journal of studies - centre for banking studies · college of asia and the pacific ... cbs journal...

The official journal of theCentre for Banking StudiesCentral Bank of Sri Lanka

Journal ofMultidisciplinary

Studies

CBS Journal of Multidisciplinary Studies

Advisory Editor

Mr. K G D D Dheerasinghe Deputy Governor

Central Bank of Sri Lanka

Managing Editor Mr. U P Alawattage

Director Centre for Banking Studies Central Bank of Sri Lanka

Editorial Board

Dr. D S Wijesinghe Former Deputy Governor

Central Bank of Sri Lanka Board Member

HDFC Bank, Sri Lanka Board Member FITCH Rating (Lanka) Ltd.

Dr. P N Weerasinghe Assistant Governor Central Bank of Sri Lanka and Alternate Executive Director for Bangladesh, Bhutan, India and Sri Lanka – International Monetary Fund

Prof. Danture Wickramasinghe Hull University Business School

The University of Hull, UK

Prof. Lalith P Samarakoon University of St. Thomas Minnesota, USA

Prof. Uditha Liyanage Director

Postgraduate Institute of Management University of Sri Jayewardenepura

Sri Lanka

Dr. D B P H Dissa Bandara University of Sri Jayewardenepura Sri Lanka Director, Financial Services Academy Securities & Exchange Commission of Sri Lanka

Prof. Premachandra Athukorale Professor of Economics

College of Asia and the Pacific Australian National University

Australia

Mr. Deepal Sooriyaarachchi Management Consultant Director, AVIVA- NDB Insurance Co. PLC and Sampath Bank PLC Sri Lanka

Mr. B D W A Silva Assistant Governor

Central Bank of Sri Lanka

Mr. K D Ranasinghe Director Economic Research and Chief Economist Central Bank of Sri Lanka

Prof. Peter Sinclair Professor of Economics

University of Birmingham, UK

Mr. R M B Senanayake General Manager Research & Development, SKM Lanka Holdings and Monetary Policy Committee Member Central Bank of Sri Lanka

Prof. Thankom Arun Gopinath Director, Institute of Global Finance and Development Professor of Development Finance and Public Policy,

Lancashire Business School The University of Central Lancashire

Honorary Senior Fellow, School of Environment and Development Brooks World Poverty institute

The University of Manchester and Research Fellow, IZA

The CBS Journal of Multidisciplinary Studies encourages multidisciplinary research which has a policy relevance. This is a peer reviewed journal published annually. Views expressed in the CBS Journal of Multidisciplinary Studies are not necessarily those of the Centre for Banking Studies (CBS) of the Central Bank of Sri Lanka. Material may be reprinted or abstracted if the journal and authors are credited. Free electronic copies are available from the CBS website: http//www.cbscbsl.lk. Printed copies are available at Centre for Banking Studies, No. 58, Sri Jayewardenepura Mawatha, Rajagiriya, Sri Lanka. Contact numbers: (9411) 2477811, (9411) 2477840, (9411) 2477821 & (9411) 2477810.

VOL. 1 NO. 1 December 2010 ISSN: 2012 – 791X


CONTENTS

PAGE

Editorial Note i - iv

1. International Stock Market Price Linkages: 1 - 16

Evidence from Sri Lanka and its Main Trading Partners S. Sivarajasingham

2. Factors Affecting the Adoption of Agricultural Forward Sales 17 - 24

Contracts in Sri Lanka K S Karunagoda, K P N S Karunagoda, H P Gunawardhane, S P R P Senanayake and Y M Wickramasinghe

3. An Empirical Examination of Informational Content of the Dividend 25 - 40

Announcements in Sri Lankan Share Market D B P H Dissa Bandara and K D I Perera

4. Interest Rate Behaviour: The Sri Lankan Case 41 – 53 W M Hemachandra

5. Multi-Elements of Exchange Rate Exposure: Evidence from 54 - 72

Japanese Sectoral Returns Prabhath Jayasinghe and Albert K Tsui

6. HR Managers of South Asian Banks: A Study of Strategic Orientation 73 – 88 Ajantha S Dharmasiri

7. Book Review: 89 - 92

The Cost of Capitalism: Understanding Market Mayhem

and Stabilising Our Economic Future P W D N R Rodrigo


i

Editorial Note

The Centre for Banking Studies (CBS) of the Central Bank of Sri Lanka for more than two decades

now, has been in the forefront of disseminating global wisdom and modern knowledge to bankers in

particular and the public in general, earning an unparalleled reputation for its knowledge

enhancement devices. Its modus operandi in this venture had principally been the use of face to face

training and learning techniques: seminars, conferences, workshops and public debates. The impact

which it has made on human capital development and organizational leadership has widely been

acknowledged. Now, we have come to the stage that we need to take our programs to the next

logical step by evolving ourselves into a centre of creation and dissemination of knowledge through

multidisciplinary research. This is considered to be crucial in assessing and understanding how far

and well we can adapt and adopt the euro-centric global knowledge to suit our idiosyncratic cultural

and economic conditions in order to achieve our development needs and goals.

CBS Journal of Multidisciplinary Studies is the result of this vision. It is based on the understanding

that economic, sociological and management research has been too compartmentalized to the extent

that we no longer see the holistic picture within which economic, sociological and organizational

dynamics evolve and are to be managed. The implication of this compartmentalization is sometimes

rather embarrassing that our collective understanding of the organizational and economic realities is

very akin to the fable of “elephant and the blind men”. However, this is not something for which we

should blame ourselves. It has been our nature over the last two centuries or so; it has been the

historical result of the way that knowledge has been institutionalized in the West and then

superimposed upon the peripheral East and South. This is very evident if we take a brief look at how

‘disciplines’ have historically been branched out to various compartments within which ‘modern’1

specializations began to take place.

Historically, all esoteric knowledge about human, social and physical behaviour was viewed as

indivisible unit called as ‘philosophy’. Over the time, this single skein of knowledge divided into two

basic strands called natural and moral philosophy. This had been the case in pre-modern institutions

of learning, especially medieval Western universities. Subsequently, natural philosophy was

transformed and re-labeled into natural science with the advent and authority of physical sciences. In

1 I use the term ‘modern’ here (and also the term modernity later on) with its sociological connotations rather

than its common usage to mean current knowledge. In that sociological sense, modern (in contrast to pre-

modern) refers to the historical phase after the Enlightenment and French Revolution, a period within which, in

the West, people are said to be emancipated from religious dogmas and the state is said to be separated from

the divine powers of the Church. More importantly, this is the period within which secularity was

institutionalised as an overriding feature of human knowledge, and secular institutions of knowledge and

sciences (such as universities and professional bodies) are established as the principal mode of knowledge

creation and dissemination.


ii

pursuit of this trend and prestige of science, moral philosophy was also re-labeled as moral science

which deals with the knowledge of human relationships (Easton and Schelling, 1991).

However, the detailed division and compartmentalization of knowledge into various disciplines (such

as economics, sociology, anthropology, management and then into further subdivisions such as

marketing, human resource management, operations and production, accounting and finance and so

on) was the result of institutional evolution of capitalism and modernity2 in the West. For Foucault

(1994), this is the genealogy of power, especially the penetration of 'pastoral power' into the secular

institutions. By the term 'pastoral power' what Foucault meant was the special form of power that

traditionally Christian pastors were exercising by virtue of their religious qualities. This power was

exercised through institutionalized rituals of confession and designated a very special form of power

which, through its penetration into secular institutions, had a far reaching effect on the division and

compartmentalization of knowledge.

The obvious question that will arise in this respect would be, how this pastoral power, which is rather

an ecclesiastical notion of power, got to do with economics and management, especially when the

ecclesiastical exercise of that power is no more a vital part of modern societies. According to

Foucault (1994), it is this pastoral power that has spread and multiplied outside the ecclesiastical

institutions but within various forms of secular institutions such as schools, universities, professional

bodies, hospitals, factories, government departments and, after all, economic enterprises such as

corporations (note that all these institutions are products of the modernity and capitalism, and was not

in existence before that). Within such secular organizations, in contrast to the Church, the pastoral

power now assumes a change in its objective: not the salvation in the next world but, rather within this

world. As such the notion of 'salvation' takes on different forms: development, poverty reduction,

employment, health and wellbeing, safety and security, education, customer satisfaction, employee

happiness and even profit maximization.

As the objective of 'pastoral power' is secularized the officials holding that pastoral power has been

multiplied. Instead of pastors in the Church, now we have doctors, lawyers, accountants, economists,

engineers, and managers whose focus is now on the development, dissemination and application of

knowledge for betterment of human lives within various organizations. For a Christian pastor,

knowing the inner soul of the confessant is the key to direct him/her to the salvation in the next world.

For these new pastors, knowing the sciences and laws of human body, human mind, society,

economy, markets, politics, culture and so on is the key to the salvation within this secular institutional

set up. As such, "this implies that power of a pastoral type, which over centuries has been linked to a

defined religious institution, suddenly spread out into the whole social network and It found support in

a multitude of institutions" (Foucault, 1994, p.335). Thus we have replaced the Church with various

2 See note 1 above.


iii

economic and social institutions and also economic enterprises as the institutional framework within

which these new modes of pastoral power are exercised. And more importantly, development,

dissemination and deployment of human knowledge have been compartmentalized into various

'faculties of knowledge' within which we produce esoteric knowledge, owners of which can qualify

themselves as 'modern pastors' - elites who are empowered by the ownership of that knowledge to

direct and lead others for collective and individual 'salvation' but only within their respective field of

specialization.

So was the story of division of knowledge. These sub disciplines are considered to be invaluable and

undoubtedly have contributed a lot by way of developing precise knowledge, concepts, theories and

skills for the purpose of systematic inquiry to improve our understanding the world around us.

