sample generalinsurers for the period under observation.However, the efficiency scores arederived using the entire panel of

Table-3 : Selected Performance Indicators

Indicators Input Indicator Output Indicator

Net premium income √ X

Operating expenses √ X

Benefits paid X √

Asset under management X √

Source : Author’s own.


35

observations (including 80 observationpoints) as the reference set. Since thereference set is the same, the efficiencyscores are comparable from one periodof time another. The results clearlypoint out that during reference timespan, mean efficiency has improved.Similar improvement is also seen inrespect of minimum efficiency scoresand standard deviation of efficiencyscores except for 2010-11. The detailedinsurer- wise efficiency scores are avai-lable in appendix Table-A1.

Returns to Scale EstimatesTable-5 provides the summary infor-mation regarding returns to scale esti-mates exhibited by the general insurersfor the period under observation. TheTable shows that during the last twoobserved years (2012-13 and 2013-14respectively), 14 out of the 15 generalinsurers exhibited decreasing returns toscale. Thus most of the general insur-ers were not operating at the optimalscale where observed productivity

Table-4 : Window Based Estimation of Efficiency

Particulars 2009-10 2010-11 2011-12 2012-13 2013-14

Mean Efficiency 0.7641 0.7852 0.9245 0.9288 0.9314

Maximum 1 1 1 1 1

Minimum 0.5228 0.3860 0.7399 0.7540 0.7582

Standard Deviation 0.1813 0.2159 0.0995 0.0840 0.0785

Source : Calculated.

Figure-1 : Inter-Temporal Movement in Efficiency

2009-10 2010-11 2011-12 2012-13 2013-14

1.2

1

0.8

0.6

0.4

0.2

0

Mean Efficiency

Maximum

Mimnimum

Standard Deviation


36

equals the optimal. The detailedinsurer-wise information about returnsto scale for the period under observa-tion is available in appendix Table-A2.

Trends in Total Factor ProductivityIn the present context, computation ofefficiency has been made without con-sidering the fact that during the periodunder consideration, the frontier itselfmight have shifted due to technologi-cal change. As indicated earlier, the

computation of Malmquist productivityindex along with its component enablesus to understand how productivitychange has been influenced by its twomajor constituents: change in efficiencyand change in the frontier itself. Table-5provides the mean bias-corrected scoresrelating to the index and its components(and sub-components) for the entireperiod under consideration. Theinsurer-wise estimates are available inappendix Tables-A3 to A8.

Table-5 : Summary Information about Returns to Scale

Particulars 2009-10 2010-11 2011-12 2012-13 2013-14

Number of general 1 1 0 1 1insurers exhibiting CRS

Number of general 3 3 2 0 0insurers exhibiting IRS

Number of general 11 11 13 14 14insurers exhibiting DRS


Table-5 : Bias-Corrected Productivity Scores Over the PeriodUnder Observation

Productivity Measure Mean Score Lower Bound Upper Bound

Malmquist Productivity Index 1.0450 0.9849 1.0985

Pure Efficiency Change 1.2718 1.1777 1.4802

Scale Efficiency Change 1.0982 1.0112 1.2907

Efficiency Change 1.4147 1.3997 1.5629

Pure Technical Change 0.9308 0.8439 0.9498

Scale Efficiency of Technical Change 0.9877 0.8888 1.0994

Technical Change 0.7838 0.7176 0.7966



37

Concluding ObservationsThe present paper made use of twointer-temporal approaches for the esti-mation of insurer performance over thefive year span 2009-10 to 2013-14.Initially we have assumed that there isno technical change during the periodunder consider and treat each insurerin a particular year as a distinct deci-sion-making unit. The windowapproach to efficiency analysis reveals thatduring the period under consideration,mean efficiency scores have improvedimplying that the insurance companieshave reached closer to the frontier.

However, we have assumed that theproduction technology is local and thusfrom the estimation process, we areunable to get any information aboutscale efficiency. Thus, in the next stage,we have computed returns to scale onthe basis of panel data. The outcomes

indicate that most of the insurers arenot operating at the optimal scale andthe proportion of insurers exhibitingdecreasing returns to scale has increasedover the observation period.

The window analysis is silent abouttechnical progress/regress and for ascer-taining total factor productivitychanges, we have estimated Malmquistproductivity index and its efficiencyand technical change components. Theresults indicate that while the insurerswitnessed positive efficiency shifts dur-ing the period, technical change hasbeen on the negative direction. Decom-position of efficiency change in to pureand scale components indicate that thepure efficiency change factor is thedominant one. However, the decom-position of technical change leads to theinfeasibility problem for six out of the15 in-sample general insurers and no

Figure-2 : Point and Interval Estimates of Productivity Indicators

Malm

quist

Pro

ducti

vity

Inde

x

1.81.61.41.2

10.80.60.40.2

0

Pure E

fficie

ncy

Chang

e

Scale

Effi

cienc

y Cha

nge

Efficie

ncy

Chang

e

Pure

Techn

ical C

hange

Self-

Efficie

ncy T

echn

ical C

hang

e

Techn

ical C

hange

123123123123123123123123123123123123

123123123123123123123123123123123123

123123123123123123123123123123123123123123123

123123123123123123123123123123123123123123123

123123123123123123123123123123123123123123

123123123123123123123123123123123123123123123123123123123

123123123123123123123123123123123123123

123123123123123123123123123123123123

123123123123123123123123123123123123123123123123123

123123123123123123123123123123123123123123123123

123123123123123123123123123123123123123123123123

123412341234123412341234123412341234123412341234123412341234123412341234123412341234

123123123123123123123123123123123

123123123123123123123123123123

1234123412341234123412341234123412341234123412341234

123123123123123123123123123123123123

123123123123123123123123123123

123123123123123123123123123123123123123123123

123123123123123123123123123

123123123123123123123123

123123123123123123123123123123123

Mean Score

Lower Bound

Upper Bound

123123123

123123123

123123123


38

conclusion should be drawn on thebasis of the truncated sample.

ReferencesAlhassan, A.L. & Biekpe, N. (2015) ‘Efficiency,

Productivity and Returns to Scale Econo-mies in the Non-life Insurance Market inSouth Africa, The Geneva Papers on Risk andInsurance-Issues and Practice, 40(3), 493-515.

Banker, R.D. (1984) “Estimating the Most Pro-ductive Scale Size Using Data EnvelopmentAnalysis,” European Journal of OperationalResearch, 17 (1), 35-44.

Banker, R.D. (1993) Maximum Likelihood,Consistency and Data EnvelopmentAnalysis : A Statistical Foundation,Management Science, 39(10), 1265-1273.

Barros, C.P., Nektarios, M. & Assaf, A. (2010)Efficiency in the Greek Insurance Industry’,European Journal of Operational Research,205(2), 431-436.

Caves, D., Christensen, L. &Diewert, E. (1982)The Economic Theory of Index Numbersand the Measurement of Input, Outputand Productivity, Econometrica, 50(6),1393–1414.

Charnes, A., Clark, C.T., Cooper, W.W. &Golany, B. (1984) A DevelopmentalStudy of Data Envelopment Analysis inMeasuring the Efficiency of MaintenanceUnits In The U.S. Air Forces, Annals ofOperations Research, 2(1), 95-112.

Cummins, J.D. & Xie, X. (2008) ‘Mergers andAcquisitions in the US Property-LiabilityInsurance Industry : Productivity andEfficiency Effects.’ Journal of Banking &Finance, 32(1), 30-55.

Efron, B.(1979) “Bootstrap Methods: AnotherLook at the Jackknife,” Annals of Statistics,7(1), 1-26.

Eling, M. & Michel, L. (2010) “FrontierEfficiency Methodologies to MeasurePerformance in the Insurance Industry :Overview, Systematization and RecentDevelopments”, The Geneva Papers on Riskand Insurance - Issues and Practice, 35(2),217-265.

Färe, R.S., Grosskopf, B. Lindgren & Roos, P.(1989) “Productivity Developmentsin Swedish Hospitals : A MalmquistOutput Index Approach,” Department ofEconomics, Southern Illinois University,Carbondale.

Färe, R., Grosskopf, S., Lindgren, B. & Roos, P.(1992) “Productivity Change in SwedishPharmacies 1980-1989 : A NonparametricMalmquist Approach”, Journal ofProductivity Analysis, 3(1), 85-101.

Färe, R., Grosskopf, S., Norris, M. & Zhang, Z.(1994) “Productivity Growth, TechnicalProgress, and Efficiency Change in Indus-trialized Countries”, American EconomicReview, 84 (1), 66-83.

Fukuyama, H. & Weber, W.L. (2001)“Efficiency and Productivity Change ofNon-Life Insurance Companies in Japan”,Pacific Economic Review, 6(1), 129-146.

Jarraya, B. & Bouri, A. (2013) “EfficiencyConcept and Investigations in InsuranceIndustry : A Survey”, Management ScienceLetters, 3(1), 39-54.

Javaheri, A.S. (2014) “Productivity Evaluationof Iranian Insurance Industry”, Iranian Journalof Economic Research, Vol. 18(57), 55-85.


39

Kasman, A. & Evrim, T. (2009) Total FactorProductivity in the Turkish InsuranceIndustry, International Journal of theEconomics of Business, 16:2, 239-247.

Korostelev, A., Simar, L. & Tsybakov, A. (1995a)Efficient Estimation of MonotoneBoundaries, Annals of Statistics, 23(2),476-489.

Korostelev, A., Simar, L. & Tsybakov, A. (1995b)“On Estimation of Monotone and ConvexBoundaries”, Publications de l’Institutde Statistique de l’Université de Paris,39(1),3-18.

Leverty, J. Tyler & Grace, M.F. (2010)“The Robustness of Output Measures inProperty-Liability Insurance EfficiencyStudies”, Journal of Banking and Finance,34(7), pp.1510-1524.

Malmquist, S. (1953) “Index Numbers andIndifference Surfaces,” Trabajos deEstatistica, 4(1) : 209-242.

Shephard, R.W. (1953) Cost and Productions,Princeton : Princeton University Press.

Shephard, R.W. (1970) Theory of Cost andProductions, Princeton : PrincetonUniversity Press.

Silverman, B.W. (1986) Density Estimation forStatistics and Data Analysis, Chapman &Hall, London – New York .

Simar, L. & Wilson, P.W. (1998) “SensitivityAnalysis of Efficiency Scores : How to Boot-strap in Nonparametric Frontier Models.”Management Science, 44(1),49-61.

Simar, L. & Wilson, P.W. (1999) “Estimatingand Bootstrapping Malmquist Indices”.European Journal of Operational Research,115(3), 459-471.

Toivanen, O. (1997) ‘Economies of Scale andScope in the Finnish Non-Life InsuranceIndustry,’ Journal of Banking & Finance,21(6), 759-779.

Vencappa, D., Paul, F. & Stephen, D. (2013)“Productivity Growth in the EuropeanInsurance Industry: Evidence from Life andNon-Life Companies,” InternationalJournal of the Economics of Business, 20(2),281-305.

Wheelock, D.C. & Wilson, P.W. (1999)“Technical Progress, Inefficiency andProductivity Change in U.S. Banking,1984-1993,” Journal of Money, Creditand Banking, 31(2), 212-234.

— x — x —


40

Table-A1: Efficiency Performance of the In-Sample General Insurers

Insurer 2010 2011 2012 2013 2014

Bajaj Allianz 0.6783 0.8279 0.8353 0.828185 0.847435

Bharti AXA 0.5847 0.3860 1 1 0.954402

Cholamandalam 0.5641 0.5684 0.769548 0.849898 0.878305

Future Generali 0.5580 0.4498 0.999999 1 1

HDFC Ergo 0.7319 0.7273 0.909503 0.836035 0.902241

ICICI Lombard 0.9413 1 1 0.966219 0.972695

IFFCO Tokio 0.8206 1 0.955501 0.8396 0.835531

Reliance 0.7150 0.7871 1 1 1

Royal Sundaram 0.6104 0.6536 0.739913 0.928925 0.861348

Shri Ram General 1 1 1 1 1

Tata AIG 0.5228 0.5640 0.752072 0.754029 0.758194

National 0.7442 0.9388 0.942914 1 0.961148

New India 1 1 1 1 1

Oriental 0.9894 0.8749 0.962079 0.943038 1

United 1 1 1 0.986481 1


Appendix


41

Table-A2 : Insurer-wise Returns to Scale for the In-Sample Years

Insurer 2009-10 2010-11 2011-12 2012-13 2013-14

Bajaj Allianz Decreasing Decreasing Decreasing Decreasing Decreasing

Bharti AXA Decreasing Increasing Decreasing Decreasing Decreasing

Cholamandalam Increasing Decreasing Decreasing Decreasing Decreasing

Future Generali Increasing Increasing Increasing Decreasing Decreasing

HDFC Ergo Increasing Decreasing Decreasing Decreasing Decreasing

ICICI Lombard Decreasing Decreasing Decreasing Decreasing Decreasing

IFFCO Tokio Decreasing Constant Decreasing Decreasing Decreasing

Reliance Decreasing Decreasing Decreasing Decreasing Decreasing

Royal Sundaram Decreasing Decreasing Decreasing Decreasing Decreasing

Shri Ram General Constant Increasing Increasing Constant Constant

Tata AIG Decreasing Decreasing Decreasing Decreasing Decreasing

National Decreasing Decreasing Decreasing Decreasing Decreasing

New India Decreasing Decreasing Decreasing Decreasing Decreasing

Oriental Decreasing Decreasing Decreasing Decreasing Decreasing

United Decreasing Decreasing Decreasing Decreasing Decreasing



42

Table-A3 : Bias Corrected Estimates of Mean Productivity,Efficiency and Technical Change

Insurer Malmquist Productivity Efficiency Change Technical Change

Bajaj Allianz 1.1734 1.1995 0.9116

Bharti AXA 1.0715 3.3732 0.3567

Cholamandalam 1.1803 1.6273 0.7172

Future Generali 1.0475 2.0312 0.5598

HDFC Ergo 0.9755 1.3574 0.7481

ICICI Lombard 0.9227 0.9760 0.8717

IFFCO Tokio 1.0045 1.0231 0.8778

Reliance 1.2297 1.4191 0.8056

Royal Sundaram 1.1549 1.3595 0.7718

Shri Ram General 0.1577 0.9025 0.1815

Tata AIG 1.1698 1.6158 0.7340

National 1.0282 1.1878 0.7718

New India 1.0063 1.0080 0.8692

Oriental 0.7770 0.9492 0.7463

United 0.8737 0.9660 0.8404



43

Table-A4 : Bias Corrected Estimates of Mean Efficiency and TechnicalChange Components

Pure Scale Pure Scale Efficiency Insurer Efficiency Efficiency Technical of Technical

Change Change Change Change

Bajaj Allianz 1.1569 0.9591 0.9119 0.9551

Bharti AXA 1.5514 1.8563 NA NA

Cholamandalam 1.4968 1.0079 NA NA

Future Generali 1.5185 1.1066 NA NA

HDFC Ergo 1.1687 1.0524 NA NA

ICICI Lombard 0.9703 0.9258 0.8474 0.9593

IFFCO Tokio 0.9547 0.9986 0.9109 0.9414

Reliance 1.2683 1.0000 0.8185 0.8570

Royal Sundaram 1.3506 0.9377 0.6926 0.9788

Shri Ram General 0.9025 1.0000 NA NA

Tata AIG 1.4080 1.0000 NA NA

National 1.1677 0.8270 0.8402 0.8442

New India 0.9194 0.7855 0.9873 0.6854

Oriental 0.9221 0.8821 0.7434 0.9224

United 0.9097 0.8296 0.8430 0.8553



44

Table-A5: Bias Corrected Lower Bounds of Productivity,Efficiency and Technical Change


Bajaj Allianz 1.1734 1.1995 0.9116

Bharti AXA 1.0715 3.3732 0.3567

Cholamandalam 1.1803 1.6273 0.7172

Future Generali 1.0475 2.0312 0.5598

HDFC Ergo 0.9755 1.3574 0.7481

ICICI Lombard 0.9227 0.9760 0.8717

IFFCO Tokio 1.0045 1.0231 0.8778

Reliance 1.2297 1.4191 0.8056

Royal Sundaram 1.1549 1.3595 0.7718

Shri Ram General 0.1577 0.9025 0.1815

Tata AIG 1.1698 1.6158 0.7340

National 1.0282 1.1878 0.7718

New India 1.0063 1.0080 0.8692

Oriental 0.7770 0.9492 0.7463

United 0.8737 0.9660 0.8404



45

Table-A6 : Bias Corrected Upper Bounds of Productivity,Efficiency and Technical Change



Bajaj Allianz 1.1569 0.9591 0.9119 0.9551

Bharti AXA 1.5514 1.8563 Infeasible LP Infeasible LP

Cholamandalam 1.4968 1.0079 Infeasible LP Infeasible LP

Future Generali 1.5185 1.1066 Infeasible LP Infeasible LP

HDFC Ergo 1.1687 1.0524 Infeasible LP Infeasible LP

ICICI Lombard 0.9703 0.9258 0.8474 0.9593

IFFCO Tokio 0.9547 0.9986 0.9109 0.9414

Reliance 1.2683 1.0000 0.8185 0.8570

Royal Sundaram 1.3506 0.9377 0.6926 0.9788

Shri Ram General 0.9025 1.0000 Infeasible LP Infeasible LP

Tata AIG 1.4080 1.0000 Infeasible LP Infeasible LP

National 1.1677 0.8270 0.8402 0.8442

New India 0.9194 0.7855 0.9873 0.6854

Oriental 0.9221 0.8821 0.7434 0.9224

United 0.9097 0.8296 0.8430 0.8553



46



Bajaj Allianz 1.2798 1.3895 1.0259

Bharti AXA 1.4066 3.6739 0.3603

Cholamandalam 1.3277 1.7339 0.7666

Future Generali 1.2879 2.1445 0.5691

HDFC Ergo 1.1201 1.4238 0.7772

ICICI Lombard 1.0087 1.1019 0.9892

IFFCO Tokio 1.0829 1.2018 1.0164

Reliance 1.3276 1.5888 0.9051

Royal Sundaram 1.2176 1.5248 0.8732

Shri Ram General 0.2277 1.2317 0.1905

Tata AIG 1.2955 1.6564 0.7735

National 1.0672 1.3604 0.8797

New India 1.0615 1.2304 1.0295

Oriental 0.8296 1.0666 0.8483

United 0.9368 1.1157 0.9453



47




Bajaj Allianz 1.4253 1.1701 1.0088 1.0623

Bharti AXA 1.8384 2.4302 Infeasible LP Infeasible LP

Cholamandalam 1.7342 1.2010 Infeasible LP Infeasible LP

Future Generali 2.3082 1.2932 Infeasible LP Infeasible LP

HDFC Ergo 1.3577 1.2320 Infeasible LP Infeasible LP

ICICI Lombard 1.1598 1.1213 0.9584 1.0982

IFFCO Tokio 1.1421 1.1109 1.0029 1.0308

Reliance 1.5826 1.2889 0.9293 1.0009

Royal Sundaram 1.5708 1.0794 0.8335 1.1338

Shri Ram General 1.2317 1.0000 Infeasible LP Infeasible LP

Tata AIG 1.5970 1.3314 NA NA

National 1.5841 1.2163 0.8925 1.1425

New India 1.2792 1.4015 1.1955 1.1465

Oriental 1.1383 1.2693 0.8024 1.1358

United 1.2534 1.2149 0.9247 1.1442



48

Impact of Privatisation on the Performanceof the Divested Firms : Appraisal of

Empirical StudiesJugal Kishore Mohapatra*

Privatisation of State-owned enterprises (SOEs) has evolved as a major tool of economicpolicy since 1980s across the globe – both in the developed as well as developing countries.This trend has continued even during the more recent period with record levels of revenuesraised from privatisation during 2015 and 2016. To a large extent, governments committedto diverse political ideologies have taken recourse to privatisation with the expectation thatchange of ownership of these enterprises would, inter-alia, bring about significantimprovement in their financial and operating performance, and raise their productivityand efficiency. Has privatisation indeed led to substantive and significant improvement inthe performance of the divested SOEs? An attempt has been made in this survey to reviewand synthesise the evidence provided by forty-six seminal empirical studies which haveinvestigated the impact of privatisation on the performance of the divested enterprises usingunivariate and multivariate analysis, parametric and non-parametric methods. Based onthe appraisal of these studies, key methodological issues and takeaways have been summarised.One of the significant learning's from these studies is that privatisation per se does notautomatically lead to improvement in the performance of the divested enterprises. Post-privatisation performance of these firms may be significantly contingent on contextual,institutional and organisational factors.

Keywords : Privatisation, State-owned Enterprises, Financial and OperatingPerformance, Efficiency.

1. Introduction : The ContextPrivatisation broadly “considered as anymaterial transaction by which State’sultimate ownership of corporate enti-ties is reduced”1 – has evolved as a signi-ficant and major tool of economicpolicy since 1980s across the globe-both in the developed and developingcountries- cutting across governmentsof diverse ideological hues and stripes.


Though the conservative Governmentled by the ‘Iron Lady’ – Mrs.Thatcher– in UK is commonly reckoned as the

* Jugal Kishore Mohapatra, IAS (Retd), FormerSecretary (Rural Development), Governmentof India, New Delhi.

Currently a doctoral scholar in the XavierUniversity, Bhubaneswar (XUB). This paper is basedon his doctoral research. The author is grateful tothe anonymous referee for valuable comments andsuggestions.