However, it is now becoming increasingly apparent that the intensity of specialization and

fragmentation of knowledge is so rapid and scholars (i.e. modern pastors) have difficulty in

maintaining of such divided knowledge within their own small niches. Rather there is a substantial

overlapping of knowledge between disciplines and when it comes to application of knowledge it

demands a holistic approach for effective results. Though still it is debatable whether heavy

specialization is conducive towards creation of knowledge and their application over generalization,

the pendulum has started swinging from the notion of heavy specialization towards convergence of

multiple disciplines, particularly in the area of policy studies. Therefore, multi-disciplinary research for

that matter has gained core attention in the recent past in the area of knowledge creation and

research in particular. In other words, the world is now seeking a proper balance and coordination

between these multitude divisions of human knowledge.

CBS Journal of Multidisciplinary Studies is such an attempt. Its primary objective is to promote

multidisciplinary research on economy, society and organizations, particularly of LDCs so that we will

have a broader understanding of social, economic and organizational conditions within which various

policy proposals have to be implemented. However, we also understand that we have been the

product of specializations and compartmentalization mentioned above, and hence it could be too

ambitious to seek broader multidisciplinary convergence within individual papers. We know that many

of us still working within our own knowledge compartments find our comfort within them. Hence, our

strategy of multi-disciplining the journal is to collect a set of papers addressing multiple issues,

adopting multiple theoretical and methodological perspectives (the way we see things) and

approaches (the way we do things). As such the scope of the journal is very broad and would

accommodate theoretical and empirical studies with no particular preference to any selected

methodological stance. However, CBS Journal of Multi-disciplinary Studies encourages policy

oriented action research to support policy discussions in macro-economic, wider social and micro

organizational issues. This is encouraged because, it is recognised, that such policy issues will

provide the necessary basis upon which we can bring hitherto compartmentalized knowledge

together.


iv

Although the journal encourages local issues particularly, and LDC issues generally its operational

apparatuses are international by nature. It has established an international network of academics and

professionals for editorial and review activities. I am proud to mention that this network of reviewers

covers such a broader spectrum of academic disciplines and methodological stances that enables the

journal to effectively project itself as truly multidisciplinary.

Udeani P. Alawattage Managing Editor

References Easton, D. and Schelling, Corinne S. (1991), "Divided Knowledge: Across Disciplines, Across Cultures, London: Sage Publications. Foucault, M. (1994), Power: essential works of Foucault 1954-1984, volume 3, (Translated by Robert Hurley and others, Edited by James D Faubion), London: Penguin Books.

1

International Stock Market

Price Linkages:

Evidence from Sri Lanka

and its Main Trading Partners

S. Sivarajasingham1

Abstract This paper investigates long term linkages among stock markets in six developed countries and five emerging-market economies in South-East Asia using daily closing stock markets price indices for the period of November 2, 1987 to December 1, 2006. Multivariate cointegration tests and other related statistical techniques were applied while paying attention to a possible structural break resulting from the 1997-98 Asian financial crisis. The results suggest that the Asian stock markets have become more interdependent and internationally integrated after the Asian financial crisis. The Sri Lankan stock market had statistically significant long run relationships with the stock markets of India, US, and UK in the pre Asian crisis period and with Hong Kong, India, South Korea and Singapore in the post-crisis period. The main policy inference is, therefore, that the authorities of Sri Lanka and other Asian countries need to coordinate macroeconomic policies with each other to avoid any spillover effects following any turmoil in one country or more countries of the region.

Key words: Stock market linkages, Multivariate Cointegration, Vector Error correction Model,

1. Senior Lecturer, Department of Economics and Statistics, University of Peradeniya.

Mr. Sivarajasingham acknowledge the contribution of Prof Jan G.De.Goojer, Prof H.P. Boswijk and Prof J.F. Kiviet, University of Amsterdam, Prof R.O Thattil, University of Peradeniya for their comments, encouragement and support.


2

1. Introduction:

Globalization and financial market liberalization have brought international stock markets to move together over the years. The Linkages among stock markets are important not only for international capital market integration but also for the integration of the goods and services markets to be effective. The empirical analysis of linkages among international stock markets is important in terms of theory, practice and policy; theoretically as an input to international portfolio diversification models, practically for international investors to know international diversification benefits, and from policy point of view for informing financial market regulators about the extent of market linkages. The issue of whether the stock markets are cointegrated carries important practical implications for portfolio diversifications. (Grubel, 1968). If stock prices are cointegrated, it implies that there is a common force, such as arbitrage activity, which brings stock markets together in the long run, and hence the gains from international asset diversification is limited. By contrast, if markets are not cointegrated, investors can obtain long run gains through international portfolio diversification (Masih and Masih, 1997). The main objective of this study is to investigate the linkages among the stock market of Sri Lanka and that of its main trading partners in the long run. It is also attempted to see whether Asian financial crisis has any influence on the degree of linkages among the selected stock markets. Currently, there exists no in-depth analysis of the linkages and interdependence structure of Sri Lankan stock market with that of its main trading partners. This paper intends to fill this gap. It Investigates long term linkages of the Sri Lankan stock market with stock markets in six developed countries and five emerging-market economies in South-East Asia using daily closing stock markets price indices for the period of November 2, 1987 to December 1, 2006. Multivariate cointegration tests and other related statistical techniques are applied while paying attention to a possible structural break resulting from the 1997-98 Asian financial crisis. If a common stochastic trend exists among stock markets, stock markets are cointegrated. If the stock markets are segmented, investors can gain potential benefits by portfolio diversification. The paper is organized in five sections. Section 2 surveys the related empirical literature in order to provide the context for the ensuing empirical analysis. The econometric methodology is discussed in Section 3. Data sources and variable construction are discussed in Section 4. Results are presented and discussed in Section 5. The paper ends in Section 6 with a summary of the key findings and policy inferences.

2. Literature Review

The available empirical evidence on international stock market linkages has shown mixed and conflicting results. Some studies conclude that international stock markets are interdependent. Example: Eun and Shim (1989), Von Furstenberg and Joen (1989), Bertera and Mayer (1990), Kasa (1992), Cheung and Mak (1992), Arshanapalli and Doukas (1993), and Masih and Masih (1997a, 1997b, 1999). While, others have found evidence of market segmentation. Dwyer and Hafer (1988), Chowdhury (1994), Kwan et al (1995), Elyasiani, Perera and Puri (1998), report that the stock markets are not interdependent. The findings of these studies are not strictly comparable as they were conducted on different stock market indices over various sample periods, and using different methodologies. There is a large empirical literature on the interdependence among stock markets in developed countries. In recent years, there has also been a sizeable literature on the interdependence of emerging stock markets and developed stock markets. However, so far only one study has dealt with this issue relating to the Sri Lankan stock market (Elyasiani, Perera, and Puri 1998). This study investigated the interdependence between Sri Lankan stock market and its main trading partners- Hong Kong, India, Japan, South Korea, Singapore, Taiwan and US by applying the VAR model to daily data for the period of 1st January 1989 to 10th June 1994. This study differs from Elyasiani, Perera, and Puri (1998) in several respects. It investigates the linkages separately for the periods before and after the 1997-98 Asian financial crisis covering the period from November 2, 1987 to December 1,

International Stock Market Price Linkages: Evidence from Sri Lanka and its Main Trading Partners

3

2006. The study includes six developed markets and five emerging Asian markets. It employs Engle-Granger bivariate cointegration test, and Johansen ‘s multivariate vector error correction methods which are preferable to VAR model in testing long-run economic relationships.

3. Analytical frame work

Cointegration is referred to as an economic concept of a long run equilibrium relationship. Let

tY be a nx1 vector of variables, multivariate process with components that are integrated of

order one. The components of tY are cointegrated if and only if there exists a non

degenerate linear combination tY with 0 , such that tY is stationary. i.e

)0(~ IYt . This stationary linear combination is called the cointegrating equation. When

0tY , variables are in long-run equilibrium. The deviation from long-run equilibrium is

called equilibrium error. It is noted as tt Yz . If the error process is stationary, then the

coefficients in can be interpreted as reflective of long-run (steady state) relationship (Engle

and Granger, 1987).

Some normalization assumption is required to uniquely identify in tY . A typical

normalization is ),........- ,- ,1( 32 n so that

)0(~)........( 221 IYYYY tnttt . The cointegrating error )0(~ Iut . In the long

run equilibrium, tu =0 and the long run equilibrium relationship is

ntntt YYY ........221

A variety of econometric methodologies are employed in this study in order to achieve robustness of the results. Multivariate cointegration method is employed to examine the long run relationships among the stock markets. The statistical notion of cointegration is well suited to study the co-movements of a set of variables in the long run. First, preliminary diagnostic tests are done by using summary statistics, contemporaneous correlations, autocorrelations and unit root analysis. Then, multivariate cointegration analysis is done to examine the dynamic linkages among those stock markets. This study uses the system based test of the Johansen procedure.

3.1 Statistical tests used in the study

a) Unit root test was performed to check whether the time series are stationary or not. The standard unit root tests used were ; augmented Dickey- Fuller test(Dickey and Fuller, 1979, 1981) and the Phillips-Perron test (Phillips and Perron 1988).

b). Multivariate cointegration test (Johansen’s Procedure) proposed by Johansen (1988, 1991) and Johansen and Jusellius (1990) was used to investigate the dynamic co-movement among stock prices and the adjustment process towards long term equilibrium. Johansen analysis is expressed in vector error correction model VECM. The Standard linear vector error correction representation of the VAR model for the n variables has the form as

1

1

1

p

i

tititt YYY

Where 1

t

Y is the error correction term. This measures the deviation from long run

equilibrium and induces a correction by its effect on t

Y towards equilibrium. is called

long run response matrix. It measures the cumulative long run effects.


4

Johansen (1989, 1990) has shown that the coefficient matrix contains sufficient information to

determine the cointegrating relationships between the variables. The rank of matrix is the

number of cointegrating relationships between the variables in Yt. Let the rank of be r. The

rank of a matrix is equal to the number of its characteristics roots or eigenvalues ( )( i The

number r can be more than one and can go up to n-1. On the basis of rank of , three distinct cases can be identified.