Impact of Privatisation on the Performance of the Divested Firms :Appraisal of Empirical Studies

49

torch-bearer of this trend, it should benoted that the first ideologically drivenlarge-scale ‘de-nationalisation’ programmewas indeed launched by KonradAdenauer’s government (Megginsonet al., 1994) in the erstwhile FederalRepublic of Germany (FRG) in the early1960s – kick started by the sale ofmajority stake in Volkswagen througha public share issues in 1961, followedby an even larger secondary share issuefor VEBA in 1965. However, admittedlythe ‘tipping-point’ tilting the politicaland ideological landscape in favourof privatisation was triggered byMrs.Thatcher’s programme of divesti-ture of the British State-Owned Enter-prises (SOEs) during the decade of1980s. Her privatisation programmewas not only viewed as audacious interms of size at that point in time; itwas also considered as bold in terms ofscope as well – covering diverse sectorsof the economy including the core sec-tors and covering enterprises operatingunder both competition and monopo-listic regimes. During the period of herpremiership, over 50 companies werereportedly privatised-including many inthe power and water industries – whichgenerated more that £50 billion for theExchequer2.

Following the path charted by Thatcher’sBritain, the next major country to embracea sizeable privatisation programme wasFrance under Jacques Chirac. The French

programme is also remarkable since itconstituted a structural break in thecountry’s long standing ‘dirigiste’ tra-dition of pervasive state intervention.Over a brief period of 15 months during1986-87, the Chirac governmentdivested 22 large companies worth $12billion (MNR, 1994). After 1987,adoption of privatisation programmesgained momentum and traction acrossthe world, particularly in the developingeconomies of South America, Africaand Asia.

This global trend of divestitures of SOEshas continued even during the first andsecond decade of the 21st century not-withstanding the world-wide financialcrisis in 2008. According to the Privati-sation Barometer Report 2015/16(PBR 2015/16) the total proceeds ofprivatisation revenue during 1988-2016was of the order $3.634 trillion3.

This report also brings out three dis-tinct and significant emerging trends.First, the fraction of privatisation reve-nue raised by EU governments in theworld-wide totals significantly declinedfrom its long-run average of about 37.5per cent to an all-time low of 14.1 percent. Secondly, for the first time theamount raised by governments throughprivatisation sales worldwide reached arecord level in 2015 exceeding the $300billion mark, shattering the previoushigh of $265.2 billion in 2009. Notably,


50

the global value of privatisation in 2016– $266.4 billion- is the second higheston record. These record levels of dives-titure revenues during the latest two

years is a pointer towards continuingmomentum of privatisation globally –though its geographical coverage interms of size and scope has not remained

Figure-1 : Worldwide Privatisation Revenue

Source : Privatisation Barometer Report 2015/2016.

350

300

250

200

150

100

50

0

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

World (Billion US$)123123 EU25 (Billion US$)

123123

121212121211

111111

11111212

121212121212

12121212

12121212121212

1212121212121212

12121212

12121212121212121212

11111

1111111111

111111

111111111111

1111111

11111111111111111111

1212121212121212

1212121212121212121212121212121212

1212121212121212

1212121212121212121212121212121212

12121212121212121212

12121212121212121212121212121212121212121212

121212121212121212

121212121212

12121212

121212121212121212

111

111111

1111

111111111111

111111111

11111111111111111

1212121212121212121212

1212121212121212121212121212

12121212121212

1212121212121212121212121212121212

12121212121212

1212121212121212121212121212

12121212121212121212

11111111111111111111111111111111

1111111

11111111111111111111111111

111111

111111111111

1111

11111111111111111111111

1212121212

1212121212121212121212121212121212121212121212

121212121212121212

1212121212121212121212121212121212121212121212121212

12121212121212121212

1212121212121212121212121212121212121212121212121212121212121212121212121212

1212121212121212121212

1212121212121212121212121212121212121212121212121212121212121212

11111

Figure-2 : Worldwide Privatisation Revenue (Billion US$)


1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

350

300

250

200

150

100

50

0


51

unaltered. Third, China has emerged asthe leading privatising country duringboth the years in 2015 and 2016. Chinaexecuted 297 sales of at least $50 million– 45 of which raised $1billions pluseach - and raised a staggering $173.2billion during 2015 and 276 sales in2016 (32 worth $1 billion+ each) raising$148.0 billion. China alone hasmobilised nearly 55 per cent of theglobal revenue from privatisationduring 2015/16.

Besides China, major economies of theOrganisation for Economic Co-opera-tion and Development (OECD) inclu-ding United States, Canada, Japan and

Australia as well as developing coun-tries like India and Malaysia have alsoresorted to privatisation on a substan-tial scale during 2015/16. Ranking ofthe top ten countries in terms of thescale of privatisation during these twoyears is shown in Table-1.

As regards the future outlook of dives-titure in 2017 and later years, the PBR2015/16 notes that seven nationalprogrammes – China, Australia, Russia,Turkey, India, Pakistan and Japan – standout either in terms of aggregate size,scope or both. Closer home, in India,emphatic announcement by the UnionFinance Minister in her speech presenting

Table-1 : Ranking of Countries by Total Privatisation Revenues 2015 and 2016

Rank 2015 CountryNo. of Value

2016 CountryNo. of Value

Deals ($million) Deals ($ million)

1. China 297 173,303 China 276 148,047

2. United Kingdom 13 34,779 Australia 5 25,705

3. Italy 11 12,383 France 9 9,596

4. Japan 3 11,947 India 35 7,393

5. India 34 11,358 Netherlands 4 7,099

6. Sweden 6 9,114 Malaysia 11 5,330

7. Australia 5 8,590 Italy 3 4,878

8. United States 6 8,230 Canada 3 4,271

9. Netherlands 3 6,208 Japan 2 4,145

10. Ireland 6 5,712 Greece 4 2,72

2015 World Total 468 319,895 2016 World Total 434 266,389



52

the Union Budget for the financial year(FY) 2019-20 that “strategic disinvest-ment of select CPSEs would continueto remain a priority of this government”and “in view of the current macro-eco-nomic parameters, government wouldnot only reinitiate the process of stra-tegic disinvestment of Air India, butwould offer more CPSEs for strategicparticipation by the private sector” hassignalled a stronger intent of thegovernment to mount a much bolderprivatisation programme than witnessedover the preceding decade4. All thesetrends suggest that privatisation hasneither lost its stream nor its relevanceand acceptability as a public policy option.

Why have governments committed tovaried political ideologies display suchcontinued and sustained zeal forprivatisation? What are the principalmotivating factors underlying this phe-nomenon? While privatisation revenues– raised through sale of assets and sharesof SOEs – does alleviate the fiscal deficitof the government and reduce thepublic sector borrowing requirements,arguably the overarching goal of almostall privatisation programme has beenthe desire to improve the efficiency ofthese enterprises. Has privatisation,indeed, led to substantive improvementin the performance of the divestedSOEs? Does it lead one-off performanceimprovement or the resultant efficiencygains are sustained over time? Does the

extent of efficiency improvementdepend on the extent of privatisation,i.e. majority or partial stake sale? Doesthe impact of privatisation on the per-formance of the divested SOEs dependon business cycles, external macro-economic factors and other exogenouspolicy changes? What happens toemployment in the privatised SOEs?

Though there is a fairly large repositoryof firm-level empirical studies whichhave explored the issues raised above,their findings are not unequivocal andfound to be at variance and inconsistenteven in respect of some focal thematicissues. Only a handful of surveys haveattempted to critically analyse and syn-thesize findings of these studies to gleana broad measure of consensus on somemore generic and contextual conclusionsrelating to the effects of privatisationon the performance of the divested SOEs.Megginson and Netter (2001), the firstmajor survey in this research area andarguably the most widely cited too,largely covers empirical studies ofprivatisation in the developed andmiddle income countries during theperiod 1980-2000. This is, of course,understandable and unexceptionablegiven the fact that few studies relatingto the developing countries were avai-lable at that point of time. Guriev andMegginson (2007) is more inclusive interms of its coverage of the less deve-loped countries, particularly of Latin


53

America and transition economies; buteven their survey is restricted to theanalysis of evidence from enterprise levelstudies up to 2002-03. Estrin et al.(2009) assesses findings of studiesrelating exclusively to the transitioneconomies of Central Europe (CEE),Common wealth of Independent States(CIS) and China covering the period1989-2006. Estrin and Pelletier (2018),evaluates more recent firm level studiesof privatisation mainly in the deve-loping countries, though it is restrictedin respect of its sectoral focus havingcovered only the regulated sectorsof banking, telecommunications andutilities.

The present survey seeks to build onthe evidence and findings of the previoussurveys and contribute to the researchin this area in the following respects.First of all, it has attempted a morerepresentative and comprehensive cove-rage of the empirical studies relating tothe developed, developing and transitioneconomies. Secondly, in terms of timeperiod of coverage, studies from themid-1990s up to more recent years havebeen included in this survey. Thirdly,methodological issues relating to the firmlevel empirical studies of privatisationhave been discussed more exhaustivelyin this paper including a criticalappraisal of the different methodologicalapproaches used in the studies.

This paper seeks to present and criticallyappraise salient findings of a set of forty-sixempirical studies which have researched–relying on credible empirical evidenceand using dependable methodology –how ownership change, throughprivatisation, affects “performance” of thedivested enterprises.Remainder of thepaper is organised as follows. Section-2contains a brief discussion of the alter-nate theories of privatisation which seekto explain the superior efficiency ofprivate/privatised enterprises vis-à-vispublic ownership. Section-3 describesthe typology of empirical studies on theimpact of privatisation on the perfor-mance of the divested enterprises whichhave been covered in this survey. It alsocontains salient findings of the studiesand a discussion of the methodologi-cal issues germane to these empiricalstudies. Section-4 summarises the keytakeaways from these studies andSection-5 concludes.

Section-2 : Privatisation : Theo-retical FoundationsSince privatisation has been pursued asa tool of economic policy by countriesacross the globe, a logical and obviousquestion that arises is whether it hasrobust theoretical foundations in main-stream economics which explain whyand how transfer of ownership of SOEsto the private sector fosters superiorproductive and allocative efficiency.


54

Theoretical studies that have examinedthe comparative efficiency of privateversus public ownership of firms canbe broadly subsumed under three dis-tinct approaches5:

1) “Agency / Property Rights Theory”

2) “Public Choice Theory”; and

3) “Organisation Theories”.

Each of these approaches, thoughdivergent in their premises and analyticalperspective, seek to provide alternativeexplanations for a commonly agreedproposition that private firms performmore efficiently that the SOEs.

The ‘Agency Theory’ identifies divorceof ownership from control as the rootcause of comparative inefficiency of theSOEs. Though it is assumed, under thistheoretical construct, that ‘managers’(the agent) of both private and publicfirms, maximise of their own utilityinstead of utility of the owners/share-holders of the firm (the principal); forthe private firms existence of the follo-wing external, market based mecha-nisms mitigate this conflict of interestand hold the managers (the agents)accountable for their performance:

i) Market for ownership rights whichprovides an exit option/window tothe owners to divest their stakes incase they are dissatisfied with themanagers’ performance;

ii) The threat and risk of bankruptcy;and

iii)A labour market for managers.

In contrast, SOEs do not have recourseto any of these performance-risk miti-gating mechanisms. Managers of theseenterprises do not have to suffer theeconomic consequences of their deci-sions as a result of which they have veryweak incentives to reduce economicwaste (overstaffing, excessive compen-sation packages and redundant or over-investment) and maximise profitability.Presence of ‘soft budget constraints’also implies that for the SOEs thebankruptcy risk virtually does not exist.Furthermore, in contrast to the privatefirms which have a ‘simple’ and directprincipal-agent relationship, SOEs haveto contend with layers of intermediateagencies intervening between “theprincipals” (i.e. the public) and “theultimate agents” (i.e. the managers ofSOEs), including elected politicians,ministers and nominated boards.Unsurprisingly, the managers (the‘bureaucrats in business’ as agents) havea significant informational advantagevis-à-vis their owners on account of thisinformation asymmetry. This may leadnot only to the problem of adverseselection of the agents; owners ofthe SOEs would also have to face theproblem of moral hazard and inhibitedfrom effective, concurrent monitoring


55

of the conduct and behaviour of themanagers. Thus, the structural andoperating environment of the SOEs donot create appropriate incentives anddisincentives to enforce accountabilityand to induce the managers to optimizetheir performance.

The main thrust of the public choicetheories is that politicians and bureau-crats/managers primarily seek tomaximise their “self-interest” instead of“public interest”, for instance, their jobsecurity, career advancement, and highercompensation packages. Their self-seeking conduct led to more expansivebudgets, excess staffing and over-supplyof public output, entailing wastage andhigher losses, thereby significantlyimpairing the enterprise efficiency. Thepoliticians tend to impose goals on theState-owned firms that may maximisetheir electoral clientele; though thesemay adversely impact the firms’ effi-ciency. For instance, this may protectand promote the interest of some stake-holders irrespective of its cost implica-tions. For the citizens, who are “theultimate owners” of the public enter-prises, the costs of monitoring thebehaviour of the “agents” may exceedthe consequential benefits (such aslower taxes or more efficient publicexpenditure). The essence of thistheoretical approach is captured anddemonstrated in a simple but elegant

model in Boycko, et.al. (1996) in whicha firm is required to choose only its levelof expenditure on labour ‘E’. It has theoption of spending an efficient amount-‘L’ or a higher amount ‘H’ (H> L). Thekey parameter in this model is who‘decides’ and ‘controls’ the level of labourspending. It is fair and reasonable toassume that in a firm that is publiclyowned and controlled, ‘E’ would bechosen by the politician. This assump-tion is realistic for the SOEs, since thegovernment exercise significant influ-ence over their major decisions, particu-larly politically sensitive ones concern-ing employment. When the politiciancontrols ‘E’, axiomatically he would setE=H, since the marginal political bene-fits additional spending on labour exceedthe marginal political costs in terms ofprofits foregone by the treasury fromsuch spending. In this way, this simplemodel demonstrates that political con-trol is likely to cause inefficiencies whichpromote the politicians’ goals to thedetriment of the public exchequer andother stakeholders.

Organisation theories draw extensivelyfrom both the above theories and high-light the organisational features of theprivate firms which are different inSOEs such as incentives and disincen-tives, systems of internal controls, cul-tural factors, organisational structures,communications / reporting systems.


56

All the three theoretical approaches,thus essentially argue that public andprivate firms are distinct in conduct andperformance because of the fundamen-tal difference in their operating envi-ronment and incentive mechanisms inthe face of information asymmetry andincomplete information.

Though a detailed critique of thesetheories, which have evolved in the con-text of the institutional land scape ofthe developed market economies, isbeyond the scope of this paper; it is per-tinent to offer some comments regard-ing their applicability and relevance tothe less developed countries and emergingmarket economies. In an influentialpaper Dharwadkar, et al. (2000) haveargued that both ‘internal and externalcontrol mechanisms’ (weak governanceand limited protection of minorityshareholders) are likely to exacerbate the‘agency problems’ in the emergingeconomies. Hence privatisation ofSOEs without addressing these risksregulatory reforms of the corporategovernance, capital market, corporatebankruptcy and mergers and acquisi-tion processes might not deliver theanticipated efficiency gains. As regardsthe public choice theory, it needs to bementioned that the public accountabilitymechanisms are likely to be far weakerin the developing countries. Theirenforcement capacity is also likely tobe constrained. Besides, SOEs in these

countries are more likely to be drivenby political mandates than the objec-tive of maximising returns to the share-holders and hence, less likely to be sub-jected to ‘hard budget constraints’. Insuch a scenario, privatisation can beexpected to improve performance of thefirms provided, of course, the privatesector itself is functioning in a well regu-lated ecosystem and not distorted bymarket failures. The approach of the‘organisation theories’, being somewhatmore generic in nature, possibly has asmuch validity tor the firms in theemerging markets as those in the deve-loped countries. To sum up, these theo-retical approaches do tend to condition-ally favour a case for divestiture even inthe developing countries contingentupon the risks flagged therein are miti-gated through appropriate institutionaland organisational reforms.

Section-3 : Privatisation andPerformance : Methodology andSalient Findings3.1 Typology of Empirical Research

3.1.1 However logical and persuasivethe theoretical studies assessing the rela-tively superior efficiency of the privateownership may be, validity andgeneralizability of such hypotheses canonly be tested through rigorous empiri-cal research. It is worth recalling anoft-cited observation of Laffont andTirole (1993) in this context who,


57

after presenting their analysis of publicand private ownership in stimulatingefficiency, came round to the view that“theory alone is thus unlikely to beconclusive in this respect”.

3.1.2 In the literature we come acrosstwo broad strands of empirical researchusing eco-matric methods which hasaddressed the question whetherprivatisation improves performance andefficiency of the divested SOEs. Thefirst set of studies examine firm levelimpact of privatisation on the perfor-mance of the divested enterprises on thebasis of empirical evidence in a multi-country, multi-sectoral framework. Theother stream of empirical researchexamines the enterprise level impact ofprivatisation on efficiency in a singlecountry and these studies are alsomostly multi-sectoral. Publication of

these studies started around mid-90ssince adequate number of privatisedfirm level observations from a cross-section of developed as well asdeveloping countries were availablefor rigorous econometric analysis andstatistical testing.

3.1.3 For the purpose of this study, wehave selected as many as 46 such studies,of which 33 are country-specific and 13are multi-country studies (including onestudy that has used case study approach6).The studies which have used econometricmethods can be further categorized intodistinct types based on the methodsused : whether they have used univariateor multivariate methods, and whether theyhave used parametric or non-parametricmethods. The following Tables-1A and1B indicate the broad classification of thestudies in terms of these filters.

Table-1A : (Typology of Empirical Studies)

Coverage Univariate Analysis Multivariate Analysis Both Total

Multi-County Studies 5 4 3 12

Single Country Studies (India) 6(1) 12(3) 15(5) 33(9)

Total 11(1) 16(3) 18(5) 45(9)

Table-1B : (Typology of Empirical Studies) Coverage Parametric Method Non-parametric Method Both Total

Multi-Country Studies 4 5 3 12

Single Country studies 15(8) 6(0) 12(1) 33(9)

Total 19(8) 11(0) 15(1) 45(9)

Note : One single country study has not used either parametric or non-parametric tests.Figures in the parentheses show the type of method used in studies relating to India.


58

3.1.4 Studies which have usedunivariate analysis largely follow themethodology pioneered by Megginsonet. al. (1994) – hereinafter “MNR(1994)” – which is based on comparisonof pre- and post-privatisation ‘averages’(computed usually for 3-5 pre-privati-sation years and equal or more numberof post-privatisation years) of selectedperformance proxies. Statistical signifi-cance of the differences in these averagesis tested by applying parametric or non-parametric tests. Studies relying on multi-variate analysis have used multivariateregression (OLS, Panel data regressionmethods -24 studies) and Data Enve-lopment Analysis (DEA-5 studies).

3.1.5 As many as 18 studies coveredin the survey have used both univariateand multivariate analysis and 15 studieshave used both parametric as well asnon-parametric methods. It may alsobe mentioned that the studies includedin the survey provide adequate repre-sentation to the privatisation programmesof both the developed and developingcountries including three large multi-

country studies covering the transitioneconomic of Eastern and CentralEurope. Since China has emerged as theleading country in terms of its scale ofprivatisation in the recent years, threecountry-specific studies of China havebeen included. As far as India isconcerned, 9 studies have been includedpurposively, since in our consideredview this has so far been a relativelyunder-researched area. An attempt hasbeen made to cover possibly all themajor papers published in the top-tierjournals which have empiricallyresearched the effects of privatisation onthe performance of the divested firms.

3.2 Salient Findings of the EmpiricalStudies3.2.1 Synopsis of the salient findingsof the multi-country and single coun-try empirical studies of privatisation,covered in this survey, have been pre-sented in respectively, along with thesample description, period analysed andmethodology used. The overall picturethat emerges from these studies is fairlymixed as indicated in Table-2.

Table-2 : Overall Findings of Empirical Studies

Overall Findings Multi-Country Studies Single Country Studies Total

Supportive 8 12 20

Not Supportive 1 6 7

Partially Supportive/Mixed/ 4 14 18Inconsistent findings

Total 13 32 45


59

It would be hard to categorically andunequivocally claim or assert, on thebasis of the conclusions derived by thesestudies, that privatisation of SOEsper se – through transfer of ownership andmanagement control, brings about sta-tistically significant efficiency improve-ments in the divested entities. That said,however, these empirical studies dobring out emphatically the need for thecomplementary institutional reformsand policy action to be pursued along-side privatisation in order to achieve thedesired efficiency outcomes.

3.2.2 Significant divergences amongthe empirical studies of privatisationsmotivated Bachiller (2017) to carry outa meta-analysis of 60 studies publishedduring 1989-2014; the sample covering48 countries for the period 1961-2010.The study examined whether themethod of privatisation and the levelof development of the country of theprivatised enterprises can explain vari-ance in the “financial performance” ofthe privatized SOEs. Results of the metaregression analysis indicated that themethod of privatisation (whether divestedthrough initial public offer- IPO- orthrough other methods) significantlydetermines the performance of theprivatised companies. Besides, the resultscontradicted the more commonly heldassumption that privatisation of SOEsin the developing countries does notimprove their financial performance.

3.2.3 Estrin and Pelletier (2018) havereviewed fifteen recent papers on theeconomic effects of privatisation on theefficiency and performance of the firmsin the developing countries. Studiesincluded in this paper have analysedevidence of the impact of privatisationin three sectors – banking, telecommu-nications and utilities- using single orcross-country samples. The overall con-clusion of this survey also reconfirmsthe proposition that transfer of owner-ship by itself is unlikely to yield theexpected efficiency gains and the impactof privatisation could well be morecontext and sector-specific contingent,inter alia, on the following factors :

• Selection effect- i.e. selection of thefirms for stake sale.