Case I: When the rank of =0, r=0, there are no cointegrating relationships. The Vector

auto-regression (VAR) polynomial )(z contains n unit roots, n stochastic trends. In which

case, all of the variables are non-stationary. There are no stationary long run relations among the elements of Yt. In this case the variables do not have common stochastic trend and hence do not move together over time. Therefore, it is not possible to obtain stationary cointegration relations between the levels of the variables.

Case II: when has full rank, r=n, all of the variables in levels are stationary and standard

inference applies. )(z has its roots outside the unit circle. There is no any stochastic

trends. All variables are I(0) , the issue of cointegration is not relevant.

Case III: When is less than full rank but not equal to zero. i.e 0 < r < n , intermediate rank. There are r linear combinations of the non-stationary variables that are stationary. i.e. there

are r cointegrating relationships. the polynomial )(z has n-r units roots/ common

stochastic trends. This indicates that 1 tY is I(0). In this case can be written as

.

3.2 Full system of estimation

The system,

1

1

1

p

i

tititt YYY , can be estimated by full

system Maximum Likelihood assuming that ),0(~ Nt and iid. Let the parameter i be

collected in a matrix ],.......[ 11 n and the variables itY in

],.........[ 11 nttt YYZ such that tttt ZYY 1 . The likelihood

function is then

tttt

TTn YZYL )(2

1exp||)2(),,,( 1

2/)2/( -------------

(1)

With .

In Johansen’s procedure, Gaussian maximum likelihood is used to estimate all parameters of the ECM. This procedure is an asymptotically nuisance parameter free. This approach allows

us to obtain estimates of the coefficients of the matrices and through a procedure

known as reduced rank regression. The estimated parameters on the ECM can be partitioned to provide information on the long run relationship and short run dynamics. The

long run relationship can be identified through testing hypothesis on . The short run

dynamics can be identified through testing hypothesis in


5

3.3 Optimal lag:

Boswijk and Franses (1992) emphasized the sensitivity of the cointegration test to the choice of the lag length. They found that underestimating the lag length can lead to the rejection of the null hypothesis of no cointegration too often, where as over parameterization of the dynamic structure leads to loss of power. Therefore, the choice of optimal lags for the VAR system is selected based on the most widely used the Akaike Information Criterion (AIC). The information criteria are computed using the full system loglikelihood. Loglikelihood value is computed assuming a multivariate normal distribution. The Akaike Information Criterion is

given by )/(2)/(2 TkTlAIC where l = the value of the log likelihood function, k=total

number of estimated parameters in the VAR. T= number of observations. By minimizing AIC , the appropriate lag lengths were selected.

3.4 Determining the Number of Cointegration Relations (rank of )

According to Johansen (1988) and Johansen and Jusellus (1990), there are two different likelihood ratio (LR) statistics to test for cointegrating vectors, namely (i) the trace test

statistic, (ii) the max test statistic. Test statistics are based on the characteristic roots (

eigenvalues, i ) obtained from the estimated matrix.

3.5 Speed of adjustment and Weak Exogeneity Test:

The factor loading matrix, , contains information the speed of adjustment towards the long

run equilibrium . The error correction coefficients can be used in exogeneity inference. Johansen (1992) developed a test based on the notion that variables that do not respond to ‘disequilibrium’ in the system of which they are a part may be considered weakly exogenous to that system. An insignificant loading coefficient thus indicates the corresponding weakly exogenous stock price.

Each of the variables for weak exogeneity can be tested on the matrix. The null

hypothesis of weak exogeneity is 0:10

iH in the case of one cointegration vector

where i=1,2,3,….11 denotes the ith

country considered in this study and

0:210

iiH in the case of two cointegration vectors. The tests are distributed

as 2 with 1 degree of freedom. Weak exogeneity implies that there is no feedback from

deviations from long run equilibria for certain variables.

4. The Data

The data used in this study consists of daily closing stock price indices over the period from November 2, 1987 to December 1, 2006 for eleven countries ( Germany, Hong-Kong, Japan, Singapore, UK, USA, India, Malaysia, South Korea, Taiwan). These data were collected from the Datastream. The sample consists of 4980 observations per country. All stock price indices are expressed in local currencies. The stock price index for each of the 11 countries are as follows: all share price index for Sri Lanka; FAZ General price index for Germany; the Hang Seng Price index for Hong Kong; the BSE National price index for India; the Nikkei 500 for Japan; The Kuala Lumpur Composite price index for Malaysia; the Singapore Straits Times Index for Singapore, the Korea Se Composite price index for South Korea; the DS-market EX TMT Price index for Taiwan; the FTSE 100 price index for UK; and the DJCMP65 for U.S. The vector of stock prices indices is denoted as follows

],,,,,,,,,,[ TAWSLSKMALINDUSUKSNGJAPHNGGERYt


6

The sample period is divided into the Pre and Post Asian financial crisis periods on the basis of 1997-1998 Asian financial crisis to address the impact of this crisis on the interdependence of the stock markets. The two sample periods are;

1. Pre Asian crisis 1997 (P1): November 2, 1987 to June 30, 1997 ( 2521 observations) 2. Post Asian crisis1997 (P2): June 1, 1998 to December 1, 2006 (2220 observations)

The Asian financial crisis started with the devaluation of Thailand’s Baht which took place on July 2, 1997, a 15 to 20 percent devaluation. Following this devaluation, a series of currencies as the Phillippine Peso, the Malaysian Ringgit, The Indonesian Ruppiah, and the Singaporean dollar started to devalue. These series of devaluations marked the beginning of the Asian financial crisis. The crisis period covered July 2, 1997 to May 30 1998 (Garay 2003). The sample countries were selected on the basis of trade share (export and imports share) of Sri Lanka. The value of share indicates as an average, calculated for 10 years, 1995-2005, exports and imports data. Table 1 reports the share percentage of imports from a country out of total imports of Sri Lanka and the share percentage of exports for a country out of total exports of Sri Lanka.

Table 1: Sri Lanka’s main trading partners trade share

Country Exports share(%) Imports share(%)

Germany (GER) 4.8 2.5

Hong-Kong (HNG) 1.3 7.4

Japan (JAP) 3.9 7.4

Singapore (SNG) 1.3 6.5

UK (UK) 11.8 4.2

USA (US) 36.3 3.4

India(IND) 2.8 12.1

Malaysia (MAL) 0.3 3.5

South Korea (SK) 0.6 5.5

Taiwan (TAW) 0.3 5.0

Source: Various issues of Annual reports, Central Bank , Sri Lanka.

5. Empirical Results:

This section focuses on the analysis of the empirical results from the performed econometric and statistical tests. All estimations have been done in EViews. Excel software was also used for ordinary calculations.

5.1 Preliminary Diagnostic Check:

In order to get a general view of the stock price series included in this study, the basic statistical properties of these stock market price indices were investigated. All stock price indices are transformed into natural logs. All indices are converted to same base period. Their time series and statistical properties are investigated through graphical representations, Summary statistics, contemporaneous correlations, auto correlation function and unit root tests.

Summary statistics of daily price changes are provided in table A1 and A2 of the Appendix. These statistics describe stylized facts of the stock price series. The kurtosis of all stock price series were much larger than the normal distribution value(3). Of the 11 countries, 8 had stock returns, negative skewness in the post crisis period while 6 countries had stock returns which were negative in the pre crisis period. The negative skewness implies that large negative returns tend to occur more often than large positive ones.

Although Sri Lanka has positive skewness in both periods, the positive magnitude decrease in period P2. These results are consistent with the literature that return series in various stock markets have fat tails and negative skewness.


7

Contemporaneous correlation matrix of stock price indices of the selected eleven countries is shown in Table A3 of the Appendix. This information gives a general understanding of the overall relationships between the stock markets. The values below diagonal of the matrix indicate the correlation coefficients for the Pre Asian financial crisis period. The values above the diagonal of the matrix indicate the correlation coefficients for the Post Asian financial crisis period. It is interesting to note that all coefficients are statistically significantly different from zero at the 1% level for both sample periods. The magnitudes of these coefficients are very small. The low correlation coefficients indicate that international diversifications among these markets are effective. The negative correlation coefficient in the pre crisis period turned to positive in the post crisis period. This implies that stock markets are linked positively after the Asian financial crisis. Very low cross correlations at lag 0 were obtained for the pairs SL-JAP, SL-TAW, SL-UK, and SL-US. However, correlation analysis cannot fully capture the long term dynamic linkages between the stock markets in a reliable way. As stock price indices are nonstationary and having serial correlation, correlation estimates are subject to statistical limitation. Therefore, contemporaneous correlation analysis does not clearly imply a great deal about international stock price linkages. Correlation techniques provide only a partial insight into the existence of the relationships between stock market prices. It is reported here for the purpose of preliminary diagnostic checking.

Sample autocorrelation of all the price series die off slowly (a clear linear decline) with very large values of first order auto correlations. All the stock price series were nonstationary. All sample PACF of all the series cuts off after lag 1. Sample PACFs for all the series show that the first PACF to be relatively significant and all the remaining ones to be not significantly different from zero. It indicates that all the price series in both sample periods are ~ I(1) processes.

The P values of ADF and PP tests indicate that all series in level form are nonstationary, I(1), in both sample periods. The ADF and PP test statistics calculated for each series in level indicate that there is no sufficient evidence to reject the unit root hypothesis at 1% level. The ADF and PP tests for the null hypothesis of a unit root test suggest that each series are stationary at their first differences (p <0.05) in both sample periods. The first difference series are strongly mean reverting. On the basis of ADF and PP test statistics, these results are broadly consistent with the hypothesis that the national stock price index series are individually integrated order one, I(1).

5.2 Test for Cointegration

The objective of this section is to take a closer look at the cointegration relationship between the selected international stock markets.