• Extent of privatisation- total or partial.Evidence from the studies includedin their survey provides evidence thatthe effect of total privatisation islikely to benefit more in terms ofefficiency gains.

• Quality and effectiveness of theregulatory framework, which islargely dependent on the prevailingpolitical and institutional setting.

• Competitiveness of the marketstructure and existence of institu-tional mechanisms to promote andenforce effective competition.


60

3.3 Methodological Issues in Empi-rical Studies of Privatisation3.3.1 Assessment of effects of privatisa-tion on enterprise performance throughempirical studies poses a complex setof methodological issues. First of all,for determination of the effect of theownership change, we need to have theright ‘control group’ comprising ofentities similar in all respects to the‘treatment group’ – i.e. the sample ofprivatised SOEs – but which were notsubjected to the policy interventionbeing investigated (which in this case isprivatisation). How do we identify themost appropriate ‘control group’?Should we take the same set of SOEs‘before’ privatisation as the ‘controlgroup’, as has been done in a numberof studies based on MNR (1994)methodology, or identify firms havingsimilar economic characteristics –matched at least in terms of relevant,major parameters – from the SOE sectoror from the private sector? In case weopt for the latter,should we identify thefirms to be included in the ‘control group’on the basis of simple and straight forwardcomparison of the descriptive statisticsor should we take recourse to moreadvanced techniques like PropensityScore Matching (PSM)? Thus, figuringout what would have happened tothe sample SOEs, had they notbeen privatised, is arguably complexand problematic. Secondly, which

performance indicators/proxies shouldbe used for evaluating the effects ofprivatisation? Should we use the con-ventional accounting measures of finan-cial and operating efficiency – whichare partial measures- or should we usemore comprehensive measures of pro-ductivity – such as ‘multi-factor pro-ductivity’ (MFP) or ‘technical effi-ciency’ (TE)? In case we opt for thepartial accounting measures, we againhave to decide which are the ‘right’ or‘most appropriate’ set of indicators orproxies. In this context, we also needto recognize that even at a conceptuallevel the objective function of the SOEsis assumed to be less profit-oriented andmore tilted towards non-profit goals;whereas enterprises under private own-ership are more likely to maximizeprofits and returns to the shareholders.Given this fundamental premise,would it be appropriate to assess theenterprise performance – “before andafter divestiture”- on the basis ofchanges in profitability? Would it notfavour privatisation abinitio if non-profit objectives, such as lower prices,or higher employment were beingdeliberately pursued under state owner-ship? Thirdly, privatisation may affect‘allocative efficiency’ and even incomedistribution besides the ‘productive ortechnical efficiency’ of the firms bychanging the relative prices both in theoutput as well as input markets. As


61

such, it may have spill over effects onthe rest of the economy which is whyassessment of its impact can be ideallycarried out in a general equilibriummodel. However, setting up GeneralEquilibrium (GE) models is enor-mously complex requiring humongouseconomy wide data on a huge set ofvariables. Unsurprisingly, therefore, fewempirical studies of privatisation haverelied on economy wide GE models7.Fourthly, how to determine the direc-tion of causality in assessing the rela-tionship between privatisation and per-formance? Just as there are studies toshow that privatisation impacts enter-prise performance; there are also stud-ies which show that better enterpriseperformance makes a firm more attrac-tive and likely candidate for divestiture.As such, can we a priori rule out ‘simul-taneous causality’ and ‘endogeneity’ ofprivatisation? If not, how do we handlethis problem in the empirical investi-gation? Fifthly, even the policyannouncement of privatisation of a SOEmay spur intense internal effort, withinthe firm, including its financial and HRrestructuring, to improve its perfor-mance. It is also not unlikely that themomentum of improved performance,generated by these efforts prior to theprivatisation, may continue even afterdivestiture. In such a scenario, do wetreat the observed improvement in

performance after privatisation as the‘privatisation effect’ or as the‘announcement effect’? In reality, thesituation may be even more complex.The observed change in the enterpriseperformance after divestiture may be acombination of these two effects. Inthat case, we have to find out whethercareful and diligent efforts have beenmade in the empirical study to decom-pose these effects.

Finally, more often than not, privati-sation is not a ‘stand-alone’ reforms.Governments pursuing privatisationprogrammes simultaneously resort toa whole range of key policy changes –particularly in the developing countries-such as de-regulation, trade liberaliza-tion, promoting competition, strength-ening corporate governance, setting upregulatory regimes for non-competitiveenterprises and freer capital movement.The observed performance change inthe divested enterprises cannot, in sucha scenario, be simply or solely attrib-uted to the ownership change. Theseexogenous, structural policy changesand business cycles are likely to affectthe enterprise performance, both inde-pendently and also by interacting withthe ownership change. How do we pre-cisely quantity the effects of privatisa-tion by factoring out the effects of theabove mentioned exogenous policychanges and macroeconomic changes?


62

3.3.2 Empirical studies reviewed inthis paper cover diverse geographies anddifferent time periods, based on a widearray of methodologies. Not surprisingly,their conclusions differ. What is moreimportant from our perspective is howrobust are the methodologies used toarrive at these conclusions. The mostwidely used univariate analysis in empi-rical studies of privatisation – MNR(1994) methodology – is vulnerable toselection bias and does not adequatelycontrol the effects of concurrent andexogenous macroeconomic and policychanges as well as institutional deve-lopments. In addition, this type ofanalysis does not capture the trend ofperformance improvement which isoften observed after the announcementof privatization, even before its actualexecution. Treating the sample SOEsbefore privatisation as the ‘controlgroup’, without sufficiently controllingfor these factors, might not, therefore,yield unbiased estimates of the effi-ciency effects of privatisation. Some,studies, based on this metho-dology,have, however, sought to mitigate thesedeficiencies by using ‘adjusted perfor-mance measures’ – benchmarking theraw performance measures to the mar-ket or industry averages or indices. Forexample, such an approach has beenused in Boubakri and Cosset (1998),Garcia and Anson (2007) and Aussenggand Jelic (2007). Notably, results based

on the adjusted performance measuresin these studies were found to be lesssignificant than those based on theunadjusted measures. Wei et al. (2003)and La Porta and Lopez-de-Silanes(1999), Omran (2004), Jerome (2008)and Jiang et al. (2009) have followed asomewhat different approach to addressthese problems- by using the firms inthe sample (i.e. the divested SOEs) asthe “treatment group” and comparedtheir performance with a matched“control group” of SOEs which werecomparable in terms of size, performanceand operating in the same industry butnot divested. Notwithstanding theseimprovisations, however, findings ofstudies solely based on the MNR (1994)methodology require further robustnesschecks and validation, preferably, bytaking recourse to a different approach,as has been done in many studies includedin this survey (which have used bothunivariate analysis and multivariateregression analysis).

3.3.3 As regards studies which haveused multivariate regressions – in mostcases advanced panel data techniques –the key issue relates to the appropriate-ness of the model used to address se-lection bias, endogeneity of privatisa-tion, simultaneous causality and thepossibility of ‘omitted variables’. Choiceof the dependent variables and indepen-dent variables for the model, includingthe control variables, also require careful


63

consideration; though in this respectmost studies, which have used thesemethods, have largely relied on priorstudies on the subject. In general, mostof these studies have also subjected theirresults to a variety of robustness checks.Notably, the multivariate regressionmodels provide flexibility to determinethe impact of a variety of factors –“economic, political, organizational andinstitutional” – on the post divestitureperformance of the firms. Besides, onecan capture, with some degree of pre-cision, the “dynamic effects” of privati-sation – the “announcement effects”,the “short-run effects” and the “long-runeffects”. Thus, we can assess whetherprivatisation brings about ‘one-off ’or ‘sustained’ improvement in theenterprise performance.

3.3.4 The third set of studies hasassessed the effects of privatisation byestimating productive or technicalefficiency of the firms using DataEnvelopment Analysis (DEA) orStochastic Frontier Analysis (SFA).Some of these studies have also used atwo-stage approach – combining DEAwith Tobit regression to analyse factorsimpacting changes in the TE of thedivested enterprises. Productive efficiencyof a firm can be estimated using twoalternative methods : parametric SFAand non-parametric DEA. Both methodshave their pros and cons. The keystrength of SFA is that it incorporates

a stochastic element in the productionprocess. Accordingly, the estimatesfrom SFA can be subjected to the con-ventional parametric hypothesis testing.On the flipside, however, SFA requiresimposition of an explicit functionalform of the production function andalso assumptions with regard to the ran-dom error term. As such, results of theSFA may be sensitive to the parametricfunctional form used in the model. Incontrast, prior specification of the formof the production function is notrequired for DEA. Hence, its sensitiv-ity to misspecification is less likely. Noassumption regarding the distributionof the error term is also necessary. Itsmain disadvantage, however, emanatesfrom its deterministic approach. Unlikestochastic frontier models, DEA assumesabsence of ‘random noise’ in the dataand attributes deviations from the esti-mated efficiency frontier to inefficiency.Thus, the estimates of DEA may be proneto errors in measurement and noise inthe data. However, as discussed in thestudy of Arocena and Oliveros (2012),the deterministic nature of DEA can beovercome by taking recourse to boot-strapping methods to generate confi-dence intervals for the TE estimates andaccordingly, subject these estimates tohypothesis testing.

Studies using two stage DEA approach –combining DEA with Tobit regression– have to address an additional


64

methodological problem arising out ofserial correlation among the DEA esti-mates. Arocena and Oliveros (2012)have demonstrated how bias-correctedand consistent efficiency estimates canbe obtained by application of boot-strapping techniques in the second stageanalysis.

Having due regard to the relative meritsand demerits of both the methods,however, the choice between SFA andDEA primarily depends on the diag-nostic tests which indicate whether thedata set has significant noise or not.

3.3.5 The major strength of methodo-logy used by Galal et al. (1994) liesin the manner in which a “counterfactual scenario” is constructed, whichenables estimation of both financial andwelfare gains from privatisation. Not-withstanding rigour and robustness ofits methodology, however, few subse-quent studies have used it, primarilybecause mapping out the ‘counterfactual’ and its long range projectionnecessitate a number of assumptions tobe made about the structure and futuretrends in the economy over a longperiod, including the future scenario inthe industries and the sectors in whichthe divested SOEs operate, which isinherently fraught with complexitiesand challenges including the risk offorecasts going way off the mark.

3.3.6 In an oft-cited paper, Rodrik(2005) had pointed out some signifi-cant limitations of cross-country empi-rical studies. Cross-country empiricalstudies of privatisation also have to con-tend with a major problem emanatingfrom data comparability and consis-tency. Accounting standards and thequality of financial reporting varieswidely across countries. In particular,quality and reliability of accountingreports of the relatively less developedcountries and also of countries in transi-tion would have to be taken care of insuch studies. This could also be anothersource of selection bias in the cross-country studies as pointed out inMegginson and Netter (2001). Sincedeveloped countries usually had betteravailability and reliability of data, moreso for the firms performing better inthese countries, better performing firmsof the developed counties were possiblydisproportionately represented in theearly cross-country empirical studiesof privatisation, somewhat bluntingtheir external validity vis-à-vis the lessdeveloped countries. Single countrystudies are, however, less susceptible theproblems arising on account of datacomparability.

Section-4 : Empirical Studies ofPrivatisation- Key TakeawayKeeping in view the advantages andlimitations of the alternate methods and


65

other methodological issues, discussedin Sub-section 3.3, some of the salienttakeaway from the studies covered inthis survey are briefly discussed below.

4.1 Firstly, does privatisation per seautomatically lead to improvements inthe enterprise performance? This,indeed, is a cross-cutting research ques-tion addressed in several studies. Logi-cally, one should expect strongest empi-rical support for privatisation in themore developed countries from whichthe contemporary wave of privatisationactually originated. Contrarily, how-ever, in both the studies relating to UK(Martin & Parker, 1995; Boussofianeet el 1997), in three of the four studiesrelating to Spain (Villalonga, 2000;Garcia & Anson 2007; and Bachiller,2009), in Alexandre and Charreaux(2004)’s study of France and inFraquelli and Erbetta (1999)’s study ofItaly we do not find statistically signifi-cant and robust evidence of efficiencygains from divestiture. As such, evenin some of the more favourable insti-tutional ecosystems of the developedcountries, with admittedly bettercorporate governance and regulatoryinstitutions, dilution of public owner-ship in the SOEs alone does not seemto be sufficient for bringing about theexpected performance improvements.

4.2 Secondly, why does privatisationby itself does not boost the efficiency

of the divested enterprises? This ques-tion – which logically flows from thefirst one raised in the preceding sub-para- has been fairly well researched inVillalonga (2000), Wu (2007) andArcas and Bachiller (2010). Findings ofthese studies emphatically show thatpost-privatisation performance of thefirms is significantly contingent oncomplementary contextual, institutionaland organizational factors. Amongother things, competition policy, appro-priate regulatory regimes for monopo-listic / oligopolistic industries, sectoralderegulation, trade liberalization, deve-lopment of capital markets, corporategovernance norms and mechanisms andbusiness cycles are some of the factorswhich may explain the variance in theobserved enterprise-level performanceoutcomes in the post-privatisation period.

4.3 Thirdly, does the assessment ofefficiency gains from privatisation dependon the choice of performance measuresused? Evidence gleaned from Dewenterand Malatesta (2001) suggests that itdoes. In that study two alternate mea-sures of profitability – net income-based measures and EBIT (earningsbefore interest and taxes) based mea-sures were used both in the univariateanalysis and in the multivariate regres-sion. In both cases, profitability wasfound to have significantly improvedin the post-privatisation period in termsof the net income-based measures; but


66

not when EBIT- based measures wereused. A similar inconsistency is alsofound in Chibber and Gupta (2017 a).Regression equations estimated in thatstudy to determine the impact ofMemorandum of Understanding(MoU) on the profitability of the SOEsindicated its significantly positiveimpact when Return on Capital (RoC)was used as the proxy; but not whenReturn on Assets (RoA) was used. Thequestion of appropriate choice of theperformance measures has also beendealt in Frydmanetal (1999)’s study ofprivatisation in the transition economicsof Central Europe. In the initial stagesof post-communist transition-which isthe period analysed in that study – theaccounting systems were still in a stateof flux; disclosure standards and mecha-nisms were very imperfect and therewere no dependable measures of the“cost of capital’. In view of these diffi-culties, the researchers considered profitsas an unreliable measure of perfor-mance for the purpose of their study.Accordingly, instead of profitability, thestudy focused on its two components– revenues and costs for which morereliable, comparable data of consistentquality were available. Thus, findingsof the empirical studies may well besensitive to the choice of the performanceindicators/proxies used.

4.4 Fourthly, does the extent ofdivestiture – i.e. whether ‘revenue’ or

‘control’ privatisation8 matter? Findingsof the three seminal cross-country studiesMNR (1994), B&C (1998) and D&M(1999)- indicate that ‘control’ privati-sations bring about significantly greaterperformance improvements than‘revenue’ privatisations. Multi-countrystudy of Aussenegg and Jelic (2007) alsofound that higher government share-holding after privatisation was associa-ted with significantly lower efficiency9.However, Naceur et al. (2007)’s study ofMiddle East and North African (MENA)countries indicated somewhat mixedresults in this regard – though subsamples of control privatisationsperformed significantly better in termsof operating efficiency and output;significant increase in profitabilitywas observed only for the subsampleof revenue privatisations, but not forcontrol privatisations.

Results from the country-specific studieson this issue are also mixed. Studies ofProcianoy and Sobrinho (2001), Weiet al. (2003) confirm that firms inwhich majority stake was transferredexperienced significantly higher effi-ciency gains. A contrarian result was,however, found in Jiang et al. (2009)’sstudy of China in which differencebetween the subsamples of control andrevenue privatisations was statisticallynot significant10. Wu (2007)’s study ofTaiwan also indicated that the extentof government’s shareholding after


67

privatization had no significant effecton performance.

Thus, while empirical evidence generallyvalidates stronger impact of controlprivatisation on enterprise performancein the developed countries; in case of theless developed countries the findingsappear to be mixed and inconsistent.

4.5 Fifthly, does the time-periodcovered by the study matter? Timingof privatisation of a SOE can have sig-nificant effects in case the transactionis carried out during a business cycle.Its short-run performance after dives-titure would depend on whether thechange of ownership was effected duringa recessionary phase or a boom phase.Also, sometimes there is a considerabletime-lag between the announcement ofgovernment’s intent to privatise a parti-cular SOE and its actual execution (incase of British Airways the time lag wasas long as 6 years). Furthermore, themanagement of the firms, after privati-sation, require sufficient time to carryout organizational, business portfolioand HR restructuring, to provide newstrategic orientation as well as to makenecessary investment to achieve improve-ment in productivity and efficiency. Forthese reasons, longitudinal coverage ofthe empirical studies of privatisationshould be sufficient both in respect ofthe pre-privatisation and the post-privatisation periods so as to control the

business cycle effects, capture its ‘tem-poral effects’ – the ‘announcementeffect’ and its short-term and long-termeffects. Besides, it should be longenough to assess if the efficiency gainsfrom privatisation are sustained ortapper off over time. If the time horizonselected by the study is too short, effi-ciency effects of privatisation may notbe observed. Villa longa (2000)’s studyempathically brings out this point. Forthe sample of Spanish firms consideredin that study, the positive effects ofprivatisation were found to be statisti-cally significantly only 7-8 years afterthe event. Alexandre and Charreaux(2001)’s study of French privatisationalsocorroborated this finding. Besides, inGarcia and Anson (2007)’s study sig-nificant improvement in performancewas found in the long-run; but not inthe short-run.

4.6 Sixthly, does performance improve-ment precede the actual event of dives-titure in case of the privatised SOEs?Strong evidence of the ‘announcementeffect’ was found in two major studies –Martin and Parker (1995), Boussofianeet al. (1997) and Dewenter and Malatesta(2001). Significant evidence of suchperformance improvement – ‘prepara-tion / announcement effect’ – prior toactual divestiture has also been foundin the studies of Ghosh (2008),Chibber and Gupta (2017 & 2019) andGunasekar and Sarkar (2014) relating


68

to India. Interestingly, Dewenter andMalatesta (2001) actually went on tofurther probe whether the performanceimprovement observed in the periodjust before privatisation represented realefficiency gains or ‘creative accounting’or ‘earnings management’11 by theGovernment. Examining long-runmarket-adjusted stock returns of theprivatised firms, the study confirmedthat the improvement in enterpriseperformance observed during theperiod before privatisation represented‘real’ and substantive and not dueto systematic manipulation of theirfinancial reports.

Keeping in view the findings of the studiescited above, it would be instructive toprobe significance of the ‘preparation/announcement effect’ in firm levelempirical studies of privatisation.

4.7 Seventhly, how sensitive is theassessment of privatisation to the methodused? We examine this issue with refer-ence to the studies of Spanish privatisa-tion included in this review. In Garciaand Anson (2007)’s study – whichconsidered a sample of SOEs privatisedduring 1985-2000, performance offirms before and after privatisation wasnot found to be significantly differentwhen industry adjusted averages wereused and the standard MNR (1994)method was applied over the ‘three yearsbefore and three years after’ (t-3 to t+3)

window. In contrast, performanceimprovement after divestiture was foundto be significant for the sample whenthe same method was applied to theunadjusted averages of the proxies. InVillalonga (2000)’s longitudinal study-which used data panel data regressionmodels-changes in both the level andgrowth rate of efficiency after privatisa-tion was found to be statistically insigni-ficant. Accordingly, the hypothesis thatprivatisation increases firm efficiencywas rejected. However, Arocena andOlivera’s (2012) study which examinedthe technical efficiency of a sample ofSOEs, privatised during 1994-2002,vis-à-vis their closest private sectorcompetitors, applying two stage DEA,found significant increase in the effi-ciency of the SOEs after privatisation;while no significant improvement wasobserved in case of their private com-petitors during the same post-privatisa-tion period. Accordingly, their studyfound some support for the claim thatefficiency of the SOEs gets significantlyenhanced after divestiture. Thus, con-clusions of the empirical studies ofprivatisation may substantially vary,even with respect to the SOEs privatisedmore or less during the same period inthe same country, depending on themethod used in the analysis.

4.8 Eighthly, what happens to employ-ment and the workers in the privatisedSOEs? Unarguably, perhaps, this is


69

the most politically sensitive dimensionof privatisation as a public policy tool.Evidence with regard to employmentin the SOEs after privatisation is quitemixed. In the earliest empirical studyof privatisation by Galal et al. (1992),workers were found to have gained inten out of the twelve cases analysed –either from appreciation of sharesoffered to them or from higher wages.Findings of the three seminal cross-country studies, however, differ in thisrespect. While MNR (1994) andBoubakri and Cosset (1998) found sig-nificant increase in employment,D’Souza and Megginson (1999) foundsignificant decline in employment afterdivestiture. Evidence from some othermulti-country studies a real so mixed.For example, Aussenegg and Jelic (2007),Cook and Uchida (2004) indicate sig-nificant decline in employment afterprivatisation; whereas Mathur andBanchuenvijit (2007) found no signifi-cant change either for the subsample ofthe developed countries or of the emerg-ing markets. Claessens and Djankov(2000) had observed lower down-sizingof labour in the firms which had beenprivatised for more than 3 years as com-pared to the state owned firms. On theother hand, Frydman et al. (1999)found a contrarian result that privatisa-tion had actually led to improvementin the employment performance of thedivested firms.

Megginson and Netter (2001)’s surveyhad also found singular lack of unanimityof the empirical studies of privatisationregarding its impact on employmentlevels in the divested SOEs. Out of 10studies reviewed in that survey, threehad documented significant increases,in two studies the observed change wasinsignificant; whereas the remainingfive studies had documented significantdecline in employment.

Findings of the single country studiesalso appear to be conflicting and incon-sistent in this regard. Bhaskar and Khan(1995)’s study of privatisation of jutemills in Bangladesh is possibly one ofthe most insightful contribution in thisregard. Results from its ‘difference-in-difference’ (DID) estimation had foundcompelling evidence of significantdecline in “white collar employment”(“clerical and managerial”) and “perma-nent manual workers”; whereas emp-loyment of casual / temporary manualworkers was found to have increasedsignificantly. La Porta and Lopez-de-Silanes (1999) had found significantdecline in the industry-adjusted measureof total employment as well as employ-ment of both white and blue-collarworkers. Okten and Arin (2006)’s studyof Turkey also provides evidence of sig-nificant decline in employment. In con-trast, Wei et al. (2003)’s study of Chinaand Garcia and Anson (2007)’s studyof Spain document no significant change


70

in employment. Similarly, estimates ofa computable general equilibrium(CGE) model used by Chisari et.al.(1999) indicate that privatisation didnot seem to be a major contributor ofthe significant increase in unemploymentin Argentina during 1993-95.

Thus, evidence from the empirical studiescovered in this survey with regard tothe impact of privatisation on emp-loyment is fairly mixed. This could bebecause of the interplay of countervailingforces which emanate from labourforce restructuring and rationalisationthat usually follows privatisation. As aresult of this, overall employmentmight expectedly decline in the initialyears. But if the new management suc-ceeds in the initial turn-around efforts,it might invest and expand output oreven diversify, creating more employ-ment opportunities. Thus, after a timelag, overall employment might berestored to the pre-divestiture level oreven higher.

A substantive point, however, thatemerges from these apparently incon-sistent and inconclusive findings is thatwhile analysing the impact of privatisa-tion on employment, the approachneeds to be more granular. Instead ofsimply comparing the aggregate emp-loyment levels before and after divesti-ture; it may be useful also to examinechanges in its composition (i.e. mix of

permanent and casual/temporaryworkers, white and blue collar workers).

Section-5 : Concluding Obser-vationsTo conclude, evidence from empiricalstudies of privatisation, covering deve-loped as well as developing countries,tend to favour a more nuanced view ofits efficiency effects. Nevertheless, thesestudies provide valuable guidance fordesigning future privatisation pro-grammes in countries of South Asia andAfrica, where SOEs still have a signifi-cant economic presence, in particularabout what to privatise (i.e. selectionof the sectors and enterprises), how toprivatise (i.e. the process and methodof divestiture), when to privatise (i.e.timing and sequencing of stake sale) andabove all, about the complementaryreforms to be undertaken to achieve thedesired goals. Given the high politicalsensitivity of privatisation programmesin liberal, open democracies, evidencefrom empirical research is likely tocontribute, in no small measure, to thepublic policy in this field.

Having said that, this survey also bringsout the ‘grey areas’ relating to empiricalprivatisation studies which merit furtherresearch. First, impact of privatisationon employment, particularly in thecontext of labour surplus developingcountries, needs to be explored further


71

in the light of mixed evidence providedby the studies covered in this survey.Does it result in an overall reduction inemployment in the firm or lead to more‘casualisation’ and recourse to contractingand outsourcing? Does it worsen labourstandards and conditions of employ-ment? More evidence based research onthese issues is likely to generate valuableinsights and inputs for providing app-ropriate safeguards in the privatisationprogrammes in the developing coun-tries, which would not only be percei-ved ‘fair’ in terms of its welfare conse-quences; but also likely to elicit stron-ger political ‘buy-in’.

Secondly, more empirical studies are alsoneeded on the distributional impact ofthe privatisation programmes in thedeveloping countries, particularly of theutilities and infrastructure sectors. Doesprivatisation of firms in these sectorshave adverse effects on the welfare ofthe consumers in terms of higher pricesand service quality, even though it mighthave had positive impact on theirfinancial and operating performance?Does it affect equitable access to ser-vices? Does privatisation yield ‘positive-sum outcomes’ for the major stake-holders? Again robust empirical evi-dence on these contentious and politi-cally sensitive issues would contributesubstantially towards our understand-ing of the likely equity-efficiency trade-offs of the divestiture programmes.

Third, there is a need for more casestudies, of both ‘successful’ and ‘failed’privatisation, in order to deepen ourunderstanding of when privatisation‘works’ and when it does not; where(sectors and industries) it ‘works’ andwhere it does not; and why and how it‘works’ in some cases and not in othercases. Also, does privatisation workmore effectively in the developingcountries in sectors and industries thatare less “institutions-intensive” as com-pared to sectors where existence ofstrong and empowered autonomousregulatory institutions is critical? Whileeconometric studies produce evidencethat enables derivation of more gene-ralizable findings; they often tend tomiss out rich institutional details whichemerge from in depth case studies. Forinstance, why privatisation of HindustanZinc- a Central Public Sector Enterprise(CPSE) of India privatised in 2002-03is acknowledged to be a ‘success’ andwhy sale of Jessop & Co. another CPSEprivatised around the same time in thesame economic environment- did not‘succeed’? Case study approach isarguably well suited to answer andexplain many such issues which havecritical policy implications.

Further research to explore these issuesis not merely of academic interest; ithas more substantive value from the


72

public policy perspective to enlightenresponsible policy makers more holis-tically not only about the efficiencyeffects of privatisation; but also aboutits equity and welfare consequences.

Notes :1. Privatisation Barometer, an agency which

provides privatisation data to OECD andWorld Bank, defines privatisation as “atransfer of ownership or voting rights fromthe ‘state’ to the private sector”.

2. As reported in in The Telegraph, April 8th,2013. Also see Edwards (2017) onMrs. Thatcher’s privatisation legacy.

3. Megginson (2016) in The PBR 2015/16.

4. Extracted from Para 99 of the speech of theUnion Finance Minister presenting UnionBudget for FY 2019-20. Para 97, 98 and100 of the speech also reaffirm the renewedpolicy thrust of the Government of Indiato scale up strategic disinvestment of theCentral Public Sector Enterprises (CPSEs).

5. For “Agency/Property Rights Theory”please refer to Alchian (1965), De Alessi(1980), Shapiro and Willig (1990) andLaffont and Tirole (1993). For “PublicChoice Theory” see Zeckhouser and Horn(1989) and Haskel and Szymanski (1992).For “Organisation Theory” see Perry andRinsey (1988) and Walker and Vasconcellos(1997).

6. Galal et al. (1994) is the only studycovered in this survey which has used casestudy method.

7. Chisari et al. (1999) have used a ComputableGeneral Equilibrium (CGE) model to esti-mate the macroeconomic and distributionaleffects of privatisation of privatisation ofutilities in Argentina.

8. Megginson et. al.(1994) distinguish bet-ween ‘revenue’ and ‘control’ privatisation.‘Control’ privatisation refers to sale sharesby the Government of a SOE to privateentities which lowers its stake below 50 percent thereby ceding control of management.‘Revenue’ privatisation refers to cases wheregovernment ‘simply sold a minority stake toprivate investors” primarily to raise revenuewithout having to cede control rights.

9. Results of Boardman and Vining (1989)similarly suggest that partial privatisationmay not be the optimal strategyfor governments which wish to reducedependence on SOEs.

10. As explained in the study of Jiang et. al.(2009), however, even in the firms of theirsample in which government diluted itsownership below 50 per cent, it stillremained as the largest and ‘controlling’shareholder. Thus, in their sample there was,in effect, no difference between the ‘revenue’and ‘control’ privatisation in terms ofgovernment’s control rights in the divestedfirms, which might be a probable explana-tion of why the extent of divestiture didnot make a difference in their study.

11. The study examined whether governmentmight have resorted to manipulation of thefinancial reports of the firms selected forprivatisation – inflating their earnings beforeprivatisation systematically in order tomislead the prospective investors and makethe firms ‘attractive’ for sale.


73

ReferencesAlchian, A.A. (1965) Some Economics of

Property Rights, IlPolitico, 30(4), 816-829.

Alexandre, H. & Charreaux, G. (2004) Effi-ciency of French Privatizations : A DynamicVision, Journal of Corporate Finance, 10(3),467-494.

Alipour, M. (2013) Has Privatization of State-Owned Enterprises in Iran Led to ImprovedPerformance? International Journal ofCommerce and Management, 23(4), 281-305.

Arcas, M.J. & Bachiller, P. (2010) OperatingPerformance of Privatized Firms in Europe :Organizational and Environmental Factors,International Journal of Public Administration,33(10), 487-498.

Arocena, P. & Oliveros, D. (2012) The Effi-ciency of State-Owned and Privatized Firms :Does Ownership Make a Difference?International Journal of Production Economics,140(1), 457-465.

Aussenegg, W. & Jelic, R. (2007) The Opera-ting Performance of Newly Privatised Firmsin Central European Transition Economies,European Financial Management, 13(5),853-879.

Bachiller, P. (2009) Effect of Ownershipon Efficiency in Spanish Companies,Management Decision, 47(2), 289-307.

Bachiller, P. (2017) A Meta-Analysis ofthe Impact of Privatization on FirmPerformance, Management Decision, 55(1).

Bishop, M. & Thompson, D. (1992) RegulatoryReform and Productivity Growth in theUK’s Public Utilities, Applied Economics,24(11), 1181-1190.

Bhaskar, V. & Khan, M. (1995) Privatizationand Employment : A Study of the JuteIndustry in Bangladesh, The AmericanEconomic Review, 85(1), 267-273.

Borcherding, T.E. (1979) Towards a PositiveTheory of Public Sector Supply Arrange-ments, Department of Economics andCommerce, Simon Fraser University, Canada.

Boubakri, N., & Cosset, J.C. (1998) TheFinancial and Operating Performance ofNewly Privatized Firms : Evidence fromDeveloping Countries, The Journal ofFinance, 53(3), 1081-1110.

Boussofiane, A., Martin, S. & Parker, D. (1997)The Impact on Technical Efficiency of theUK Privatization Programme, AppliedEconomics, 29(3), 297-310.

Boycko, M., Shleifer, A. & Vishny, R.W. (1996)A Theory of Privatisation, The EconomicJournal, 106 (435), 309-319.

Brown, J.D., Earle, J.S. & Telegdy, A. (2006)The Productivity Effects of Privatization :Longitudinal Estimates from Hungary,Romania, Russia, and Ukraine, Journal ofPolitical Economy, 114(1), 61-99.

Chhibber, A., & Gupta, S. (2017) PublicSector Undertakings : Bharat’s otherRatnas, International Journal of Public SectorManagement, (just-accepted), 00-00.

Chhibber, A. & Gupta, S. (2019) BolderDisinvestment or Better PerformanceContracts? Which Way Forward for India’sState-Owned Enterprises, ICPE PublicEnterprise Half-Yearly Journal, 24(1).

Chisari, O., Estache, A. & Romero, C. (1999)Winners and Losers from the Privatizationand Regulation of Utilities : Lessons from aGeneral Equilibrium Model of Argentina,The World Bank Economic Review, 13(2),357-378.


74

Claessens, S. & Djankov, S. (2002) PrivatizationBenefits in Eastern Europe, Journal ofPublic Economics, 83(3), 307-324.

Cook, P. & Uchida, Y. (2004) An Appraisal ofthe Performance of Privatized Enterprises inDeveloping Countries (No. 30657), Uni-versity of Manchester, Institute for Develop-ment Policy and Management (IDPM).

Dewenter, K.L. & Malatesta, P.H. (2001) State-owned and Privately Owned Firms : AnEmpirical Analysis of Profitability, Leverage,and LaborIntensity, The American EconomicReview, 91(1), 320-334.

Dharwadkar, B., George,. & Brandes, P. (2000)Privatization in Emerging Economies : AnAgency Theory Perspective, Academy ofManagement Review, 25(3), 650-669.

D’Souza, J. & Megginson, W.L. (1999) TheFinancial and Operating Performance ofPrivatized Firms during the 1990s, TheJournal of Finance, 54(4), 1397-1438.

Edwards, C. (2017) Margaret Thatcher’sPrivatization Legacy, Cato Journal, 37(1),89-101.

Estrin, S., Hanousek, J. Kocenda, E. & Svejnar,J. (2009) The Effects of Privatization andOwnership in Transition Economies,Journal of Economic Literature, 47(3), 699-728.

Estrin, S. & Pelletier, A. (2015) Privatisation inDeveloping Countries : What are the Lessonsof Experience, EPSPeaks, November.

Estrin, S. & Pelletier, A. (2018) Privatization inDeveloping Countries : what are the Les-sons of Recent Experience? The World BankResearch Observer, 33(1), 65-102.

Fraquelli, G., & Erbetta, F. (1999) Privatizationin Italy : An Analysis of Factors of Productivityand Technical Efficiency, Ceris-CNR.

Frydman, R., Gray, C., Hessel, M. &Rapaczynski, A. (1999) When doesPrivatization Work? The Impact of PrivateOwnership on Corporate Performance inthe Transition Economies, The QuarterlyJournal of Economics, 114(4), 1153-1191.

Gakhar, D.V. & Phukon, A. (2018) From Wel-fare to Wealth Creation : A Review of theLiterature on Privatization of State-OwnedEnterprises, International Journal of PublicSector Management, (just-accepted), 00-00.

Galal, A., Jones, L., Tandon, P. & Vogelsang, I.(1994) Welfare Consequences of SellingPublic Enterprises : An Empirical Analysis,Washington, DC, The World Bank.

García, L. C. & Ansón, S. G. (2007) TheSpanish Privatisation Process : Implicationson the Performance of Divested Firms,International Review of Financial Analysis,16(4), 390-409.

Ghosh, S. (2008) Does Divestment Matter forFirm Performance? Evidence from the IndianExperience, Economic Systems, 32(4), 372-388.

Gunasekar, S. & Sarkar, J. (2014) DoesAutonomy Matter in State-Owned Enter-prises? Evidence from PerformanceContracts in India, WP-2014-034, IndiraGandhi Institute of DevelopmentResearch, Mumbai.

Gupta, N. (2005) Partial Privatization and FirmPerformance, The Journal of Finance, 60(2),987-1015.

Gupta, N. (2008) Privatization in South Asia,Privatization : Successes and Failures, ColumbiaUniversity Press, New York, NY, 170.


75

Gupta, N. (2013) Selling the Family Silver toPay the Grocer’s Bill? Reforms and Eco-nomic Transformation in India, In JagdishBhagwati and Arvind Panagariya (Eds.),Reforms and Economic Transformation inIndia (141-167), Oxford University Press :New York, NY.

Guriev, S. & Megginson, W. (2007) Privati-zation : What Have We Learned? InAnnual World Bank Conference onDevelopment Economics—Regional 2007:Beyond Transition (pp. 249-96).

Haskel, J. & Szymanski, S. (1992) A BargainingTheory of Privatisation, Annals of Publicand Cooperative Economics, 63(2), 207-227.

Jain, R. (2017) Public Sector Enterprise Disin-vestment in India : Efficiency Gains ina Political Context, Journal of AsianEconomics, 53, (C) 18-36.

Jerome, A. (2008) Privatization and EnterprisePerformance in Nigeria : Case Study ofSome Privatized Enterprises, AERCResearch Paper 175, African EconomicResearch Consortium, Nairobi.

Jiang, G., Yue, H. & Zhao, L. (2009) A Re-Exa-mination of China’s Share Issue Privati-zation, Journal of Banking and Finance,33(12), 2322-2332.

Kay, J.A. & Thompson, D.J. (1986) Privati-sation : A Policy in Search of a Rationale,The Economic Journal, 96(381), 18-32.

Laffont, J.J. & Tirole, J. (1993) A Theory ofIncentives in Procurement and Regulation,MIT Press, Cambridge, MA.

La Porta, R. & Lopez-de-Silanes, F. (1999) TheBenefits of Privatization : Evidence fromMexico, The Quarterly Journal of Economics,114(4), 1193-1242.

Letza, S.R., Smallman, C. & Sun, X. (2004)Reframing Privatisation : Deconstructingthe Myth of Efficiency, Policy Sciences,37(2), 159-183.

Loc, T. D. & Tran, N. M. (2016) Impact ofEquitization on Performance of Enterprisesin Vietnam, Journal of EconomicDevelopment, 23(3), 36-56.

Mandiratta, P. & Bhalla, G.S. (2017) Pre andPost-Disinvestment Performance Evaluationof Indian CPSEs, Management and LabourStudies, 42(2), 120-134.

Martin, S. & Parker, D. (1995) Privatizationand Economic Performance Throughoutthe UK Business Cycle, Managerial andDecision Economics, 16(3), 225-237.

Mathur, I. & Banchuenvijit, W. (2007) TheEffects of Privatization on the Performanceof Newly Privatized Firms in EmergingMarkets, Emerging Markets Review, 8(2),134-146.

Megginson, W.L., Nash, R.C. & Randenborgh,M. (1994) The Financial and OperatingPerformance of Newly Privatized Firms :An International Empirical Analysis, TheJournal of Finance, 49(2), 403-452.

Megginson, W.L. & Netter, J.M. (2001) FromState to Market : A Survey of EmpiricalStudies on Privatization, Journal of EconomicLiterature, 39(2), 321-389.


76

Megginson, W.L. (2016) Major PrivatizationDeals and Trends of 2015 and 2016,Privatisation Barometer, Milano – Italy.

Naceur, S.B., Ghazouani,S. & Omran, M.(2007) The Performance of Newly Priva-tized Firms in Selected MENA Countries :The Role of Ownership Structure,Governance and Liberalization Policies,International Review of Financial Analysis,16(4), 332-353.

Naib, S. (2003) Partial Divestiture and Perfor-mance of Indian Public Sector Enterprises,Economic and Political Weekly, 38(29),3088-3093.

Nellis, J. (2005)Privatization in Africa : Whathas Happened? What is to be Done?,Working Paper, The Center for GlobalDevelopment, Washington, DC.

Omran, M. (2004) The Performance of State-Owned Enterprises and Newly PrivatizedFirms : Does Privatization Really Matter?World Development, 32(6), 1019-1041.

Okten, C. & Arin, K.P. (2006) The Effects ofPrivatization on Efficiency : How DoesPrivatization Work? World Development,34(9), 1537-1556.

Oqdeh, L.N. & AbuNassar, M.A.N. (2011)Effects of Privatization on Firms Financialand Operating Performance : Evidencefrom Jordan, Dirasat : AdministrativeSciences, 38(1).

Pakes, A. & Olley, G.S. (1996) The Dynamicsof Productivity in the TelecommunicationsEquipment Industry, Econometrica, 64(6),1263-1297.

Parker, D. (1999) The Performance of BAABefore and After Privatisation : A DEAStudy, Journal of Transport Economics andPolicy, 33(2),133-145.

Parker, D. & Kirkpatrick, C. (2005) Privatisationin Developing Countries : A Review of theEvidence and The Policy Lessons,Journal ofDevelopment Studies, 41(4), 513-541.

Perry, J.L. & Rainey, H.G. (1988) The Public-Private Distinction in OrganizationTheory : A Critique and Research Strategy,Academy of Management Review, 13(2),182-201.

Procianoy, J.L. & Sobrinho, J.O.F. D. C. (2001)Does the Privatization of State CompaniesImprove their Performance? Evidence fromthe Brazilian Case, Latin American BusinessReview, 2(3-4), 5-36.

S. & Pakes, A. (1992) The Dynamics ofProductivity in The TelecommunicationsEquipment Industry (No.W3977),National Bureau of Economic Research.

Sappington, D.E. & Stiglitz, J.E. (1987) Privati-zation, Information and Incentives,Journal of Policy Analysis and Management,6(4), 567-585.

Sakr, A. (2014)The Impact of Privatisation onthe Performance of Firms in Egypt, ResearchJournal of Finance and Accounting, 5(15),73-81.

Seema, G., Jain, P.K., Yadav, S.S. & Gupta, V.K.(2011) Financial Performance of Disin-vested Central Public Sector Enterprises inIndia : An Empirical Study on SelectDimensions, Journal of Applied Finance andBanking, 1(4), 57.

Shapiro, C. & Willing, D.R. (1990) EconomicRationales for the Scope of Privatization,Discussion Paper-41, Woodrow WilsonSchool, Princeton.


77

Sun, Q. & Tong, W.H. (2003) China ShareIssue Privatization : The Extent of itsSuccess, Journal of Financial Economics, 70,183-222.

Sun, Q. & Tong, W.H. (2002) MalaysiaPrivatization : A Comprehensive Study,Financial Management, 31(4), 79-105.

Tran, N.M., Nonneman,W. & Jorissen, A.(2015) Privatization of VietnameseFirms and its Effects on Firm Performance,Asian Economic and Financial Review,5(2), 202.

Vickers, J. & Yarrow, G.K.(1988) Privatization :An Economic Analysis (Vol.18), MIT Press,Cambridge.

Vickers, J. & Yarrow, G. (1991) EconomicPerspectives on Privatization, The Journal ofEconomic Perspectives, 5(2), 111-132.

Villalonga, B. (2000) Privatization andEfficiency : Differentiating OwnershipEffects from Political, Organizational, andDynamic Effects, Journal of EconomicBehavior & Organization, 42(1), 43-74.

Walker, J.S. & Vasconcellos, G.M. (1997) AFinancial-Agency Analysis of Privatization :Managerial Incentives and FinancialContracting Contracting, Lehigh UniversityPress, Bethlehem.