5.2.1 E-G Cointegration Test Results:

The Engle-Granger (1987) residual-based approach is also implemented in order to test for cointegration bilaterally. The long run relationship between the stock market of Sri Lanka and that of its main trading partners is examined here pair wise by employing Engle and Granger’s two step procedure. Sri Lanka’s stock price index is taken as the base index. OLS estimation of the cointegrating regression is performed to examine the stationarity of the cointegrating regression residuals. If the residuals are stationary, then it implies that those series are

cointegrated. Consider the following co integrating regression as ttt uYX 10

where Xt is log of Sri Lanka stock market price index, Yt is log of an other selected country stock price index.

The ADF and PP tests are performed for u series, estimated from a regression of the form

given above. Since this is a test on residuals of a model , tu , the critical values of ADF and

PP tests on this residual series are not the same as the critical values of ADF and PP tests on a series of a raw data. Therefore, P values for ADF and PP tests for this residual series are calculated from James G MacKinnon’s computer program (1999).


8

Table 2 given below reports the results of the unit root test for the residuals of the cointegration regression for both sample periods.

Table 2- Unit root test for the EG cointegration regression residual

(P values)

LGER LHNG LJAP LSNG LUK LUS LIND LMAL LSK LTAW

ADF P1 0.32936 0.25823 0.02314* 0.12890 0.27840 0.40096 0.01501* 0.13230 0.34436 0.23673

TEST P2 0.88610 0.59336 0.92333 0.34090 1.000 0.28956 0.06663 0.17552 0.02629 0.6114

PP P1 0.30587 0.21187 0.04797 0.10833 0.22673 0.36130 0.00904 0.09913 0.33704 0.24256

TEST P2 0.90265 0.57676 0.90746 0.37752 0.91031 0.28956 0.07529 0.19250 0.04352 0.62422

As the ADF and PP statistics for cointegration residuals are less than the relevant critical values at 5% levels for many stock markets, null hypothesis of unit roots is not rejected. It indicates that stock price series are not cointegrated. This implies that there is lack of cointegration between the Sri Lankan stock market and its main trading partners. According to ADF and PP test evidence, Sri Lankan stock market is cointegrated with Japan’s stock market and Indian’s stock market in the period. Whereas in the P2, Sri Lankan stock market is linked with South Korean’s market. Sri Lankan stock market is a relatively smallest in terms of market capitalization, listed companies, among its trading partners. Sri Lankan market behaves completely through its own internal dynamics. This is in accordance with the study by Elyasiani et al. This lack of integration allows Sri Lankan stock market to provide additional profit opportunities and diversification benefits to global investors. Although the EG approach is very easy to understand and to implement, there are important shortcomings of this method. This is a single equation method. It can estimate only up to one cointegrating relationship between the variables. However there are 11 variables in the system. It does not give us the number of cointegration vector. It does not identify joint evolution of the variables. Therefore, it is appropriate to use Johansen Cointegration test.

5.2.2 Johansen Cointegration Test:

The Johansen approach incorporates the relationships between the system of variables. Therefore, this approach can identify several equilibrium relationships governing the joint evolution of all the variables. First step in the analysis is to determine the optimal Lag: In this study, the AIC criteria was used for the model’s lag length selection, and the smallest value of AIC is preferable and helps to select the best VAR specification. The results from table 3 show that the optimal lag length is p=2 for the pre-crisis period and p=3 for the post-crisis period to describe the stock price behaviour. These results further indicate that the Asian financial crisis had an obvious influence on the orders of the VAR model selected for this study.

Table 3: VAR lag order selection criteria

Lag AIC(P1) AIC(P2)

0

1

2

3

4

-16.04845

-68.39962

-68.95444*

-68.92848

-

-19.33381

-65.0967

-65.56151

-65.58636*

-65.56354

*-indicates lag order selected by the criterion


9

After selecting the optimal order of VAR model, the model is estimated by regressing tY

against the lagged differences of tY and 1tY . The eigenvalues of are calculated in

order to find cointegrating vectors. the rank of is denoted by r and decided based on the statistical significance of its eigenvalues.

5.2.2.1 LR Test for Cointegration:

The test for cointegration among the 11 stock price series is calculated by looking at the rank

of the long run impact matrix via its eigenvalues. Using trace statistic and Maximum Eigenvalue statistics, the sequential procedure is done to find the number of cointegrating relations. Table 4 and 5 provide the evidence from both trace and Maximal Eigenvalue tests for rank determination.

Table 4: Johansen and Juselius cointegration tests (P1)

Ho H1 Max Prob** Ho H1

Trace Prob**

r=0

r=1

r=2

r=3

r=1

r=2

r=3

r=4

80.359

62.678

36.665

29.771

0.005

0.074

0.931

0.968

r=0

r 1

r 2

r 3

r 1

r 2

r 3

r 4

306.859

226.500

163.822

127.166

0.005

0.161

0.634

0.693

**- Mackinnon-Haug-Michelis(1999) P values

From Table 4, The trace statistic Trace , and Maximum Eigenvalue statistic, Max , indicate

that 0:0 rH null hypothesis is rejected at 1% level. These Trace and Max results

suggest that these stock markets possess at least one cointegrating vector(r=1), that is the stock price indices of eleven markets share a long run equilibrium, where as the independent common stochastic trends are identified in the pre Asian financial crisis period. This result is consistent with studies by Jeon and Chiang (1991), Corhay et al (1993) and Leachman and Francis (1995).

Table 5 - Johansen and Juselius cointegration tests (P2 )

Ho H1 Max Prob** Ho H1

Trace Prob**

r=0

r=1

r=2

r=3

r=1

r=2

r=3

r=4

89.208

69.732

48.890

36.071

0.000

0.015

0.315

0.737

r=0

r 1

r 2

r 3

r 1

r 2

r 3

r 4

361.913

272.704

202.971

154.081

0.000

0.001

0.026

0.095

**- Mackinnon-Haug-Michelis(1999) P values

From Table 5, The trace statistic, Trace indicates that there are two cointegration relations

at 1% level in the post Asian crisis period. Maximum eigenvalue statistic, Max , also rejects

1:0 rH for the alternative 2:1 rH at the 1% level. This indicates that there are two

cointegration vectors among these stock markets. However, at 5% level, in the post Asian financial crisis period, the trace statistic test results indicates that there are three cointegration vectors. The previous studies also had this contradictory result.


10

The maximum eigenvalue statistic indicates only two cointegration vectors. Mayasami and Koh(2000) study is an example, where 1% level results were considered.

The above result suggests that the Asian financial crisis had an impact on strengthening stock market linkages by increasing the number of cointegrating vectors within this particular system. South east Asia as a region has undergone rapid market liberalization in the past decade resulting in increased investment flows. This financial openness increased in the causal linkages. In particular, Asian financial crisis (1997) may have changed the direction and strength of the causal relationships among the markets under study.

Sri Lankan stock market (LSL) is deliberately chosen as the dependent variable to see if there is a long run link between Sri Lankan stock market and its main trading partners’ stock markets. Short run and long run relationships among the stock markets are also examined by estimating the VECM. The long run relationship can be identified through testing hypothesis

on cointegrating vector .

5.3 Long Run Relationship

Long run impact matrix, , contains information about the long run relationships between the

stock markets price indices. The rank ( r ) of indicates the number of cointegrating vectors.

The size of the eigenvalues j in the log-likelihood for the VECM are used as measure of

relative strength of long run relationship (Johansen and Juselius, (1992)). The square root,

j of these eigenvalues are called Cannonical correlation coefficients of the two vectors

tY and 1tY . The larger value indicates that the corresponding cointegrating vector is

more correlated with the stationary part of the process.

Johansen and Juselius (1992) point out that the size of the eigenvalue can be used as a measure of the relative strength of the long run relationship(s), with larger values indicating that the cointegrating vector(s) are more correlated with the stationary part of the process. Tables 6 contains estimates of the eigenvalues for both sample periods.

Table 6 - Eigenvalues

P1 (r=1)

0.0314

0.0224

0.0144

0.0117

0.0115

0.0102

0.0065

0.0043

0.0042

0.0016

0.00004

P2 (r=2)

0.0395

0.0309

0.0218

0.0161

0.0142

0.0133

0.0097

0.0072

0.0057

0.0027

0.0001

We can compare the strength of each equilibrium for both periods by examining the eigenvalues. Evidence indicates that the cointegration relationship for the post asian crisis period is slightly stronger than the equilibrium relationship characterizing the Asian pre crisis period. The eigenvalue 0.0395 corresponding to the single cointegrating vector in the post asian crisis period is larger than that (0.0314) of the pre asian crisis period.

The most frequently used method to identify the long run relationship among stock markets is to estimate the number of cointegrating vectors.

It was shown that there are two cointegrating vectors in the P2 period and one cointegrating vector in P1 period. It implies that there were two distinct long run relationships in the P2 period. But, there is only one long run relationship in the P1 period. The increase of the number of cointegrating vectors in the P2 period indicates that Asian financial crisis had an impact on the relationships among stock markets and the national stock markets have


11

become more interdependent in the P2 period compared to the P1 period. This is consistent with the increasing liberalization and globalization of financial markets. In the P2 period, the markets possess nine (n-r) common stochastic trends while P1 period exhibits ten common stochastic trends. These common stochastic trends represent non stationary dynamics.

The coefficients of the cointegrating equation can be interpreted as long run elasticities as the variables are in natural log form. In the P1 period, Sri Lankan stock market had negative long term relation with Germany, Hong Kong, Japan, Malaysia, Singapore, US stock markets. India, South Korea, Taiwan and UK stock markets had positive long term relationship with the Sri Lankan stock market.

The statistical significance of each variable in the cointegrating vector is tested by using the

likelihood ratio test statistic (LR), which is asymptotically distributed as 2

)1( . Using likelihood

ratio test, linear restrictions on are tested. In this study, it is tested whether Sri Lankan

stock market has a significant long run relationship with its main trading partners. Each trading partner is tested one by one. The test is done for both periods. The null hypothesis is that each cointegrating coefficient is equivalent to zero. For example, of pre asian financial

crisis period, 0:0 LGERH . LR statistic , )1(2 =0.314860 and P value is 0.57471,

greater than 0.05. we can conclude that this hypothesis is not rejected at 5% level. Similarly, each partner’s stock market was tested. The estimated results of P values for the LR test statistics are given in the Table 7 for both sample periods.