Wei, Z., Varela, O., Hassan, M.K. & D’Souza, J.(2003) The Financial and OperatingPerformance of China’s Newly PrivatizedFirms, Financial Management, 32(2),107-126.

Wu, H.L. (2006) The Policy-Fit View on theEfficiency Effects of Privatization, Journalof Policy Modeling, 28(3), 281-292.

Wu, H.L. (2007) Exploring the Sources ofPrivatization-Induced PerformanceChanges, Journal of Organizational ChangeManagement, 20(1), 44-59.

Zeckhauser, R.J., Horn, M., Murphy, K.J. &Peltzman, S. (1989) The Control and Per-formance of State-Owned Enterprises, In :Macavoy, P.W., Stanbury, W.T., Yarrow, G.& Zeckhauser, R. (Eds.), Privatization andState-Owned Enterprises, Rochester Studiesin Economics and Policy Issues, Vol.6,Springer, Dordrecht.

— x — x —


78

Corporate Governance Practices and itsImpact on Non-Performing Assets ofSelected Commercial Banks in India

S.K.Chaudhury* & Devi Prasad Misra**

When India is marching ahead for larger market presence with the liberalized business, theIndian banks are today facing greater challenge of realization of Non-Performing Assets(NPAs). Due to enormous increase in the NPAs level of banks, i.e. non-performance of aportion of loan portfolio, has becomes a daunting task before them to recover the same. Asa result, not only the performance of banks goes awry but also has an adverse effect on theIndian economy. NPAs are now considered as burden on the Indian banking industry.Hence, the methods/procedure of managing NPAs and keeping them within the tolerancelevel is the need of the hour. This research study tries to focus on the Corporate Governanceaspect as an appropriate tool to bring down the mounting NPAs in Indian banks. For thispurpose, the statistical tools such as correlation and regression analysis are used to i) ascertainwhether there exists the correlation between corporate governance measures and NPA ratioof public and private sector banks in India and ii) to determine the impact of corporategovernance practices on NPAs in Indian public and private sector banks. The findings ofthe study reveal that there is no correlation between corporate governance measures andNPA ratio in public sector and private sector banks. Further, the regression analysis revealsthat there is no impact of corporate governance practices on NPAs in public sector andprivate sector banks.

Keywords : Asset Quality, Bad Loans, Corporate Governance, Non-Performing Assets

IntroductionThe evaluation of the Indian bankingindustry during the pre-liberalization erarevealed the presence of several short-comings such as reduced productivity,deteriorated asset quality and efficiencyand increased cost structure due to tech-nological backwardness. Among thesedeficiencies, policy makers identifiedthe erosion of asset quality as the most


* Dr.S.K.Chaudhury, Faculty Member, Depart-ment of Business Administration, BerhampurUniversity, Berhampur, Odisha, India.

** Prof.Devi Prasad Misra, Professor, Departmentof Business Management, F.M.University,Balasore, Odisha, India.

significant obstacle for the developmentof a sound and efficient banking sector.In fact, the various practices that werefollowed during pre-liberalizationperiod that includes asset classification

Corporate Governance Practices and its Impact on Non-Performing Assets ofSelected Commercial Banks in India

79

using health code system, accrual basisused to book interest in bank accountsetc., concealed the gravity of asset qualityissues of the banking sector. The assetquality is a prime concern and impactsvarious performance indicators, i.e.,profitability, intermediation costs,liquidity, credibility, income generatingcapacity and overall functioningof banks. The reduction in assetquality results in accumulation ofnon-performing assets.

In the recent past, non-performing assetshas become a matter of great concernfor the Indian bankers. The bankingsector’s asset quality was worsened inthe recent years, with Gross Non-Per-forming Assets (GNPAs) ratio crawlingto 4.45 per cent as on March 2015, asagainst 4.1 per cent in March 20141, asper the latest data released by theReserve Bank of India. Both gross andnet NPAs ratios2 are strong indicators ofthe asset quality of a bank; the higherthe gross/net NPA ratio, the lower its assetquality and vice-versa. In this context,private sector banks remain within thecontrolled limits, and importantly,seem to have provision enough to keepnet NPAs within 1 per cent. But thepublic sector banks, sadly, have not3.

Statement of the ProblemThe need for corporate governancewhich emerged as a result of corporatefailures/bank failures as well as wide

spread dissatisfaction in the way manysuch institutions function, has becomeone of the wide and deep discussionsacross the globe. Non-performingassets is a key concern for banks in India.NPA is said to be an important indicatorto judge the health of the bankingindustry. During the last one decade,the public sector banks have displayedexcellent performance and are one stepahead of private sector banks in finan-cial operations. However, the onlyproblem with public sector banks is theincreasing level of NPA year after year.In this connection, the application ofcorporate governance can be a potentialtool to contain NPAs of banks becausethe corporate governance primarily centreson complete transparency, integrity andaccountability of the management.Further, it focuses on investors’ protec-tion and public interest. Apart fromthat, corporate governance is concernedwith the values, vision and visibility. Inview of the above issues, a thoughtprovoking ideas need to be diagnosedand implemented for better governanceand to make the banking sector moredynamic and vibrant so that it can takecare of the interest of various stakeholders.In this research article, an attempt hasbeen made to examine whether thereexist any correlation between corporategovernance practices and the NPAs ofbanks and also to find out the impactof corporate governance practices onNPAs of banks.


80

Review of Select LiteratureThe review of literature is an integralpart of the research work. Review ofpast studies primarily reveals the researchworks done by individual researchersand institutions and also facilitates tocreate the base for further research.Various studies related to corporategovernance, particularly in bankingsector have been carried out by manyscholars on different aspects at nationaland international level. However, someof the notable research works carriedout in this area is noted below.

Indian ContextBalasubramaniam (2001) in his workon non-performing assets and profita-bility of commercial banks in Indiastressed on the point that the level ofNPAs is high with all banks and thebanks are supposed to bring it down.He suggested that effective internalcontrol systems, good credit appraisalprocedures, along with the improve-ment in asset quality in the balancesheets have the potential to bring downNPA in banking sector. Bhabani andVeena (2011) reveals that the publicsector banks, which are perceived as thefoundation of the Indian banking sys-tem, are unfortunately burdened withexcessive NPAs, huge manpower andlack of advanced technology. Chaudharyand Sharma (2011) in their comparativestudy on the performance of Indian

public and private sector banks statedthat it is high time to take appropriateand stringent measures to get rid ofNPA problem. An efficient Manage-ment Information System (MIS) shouldbe developed whose task is mainly totrain the bank staff involved in sanc-tioning the loans and advances withproper documentation and charge ofsecurities. The bank staff should bemotivated to take methods to preventadvances turning into NPA. Moreover,they suggested that public sector banksshould pay adequate attention on theirfunctioning to compete with privatesector banks. Zafar, Maqbool andKhalid (2013) opine in their study thatin Indian banking system, public sec-tor banks are worst affected by NPAs.NPAs reflect the poor performance ofbanks and its failure adversely affectstheir banking sector’s health. Josephand Prakash (2014) suggested thatfinancial institutions particularly thebanks should be proactive to adopt apractical and structured non-perform-ing assets management system whereprevention of NPA should be accordedthe top priority. They also added thatthe NPA level is much higher in publicsector banks in India as compared tothe private banking sector and foreignbanks. They also suggested that thepublic sector banks should take utmostcare and avoid loan/advances leading toNPA by taking suitable preventive


81

measures in an efficient way. Goel andSahu, et al. (2017) point out that themain reason of poor performance ofmany banks in recent past was thenon-performing assets of the banks.They further add that corporate gover-nance and non-performing assets areinter-related and serious research shouldbe done to find out the relationbetween corporate governance andnon-performing assets.

Global ContextBerger et al. (2005) in their studyviewed that state-owned banks havepoor long-term performance (staticeffect), those undergoing privatizationhad particularly poor performancebeforehand (selection effect), and thesebanks dramatically improved follow-ing privatization (dynamic effect),although much of the measuredimprovement is likely due to placingnon-performing loans into residualentities. Ennobakhare (2010) inspectedthe relationship between corporategovernance practices and its impact onprofitability and found that there wasa significant relationship between banksoperation and the corporate governancepractices. They also proved that theownership style of banks has significantimpact on the size of NPA in banks.

Bebeji (2010) conducted a study tofind out the impact of different creditmanagement strategies on NPA andsuggested that poor management,ineffective monitoring of debts inaddition to liberal credit policy arehaving good relation with corporategovernance practices. They are also sig-nificantly related to non-performingloans. Nyor and Mejabi (2013) in theirresearch on the effect of corporate gov-ernance practices on non-performingloans of Nigerian banks concluded thatcorporate governance variables likeboard size, board composition, compo-sition of audit committee and powerseparation do not have significantimpact on NPA. Moreover, they saythat these variables cannot be reliedupon to solve the problem of NPAmanagement. They suggested that thebanks should shift the focus from theexplanatory variables to other corporategovernance related variables like trans-parency, insider abuse, disclosure prac-tices and accountability. HifzaInamand Aqeel Mukhtar (2014) stated thatgood corporate governance aids inimproving the quality of assets. As theoperational efficiency of a bank isappraised in terms of amount of non-performing assets and default loans, thebanks are required to effectively manageits non-performing loans. Goodcorporate governance helps to efficientlymanage non-performing assets of a


82

bank. Thus, there is positive linkbetween corporate governance andoperational efficiency. Paul and Simon(2014) stated that the corporate gover-nance variable like board size, boardcomposition, composition of auditcommittee and power separation arenot the factors responsible for the risingfigure of non-performing assets ofNigerian Banks. Hence, they suggestedgiving focus on the key variables likeinsider abuse, transparency andaccountability and so on for betterperformance of the banks.

Objectives and Scope of the StudyThe study outlines the followingobjectives for the present research work.

• To examine whether there is anyimprovement in the corporategovernance practices of selectedIndian banks over the last five yearsand if yes, then does it have anyimpact on the NPA of such banks?

• To explore the relationship betweencorporate governance practicesand NPA ratios in selected Indianpublic and private sector banks.

• To assess the impact of corporategovernance practices on non-performing assets in Indian publicand private sector banks.

The scope of the study is limited topublic and private sector banks only.

The foreign banks are excluded fromthe study.

Hypotheses of the StudyIn the backdrop of above objectives,following hypotheses have beenformulated by the researchers.

1. H0 = There is no significant correlationbetween corporate governance andNPA of public sector banks.

2. H0 = There is no significant correlationbetween corporate governance andNPA of private sector banks.

3. H0 = There is no significant impactof corporate governance practices onNPA ratio in Indian public sectorbanks.

4. H0 = There is no significant impactof corporate governance practices onNPA ratio in Indian private sectorbanks.

MethodologyThe study is empirical in nature. Datapertaining to the study were collectedfrom secondary sources and analysedthrough SPSS package version (21).For this purpose 12 sample banks, 6each from public and private sectors arerandomly selected. Public sector banksinclude State Bank of India, Bank ofBaroda, Punjab National Bank, Bankof India, Central Bank of India andUnited Bank of India. Similarly,


83

private sector banks namely, FederalBank, ICICI bank, HDFC bank,Kotak Mahindra Bank, IndusInd Bankand Axis Bank are included as samplesfor the study. The period of study covers6 years i.e. from 2010-11 to 2015-16.

Net NPA ratios are calculated usingrelevant data collected from RBI reports.To calculate the average weighted cor-porate governance score of the samplebanks, 59 point rating scale was com-piled referring different score cards usedby different researchers, committees oncorporate governance and its practicesas mentioned in the annual reports ofIndian public and private sector banks.Binary feeding method was adopted toassign scores against each of the itemsin the scale. In other words, if a parti-cular practice mentioned in the scale isfollowed by the concerned bank andthe same is reflected in its annual reportthen ‘1’ was assigned otherwise ‘0’ wasassigned. Then sum was calculated andthen divided by 59 to obtain the weightedcorporate governance score. Further,statistical tools namely correlationanalysis and regression analysis wereapplied to find out i) whether thereexists the correlation between corporategovernance measures and NPA ratio ofselected public and private sector banksin India and ii) the impact of corporategovernance practices on NPAs suchbanks.

Data Analysis and InterpretationNet NPA ratio of the sample banks aresourced from the RBI reports. Corpo-rate governance scores of public andprivate sector banks are calculated byusing the 59 point scale as discussed inthe methodology. The weightedcorporate governance score close to ‘one’indicates higher corporate governancemeasures taken up by the bank andscore closer to ‘zero’ indicates lowcorporate governance measures takenup by sample banks in both the groups.

The compiled data (net NPA ratiosand weighted corporate governancescores) are presented in the Tables inthe next page. Table-1 and Table-2present the computed data pertainingto weighted average corporate gover-nance score and net NPAs ratio ofsample public sector and privatesector banks respectively.

Relationship between Mean NetNPAs and Average WeightedCorporate Governance ScoresWith a view to ascertaining the relation-ship between net NPA ratios andweighted average corporate governancescores, correlation analysis has beenperformed separately for both publicand private sector banks and they arepresented in the succeeding section.


84

Tab

le-1

: N

et N

PA R

atio

s an

d W

eigh

ted

Ave

rage

Cor

pora

te G

over

nanc

e Sc

ores

of

Publ

ic S

ecto

r B

anks

Nam

e of

SBI

B

oB

PN

B

B

oI

C

B

U

BI

Avg

.A

vg.

the

Ban

kN

PAW

eigh

-

Net

Wei

gh-

Net

Wei

ghN

etW

eigh

-N

etW

eigh

-N

etW

eigh

-N

etW

eigh

-R

atio

ted

CG

Y

ear

NPA

ted

CG

NPA

ted

CG

NPA

ted

CG

NPA

ted

CG

NPA

ted

CG

NPA

ted

CG

(Sam

ple

Scor

e

Rat

ioSc

ore

Rat

ioSc

ore

Rat

ioSc

ore

Rat

ioSc

ore

Rat

ioSc

ore

Rat

ioSc

ore

Ban

ks)

(Sam

ple

Ban

ks)

2010

-11

1.63

0.62

70.

350.

593

0.85

0.67

80.

910.

559

1.10

0.61

01.

190.

780

1.00

50.

641

2011

-12

1.82

0.66

10.

540.

610

1.52

0.69

51.

470.

576

1.46

0.62

71.

700.

797

1.41

80.

661

2012

-13

2.10

0.67

81.

280.

644

2.35

0.74

62.

060.

610

2.18

0.64

41.

610.

797

1.93

00.

686

2013

-14

2.57

0.69

51.

520.

627

2.85

0.74

62.

000.

593

1.98

0.64

42.

330.

814

2.20

80.

686

2014

-15

2.12

0.69

51.

890.

644

4.06

0.76

33.

360.

610

2.65

0.64

42.

710.

831

2.79

80.

698

2015

-16

3.81

0.69

55.

060.

644

8.61

0.76

33.

350.

610

6.42

0.64

45.

250.

831

5.41

60.

697

Sour

ce :

RB

I re

port

s (N

PA) a

nd A

utho

rs’ co

mpi

latio

n (C

orpo

rate

Gov

erna

nce

scor

es) f

rom

Ann

ual r

epor

ts of

ban

ks.


85

Tab

le-2

: N

et N

PA R

atio

s an

d W

eigh

ted

Ave

rage

Cor

pora

te G

over

nanc

e Sc

ores

of

Priv

ate

Sect

or B

anks

Nam

e of

FB

ICIC

I

H

DFC

KM

B

I

ND

US

IND

AX

ISA

vg.

Avg

.

the

Ban

kN

PAW

eigh

-

Net

Wei

gh-

Net

Wei

ghN

etW

eigh

-N

etW

eigh

-N

etW

eigh

-N

etW

eigh

-R

atio

ted

CG

Y

ear

NPA

ted

CG

NPA

ted

CG

NPA

ted

CG

NPA

ted

CG

NPA

ted

CG

NPA

ted

CG

(Sam

ple

Scor

e

Rat

ioSc

ore

Rat

ioSc

ore

Rat

ioSc

ore

Rat

ioSc

ore

Rat

ioSc

ore

Rat

ioSc

ore

Ban

ks)

(Sam

ple

Ban

ks)

2010

-11

0.60

0.76

31.

110.

678

0.19

0.67

80.

590.

576

0.28

0.81

40.

260.

695

0.50

50.

701

2011

-12

0.53

0.78

00.

730.

695

0.18

0.69

50.

510.

593

0.27

0.84

70.

250.

712

0.41

20.

720

2012

-13

0.98

0.78

00.

640.

712

0.20

0.71

20.

550.

610

0.31

0.84

70.

320.

712

0.50

00.

729

2013

-14

0.74

0.81

40.

820.

695

0.27

0.69

50.

880.

610

0.33

0.84

70.

440.

729

0.58

00.

732

2014

-15

0.73

0.79

71.

400.

695

0.25

0.69

50.

790.

627

0.31

0.83

10.

460.

712

0.67

80.

726

2015

-16

1.64

0.79

72.

670.

695

0.28

0.71

20.

930.

627

0.36

0.83

10.

730.

712

1.1

010.

726

Sour

ce :

RB

I re

port

s (N

PA) a

nd A

utho

rs’ co

mpi

latio

n (C

orpo

rate

Gov

erna

nce

scor

es) f

rom

Ann

ual r

epor

ts of

ban

ks.


86

Table-3 : Public Sector Banks

Mean Net NPA Ratio Weighted Average CGof Sample Public Score of Sample Public

Sector Banks Sector Banks

Mean Net NPA Ratio Pearsonof Sample Public Correlation 1 .722Sector Banks Sig. (2-tailed) .106

N 6 6

Average Weighted CG PearsonScore of Sample Public Correlation .722 1Sector Banks Sig. (2-tailed) .106

N 6 6

* Correlation is significant at the 0.05 level (2-tailed).

From Table-3, it is evident that Pearson’sr = 0.722 (p value = 0.106) which is notsignificant and hence, the null Hypo-thesis 1 is accepted. It indicates thatthere is no significant relationshipbetween net NPA ratio and weightedaverage corporate governance score ofpublic sector banks. Even though the

correlation is not significant but theyhave positive relationship.

From Table-4, it is seen that Pearson’sr = 0.269 (p value = 0.606) which is notsignificant and hence, null Hypothesis2 is accepted. It implies that there is nostatistically significant relationship

Table-4 : Private Sector Banks

Mean Net NPA Ratio Weighted Average CGof Sample Private Score of Sample Private

Sector Banks Sector Banks

Mean Net NPA Ratio Pearsonof Sample Private Correlation 1 .269Sector Banks Sig. (2-tailed) .606

N 6 6

Average Weighted CG PearsonScore of Sample Private Correlation .269 1Sector Banks Sig. (2-tailed) .606

N 6 6


87

between NPA ratio and weighted ave-rage corporate governance score ofprivate sector banks. In this case also, eventhough the correlation is not significantbut they have positive relationship.

Impact of Corporate GovernancePractices on Net NPAs RatioTo examine the impact of corporategovernance practices on net NPA ratioof banks, regression analysis is used byconsidering mean net NPA ratio ofsample banks as dependent variable andweighted average corporate governancescore of sample banks as independentvariable. Regression analysis is per-formed separately for both public andprivate sector banks and the results arementioned below.

Regression Analysis : PublicSector BanksThe value of R2 equals 0.521 (Table-5),indicates that 52.1 percent of the varia-tions in net NPA ratio in public sectorbanks is accounted for its weighted ave-rage corporate governance score. Inother words, net NPA ratio in publicsector banks is moderately influencedby weighted average corporate gover-nance score. However, the value of R2

equals 0.521 is not significant (at 5 percent level) as indicated by p value(0.106) of F statistics4 as obtained inAnova Table. It indicates poor goodnessof fit of the model. Hence, Hypothesis 3is accepted. It means there is no signifi-cant impact of corporate governancepractices on NPAs in selected publicsector banks in India.

Table-5 : Model Summary

Model R R Square Adjusted R Square Std. Error of the Estimate

1 .722a .521 .401 1.2190847

a. Predictors : (Constant), Weighted average corporate governance score of sample public sector banks.

Table-6 : Anovaa

Model Sum of Squares Df Mean Square F Sig.

Regression 6.445 1 6.455 4.343 .106b

1 Residual 5.945 4 1.486

Total 12.399 5

a. Dependent Variable : Mean Net NPA ratio of sample public sector banks.b. Predictors : (Constant), Weighted average CG score of sample public sector banks.


88

The estimated regression equation asobtained in Table-7 may be written as :

Net NPA ratio = -31.678 +50.343*AWCG

t = (-1.933) (2.084)

Regression Analysis : PrivateSector BanksThe value of R2 equals 0.072 (Table-8),which indicates that only 7.2 per cent

of the variations in net NPA ratio isaccounted for weighted average corpo-rate governance score in private sectorbanks. In other words, net NPA ratioin private sector banks is not significantlyinfluenced by its weighted average cor-porate governance score. The value ofR2 equals 0.106 is not significant (at 5 percent level) as indicated by p value (0.606)of F statistics5 and obtained in AnovaTable. It also indicates low goodness of

Table-7 : Coefficientsa

Unstandardized Standardized Model Coefficients Coefficients T Sig.

B Std. Error Beta

(Constant) -31.678 16.390 -1.933 .125

1 Weighted Average CGScore of Sample Public 50.343 24.156 .722 2.084 .106Sector Banks

a. Dependent Variable: Mean net NPA ratio of sample public sector banks

Table-8 : Model Summary

Model R R Square Adjusted R Square Std. Error of the Estimate

1 .269a .072 -.159 .2666751

a. Predictors : (Constant), Weighted average corporate governance score of sample public sector banks.

Table-9 : Anovaa

Model Sum of Squares Df Mean Square F Sig.

Regression .022 1 .022 .313 .606b

1 Residual .284 4 .071

Total .307 5

a. Dependent Variable : Mean Net NPA ratio of sample public sector banks.b. Predictors: (Constant), Weighted average CG score of sample public sector banks.


89

fit of the model. Hence, Hypothesis 4is accepted. That means there is nosignificant impact of corporate gover-nance practices on NPAs in selectedprivate sector banks in India.

The estimated regression equation asobtained in Table-10 may be writtenas :

Net NPA Ratio = -3.677 +5.962AWCG

t = (-0.477) (0.559)

From the foregoing analysis, it is revea-led that there is no statistically signifi-cant relationship between the net NPAsratio and weighted average corporategovernance score of both the selectedpublic and private sector banks. Simi-larly, the regression analysis indicatesthat weighted average corporate gover-nance practices do not have any impacton NPAs in both public and privatesector banks. It is quite surprising thatthe implementation of corporategovernance measures is not resulting in

minimization of NPAs ratio6 in Indianpublic and private sector banks.

Limitation and Future Scope ofthe StudySo far as the limitations are concerned,the sample size taken for this researchwork is limited to six public and sixprivate sector banks. The foreign banksare excluded from the study. The dataconsidered for the study is only six yearsonly. Study conducted with largesample size and data long time periodmay yield different results. Further, thescale developed to measure corporategovernance score is limited to 59 pointscale. Inclusion of more items incorporate governance score card mayproduce different results.

Future studies may be conducted byincluding more public, private and for-eign banks in the list of sample size.Apart from this, NPA data of a longtime frame say 10-15 years may betaken up for obtaining accurate resultand inference.

Table-10 : Coefficientsa

Unstandardized Standardized Model Coefficients Coefficients T Sig.

B Std. Error Beta

(Constant) -3.677 7.704 -.477 .658

1 Weighted Average CGScore of Sample Public 5.962 10.664 .269 .559 .606Sector Banks

a. Dependent Variable: Mean net NPA ratio of sample public sector banks.


90

Suggestions and ConclusionIn the last few years, Indian bankingsector has witnessed high volume ofNPAs. Now managing bad loans andcontrolling and keeping them at lowerlevel has become imperative for thebanking industry. The analysis revealsthat corporate governance measures didnot bring any desired result in control-ling the NPA in Indian commercialbanks. To reduce the level of NPAs inthe loan portfolio, comprehensive pre-ventive monitoring mechanism has tobe explored. Further, there should amechanism to maintain a sound andhealthy loan portfolio. The approachto NPA management by the banks hasto be multipronged, necessitating variedstrategies suitable to different stages ofthe passage of credit. Every commercialbank has to embark upon strategic planto prevent/control the occurrence of theNPAs. An enduring solution to theproblem of NPAs can be attained onlyby adopting clear-cut policy guidelinesin respect of credit appraisal withminute, proper assessment of credit andrisk management mechanism alongwith credit rating of the potential bor-rower. Indian bankers may go for theabove suggestive measures in orderto achieve and maintain a lower NPAratio.

ReferencesBalasubramaniam, C.S. (2001) “Non-Perform-

ing Assets and Profitability of CommercialBanks in India : Assessment and Emergingissues”, Abhinav Journal, Vol.1, Issue 7,pp.91-103.

Bebeji, A. (2010) “An Accessment of CreditManagement Strategies and Non-Perfor-ming Loans of Banks in Nigeria”, Paperpresented at a maiden annual conference,organised by School of Business andFinancial Studies, Kaduna Polytechnic.

Berger, N.A., Clarke, G., Cull, R., Klapper, L. &Udell, G.F. (2005) “Corporate Governanceand Bank Performance : A Joint Analysis ofthe Static, Selection and Dynamic Effectsof Domestic, Foreign and State Ownership’’,Journal of Banking & Finance, Vol.29,No.8-9, pp.2179-2221.

Bhabani Prasad, G.V. & Veena, D. (2011)“NPAs Reduction Strategies for Commer-cial Banks in India”, International Journalof Management & Business Studies, Vol.1,Issue 3, pp.47-53.

Chaudhary, K. & Sharma, M. (2011) “Perfor-mance of Indian Public Sector Banks andPrivate Sector Banks : A ComparativeStudy”, International Journal of Innovation,Management and Technology, Vol.2, No.3,pp.45-58.

Enobakhre, A. (2010) “Corporate Governanceand Bank Performance in Nigeria”, MBAResearch Report, University ofStellenbossch, Retrived from: http://hdl.handle.net/10019.1/8439.


91

Goel, I. (2014) “Impact of Corporate Gover-nance on Non-Performing Assets in Banks”,ZENITH International Journal of BusinessEconomics & Management Research, Vol.4,No.3, pp.15-28.

HifzaInam & Aqeel Mukhtar (2014) “Corpo-rate Governance and its Impact on Perfor-mance of Banking Sector in Pakistan”,International Journal of Information,Business and Management, Vol.6, No.3,pp.106-117.

Joseph, A.L. & Prakash, M. (2014) “A Study onAnalyzing the Trend of NPA Level in PrivateSector Banks and Public Sector Banks”,International Journal of Scientific andResearch Publications, Vol.4, No.7, pp.1-9.

Nworji, I.D., Adebayo, O. & Adeyanju, O.D.(2011) “Corporate Governance and BankFailure in Nigeria : Issues, Challenges andOpportunities”, Research Journal of Financeand Accounting, Vol. 2, No.2, pp.24-37.

Paul, A.A. & Simon, K.M. (2014) “The Impactof Corporate Governance Variables onNon-performing Loans of Nigerian DepositMoney Banks”, Asian Economic and Finan-cial Review, Vol.4, No.11, pp.1531-1544.

Sahu, M. M. K., Maharana, N. & Chaudhury,S. K. (2017) “Impact of CorporateGovernance Practices on Non-PerformingAssets (NPA) Management in IndianPublic and Private Sector Banks”, PacificBusiness Review International, 10(4),67-79.

Siraj, K.K. (2014) “A Study on Non-Perform-ing Assets of Public Sector Banks in Indiawith Special Reference to State Bank ofTravancore”, Ph.D Thesis, Submitted toSchool of Management Studies, CochinUniversity of Science and Technology.

Terzungwe, N, & Simon, K.M. (2013)“Impact of Corporate Governance on Non-Performing loans of Nigerian DepositMoney Banks”, Journal of Business andManagement, Vol.2, No.3, pp.12-21.

Zafar, S.M.T., Maqbool, A. & Khalid, S.M.(2013) “Non-Performing Assets and itsImpact on Indian Public Sector Banks”,International Journal of Marketing,Financial Services & Management Research,Vol.3, No.2, pp.68-87.

— x — x —


92

Persistence in Performance of IndianMutual Fund Schemes : An Evaluation

Joyjit Dhar* & Kumarjit Mandal**

Persistence refers to the ability of a fund to maintain its relative performance ranking overtime In the prospectus pertaining to mutual funds is very common to found the phrase that“past performance is not an indicator of future performance”. On the other hand, investorsexpect a consistent return on their investments and persistence of performance is of course agreat concern for them. The fund managers also try to maintain their good performanceand try to improve it when it is below satisfactory level. This is because, on the one hand, itis related to their compensation package and on the other, it is extremely difficult to sell amutual fund scheme that has poor track record. From an academic perspective, the existenceof performance persistence actually challenges semi-strong form of the Efficient MarketHypothesis (Fama, 1970). In this backdrop, the present study attempts to address the issueof ‘persistence’ in equity mutual funds in India across two different time horizons withrespect to three different performance measures applying a non-parametric contingencytable approachand the Spearman rank correlation coefficient (SRCC) test. The resultsindicate that though there is some evidence of short-run performance persistence in theIndian mutual funds marketthere is no long-run performance persistence during the periodunder consideration. Thus, on the whole this analysis suggests that past performance ofmutual funds cannot be used as indicators of future performance.

Keywords : Mutual Funds, Performance Persistence, Contingency Table, EfficientMarket, Spearman Rank Correlation Coefficient (SRCC) Test

IntroductionIn the performance evaluation literature,persistence refers to the ability of afund to maintain its relative perfor-mance ranking over time. In the mutualfunds prospectus it is very common tofound the phrase that “past performanceis not an indicator of future perfor-mance”. However, investors expect a


* Dr.Joyjit Dhar, Associate Professor of Economics,West Bengal Education Service (W.B.E.S.)Government of West Bengal, A.B.N.SealCollege, P.O. Coochbehar, Dist.-Coochbehar,PIN- 736101, West Bengal, India.

** Dr.Kumarjit Mandal, Associate Professor ofEconomics, Department of Economics,University of Calcutta, 56A, BarrackporeTrunk Road, Kolkata-700050, West Bengal,India.

Persistence in Performance of Indian Mutual Fund Schemes :An Evaluation

93

consistent return on their investmentsand persistence of performance is ofcourse a great concern for them. Patel,Zeckhauser and Hendricks (1992) haveshown that investors keep their moneyto funds which are superior performersin recent times. Both individual andinstitutional investors are looking intothe performance evaluation methodswhich will ultimately help them to selectthose funds which are superior per-formers. The fund managers also try tomaintain their good performance andtry to improve it when it is belowsatisfactory level. This is because, on theone hand, it is related to their compen-sation package and on the other, it isextremely difficult to sell a mutualfund scheme that has poor track record.From an academic perspective, theexistence of performance persistenceactually challenges semi-strong form ofthe Efficient Market Hypothesis (Fama,1970). This form of market efficiencystates that history of past prices cannotbe used to predict future performances,while persistence in performanceactually means predictive ability of thefund managers. According to Grossmanand Stiglitz (1980) security pricesshould not reflect the full informationpossessed by the informed individuals.There must be a reward for the investorsfor their costly endeavor of seeking newinformation. In the context of mutualfund performance evaluation some

mutual fund managers are expec-ted to have an informational advantage.Berk and Green (2004), however hasshown theoretically that such an infor-mational advantage will be temporarywhen the investors direct their capitalto recent winners, thus making theindustry competitive. The objective ofthis study is to determine empiricallyif such managerial ability persists overa time horizon.

Several studies have examined persis-tence in fund performance and havefound a ‘hot hands’ phenomenon. Forexample, Hendricks et al. (1993) andGoetzmann and Ibbotson (1994) haveshown that past mutual fund returnspredict future returns. This type ofevidence is not only inconsistent withefficient markets, which advocates thatpast performance cannot be a guide tofuture performance, but also influencesinvestors by suggesting that they mayrealize superior returns by purchasingfunds which have performed wellrecently. Malkiel (1995) found evidenceof persistence in the 1970s which weredisappearing in the 1980s. However,Gruber (1996) found very strongevidence of persistence looking at equitymutual funds from 1985 to 1994 andargued that persistence affected thegrowth of active mutual funds duringthis period. Studies of Brown andGoetzmann (1995), Wermers (1997),Carhart (1997) and Droms and Walker


94

(2001) found evidence of short-termpersistence in mutual funds. Otten andBams (2002) evaluated the perfor-mance persistence of 506 funds of fiveEuropean countries such as U.K., France,Germany, Italy and Netherlands.Their results indicated that mostEuropean funds provide only weakevidence of persistence in performance,except U.K. Babalos et al. (2008)examined the performance persistenceof domestic equity funds in Greece.They argued that persistence is evidentfor specific periods and it was notsignificant for the overall sample period.Grinblatt and Titman (1992), Elton et al.(1996), Volkman and Wohar (1996),Allen and Tan (1999), Filip (2011)observed persistence over longer periods.On the other hand, Jensen (1968),Kritzman (1983), Dunn and Theisen(1983), Elton et.al (1990), Bauer et al.(2006), Casarin et al. (2008), Barras(2010), Fama and French (2010),Busse et al. (2010) had evidence of littleor no evidence of persistence in the per-formance of mutual funds. Thus, theavailable studies mentioned earlierpresent contrasting results and hence atbest, be called inconclusive.

The issue of performance persistence ofIndian mutual funds is not adequatelyaddressed in the academic literature.While Roy and Deb (2003) foundsignificant evidence of persistence, thestudy made by Guha Deb (2006)

revealed moderate evidence of short-term performance persistence. Sehgaland Jhanwar (2008) showed no evidenceof persistence for monthly data butfound evidence of persistence usingdaily data for only one model. However,the study of Mondal and Khan (2014)has shown no evidence of persistencefor both short-run and long-run timeperiods. Thus, there is no conclusiveevidence so far on the issue of perfor-mance persistence of mutual funds indifferent economies including India. Inthis backdrop, the present study hasmade an attempt to examine the issueof performance persistence for theIndian mutual fund schemes during2000-2012.

Data and MethodologyDataThe present study uses a sample of 80mutual fund schemes. The details ofthese schemes are given in Table-A- 1.1in Appendix-I. Out of eighty schemesthe sample comprises of sixty six growthschemes and fourteen equity linked sav-ings schemes (ELSS). Since balancedschemes of the sample are basically equityoriented they are also treated as equityschemes. The data used in the studymainly comprise of weekly Net AssetValues (NAV) for the eighty mutual fundsschemes during May 2000 to March2012. These NAV data are collectedfrom www.mutualfundsindia.com.


95

This study has used Sensex as themarket proxy. Sensex data in weeklyfrequency are collected from BSEwebsite. In this study, weekly yield on91 days treasury bills of Governmentof India (GoI) is used as a proxy forrisk-free return. These data are collectedfrom RBI website.

Study Methodology

The present study evaluates the issueof ‘persistence’ for equity mutual fundsin India across two different time hori-zons with respect to three different per-formance measures. Different choicesof time horizons are needed to checkthe impact (if any) of the length of thetime horizon on the persistence of ascheme. In order to test the persistenceof mutual fund schemes this study hasadopted a non-parametric contingencytable approach. According to thismethod the measurement of fund per-formance requires two consecutive timeperiods : first one is the current periodand the second one is the test period.This study has used two different timeperiods of six months and one-year justto explore the issue of persistence overshort-term and long-term horizons. So,it has twenty-three pairs of half-yearlytime periods of current period - testperiod and eleven pairs of annual timeperiods of current period - test period.Within each period the schemes areclassified as winners or losers depending

on their performance in that period. Afund is said to be winner in the currentperiod if it is above or equal to themedian performance of all funds inthat period and it is loser if the reversehappens. Similarly, the other categoriescan be defined.

Once this classification is done, a two-way contingency table is created withelements such as WW (winner in suc-cessive periods), LL (loser in successiveperiods), WL (winner in the currentperiod and loser in the test period) andLW (loser in the current period andwinner in the test period) and is givenbelow :

Period(t)

Period (t+1)

Winner Loser

Winner WW WL

Loser LW LL

For each pair of current period and testperiod, number of funds belonging toeach category as mentioned above areobtained for each of the three measures -(i) Benchmark adjusted return (ii) Sharperatio and (iii) Jensen alpha. If there isevidence of positive persistence, then wewould observe more number of schemesin the WW or LL categories. If there isreversal there would be more cases inthe WL or LW categories. Evidence forpersistence is statistically tested usingrepeat winner approach (Malkiel, 1995),


96

the odds ratio test (Brown & Goetzmann,1995) Kahn and Rudd’s Chi-square (χ2)test (1995) and the Spearman RankCorrelation Coefficient (SRCC) test.

In the present study returns refer to theaverage weekly return achieved by theconcerned mutual funds due to thechange in the net asset value from periodt-1 to period t. Income of any associateddividends is assumed to be reinvestedthus incorporated in the fund NAV.

The present study has used threedifferent performance proxies for testingpersistence which are given below :

(i) Benchmark adjusted return :

Re = Rp – Rb .....(1)

(ii) Sharpe ratio :

SR =

R Rp f

p

−σ

.....(2)

(iii) Jensen’s Alpha (ααααα) :

R R R Rp f p p m f p− = + ∗ −( ){ } +α β ε .....(3)

where,

Rp = the return of the scheme p for theconcerned period

Rb = benchmark return

Rf = the risk- free return for the sameperiod

σp = the total risk of the scheme p forthe same period

Rm = the market return for the sameperiod

βp = the systematic risk of the schemep

εp = the error term

The null hypotheses of these tests arethat there is no evidence of performancepersistence i.e., there is no relationbetween fund performance in currentperiod and in subsequent period or testperiod.

Malkiel’s Z – TestMalkiel’s Z test or repeat winner testshows the percentage of repeat winners(WW) to winner–losers (WL). Let, ‘p’be the probability that a winning fundcontinues to be a winning fund in the

next period, then p = 12

if there would

be no persistence. Since the randomvariable Y of the number of persistentlywinning funds is binomially distribu-ted, a binomial test to see if ‘p’ > ½ i.e.the test for repeat winner when n is rea-sonably large (n>20) is given as follows :

Z Y npnp p

= −−( )1

.....(4)

In equation (4) Z will be distributednormally with mean zero and standarddeviation one, A percentage of repeatwinner above 50 per cent and a Z-statistic above zero show performancepersistence.


97

Brown and Goetzmann’s Odds Ratioor Cross Product Ratio (B&G CPR)Brown and Goetzmann (1995) proposedCross Product Ratio (CPR) or OddsRatio which is defined as the numberof repeat performers to the number ofthose that do not repeat and is given as(WW*LL) / (WL*LW). The statisticalsignificance of the CPR can be examinedby a Z-statistic which is given as :

ZCPRCPR

= ( )( )

loglogσ

.....(5)

In large samples with independentobservations, the standard error of thenatural log of the cross product ratio isapproximated as (Christensen, 1990)

σlog CPR WW WL LW LL( ) = + + +1 1 1 1.....(6)

For large samples, the Z statistic is nor-mally distributed with mean zero andstandard deviation one. A CPR aboveone and a positive Z statistic indicatepersistence in fund performance.

Kahn and Rudd’s Chi-square (χχχχχ2)Test

This study has also used Chi-square (χ2)test of Kahn and Rudd (1995) for per-formance persistence and is given as :

χ2 =−( )∑

O E

Eij ij

ij

i, j = 1, 2 ….n .....(7)

where, Oij = actual frequency of theith row and jth column in the contingencytable.

Eij = expected frequency of the ith row

and jth column in the contingency table.

However, the χ2 test suffers from oneserious limitation: though the devia-tions from expected frequencies areconsidered by the test, no definite con-clusions are drawn in terms of persis-tence or mean reversion properties ofthese deviations. Thus to test persis-tence this study has also examined theconditional probabilities apart from thesignificance of the chi-square values.

Spearman Rank Correlation Coeffi-cient (SRCC) TestSpearman Rank Correlation Coefficientdenoted by Rs is defined as follows :

R dn ns

i= − ∑−( )1 61

2

2 .....(8)

di = difference between rank of a schemein current period and test period

n = number of observations or num-ber of sample mutual fund schemes

For large samples, the sampling distri-bution of Rs is approximately normallydistributed with mean zero and standard

deviation 11n − . Then the corres-

ponding Z statistic is given as follows :


98

Z RnS=−1 .....(11)

If the computed Z value is greater thanthe critical Z value at chosen level ofsignificance then null hypothesis of nopersistence is rejected.

Results and ConclusionThe objective of the study is to figureout whether performance is persistentfor the sample equity mutual fundschemes applying a 2x2 non-parametriccontingency table approach. Bench-mark adjusted return, Sharpe ratio andJensen Alpha are used as performanceproxies to rank all schemes based onthe past half-yearly return and annualreturn during May 2000 to March 2012.The significance of these results are alsotested by three different tests of repeatwinner approach (Malkiel, 1995), theodds ratio test (Brown & Goetzmann,1995) and Kahn and Rudd’s Chi-square(χ2) test (1995). Besides, contingencytable approach the Spearman rankcorrelation coefficient (SRCC) test isalso used in the study to examine theperformance persistence of samplemutual funds. The results are given inTables A-1.1 to A-1.8 in the Appendix.

Performance Persistence of Bench-mark Adjusted ReturnAccording to the results given inTable-A-1.1 for the benchmarkadjusted return there is some evidence

of persistence in terms of odds ratiotest, repeat winner approach and χ2 testfor a time horizon of six months. It isfound from Table A-1.1 that amongtwenty three number of half-year periodsthere are seven numbers of persistenceand three numbers of reversals in termsof odds ratio test and eight numbers ofpersistence and three numbers of rever-sals according to repeat winner app-roach at 5 per cent level of significance.This study has also examined the condi-tional probabilities apart from the sig-nificance of the χ2 values to determinewhether it is persistence or reversal. Ifit is persistence then there will be higherconditional probabilities for WW andLL than WL and LW. In case of rever-sals the results will be opposite. Basedon these considerations seven cases ofpersistence and four cases of reversalsare reported for χ2 test at 5 percent levelof significance in Table-A-1.1. Thus, forthe shorter time horizon (six months)there is some evidence of persistence ofperformance for the sample mutualfunds. Table-A-1.1 also documentssome cases of reversals in respect of thesetests. However,such short-run perfor-mance persistence vanishes whenSpearman rank correlation coefficienttest (SRCC Test) is used to examine it.The results of this test are given inTables-A-1.7 to A-1.8.

However, as the time horizon increasesto one year the number of persistence


99

cases is significantly reduced for all thetests (except SRCC test) and disappearedfor repeat winner approach as revealedby Table-A-1.2. This actually supportsthe strong case of market efficiency ofthe mutual funds industry in the longrun.