Table 7 - P values : Restrictions on the cointegrating vectors, B(r,n)=0

LGER LUK LUS LHNG LIND LJAP LMAL LSK LSNG LTAW

P1 0.5747 0.0003* 0.0086* 0.9738 0.0255* 0.2365 0.2742 0.8141 0.4653 0.5697

P2 0.1778 0.4912 0.1702 0.000* 0.0014* 0.3730 0.6135 0.000* 0.0029* 0.2204

The table 7 shows that Sri Lankan stock market has statistically significant long run relationship with the stock markets of India, US, and UK and there is no evidence of long run relationship between Sri Lanka and that of Germany, Hong Kong, Japan, Malaysia, South Korea, and Taiwan in the P1 Period.

In the P2 period, India, Hong Kong, South Korea, and Singapore stock markets have statistically significant long run relationship with the Sri Lankan stock market. This indicates that the pattern of linkages has changed from P1 period to P2 period. Before Asian financial crisis, Sri Lankan stock market was linked with more western developed countries, such as US UK. But, after the Asian financial crisis, Sri Lankan stock market was interlinked with the eastern South Asian countries; Hong Kong, India, South Korea, and Singapore.

5.4 Speed of Adjustment:

The speed of adjustment is represented by the absolute value of the coefficient of error correction term (The coefficient of the lagged error correction term) and it measures the (convergence speed) proportion by which the long run equilibrium in the stock price of the Sri Lankan stock market is being corrected each day. The relative speed of adjustment of each series to the long run equilibrium (revert to their long run values) is given by the values of

ir . Larger values indicate a more rapid speed adjustment. Reversion to the long run

equilibrium will be quick. Since all the variables in the ECM are I(0), statistical inferences

using standard t and F tests is valid. Estimates of the ij are presented in the table 8.


12

Table 8: Estimates of Speed of adjustment coefficient,

Period DLSL DLGER DLUK DLUS DLHNG DLIND DLJAP DLMAL DLSK DLSNG DLTAW AVER

P1 0.0014* -0.0005 0.0007 -0.0014 -0.0004 0.0045* -0.0015* -0.0004 -0.0025* -0.0009 -0.0027 0.0015

P2 0.0008* 1.94E-05 -0.0002 0.0004 -0.0003 0.0018* 0.0002 0.0004 -9.08E-05 0.0021 4.72E-05 0.0006

“*” Indicates the statistical significant at the 5 % level.

In Pre Asian financial crisis period, P1, coefficient values for error correction term, , of Sri

Lanka, India, Japan, South Korea, Taiwan and US are statistically significant at the 5% level.

In Post Asian financial crisis period II, P2, the coefficient values of Sri Lanka, India and Singapore are statistically significant at the 5% level. The positive adjustment coefficient value indicates upward adjustment and negative value indicates downward adjustment in the direction of equilibrium. There was an upward adjustment of LSL in the direction of equilibrium in the Pre Asian financial crisis period (of 0.14 percent of the difference) and in the Post Asian financial crisis period, (0.08 percent of the difference) of Sri Lankan stock market price.

We can notice two points from the results. 1) the speed of adjustment per day is very low in these countries in both sample periods. It indicates that adjustment takes long time to reach equilibrium. 2) In the case of Sri Lanka, during the P2 period, the speed of adjustment became slower compared to p1.

5.5 Test for the Weak Exogeneity and Restrictions of Weak Exogeneity is tested via the ECM weights. Test results show that Stock markets of Sri Lanka, US , India, Japan and South Korea are weak exogenous for other stock markets in the P1 period while India Malaysia and Singapore are weak exogenous for other markets in the P2 period.

6. Summary and Conclusions: Following globalization, investigation of the dynamic linkages between Sri Lankan stock market and that of its main trading partners has become an important issue. Hence, this paper investigates the dynamic stock market linkages between the Sri Lankan stock market and that of its main trading partners for the period of 1987-2006. In this study, the Engle- Granger residual based approach and Johansen cointegration test have been employed to test whether the national stock markets of Sri Lanka and that of its main trading partners are interlinked for the two periods: Pre Asian Financial crisis and Post Asian Financial crisis.

The ADF and PP test reveal that all stock price series in level are non stationary, I(1). The pair wise cointegration E-G test shows that Sri Lankan stock market is linked with Japan’s and Indian’s stock markets in period 1. In the period II, only the South Korean stock market is related with the Sri Lankan stock market. The lack of cointegration with its trading partners indicate a modest degree of attraction to major investors in Sri Lanka.

The Johansen cointegration test results indicate that there is one cointegration vector in the Pre Asian financial crisis period and there are two co integrating vectors in the Post Asian financial crisis period. The high degree of cointegration between the examined stock markets leads to a total absence of long term diversification gains when investing across the national stock markets. In particular, Sri Lankan stock market has significant long run relationship with India , US and UK in the period 1. In the Post Asian financial crisis period, Sri Lankan stock market is linked with Hong Kong, South Korea, India and Singapore. Terefore, the authorities of the Asian countries need to coordinate with each other to avoid any spillover effects following any turmoil in one or more countries of the region.


13

References

Arshanapalli, B and J.Doukas (1993). “International Market Linkages: Evidence from the pre- and post October 1987 period”, Journal of Banking and Finance 17, 193-208

Bertera E and C.Mayer, (1990), “Structure and performance: Global interdependence of stock markets around the crash of October 1987”, European Economic Review, Volume 34, Issue 6, September 1990, Pages 1155-1180.

Boswijk and Franses (1992). “Dynamic Specification and Cointegration”, Oxford Bulletin of Economics and Statistics, 54, 369-381

Cheung, Y.L, Mak, S., (1992) “The international transmission of stock market fluctuations between the developed markets and the Asian – Pacific markets. Applied Financial Economics 2, 43-47

Chowdhury, A.R. (1994), “Stock market independencies: Evidence from the Asian NIEs”. Journal of Macroeconomics 16 629-651.

Dickey, D.A., and W.A. Fuller (1981). “Likelihood Ratio Statistics for Autoregressiove Time Series with a Unit root” , Econometrica 49, 1057-1072.

Engle,R.F. and Granger, C.W.J , (1987) “Cointegrationn, and Error Correction: Representation”, Estimation, and testing, Econometrica 55, 251-76.

Elyasiani, E., and P. Perera and T.N. Puri ( 1998). “Interdependence and dynamic linkages between stock markets of Sri Lanka and its trading partners”, Journal of Multinational Financial management 8, 89-101.

Eun C and S.Shim (1989). “International Transmission of stock markets”, Journal of Financial and Quantitative Analysis, 24, 241-256.

Garay, U (2003), “The Asian Financial Crisis of 1997 - 1998 and the Behavior of Asian Stock Markets”, A web Journal of Applied Topics in Business and Economics

Gonzalo, J., (1994), “Five alternative methods of estimating long run relationships”. Journal of Econometrics 60, 203-233.

Grubel, H.G. (1968). “Internationally diversified portfolios: welfare gains and capital flows”, American Economic Review 58, 1299-1314.

Johansen, S . (1988a). “The mathematical structure of error correction models”, Contemporary Mathematics , 80, 359-386

Johansen, S . (1988). “Statistical analysis of conintegration vectors”, Journal of Economic Dynamics and Control, 12, 231-254

Johansen, S . (1991). “Estimation and hypothesis of conintegration vectors in Gaussian vector autoregressive models”. Econometrica, 59(6), 1551-1580.

Johansen, S . (1992). “Determination of conintegration rank in the presence of a linear trend”, Oxford Bulletin of Economics and Statistics, 54 383-97.

Johansen, S. (1997), Likelihood – “Based Interference in cointegrated Vector Autoegressive Models”. Oxford, Oxford University Press.

Johansen, S. (1998). “Statistical analysis of cointegration vectors”. Journal of Economic Dynamics and control, 12, 231-254


14

Johansen, S and Juselius, K. (1990). “Maximum likelihood estimation and inference on cointegration : with applications to the demand for money”, Oxford Bulletin of Economics and Statistics, 52, 169-210.

Johansen, S., and K. Juselius (1992), “Testing structural hypotheses in a multivariate cointegration analysis of the PPP and the UIP for UK”. Journal of Econometrics 53, 211-244.

Kasa, Kenneth. (1992), “Common Stochastic Trends in International Stock Markets”. Journal of Monetary Economics 29, 95-124

Kwan, A.C.C, Sim, A.B. Cotsomitis, J.A. , (1995), “The causal relationships between equity indices on world exchanges”. Applied Economics, 27, 33-37.

Mackinnon, J., A.A. Huag, And L. Michelis (1999) “Numerical Distribution Function of Likelihood Ratio Test for cointegration”, Journal of Applied Econometrics, 14, 563-577.

Masih, A.M.M. and Maish, R. (1997a). “Empirical tests to discern the dynamic causal chain in macroeconomic activity: new evidence from Thailand and Malaysia based on a multivariate cointegration /vector error correction modeling approach”. Journal of Policy Modeling 19 (4) 441-468.

Masih, A.M.M. and Maish, R. (1997b). “A comparative analysis of the propagation of stock market fluctuations in alternative models of dynamic causal linkages”. Applied Financial Economics, 7(1) 59-74.

Masih, A.M.M. and Maish, R. (1999). “Are Asian Stock market fluctuations due mainly to intra – regional contagion effects? Evidence based on Asian emerging stock markets”, Pacific – Basin finance Journal, 7, 251-282.

Mayasami, R.C., and Koh, T.S, (2000) “A vector Error Correction Model of the Singapore Stock market”, International Review of Economics and Finance 9, 79-96.

Phillips P.C.B. and P. Perron (1988). “Testing for a unit root in time series regression”, Biometrika, 75, 335-346

Von Furstenburg, G.M., Jeon. B.N., (1989), “International Stock Price movements: links and

messages”. Brookings Papers on Economic Activity. 1, 125-179.


15

APPENDIX

Table A1 - Summary Statistics(P1)

Country Sample

size

Mean median std skewness kurtosos JB prob

Germany 2520 0.000384 0.000158 0.01457 -1.504329 25.20795 0.0000

Hong Kong 2520 0.000766 0.000150 0.014670 -2.314551 39.57283 0.0000

Japan 2520 -3.92E-05 0.0000 0.01115 0.21002 11.5608 0.0000

Singapore 2520 0.000371 0.0000 0.01068 -0.59499 11.59701 0.0000

UK 2520 0.000390 0.000175 0.008088 -0.03298 6.0036 0.0000

US 2520 0.000457 0.000256 0.007998 -0.480205 8.432832 0.0000

India 2520 0.000814 0.0000 0.015575 0.73218 15.52587 0.0000

Malaysia 2520 0.000544 0.000108 0.011856 -0.569256 13.9216 0.0000

South Korea 2520 0.000150 0.0000 0.013702 0.28449 5.590256 0.0000

Sri Lanka 2520 0.000503 0.0000 0.009790 2.473415 39.69408 0.0000

Taiwan 2520 0.000422 0.0000 0.02197 0.022608 5.38064 0.0000

Table A2 - Summary Statistics(P2)

Country Sample

size

Mean median std skewness kurtosos JB prob

Germany 2219 1.3E05 0.000463 0.015551 -0.304878 6.942685 0.0000

Hong Kong 2219 0.000349 0.00000 0.014635 0.039749 7.299422 0.0000

Japan 2219 0.000106 0.0000 0.012872 -0.284901 5.127769 0.0000

Singapore 2219 0.000410 7.33E05 0.012752 -0.017430 7.849200 0.0000

UK 2219 1.40E-05 0.00000 0.059026 0.011621 5.703508 0.0000

US 2219 0.000170 0.00000 0.010412 -0.192288 6.890234 0.0000

India 2219 0.000658 0.001026 0.016573 -0.486742 7.072918 0.0000

Malaysia 2219 0.000331 0.0000 0.014576 -0.401568 67.27159 0.0000

South Korea 2219 0.000653 0.000473 0.020110 -0.139930 6.368313 0.0000

Sri Lanka 2219 0.000644 0.0000 0.013267 0.361308 34.74985 0.0000

Taiwan 2219 8.25E05 0.0000 0.015738 0.014946 5.5888 0.0000


16

Table A3 - Contemporaneous correlation- P1 ( below diagonal) , P2 (above diagonal)

Country GER HNG JAP SNG UK US IND MAL SK SL TAW

GER 0.59 0.84 0.37 0.92 0.46 0.11 0.12 0.07 -0.25 0.47

HNG 0.84 0.77 0.92 0.59 0.82 0.72 0.77 0.70 0.46 0.647

JAP -0.32 -0.66 0.67 0.85 0.62 0.43 0.43 0.45 0.07 0.72

SNG 0.85 0.93 -0.52 0.46 0.86 0.83 0.88 0.88 0.64 0.65

UK 0.91 0.96 -0.57 0.92 0.58 0.19 0.11 0.18 -0.13 0.53

US 0.91 0.94 -0.54 0.87 0.98 0.83 0.65 0.76 0.65 0.54

IND 0.66 0.85 -0.76 0.85 0.79 0.73 0.78 0.91 0.90 0.50

MAL 0.86 0.95 -0.57 0.98 0.94 0.89 0.87 0.84 0.68 0.58

SK 0.29 0.31 0.12 0.51 0.30 0.24 0.34 0.45 0.82 0.52

SL 0.52 0.64 -0.72 0.72 0.62 0.52 0.91 0.73 0.23 0.31

TAW 0.44 0.18 -0.03 0.28 0.25 0.28 -0.03 0.25 0.53 -0.19

17

Factors Affecting

the Adoption of

Agricultural Forward Sales

Contracts in Sri Lanka

K.S. Karunagoda1, K. P. N. S. Karunagoda

2, H.P. Gunawardhane

3,

S.P.R.P. Senanayake4, Y.M. Wickramasinghe

5

Abstract As Forward Sale Contracts (FSCs) offer a mechanism to reduce only the price risk, a few percentage of farmers have joined the program. The factors affecting the adoption of FSCs are evaluated using a sample of rice, maize and onion growers in Anuradhapura and Matale districts. The analysis indicates that the scale of operation, contracts with village extension agents and participation in social activities positively influence adoption decision, while farmers with off-farm income and higher household income are most unlikely to participate in the FSCs. The results indicate that FSCs provide a reasonable mechanism to farm households to reduce price risk and uncertainty of farmers who depend more on agricultural income and have invested more in agricultural activities. Albeit the lower share of contracts compared to the farmer population, FSCs have provided to reduce the risk of more vulnerable groups. Promotional campaign of FSCs would effectively be carried out through the agricultural extension system. The higher capital requirement and price instability may act as main barriers for further promotional effort of the system and its sustainability.

Keywords: Adoption, Socio-economic factors, Rice, Maize, Onion

1. Socio-economics and Planning Center, Department of Agriculture 2. Welfare Department, Central Bank of Sri Lanka 3. Faculty of Agriculture, University of Rajarata, Pulliyankulama, Sri Lanka 4. Faculty of Agriculture, University of Peradeniya, Peradeniya, Sri Lanka 5. Faculty of Agriculture, University of Rajarata, Pulliyankulama, Sri Lanka


18

1. Introduction

Development policies of the agricultural sector targeted at altering the biological, institutional and technological progress helps to bring about a sustainable increase in agricultural production as well as increase in farmers’ living standards in Sri Lanka. Research and development efforts have helped in increasing output of agriculture but less emphasis has been placed on the improvement of the marketing and distribution system. The poor development of marketing of agricultural products has become a bottleneck to the growth efforts of agriculture in Sri Lanka. Production of rice, field crops and vegetables is significantly influenced by weather and the farmers take their production decisions mainly based on the expected prices. The fluctuating production and prices go in a vicious circle and seasonality adds a high degree of uncertainty. Therefore, the development of a mechanism to minimize uncertainty of future prices would help to overcome the undesirable effects of fluctuations in agricultural production and prices. In 1999, the Central Bank of Sri Lanka (CBSL) introduced ‘Forward Sale Contracts’ (FSCs) to overcome the price and marketing uncertainties faced by the farmers and uncertainty of input availability faced by processors. Though the FSC promises benefits to all stakeholders, the level of adoption has been far from satisfactory (CBSL, 2006). A plethora of socio-economic factors could influence the farmers’ decision on adoption of FSCs, and the main objective of this study is to identify such contributory socio economic characteristics of farm households.

2. Forward Sale Contracts

A FSC is a legally binding agreement between a buyer and a seller and it guarantees an acceptable price for both seller and buyer in the market. The seller agrees to sell a specified quantity of an agricultural product of a specified quality on a given future date at a predetermined price. Such contracts lead to marketing agreements in advance, which would secure a confirmed order to the seller and on the other hand, an assured supply to the buyer. A provision has been made for a bank to participate as a facilitator of the FSC. The bank helps in bringing a buyer and a seller together and in determining the contract price. The FSCs operate through the coordination of different stakeholders including participating financial institutions (PFIs), buyers, farmers, farmer organizations, government organizations etc. The FSCs help in strengthening the agricultural supply chain and the reduction in marketing costs and stabilization of prices.

2.1 Determination of Contract Price

The contract price is determined based on the cost of production (50% margin to cost of cultivation), the cost of production and retail prices (for commodities with high seasonal price variation), or the border price of the commodity (import competing products). The choice of the most appropriate method depends upon the commodity itself and various other factors such as the variation of farm gate prices and retail prices etc. (Govi Sahanaya, 2000).

2.2 Progress of FSCs

The progress of the FSCs during the period of 2004 – 2006 is presented in the Tables 1, 2 and 3. Accordingly, FSCs are in operation for maize, paddy, soyabean, green gram, finger millet, ginger, sesame, papaya, sunflower and vegetables. When compared with the total number of contracts signed in 2004, the number of contracts signed in 2005 has increased by 111%. The number of farmers benefited by FSCs during the same period has increased by 132%. FSCs of Maize shows the highest growth (383%) while FSCs of paddy shows a marginal growth of 9% during the period of 2004 to 2005. The contract grower system is popular in maize and soybean. The development of maize and soy-based food industry; increase in world market prices and the concern on genetically modified organisms may have encouraged the local food processors to rely more on local suppliers. In these crops, the firms provide inputs and extension services to farmers and the presence of FSCs may have encouraged processors to initiate FSCs. In 2006, a positive development is seen only in paddy whereas all other crops show a decline. The increase in area under maize in 2005 and

Factors Affecting the Adoption of Agricultural Forward Sales Contracts in Sri Lanka

19

2006 may have induced the processors not to enter in FSCs due to lower risk of availability and higher price risk (Table 1). When compared with the number of farmers who cultivate these crops, the level of adoption of FSCs is very low. Contract prices in a given year show a variation and it is mainly attributed to the geographical location and variety (Table 2). The buyers are entitled for low interest (7% per annum in 2006) bank loans and the maximum limit of the loan was revised from Rs. 25 million in 2004 to Rs. 50 million in 2005 (CBSL). The amount of loans granted were Rs. 2,412 and 2,637 million in 2005 and 2006 respectively (CBSL) (Table 3). The credit per contract is quite large and it indicates that the credit component also plays a significant role in the FSCs.