Performance Persistence of SharpeRatio

Table-A-1.3 displays the contingencytable and different test results whensample schemes are ranked as winnersor losers based on Sharpe ratio. Theresults show that out of twenty threenumber of half-yearly periods there arefourteen numbers of persistence andseven numbers of reversals in terms ofodds ratio test, fifteen numbers of per-sistence and seven numbers of reversalsaccording to repeat winner approachand sixteen numbers of persistence andseven numbers of reversals according toχ2 test at 5 per cent level of significance.Thus, the number of persistenthalf-yearly periods is much higherfor Sharpe ratio than showed bybenchmark adjusted return in terms ofodds ratio test, Malkiel test (repeatwinner approach) and Kahn and Ruddχ2 test. But, Tables-A-1.7 to A-1.8reveal that likewise benchmark adjustedreturn such short-run performance per-sistence vanishes when Spearman rankcorrelation coefficient test (SRCC test)is used.

Nevertheless, for annual time horizonthough the numbers of persistence andreversal cases are significantly reducedfrom earlier half-yearly time horizon(except SRCC test), they did notdisappear and are equal for odds ratiotest and χ2 test applied in the study.These results are shown in Table-A-1.4.

Performance Persistence of JensenAlphaIn respect of Jensen Alpha, the type ofpersistence is almost similar as observedfor benchmark adjusted return for thesample mutual fund schemes accordingto odds ratio Test, Malkiel Test (repeatwinner approach) and Kahn and Ruddχ2 test. The results obtained fromTable-A-1.5 indicate that for odds ratiotest numbers of persistence are six outof twenty three half-yearly periods andthere is only one case of reversal. Thecorresponding figures for repeat winnerapproach and χ2 test are six numbersof persistence and two numbers ofreversals and six numbers of persistenceand one number of reversal only. Now,as before the number of persistence andreversals are reduced in the case of one yeartime horizon (Table-A-1.6). Finally, theSRCC test has documented no persis-tence for both the time horizons.

Thus, in conclusion, this can be arguedthat though there is some evidence ofshort-run performance persistence inthe Indian mutual funds market there


100

is no long-run performance persistenceduring the period under consideration.This actually signifies that mutual fundsmarket in India is efficient in the long-run. This is because market efficiencyof any form implies historical perfor-mance cannot be used to select fundsthat will be superior performers in thefuture. Thus, on the whole this analy-sis suggests that past performance ofmutual funds cannot beused as indi-cators of future performance. This find-ing is consistent with the earlier Indianstudies of Guha Deb (2006), Sehgaland Jhanwar (2008) and Mondal andKhan (2014) and also with studies ofdeveloped capital markets like Jensen(1968), Kritzman (1983), Dunn andTheisen (1983), Elton et.al (1990),Bauer et. al. (2006), Casarin et al. (2008),Barras (2010), Fama and French(2010), Busse et al. (2010).

ReferencesAllen, D.E. & Tan, M.L. (1999) A Test of the

Persistence in the Performance of UKManaged funds, Journal of Business Finance& Accounting, 26, 559-593.

Babalos, V., Caporale, G.M., Kostakis, A., &Philippas, N. (2008) Testing for Persistencein Mutual Fund Performance and theEx-Post Verification Problem : Evidencefrom the Greek Market, European Journalof Finance, 14 (8), 735-753.

Bauer, R., Otten, R. & Rad, A.T. (2006) NewZealand Mutual Funds : Measuring Per-formance and Persistence, Accounting andFinance, 46, 347-363.

Barras, L., Scaillet, O. & Wermers, R. (2010)False Discoveries in Mutual Fund Perfor-mance : Measuring Luck in EstimatedAlphas, Journal of Finance, 65, 179-216.

Berk, J.B. & Green, R.C. (2004) Mutual FundFlows and Performance in RationalMarkets, Journal of Political Economy, 112,1269-1295.

Brown, S. & Goetzmann, W. (1995) Perfor-mance Persistence, Journal of Finance, 50,679-698.

Carhart, M.M. (1997) On Persistence in MutualFund Performance, Journal of Finance,52(1), 57-82.

Casarin, R, Pelizzon, L. & Piva, A. (2008)Italian Equity Funds : Efficiency andPerformance Persistence, Working Paper12/WP/2008, Department of EconomicsCa’Foscari University, Venice.

Dunn. P. & Theisen, R. (1983) How Consis-tently do Active Managers Win?, Journal ofPortfolio Management, 9, 47-53.

Droms, W. & Walker, D. (2001) Persistence ofMutual Fund Operating Characteristics :Returns, Turnover Rates and ExpenseRatios, Applied Financial Economics, 11(44), 457-466.

Elton, E.J., Gruber, M.J. & Rentzler, J. (1990)ThePerformance of Publicly OfferedCommodity Funds, Financial AnalystsJournal, 46, 23-30.

Elton, E.J., Gruber, M.J., Das, S. & Blake. C.(1996) The Persistence of Risk-AdjustedMutual Fund Performance, Journal ofBusiness, 69, 133-157.

Fama, E. (1970) Efficient Capital Markets : AReview of Theory and Empirical Work,Journal of Finance, 25, 383-417.


101

Fama, E. & French, K.R. (2010) Luck VersusSkill in the Cross Section of Mutual FundReturns, Journal of Finance, 65, 1915-1947.

Filip, D. (2011) Performance Persistence ofEquity Funds in Hungary, ContemporaryEconomics, 5(1), 18-34.

Goetzmann, W.N. & Ibbotson, R.G. (1994)Do Winners Repeat? Journal of PortfolioManagement, 20(2), 9-18.

Grinblatt, M, & Titman, S. (1992) The Persis-tence of Mutual Fund Performance,Journal of Finance, 47(5), 1977-1984.

Grossman, S.J. & Stiglitz, J.E. (1980) On theImpossibility of Informationally EfficientMarkets, American Economic Review, 70(3),393- 408.

Gruber, M.J. (1996) Another Puzzle : TheGrowth in Actively Managed MutualFunds, Journal of Finance, 51(3), 783-810.

Guha Deb, S. (2006) Do Winner Funds Repeat?An Empirical Exploration with IndianEquity Mutual Funds, Paper Presented inICFAI Conference, Hyderabad.

Hendricks, D., Patel, J. & Zeckhauser, R. (1993)Hot hands in Mutual Funds : Short-RunPersistence of Relative Performance, 1974-1988, Journal of Finance, 48(1) : 93-130.

Jensen, M.C. (1968) The Performance ofMutual Funds in The Period 1945-1964,Journal of Finance, 23, 389-416.

Kritzman, M. (1983) Can Bond ManagersPerform Consistently? Journal of PortfolioManagement, 9, 54-56.

Malkiel, B.G. (1995) Returns from Investingin Equity Mutual Funds 1971 to 1991,Journal of Finance, 50(2), 549-572.

Mondal, T., & Khan, T.L. (2014) A Study onthe Performance Persistence of IndianEquity Mutual Funds, Indian Journal ofAccounting, 46(1), 42-50.

Otten, R. & Bams, D. (2002) European MutualFund Performance, European FinancialManagement, 8, 75-101.

Roy, B. & Deb, S.S. (2003) The ConditionalPerformance of Indian Mutual Funds : AnEmpirical Study, Working Paper, SocialScience Research Network.

Sehgal, S. & Jhanwar, M. (2008) On StockSelection Skills and Market Timing Abili-ties of Mutual Fund Managers in India,International Research Journal of Finance &Economics, 15, 299-309.

Volkman, D.A. &Wohar, M.E. (1995)Determinants of Persistence in RelativePerformance of Mutual Funds, Journal ofFinancial Research, 18, 415-30.

Wermers, R. (1997) Momentum InvestmentStrategies of Mutual Funds, PerformancePersistence and Survivorship Bias, Work-ing Paper, Graduate School of Business andAdministration, University of Colorado atBoulder, Colorado, USA.

— x — x —


102

Ap

pen

dix

: R

esu

lts

of

Perf

orm

an

ce P

ers

iste

nce

Ba

sed

on

Thre

e P

erf

orm

an

ce M

ea

sure

sTa

ble-

A-1

.1 :

Con

tinge

ncy

Tabl

e fo

r Pe

rfor

man

ce P

ersis

tenc

e of

Ben

chm

ark

Adj

uste

d R

etur

n (6

Mon

ths

Lag)

Tim

ePe

rcen

tage

Mal

kiel

B &

GS.

No.

Peri

odW

WW

LLW

LLR

epea

tZ

Sco

reC

PR

Z S

core

χχχχ χ2

Win

ner

1H

1-H

225

1416

2564

.10

1.76

1409

7*2.

7901

792.

2184

91*

5.1*

(62.

5%)

(35%

)(4

0%)

(62.

5%)

2H

2-H

322

1916

2353

.66

0.46

8521

31.

6644

741.

1278

941.

5(5

5%)

(47.

5%)

(40%

)(5

7.5%

)

3H

3-H

419

2121

1947

.5-0

.316

228

0.81

8594

-0.4

4703

0.2

(47.

5%)

(52.

5%)

(52.

5%)

(47.

5%)

4H

4-H

518

2222

1845

-0.6

3245

60.

6694

21-0

.892

930.

8(4

5%)

(55%

)(5

5%)

(45%

)

5H

5-H

616

2524

1539

.02

-1.4

0556

40.

4-1

.995

55*

4.1*

(40%

)(6

2.5%

)(6

0%)

(37.

5%)

6H

6-H

727

1313

2767

.52.

2135

944*

4.31

3609

3.06

1887

*9.

8*(6

7.5%

)(3

2.5%

)(2

7%)

(67.

5%)

7H

7-H

819

2121

1947

.5-0

.316

228

0.81

8594

-0.4

4703

0.2

(47.

5%)

(52.

5%)

(52.

5%)

(21%

)

8H

8-H

913

2727

1332

.5-2

.213

594*

0.23

1824

-3.0

6189

*9.

8*(3

2.5%

)(6

7.5%

)(6

7.5%

)(3

2.5%

)

9H

9-H

1025

1126

1869

.44

2.33

3333

3*1.

5734

270.

9557

047.

3(6

2.5%

)(2

7.5%

)(6

5%)

(45%

)

10H

10-H

1126

1414

2665

1.89

7366

6*3.

4489

82.

6409

12*

7.2*

(65%

)(3

5%)

(35%

)(6

5%)

(Con

td...

)


103

11H

11-H

1225

1515

2562

.51.

5811

388*

2.77

7778

2.21

194*

5(6

2.5%

)(3

7.5%

)(6

2.5%

)(6

2.5%

)

12H

12-H

1322

1619

2357

.89

0.97

3328

51.

6644

741.

1278

941.

5(5

5%)

(40%

)(4

7.5%

)(5

7.5%

)

13H

13-H

1424

1616

2460

1.26

4911

12.

251.

7766

593.

2(6

0%)

(40%

)(4

0%)

(60%

)

14H

14-H

1524

1617

2360

1.26

4911

12.

0294

121.

5575

812.

5(6

0%)

(40%

)(4

2.5%

)(5

7.5%

)

15H

15-H

1611

2929

1127

.5-2

.846

05*

0.14

3876

-3.8

7154

*16

.2*

(27.

5%)

(72.

5%)

(72.

5%)

(27.

5%)

16H

16-H

1727

1313

2767

.52.

2135

944*

4.31

3609

3.06

1887

*9.

8*(6

7.5%

)(3

2.5%

)(3

2.5%

)(6

7.5%

)

17H

17-H

1818

2222

1845

-0.6

3245

60.

6694

21-0

.892

930.

8(4

5%)

(55%

)(5

5%)

(45%

)

18H

18-H

1922

1226

2064

.71

1.71

4985

9*1.

4102

560.

7375

495.

2*(5

5%)

(30%

)(6

5%)

(50%

)

19H

19-H

2028

1212

2870

2.52

9822

1*5.

4444

443.

4728

88*

12.8

*(7

0%)

(30%

)(3

0%)

(70%

)

20 H

20 -H

2119

2121

1947

.5-0

.316

228

0.81

8594

-0.4

4703

0.2

(47.

5%)

(52.

5%)

(52.

5%)

(47.

5%)

21 H

21-H

2214

2325

1837

.84

-1.4

7959

10.

4382

61-1

.798

193.

7(3

5%)

(57.

5%)

(62.

5%)

(45%

)

22 H

22-H

2328

816

2877

.78

3.33

3333

3*6.

125

3.56

1713

*14

.4*

(70%

)(2

0%)

(40%

)(7

0%)

23 H

23-H

2413

2727

1332

.5-2

.213

594*

0.23

1824

-3.0

6189

*9.

8*(3

2.5%

)(6

7.5%

)(6

7.5%

)(3

2.5%

)

Sour

ce :

Cal

cula

ted.

Figu

res

in t

he p

aren

thes

es a

re c

ondi

tion

al p

roba

bilit

ies.

; *

Den

otes

5%

leve

l of s

igni

fican

ce.


104

Tab

le A

-1.2

: C

onti

ngen

cy T

able

for

Per

form

ance

Per

sist

ence

of

Ben

chm

ark

Adj

uste

d R

etur

n (1

year

Lag

)

Tim

ePe

rcen

tage

Mal

kiel

B &

GS.

No.

Peri

odW

WW

LLW

LLR

epea

tZ

Sco

reC

PR

Z S

core

χχχχ χ2

Win

ner

1Y1

-Y2

2020

2020

500

10

0(5

0%)

(50%

)(5

0%)

(50%

)

2Y2

-Y3

2119

1921

52.5

0.31

6227

81.

2216

070.

4470

270.

2(5

2.5%

)(4

7.5%

)(4

7.5%

)(5

2.5%

)

3Y3

-Y4

2317

1723

57.5

0.94

8683

30.

5463

14-1

.336

551.

8(5

7.5%

)(4

2.5%

)(4

2.5%

)(5

7.5%

)

4Y4

-Y5

2317

1723

57.5

0.94

8683

30.

5463

14-1

.336

551.

8(5

7.5%

)(4

2.5%

)(4

2.5%

)(5

7.5%

)

5Y5

-Y6

2317

1723

57.5

0.94

8683

30.

5463

14-1

.336

551.

8(5

7.5%

)(4

2.5%

)(4

2.5%

)(5

7.5%

)

6Y6

-Y7

2515

1426

62.5

1.58

1138

83.

0952

382.

4279

55*

6.1*

(64%

)(3

8%)

(34%

)(6

3%)

7Y7

-Y8

1821

2120

46.1

5-0

.480

384

0.81

6327

-0.4

5289

0.3

(46%

)(5

4%)

(51%

)(4

9%)

8Y8

-Y9

1722

2417

43.5

9-0

.800

641

0.54

7348

-1.3

3183

1.9

(41%

)(5

4%)

(62%

)(4

4%)

9Y

9-Y

1017

2416

2341

.46

-1.0

9321

60.

4927

54-1

.557

582.

5(4

3%)

(60%

)(5

8%)

(40%

)

10Y

10-Y

1125

1516

2462

.51.

5811

382.

51.

9955

55*

4.1*

(61%

)(3

7%)

(41%

)(6

2%)

11Y

11-Y

1223

1817

2256

.09

0.78

0868

1.65

3595

1.11

5436

1.3

(58%

)(4

5%)

(43%

)(5

5%)

Sour

ce :

Cal

cula

ted.

Figu

res i

n th

e pa

rent

hese

s are

con

ditio

nal p

roba

bilit

ies;

* D

enot

es 5

% le

vel o

f sig

nific

ance

.


105

Tab

le-A

-1.3

: C

onti

ngen

cy T

able

for

Per

form

ance

Per

sist

ence

of

Shar

pe R

atio

(6

Mon

ths

Lag)

Tim

ePe

rcen

tage

Mal

kiel

B &

GS.

No.

Peri

odW

WW

LLW

LLR

epea

tZ

Sco

reC

PR

Z S

core

χχχχ χ2

Win

ner

1H

1-H

227

1313

2767

.52.

2135

94*

4.31

3609

3.06

1887

*9.

8*(6

7.5%

)(3

2.5%

)(3

2.5%

)(6

7.5%

)

2H

2-H

330

1023

1775

3.16

2278

*2.

2173

911.

6404

9110

.9*

(75%

)(2

5%)

(57.

5%)

(42.

5%)

3H

3-H

420

3320

737

.73

-1.7

8569

*0.

2121

21-2

.966

81*

16.9

*(5

0%)

(82.

5%)

(50%

)(1

7.5%

)

4H

4-H

523

173

3757

.50.

9486

8316

.686

274.

1379

25*

29.8

*(5

7.5%

)(4

2.5%

)(7

.5%

)(9

2.5%

)

5H

5-H

621

915

3570

2.19

089*

5.44

4444

3.36

2609

*18

.6*

(52.

5%)

(22.

5%)

(37.

5%)

(87.

5%)

6H

6-H

733

108

2976

.74

3.50

7467

*11

.962

54.

6102

47*

24.7

*(8

2.5%

)(2

5%)

(20%

)(7

2.5%

)

7H

7-H

82

3838

25

-5.6

921*

0.00

277

-5.7

3977

*64

.8*

(5%

)(9

5%)

(95%

)(5

%)

8H

8-H

93

3739

17.

5-5

.375

87*

0.00

2079

-5.2

4585

*65

*(7

.5%

)(9

2.5%

)(9

7.5%

)(2

.5%

)

9H

9-H

1038

32

3792

.68

5.46

6082

*23

4.33

335.

7949

69*

61.3

*(9

5%)

(7.5

%)

(5%

)(9

2.5%

)

10 H

10-H

1138

22

3895

5.69

21*

361

5.73

9768

*64

.8*

(95%

)(5

%)

(5%

)(9

5%)

11 H

11-H

1236

44

3690

2.21

3594

*81

5.89

5772

*51

.2*

(90%

)(1

0%)

(10%

)(9

0%)

(Con

td...

)


106

12 H

12-H

1337

43

3690

.24

3.16

2278

*11

15.

8954

94*

54.5

*(9

2.5%

)(1

0%)

(7.5

%)

(90%

)

13 H

13-H

1439

22

3795

.12

-1.7

8569

360.

755.

7389

94*

64.9

*(9

7.5%

)(5

%)

(5%

)(9

2.5%

)

14 H

14-H

1539

11

3997

.50.

9486

8315

215.

1158

85*

72.2

*(9

7.5%

)(2

.5%

)(2

.5%

)(9

7.5%

)

15 H

15-H

162

3938

14.

878

2.19

089*

0.00

135

-5.3

0433

*68

.5*

(5%

)(9

7.5%

)(9

5%)

(2.5

%)

16H

16-H

1739

11

3997

.53.

5074

67*

1521

5.11

5885

*72

.2*

(97.

5%)

(2.5

%)

(2.5

%)

(97.

5%)

17 H

17-H

181

3939

12.

5-5

.692

1*0.

0006

57-5

.115

89*

72.2

*(2

.5%

)(9

7.5%

)(9

7.5%

)(2

.5%

)

18 H

18-H

1940

00

4010

0-5

.375

87*

unde

fined

unde

fined

80*

(100

%)

(0%

)(0

%)

(100

%)

19 H

19-H

2037

33

3792

.55.

4660

82*

152.

1111

5.91

86*

57.8

*(9

2.5%

)(7

.5%

)(7

.5%

)(9

2.5%

)

20 H

20 -H

2137

33

3792

.55.

6921

*15

2.11

115.

9186

*57

.8*

(92.

5%)

(7.5

%)

(7.5

%)

(92.

5%)

21H

21-H

2239

11

392.

5-6

.008

33*

0.00

0657

-5.1

1589

*72

.2*

(97.

5%)

(2.5

%)

(2.5

%)

(97.

5%)

22H

22-H

2337

33

3792

.55.

3758

72*

152.

1111

5.91

86*

57.8

*(9

2.5%

)(7

.5%

)(7

.5%

)(9

2.5%

)

23H

23-H

2437

32

387.

5-5

.375

87*

0.00

4267

-5.7

9497

*61

.3*

(92.

5%)

(7.5

%)

(5%

)(9

5%)

Sour

ce :

Cal

cula

ted.

Figu

res i

n th

e pa

rent

hese

s are

con

ditio

nal p

roba

bilit

ies;

* D

enot

es 5

% le

vel o

f sig

nific

ance

.


107

Tab

le-A

-1.4

: C

onti

ngen

cy T

able

for

Per

form

ance

Per

sist

ence

of

Shar

pe R

atio

(1y

ear

Lag)

Tim

ePe

rcen

tage

Mal

kiel

B &

GS.

No.

Peri

odW

WW

LLW

LLR

epea

tZ

Sco

reC

PR

Z S

core

χχχχ χ2

Win

ner

1Y1

-Y2

337

373

7.5

-5.3

7587

*0.

0065

74-5

.918

6*57

.8*

(7.5

%)

(92.

5%)

(92.

5%)

(7.5

%)

2Y2

-Y3

1525

2515

37.5

-1.5

8114

0.36

-2.2

1194

*5*

(37.

5%)

(62.

5%)

(62.

5%)

(37.

5%)

3Y3

-Y4

1723

2317

42.5

-0.9

4868

0.54

6314

-1.3

3655

1.8

(42.

5%)

(57.

5%)

(57.

5%)

(42.

5%)

4Y4

-Y5

382

238

955.

6921

*36

15.

7397

68*

64.8

*(5

%)

(95%

)(9

5%)

(5%

)

5Y5

-Y6

392

138

95.1

25.

7784

29*

741

5.30

4327

*68

.5*

(97.

5%)

(5%

)(2

.5%

)(9

5%)

6Y6

-Y7

355

535

87.5

4.74

3416

*49

5.75

608*

45*

(87.