Table 1. Forward Sale Contracts: 2004-2006

2004 2005 2006

Crop

No. of

Contracts

No. of

farmers

benefited

No. of

Contracts

No. of

farmers

benefited

No. of

Contracts

No. of

farmers

benefited

Maize 7,048 7,102 34,071 40,500 6,667 7,194

Paddy 13,130 13,470 14,270 16,200 23,033 23,585

Soybean 5,442 5,520 6,700 6,700 4,325 4,325

Green Gram 5,335 5,443 1,200 1,200 988 1,038

Finger millet 1,813 1,890 540 540 101 832

Other crops* 9,075 9,940 12,410 12,410 553 847

Total 32,768 33,425 77,550 77,550 35,667 42,821

* includes ginger, sesame, papaya, sunflower and vegetables. Source: Central Bank of Sri Lanka; Annual Reports (2005, 2006 and 2007). Rural Development Department, Central Bank of Sri Lanka

Table 2. Forward Sale Prices (Rs/kg): 2004-2006

Crop 2004 2005 2006

Maize 14-18 16-18 14-18

Paddy 13-16 14-17 11-18.5

Soybean 20-25 30-38 25-30

Green Gram 27-30 47-52 50-65

Finger millet 45-50 20-25 35-45

Other crops* 25-375 20-370 20-65

* includes ginger, sesame, papaya, sunflower and vegetables. Source: Central Bank of Sri Lanka, Annual Reports (2005, 2006 and 2007). Regional Development Department, Central Bank of Sri Lanka

Table 3. Loans Granted for Forward Sale Contracts: 2005 and 2006

Crop 2005 2006

Rs. million Average per

contract (Rs)

Rs. Million Average per

contract (Rs)

Maize 687 16,963 132.5 18,348

Paddy 745 45,987 2,243.0 78,467

Soybean 375 55,970 178.1 41,179

Green Gram 51 42,500 59.2 57,032

Finger millet 14 25,926 11.0 13,221

Other crops* 540 43,513 23.7 16,175

Total 2,412 31,102 2,637.5 61,594

Note: * includes ginger, sesame, papaya, sunflower and vegetables. Source: Central Bank of Sri Lanka, Annual Reports 2006 and 2007. Regional Development Department, Central Bank of Sri Lanka.


20

3. Previous Studies

The economic decision units faced with the possibility of adopting an innovation may be understood by using economic theory to derive theoretical results predicting the qualitative effect of factors such as farm size, liquidity constraints and risk attitudes on the decision to adopt or not to adopt the innovation or through empirical studies which identify the factors or characteristics correlated with adoption decisions (Dorfman, 1996). Goodwin and Schroeder (1994) examined the role of participation in public and private risk management and marketing seminars on adopting forward pricing methods and pricing strategies used by wheat, corn, grain sorghum, soybean, cattle and hog producers. It was noted that agricultural producers face uncertain outcomes when adopting innovations. The decision on adoption of some marketing techniques (futures hedging, forward contracts and options markets) in order to manage price risk depends on farmers’ experience, education, management rating, leverage, farm size, off farm income, expected income change from adopted technique and their beliefs on adopted technique could stabilize income. When adoption is discrete and when adoption and non adoption decisions are mutually exclusive, farmers would choose to adopt a new technique, if their expected utility of profits with the technique, exceeds their expected utility of profits without adoption. Costs associated with adopting futures and forward pricing techniques include variables associated with the information gathering costs and the efficiency of adoption as well as commissions and brokerage fees. Goodwin and Schroeder (1994) indicated that adoption of forward pricing methods decreases with increase in experience and it increases with increases in farm size, crop intensity, input intensity, leverage and education. The results indicate that the participation in marketing/risk management educational programs significantly increases farmers’ adoption of forward pricing methods. Many studies by Lowell and Kau (1973), Rahm and Huffman (1984), Byerlee and Polanco (1986), Dorfman (1996) and Shiyani at el. (2000) have focused on factors influencing adoption of an innovation. The characteristics of farmers have been the dominant theme of many of these studies. Lowell and Kau (1973) indicated that decision on adoption of an innovation can be explained using a threshold theory of decision

1 in which a reaction occurs

only after the strength of the stimuli increases beyond farmers’ reaction threshold. Rahm and Huffman (1984) concluded that the rate of adoption of an innovation increases with increased education level. Byerlee and Polanco (1986) have used a stepwise approach to the adoption of a package that reflects the characteristics of each component and the interactions between them and identified profitability, risk, divisibility, initial capital requirements, complexity and availability of the technologies as significantly influencing factors.

4. Method

The participation in the FSCs adoption decision is dichotomous in nature. Therefore, binary choice models are applicable. In the present study, the factors affecting adoption of FSCs were related to the socio-economic characteristics of the farmers. The data published by CBSL were mainly used for the selection of the sample.

4.1 Theoretical Background

Qualitative response models have been widely used in economics to investigate factors affecting an individual’s choice among two or more alternatives. Maximum-likelihood logistic regression (logit) method was employed to analyze the use of FSC rather than ordinary least squares (OLS) method because the dependent variable is binary (0,1). The dependent variable, FSC (Yi), takes a value of one if the farmer reported using FSC and assumes a value of zero or otherwise. Specifically, the logit is defined as the natural logarithmic value of

1 It was hypothesized that there exists a “breaking point” or threshold in the dimension of the explanatory variable

bellow which a stimuli elicits no observable response. Only when the strength of the stimuli reaches the threshold level does a reaction occur.


21

the odds in favor of a positive response (in this case, FSC), i.e.: Yi = 1 if producer participates in FSC, 0 otherwise. An empirical representation of the FSC (Yi) model by producer I to observable explanatory

variables is given by: where Xi is a vector of explanatory variables relevant to

producer I’s use of FSC, is a vector of unknown parameters, and i is a residual error

assumed normally distributed with a zero mean and constant variance. In a binary logit model, the marginal effect of a variable Xj on the response probability (P);

. The log odds-ratio can be expressed as a linear function of the

explanatory variables (equation 1) (Gujarati, 1995). The left hand side of the equation

indicates the odd-ratio.

k

log Pi

= β0 + ∑ βj χij ……………………….(1)

1 – Pi j = 1

4.2 Sample

The preliminary information collected from the Central Bank of Sri Lanka (CBSL) indicated that the FSCs are popular among growers of rice, big onion, maize, soybean and the rate of diffusion of FSCs is high in Polonnaruwa, Anuradhapura, Monaragala, Ampara, Matale and Hambanthota districts. Anuradhapura and Matale districts are purposely selected for the study because of the proportion of adopters in these districts is relatively high. Two villages were purposely selected from Matale (Wewala) and Anuradhapura (Kudapattiya) districts using the information provided by the Central Bank’s regional office of Anuradhapura and Kandurata Development Bank of Dambulla. A random sample of 150 farmers was selected for three crops; onion, paddy and maize (50 each) from Wewala (onion) and Kudapattiya (maize and paddy). A pre-tested structured questionnaire was used to collect socio-economic information from the farmers of the selected villages.

4.3 Data

The data collected included information of the farmer (age, education, occupation, farming experience), income, particulars of land ownership, information of social participation (participation in common village activities) and contacts with extension agents, storage facilities, knowledge of farmer on FSCs and problems encountered by the farmer. The key informants such as senior managers and regional managers of the CBSL, Agricultural Instructors, and office bearers of the farmer organizations were also interviewed to obtain their perception on the FSCs. Data collection was carried out in 2006.

5. Results and Discussion

The characteristics of the farmers surveyed in the selected villages are provided in the Table 4. Education up to primary level is prominent in all three cases while the majority of farmers had formal education below the tertiary level. The land holding sizes are varied with the crops investigated and it is less than 3 acres for majority of onion and maize farmers, while the sample of paddy included a group of farmers with more than 3 acres of land holding. The logit model was used to determine the factors that influence farmer’s participation in the FSCs (1 = farmer joined, 0 = farmer did not join). The independent variables were farmer’s age (Average years), land holding area (land extent), level of farmer education (above secondary = 1, below secondary = 0), access to extension services and social participation (0 for farmers with score less than 2 and 1 for farmers with score more than 2) and income level and source of income. The results of the three logistic models are summarized in Table 5.


22

For paddy and maize farmers, off-farm employment, land holding size, participation in social activities, and access to extension service act as determining factors of participation of the FSCs. In case of maize farmers, other than above factors, indebtedness and farming experience show a significant relationship. Farming experience, land holding sizes (low land and high land) and household income are the determinants of participation of FSCs by Onion growers.

Table 4. Characteristics of the Study Samples

Anuradhapura Matale

Paddy Maize Onion

Sample size 50 50 50

Age (Average yrs) 40 41 43

Family size

(members)

(mode)

3-5 3-5 3-5

Education

Primary

Secondary

Tertiary

54% 36% 10%

50% 38% 12%

34% 54% 12%

Land holding size

<1ac

1-3ac

>3ac

20% 34% 46%

22% 64% 14%

64% 36%

Farming experience

>10 yrs

<10 yrs

62% 38%

58% 42%

90% 10%

Involved in Off-farm

employment

Yes

No

44% 56%

54% 43%

42% 58%

Access to extension

High

Low

48% 52%

54% 46%

34% 63%

It was also found that farmers whose livelihood is mainly dependent upon off-farm employment (paddy and maize) or with high household income (onion) are reluctant to adopt the FSCs. This indicates that the tendency to join the program decreases when the farmer has non-agricultural income opportunities. Higher social participation is found to be positively related with the adoption of FSCs by paddy (odds in favor increase by 11) and maize farmers (odds in favor increase by 7.4). This may be due to the fact that higher social participation makes farmers aware and it persuades farmers to join the FSCs. Access to extension service was found significant with a positive sign for paddy and maize samples. Land holding size increases the odds in favor of adopting FSCs in all crops. In case of onion farmers, the separation of land holdings into lowland and highland shows a different result and a higher cultivation extent of onion and maize favors adopting decision. The higher extent associated with higher output and therefore, higher risk on returns. As a result, farmers may have opted to hedge against the price risk by adopting the FSCs. Though education plays a significant role in the adoption decision, it is not significant. Farming experience shows a positive and significant relationship for maize and onion growers.