5%)

(12.

5%)

(12.

5%)

(87.

5%)

7Y7

-Y8

337

733

82.5

4.11

0961

*22

.224

495.

2697

5*33

.8*

(82.

5%)

(17.

5%)

(17.

5%)

(82.

5%)

8Y8

-Y9

735

335

16.6

6-4

.320

49*

0.03

0303

-5.5

1654

*39

.4*

(17.

5%)

(87.

5%)

(82.

5%)

(12.

5%)

9Y

9-Y

104

3636

410

-5.0

5964

0.01

2346

-5.8

9577

*51

.2*

(10%

)(9

0%)

(90%

)(1

0%)

10Y

10-Y

1136

74

3383

.72

4.42

2459

42.4

2857

5.58

111*

42.5

*(9

0%)

(17.

5%)

(10%

)(8

2.5%

)

11Y

11-Y

1214

2725

1434

.14

-2.0

3026

0.29

037

-1.9

5899

*7.

3*(3

5%)

(67.

5%)

(62.

5%)

(35%

)

Sour

ce :

Cal

cula

ted.

Figu

res i

n th

e pa

rent

hese

s are

con

ditio

nal p

roba

bilit

ies;

* D

enot

es 5

% le

vel o

f sig

nific

ance

.


108

Tab

le-A

-1.5

: C

onti

ngen

cy T

able

for

Per

form

ance

Per

sist

ence

of

Jens

en A

lpha

(6

Mon

ths

Lag)

Tim

ePe

rcen

tage

Mal

kiel

B &

GS.

No.

Peri

odW

WW

LLW

LLR

epea

tZ

Sco

reC

PR

Z S

core

χχχχ χ2

Win

ner

1H

1-H

218

2123

1846

.15

-0.4

8038

0.67

0807

-0.8

8791

0.9

(45%

)(5

2.5%

)(5

7.5%

)(4

5%)

2H

2-H

319

2222

1746

.34

-0.4

6852

0.66

7355

-0.8

9904

0.9

(47.

5%)

(55%

)(5

5%)

(42.

5%)

3H

3-H

420

2120

1948

.78

-0.1

5617

0.90

4762

-0.2

2365

0.1

(50%

)(5

2.5%

)(5

0%)

(47.

5%)

4H

4-H

517

2323

1742

.5-0

.948

680.

5463

14-1

.336

551.

8(4

2.5%

)(5

7.5%

)(5

7.5%

)(4

2.5%

)

5H

5-H

618

2222

1845

-0.6

3246

0.66

9421

-0.8

9293

0.8

(45%

)(5

5%)

(55%

)(4

5%)

6H

6-H

724

1616

2460

1.26

4911

2.25

1.77

6659

3.2

(60%

)(4

0%)

(40%

)(6

0%)

7H

7-H

814

2726

1334

.15

-2.0

3026

*0.

2592

59-2

.853

19*

8.5*

(35%

)(6

7.5%

)(6

5%)

(32.

5%)

8H

8-H

914

2620

2035

-1.8

9737

*0.

5384

62-1

.351

213.

6(3

5%)

(65%

)(5

0%)

(50%

)

9H

9-H

1027

1313

2767

.52.

2135

94*

4.31

3609

3.06

1887

*9.

8*(6

7.5%

)(3

2.5%

)(3

2.5%

)(6

7.5%

)

10H

10-H

1126

1414

2665

2.21

3594

*3.

4489

82.

6409

12*

7.2*

(65%

)(3

5%)

(35%

)(6

5%)

(Con

td...

)


109

11H

11-H

1220

2020

2050

01

00

(50%

)(5

0%)

(50%

)(5

0%)

12 H

12-H

1321

1919

2152

.50.

3162

281.

2216

070.

4470

270.

2(5

2.5%

)(4

7.5%

)(4

7.5%

)(5

2.5%

)

13 H

13-H

1423

1817

2256

.09

0.78

0869

1.65

3595

1.11

5436

1.3

(57.

5%)

(45%

)(4

2.5%

)(5

5%)

14 H

14-H

1521

1919

2152

.50.

3162

281.

2216

070.

4470

270.

2(5

2.5%

)(4

7.5%

)(4

7.5%

)(5

2.5%

)

15 H

15-H

1616

2424

1640

-1.2

6491

0.44

4444

-1.7

7666

3.2

(40%

)(6

0%)

(60%

)(4

0%)

16H

16-H

1722

1917

2253

.66

0.46

8521

1.49

8452

0.89

9041

0.9

(55%

)(4

7.5%

)(4

2.5%

)(5

5%)

17 H

17-H

1817

2323

1742

.5-0

.948

680.

5463

14-1

.343

921.

8(4

2.5%

)(5

5%)

(55%

)(4

2.5%

)

18 H

18-H

1925

1515

2562

.51.

5811

392.

7777

782.

2119

4*5

(62.

5%)

(37.

5%)

(37.

5%)

(62.

5%)

19H

19-H

2031

810

3179

.48

3.68

2948

*12

.012

54.

6201

45*

24.3

*(7

7.5%

)(2

0%)

(25%

)(7

7.5%

)

20H

20-H

2127

1313

2767

.52.

2135

94*

4.31

3609

3.06

1887

*9.

8*(6

7.5%

)(3

2.5%

)(3

2.5%

)(6

7.5%

)

21H

21-H

2228

1211

2970

2.52

9822

*6.

1515

153.

6744

91*

14.5

*

22 H

22-H

2325

1416

2564

.10

1.76

141*

2.79

0179

2.21

8491

*5.

1*

23 H

23-H

2418

2322

1743

.90

-0.7

8087

0.60

4743

-1.1

1544

1.3

Sour

ce :

Cal

cula

ted.

Figu

res i

n th

e pa

rent

hese

s are

con

ditio

nal p

roba

bilit

ies;

* D

enot

es 5

% le

vel o

f sig

nific

ance

.


110

Tab

le-A

-1.6

: C

onti

ngen

cy T

able

for

Per

form

ance

Per

sist

ence

of J

ense

n A

lpha

(1

year

lag)

Tim

ePe

rcen

tage

Mal

kiel

B &

GS.

No.

Peri

odW

WW

LLW

LLR

epea

tZ

Sco

reC

PR

Z S

core

χχχχ χ2

Win

ner

1Y1

-Y2

1922

2019

46.3

4-0

.468

520.

8204

55-0

.441

720.

3(4

8%)

(55%

)(5

0%)

(48%

)

2Y2

-Y3

2317

1723

57.5

0.94

8683

1.83

045

1.33

6546

1.8

(58%

)(4

3%)

(43%

)(5

8%)

3Y3

-Y4

1823

2019

43.9

0-0

.780

870.

7434

78-0

.660

060.

7(4

5%)

(58%

)(5

0%)

(48%

)

4Y4

-Y5

2316

1724

58.9

71.

1208

972.

0294

121.

5575

812.

5(5

9%)

(41%

)(4

1%)

(59%

)

5Y5

-Y6

2218

1822

550.

6324

561.

4938

270.

8929

280.

8(5

6%)

(46%

)(4

4%)

(54%

)

6Y6

-Y7

1921

2020

47.5

-0.3

1623

0.90

4762

-0.2

2365

0.1

(49%

)(5

4%)

(49%

)(4

9%)

7Y7

-Y8

1921

2020

48.7

1-0

.160

130.

9047

62-0

.223

650.

1(4

9%)

(51%

)(5

1%)

(49%

)

8Y8

-Y9

2020

2020

500

0.90

4762

-0.2

2365

0(5

0%)

(50%

)(5

0%)

(50%

)

9Y

9-Y

1020

2220

1847

.62

-0.3

0861

10

0.4

(50%

)(5

5%)

(50%

)(4

5%)

10Y

10-Y

1127

1514

2464

.29

1.85

164

3.08

5714

*2.

4199

8*6.

3*(6

6%)

(37%

)(3

6%)

(62%

)

11Y

11-Y

1226

1414

2665

1.89

7367

3.44

898*

2.64

0912

*7.

2*(6

3%)

(34%

)(3

6%)

(67%

)

Sour

ce :

Cal

cula

ted.

Figu

res i

n th

e pa

rent

hese

s are

con

ditio

nal p

roba

bilit

ies;

* D

enot

es 5

% le

vel o

f sig

nific

ance

.


111

Tab

le-A

- 1.

7 : R

esul

ts o

f Sp

earm

an R

ank

Cor

rela

tion

Coe

ffic

ient

Tes

t (6

Mon

ths

Lag)

S.N

o.T

ime P

erio

dB

AR

Shar

pe R

atio

Jens

en A

lpha

1H

1-H

2Z

=0.0

2742

3028

5837

525

Z=0

.039

3147

8044

7970

6Z

=0.0

0198

5914

6493

1386

2H

2-H

3Z

=-0.

0184

4025

9266

9357

Z=-

0.00

6108

0721

4848

728

Z=-

0.00

7698

3862

7004

937

3H

3-H

4Z

=0.0

0341

7988

5597

7527

Z=-

0.02

0167

7149

0478

5Z

=-0.

0083

0761

1082

7871

1

4H

4-H

5Z

=-0.

0253

5008

1818

3332

Z=0

.018

8912

4386

8572

7Z

=-0.

0210

0111

3349

9154

5H

5-H

6Z

=-0.

0205

8441

4127

3502

Z=0

.017

8442

2113

8456

4Z

=-0.

0180

2356

0044

3705

6H

6-H

7Z

=0.0

3284

8030

4876

55Z

=0.0

6999

4920

0435

459

Z=0

.031

0572

7876

5365

3

7H

7-H

8Z

=-0.

0111

4011

0861

4897

Z=-

0.07

7039

2467

7437

9Z

=-0.

0429

3320

6608

4734

8H

8-H

9Z

-0.0

4652

2622

0636

077

Z=-

0.08

8896

7132

5952

53Z

=-0

.062

7527

9304

8836

9

9H

9-H

10Z

=0.0

5781

3061

1256

432

Z=0

.089

3846

2057

7085

8Z

=0.0

6195

8954

6564

817

10H

10-H

11Z

=0.0

3978

4226

4075

693

Z=0

.089

4874

7671

4301

3Z

=0.0

5008

0389

4765

22

11H

11-H

12Z

=0.0

2382

3063

7812

114

Z=0

.079

2308

7369

8123

8Z

=0.0

1810

5317

4867

725

12 H

12-H

13Z

=0.0

0542

7639

2407

5423

Z=0

.066

6006

6751

5435

7Z

=0.0

1067

5939

5755

943

13 H

13-H

14Z

=0.0

1619

0610

9324

54Z

=0.0

8191

3045

2762

808

Z=0

.027

1065

4816

1551

14 H

14-H

15Z

=0.0

2841

7304

5768

352

Z=0

.083

2686

3641

8043

5Z

=0.0

1986

7058

5036

937

(Con

td...

)


112

15 H

15-H

16Z

=-0.

0638

9475

9905

6136

Z=-

0.09

0814

0571

5069

55Z

=-0.

0428

2243

8460

7029

16H

16-H

17Z

=0.0

4049

1032

6838

191

Z=0

.083

6906

1031

4312

Z=0

.022

4621

9796

5745

3

17 H

17-H

18Z

=-0.

0017

5910

3680

0695

2Z

=-0.

0900

5714

1474

2638

Z=-

0.01

7361

5884

9459

92

18 H

18-H

19Z

=-0.

0114

6186

5957

3945

Z=0

.084

9802

6803

4782

8Z

=0.0

3963

6535

5438

753

19H

19-H

20Z

=0.0

5683

1971

8168

188

Z=0

.085

5921

3018

4372

2Z

=0.0

7762

7372

8923

033

20H

20-H

21Z

=0.0

0869

2662

2631

3214

Z=0

.075

9658

5067

5745

9Z

=0.0

3915

6540

2368

699

21H

21-H

22Z

=-0.

0204

4463

5274

2113

Z=-

0.07

3668

7302

7793

4Z

=0.0

3689

1067

8812

781

22 H

22-H

23Z

=0.0

6673

2534

3580

197

Z=0

.085

5710

3148

9558

8Z

=0.0

3899

5662

6889

175

23 H

23-H

24Z

=-0.

0671

7033

2275

3983

Z=-

0.08

3904

2345

9929

8Z

=-0.

0188

8069

4521

166

Sour

ce :

Cal

cula

ted.

BA

R :

Ben

chm

ark

Adj

uste

d R

etur

n.


113

Tab

le-A

-1.8

: R

esul

ts o

f Sp

earm

an R

ank

Cor

rela

tion

Coe

ffic

ient

Tes

t (1

yea

r La

g)

S.N

o.T

ime P

erio

dB

AR

Shar

pe R

atio

Jens

en A

lpha

1Y1

-Y2

Z=-

0.00

3003

9266

7406

175

Z=-

0.05

5347

1511

6932

38Z

=-0.

0174

3807

1263

2979

2Y2

-Y3

Z=-

0.00

1400

4258

6824

125

Z=-

0.02

2103

5201

5391

7Z

=0.0

1426

7992

3675

804

3Y3

-Y4

Z=0

.014

7084

2762

1810

7Z

=-0.

0080

9398

6797

8011

5Z

=-0.

0022

2854

9639

6682

9

4Y4

-Y5

Z=0

.024

8806

3585

8734

4Z

=0.0

7854

2528

7798

357

Z=0

.039

3965

3789

0372

6

5Y5

-Y6

Z=0

.031

6480

4222

0141

3Z

=0.0

7436

7624

5436

287

Z=0

.040

9472

9195

9159

5

6Y6

-Y7

Z=0

.030

0761

8945

6541

Z=0

.071

4533

6732

2524

1Z

=0.0

2238

5715

1970

466

7Y7

-Y8

Z=0

.000

1345

0417

9435

599

Z=0

.050

1885

2028

7440

8Z

=0.0

1088

1651

8500

253

8Y8

-Y9

Z=-

0.01

4631

9448

5311

2Z

=-0.

0456

5757

5576

2572

Z=-

0.00

2603

0514

7260

663

9Y

9-Y

10Z

=-0.

0378

8798

1211

2125

Z=-

0.06

8813

3931

3399

39Z

=-0.

0202

0727

4957

5602

10Y

10-Y

11Z

=0.0

1961

6511

5027

843

Z=0

.067

6767

0095

0920

5Z

=0.0

4448

6598

0141

12

11Y

11-Y

12z=

0.02

2894

7212

0942

06Z

=-0.

0085

2123

5367

7730

6z=

0.02

4959

7559

6428

48

Sour

ce :

Cal

cula

ted.

BA

R :

Ben

chm

ark

Adj

uste

d R

etur

n


114

Table-A-1.8 : Sample Mutual Fund Schemes

S. No. Name of the Scheme Aim

1. Baroda Pioneer ELSS 96 TP

2. Birla Sun Life 95 G

3. Birla Sun Life Advantage Fund G

4. Birla Sun Life buy India Fund G

5. Birla Sun Life Equity Fund G

6. Birla Sun Life India Opportunities Fund G

7. Birla Sun Life MNC Fund G

8. Birla Sun Life New Millennium G

9. CanaraRobeco Balance B

10. DSP BlackRock Balanced Fund B

11. DSP BlackRock Opportunities Fund G

12. DSP BlackRock Technology.com Fund G

13. Escorts Tax Plan TP

14. Franklin India Bluechip G

15. Franklin India Opportunity Fund G

16. Franklin India Prima Fund G

17. Franklin India Prima Plus G

18. Franklin India Taxshield TP

19. Franklin Infotech Fund G

20. FT India Balanced Fund B

21. HDFC Balanced Fund B

22. HDFC Capital Builder Fund G

23. HDFC Equity Fund G

24. HDFC Growth Fund G

25. HDFC Prudence Fund B

26. HDFC Taxsaver TP

(Contd...)


115


27. HDFC Top 200 G

28. ICICI Prudential Balanced B

29. ICICI Prudential FMCG G

30. ICICI Prudential Tax Plan TP

31. ICICI Prudential Top 100 Fund G

32. ICICI Prudential Top 200 Fund G

33. ICICI Prudential Technology Fund G

34. ING Balanced Fund B

35. ING Core Equity Fund G

36. JM Balanced B

37. JM Basic Fund G

38. JM Equity G

39. Kotak 50 G

40. Kotak Balance B

41. L&T Opportunities Fund G

42. LIC Nomura Equity Fund G

43. LIC Nomura MF Growth Fund G

44. LIC Nomura Tax Plan TP

45. PRINCIPAL Balanced Fund B

46. PRINCIPAL Index Fund G

47. PRINCIPAL Growth Fund G

48. Reliance Growth G

49. Reliance Vision G

50. Sahara Taxgain TP

51. SBI Magnum Balanced Fund B

52. SBI Magnum Equity Fund G

53. SBI Magnum Global Fund 94 G

54. SBI Magnum Multiplier Plus 93 G

(Contd...)


116


55. SBI Magnum Sector Funds Umbrella – Contra G

56. SBI Magnum Sector Funds Umbrella – Pharma G

57. SBI Magnum Tax Gain Scheme 93 TP

58. Sundaram Balanced Fund B

59. Sundaram Growth Fund G

60. SundaramTaxsaver TP

61. Tata Balanced Fund B

62. Tata Ethical Fund G

63. Tata Life Sciences and Technology Fund G

64. Tata Pure Equity Fund G

65. Tata Tax Saving Fund T P

66. Taurus Bonanza Fund G

67. Taurus Discovery Fund G

68. Taurus Starshare Fund G

69. Taurus Taxshield TP

70. Templeton India Growth Fund G

71. UTI Balanced Fund B

72. UTI Energy Fund G

73. UTI Equity Fund G

74. UTI Equity Tax Savings Plan TP

75. UTI Masterplus Unit Scheme 91 G

76. UTI MNC Fund G

77. UTI Pharma and Healthcare Fund G

78. UTI Nifty Fund G

79. UTI Top 100 Fund G

80. UTI Services Industries Fund G

G-Growth, B-Balanced, TP- Tax Planning

INSTITUTE OF PUBLIC ENTERPRISE

Subscription FormToThe Publications Cell,Institute of Public Enterprise,Osmania University Campus,Hyderabad - 500 007, Telangana, India.

Dear Sir/Madam,I/We wish to subscribe to the Journal of Institute of Public Enterprise (JIPE)

Please find enclosed herewith a Demand Draft/Cheque, bearing No.______________

drawn on Bank _________________________ dated __________ for /$ _________

(in words) _________________________________________________________

drawn in favour of “Institute of Public Enterprise” payable at Hyderabad. Please

effect the subscription from next issue onwards.

Journal Dispatch Address :

Name of the Individual / Organization ______________________________________

Address ___________________________________________________________

City_____________Pin_________Country____________Ph : No.______________

Mobile No.________________e-mail Address______________________________

Signature Office Seal

Annual Subscription Rates*

For IndiaFor OtherCountries

S.Individuals Academic/ Corporates Foreign

No. Journal Title Libraries Subscribers

( ) ( ) ( ) (US$)

1 The Journal of Institute of Public Enterprise 1,500/- 2,000/- 3,000/- 200

✂✂

✂Attention SubscribersAttention SubscribersAttention SubscribersAttention SubscribersAttention Subscribers

We invite attention to the subscribers of the The Journal of Institute of PublicThe Journal of Institute of PublicThe Journal of Institute of PublicThe Journal of Institute of PublicThe Journal of Institute of PublicEnterprise’ (JIPE)Enterprise’ (JIPE)Enterprise’ (JIPE)Enterprise’ (JIPE)Enterprise’ (JIPE) that owing to the price escalation in the printing, paper and postage cost, theannual subscription rates of the JIPE have been revised with effect from 01-Jan-2015. In keepingwith the exemplary tradition of sustaining this scholarly journal and consistent with contemporarybest practice, we hereby assure the subscribers and readers of the JIPE that the Editorial Team,in tune with the revised rates, will continue to provide in its publication increased coverage ofquality articles enhancing the number of pages in terms of value addition for the professors,policy-makers, practitioners, researchers, teachers and scholars as a must read journal.

Prof.K.TrivikramProf.K.TrivikramProf.K.TrivikramProf.K.TrivikramProf.K.TrivikramEditor

***** With effect from 01-01-2015.

Vol : 42 January – June, 2019 No : 1Content

Research Papers★ Effect of Work/Family Demand, Work/Family Conflict on Job

Satisfaction : A Case of SAIL Women EmployeesAnuradha Nayak & Mrinalini Pandey

★ An Analysis of Households’ Access to Toilet Facility in RuralBirbhum and it’s Implications for Public PolicySubhasish Das, Soumyadip Chattopadhyay & Raktimava Bose

★ Testing of Efficient Market Hypothesis in Indian Emerging Marketwith Reference to the Public Sector EnterprisesNenavath Sreenu

★ E-governance Best Practices in Higher Education Institutions inTelangana StateT.Papi Reddy, V.Venkata Ramana, Rajesh Ittamalla &Mohan Venkatesh Palani

★ Working of State Universities in Punjab : An Exploratory ReportMeenakshi Sharma & Anjali Mehra

Perspectives★ Contribution of Education to Income in Service Sector of Odisha :

A Gender PerspectiveAnanya Mitra & Himanshu Sekhar Rout

★ Corporate Social Responsibility (CSR) Reporting Practices ofIndian CompaniesSushanta Kumar Nandi, Meena Sunar & Kago Taping

The Journal of Institute of Public Enterprise

Vol : 42 July – December, 2019 No : 2...Vol : 42 July – December, 2019 No : 2 Content Research...

Documents

Transcript of Vol : 42 July – December, 2019 No : 2...Vol : 42 July – December, 2019 No : 2 Content Research...