5.1 Adoption of FSCs

About 22% of farmers in Wewala village have joined FSCs since its inception in 1999. The percentage of farmers who joined the program in Kudapattiya village was about 26%. The majority of farmers were aware of the FSCs and only 26% of the non-adopters had no idea about the system. The doubt about the agreement (36% of non-adopters) was one of the main reasons for non-adoption and it is mainly due to less acquaintance with the buyers and


23

the suspicion on the strength of the agreement in fulfilling stipulated conditions by buyers under more favourable (unfavourable) market prices. The majority of farmers depend more on the conventional marketing channels and a very small percentage of farmers have concentrated on possible alternatives for the reduction of risk through new marketing strategies.

Table 5: Determinants of Participation of FSC

The majority (75%) of farmers do not have sufficient storage facilities and they are unable to keep the harvest for sometime until the buyers come to fulfill the contractual agreement. The FSC envisages the determination of contract prices by various methods but the price determination of products seems highly influenced by the prices that prevailed in the past season. The violation of the agreement was observed in both parties (buyers and sellers) and the main reason for the violation was unfavourable price change at the market. These observations indicate that the price stability, forecasting system, improvement of information availability, etc. would improve the benefits of the FSCs.

6. Concluding Remarks

Although FSCs offer a mechanism to reduce producer’s price risk, the proportion of farmers who joined the program was comparatively low. The analysis indicates that the land extent, participation in social activities, contacts with village extension agent, positively influences the adoption decision. Higher level off-farm income or household income reduces the chances of participation in the FSCs. The average loan per contract is fairly high and the availability of loans may have acted as a barrier for FSCs and it needs further investigation. The determinants of participation indicate that FSCs provide a reasonable hedging mechanism to farm households who depend more on agricultural activities. Therefore, farmers have identified the FSCs as a mechanism to reduce the price (income) and marketing risks. A promotional campaign of FSCs may be effectively carried out through the agricultural extension system. The uncertainty of future prices could further be reduced through the consistent trade policies and development of institutional support to the three parties involved FSCs. Marketing of agricultural products is still a major concern of the majority of farmers in the villages surveyed. The marketing strategies adopted by farmers in the selected villages are not diversified due to various constraints. Diversification of marketing strategies such as focusing on specific market windows, aiming at specific markets, selling of value added products etc. together with forward contacting would provide better solution to the pricing and marketing problems faced by farmers.

Onion Maize Paddy

Variables Co-

efficient

Variables Co-

efficient

Variables Co-efficient

Intercept 5.55

Intercept -13.73**

Intercept -10.25**

Age 0.16

Off-farm employment

-4.22***

Off-farm employment

-2.7**

Family size 0.76

Land extent (Total)

1.99***

Land extent (total)

2.6**

Level of education -0.21

Social participation

2.00*

Social participation

2.43**

Farming Experience

0.23**

Access to extension service

1.99 Access to extension service

2.82

Land Holding-High land

2.42*

Farming experience

2.42*

Household income level

-0.42***

Indebtedness 1.83

Education 1.29

Note: *,** and *** indicate significance levels 10%, 5% and 1% respectively.


24

References

Byerlee, D. and E.H.D. Polanco. 1986. “Farmers’ Stepwise Adoption of Technological

Packages: Evidence from the Mexican Altiplano”. American Journal of Agricultural Economics, 68(3):519-527

CBSL (Central Bank of Sri Lanka), Annual Reports (annual). “Various reports from 1999-

2007”. Central Bank of Sri Lanka, Colombo. Dorfman, J.H. 1996. “Modeling Multiple Adoption Decision in a Joint Framework”, Amrican

Journal of Agricultural Economics 78: pp 547-557.

Goodwin, B.K., and T.C. Schroeder, 1994. “Human Capital, Producer Education Programs,

and the Adoption of Forward Pricing Methods”, American Journal of Agricultural

Economics, 76(4):936-947. Govi Sahanaya (Sinhala). 2000. “Regional Development Department”, Central Bank of Sri

Lanka. Gujarati, D.N. 1995. “Basic Econometrics (Third edition)”, McGraw-Hill, Inc. New York. Lowell H. and P. Kau. 1973. “Application of Multivariate Probit to a Threshold Model of Grain

Dryer Purchasing Decisions”, American Journal of Agricultural Economics, 55(4):18-27.

Rahm, M.R. and W.W. Huffman. 1984. “The Adoption of Reduced tillage: The Role of Human

Capital and Other Variables”, American Journal of Agricultural Economics, 66(4) 405-418.

Shiyani, R.L., P.K. Joshi, M. Asoken and M.C.S. Bantilan. “Adoption of Improved Chickpea

Verities: Evidences from tribal Region of Gujarat”. Indian journal of Agricultural Economics, 55(2): 158-171.

25

An Empirical Examination of

Informational Content

of Dividend Announcements

in the Sri Lankan Share Market

D.B.P.H.Dissa Bandara1, K.D.I.Perera

2

Abstract

This research paper examines the informational content of dividend announcements and tests the semi-strong form of market efficiency in the Sri Lankan share market. This study attempts to answer the research question of: How does the Sri Lankan share market respond to dividend announcements? This issue is examined using an overall sample of 100 events relating to 12 listed companies covering the period from January 2000 to December 2008. The Standard Event Study Methodology (ESM) is adopted in order to carry out the study. The study finds a considerable informational content of dividend hypothesis. The results show significant characteristics between the emerging Sri Lankan and other developed capital markets. Further the evidence finds that considerable anticipation before the event date and prices adjust even after the event date. However, the evidence is not consistent with the semi-strong form Efficient Market Hypothesis (EMH). Key words: Informational Content of Dividend Announcements, Sri Lankan Share Market, Event Study Methodology, Capital Market Efficiency

1 Dr.D.B.P.H.Dissa Bandara, Senior Lecturer in Corporate Finance and Corporate Governance, Department of Finance, University of Sri Jayewardenepura and Director of Securities and Exchange Commission of Sri Lanka (SEC), E-mail : [email protected] 2 Mr.K.D.I.Perera, Officer of SEC, E-mail : [email protected] The authors thank to valuable suggestions from anonymous referees and editors of the research journal committee of the Center for Banking Studies of Central Bank of Sri Lanka. Mistakes are the sole responsibility of the authors. The views expressed on the research paper are those of the authors and do not reflect the view of mentioned organizations.

mailto:[email protected]



26

1. Introduction

The primary objective of this paper is to examine the impact of informational content of dividend announcements in Sri Lanka. Secondly, it tests the semi-strong form of EMH in Sri Lankan share market. The Standard Event Study Method (SESM) introduced by Brown and Warner (1980, 1985) is used as a methodology to measure the market response in Sri Lanka. The goal of corporate management is to maximize the wealth of shareholders’ investment in the firm. Corporate managers’ endeavor to achieve this goal through their investment and financing decisions. While achieving this goal they always try to balance the risk–return tradeoff which almost all financial decisions are involved. Apart from the investment and financing decisions, managers need to decide whether to payout the earning to shareholders on regular basis. The dividend decision is an integral part of the firm’s financial strategies. However, the unresolved puzzle remains whether paying out of earnings would essentially create value for the shareholder or not. Theoretically, in an ideal world without tax and any restrictions, dividend payments would have no impact on the shareholders’ value (Modigliani and Miller, 1961). According to the Dividend Irrelevancy Theory, dividend should be paid out of funds which are available after financing all positive net present value projects. Contrary to the Modigliani and Miller (1961), shareholders do indeed prefer dividends to future capital gains. According to the Dividend Signaling Theory, Firm’s dividend policy can be observed to have an effect on its share price. An informational content of dividend announcements is highly considerable in financing decisions. The Signaling Theory argues that shareholder and investing community understand that managers have more information about firms’ future prospects. Bhattacharya (1979), John & Williams (1985), Miller and Rock (1985) noted that managers know more about the real value of the firm than investors and they direct the information in the market for dividend policy. Dividend announcements issue positive and negative signals to share holders. Especially it affects the financing and investment decisions of existing and prospective shareholders. Market participants are cognizant that informational content of dividend announcements can be used as a signaling device to transfer various types of signals by the management. Market efficiency will depend on the reflection of the informational content on the capital assets prices. According to Pettit (1972), the allocative efficiency of capital markets depends on the extent to which capital assets prices fully reflect information that affects their value. In an efficient capital market this information is fully, instantaneously and unbiasedly reflected in share prices subsequent to a public announcement of dividends (Dissa Bandara, 2001). Managers use the dividend announcements as a signaling tool that conveys grater informational content accordingly. Obviously it implies broad scope to think the validity of informational content and it invites to test the EMH in different forms. In the real world, dividend announcements are important to the shareholders because of its tax effect and its informational content. Many scholars initially like Fama (1970), subsequently Pettit (1972, 1976), Aharony and Swary (1980), Kalay and Loewenstein (1985, 1986), Elfakhani (1995), Below and Johnson (1996), Ryan, Besley and Lee (2000), Fung and Patel (2005) have empirically examined the informational content of dividend announcements in Anglo-Saxon countries such as the US and the UK markets. Uddin (2003), has empirically tested South Asian market like Bangladesh. Adelegan (2003), has empirically supported African market like Nigeria. Cheng, Fung and Leung (2004), have investigated in Hong Kong. Saens and Sandoval (2005), have investigated security performance based on Latin American countries like Chilean. Valipor, Rostami and Salehi (2009), have studied in Iran. In Sri Lankan context, Dissa Bandara (2001), Dissa Bandara and Samarakoon (2002) have directly supported informational content of dividend announcements in Sri Lankan share market when compared with the evidence of developed countries, a very limited number of published papers are available on the Sri Lankan market. In order to enhance the studies of Dissa Bandara (2001), Dissa Bandara and Samarakoon (2002); this paper attempts to test

An Empirical Examination of Informational Content of Dividend Announcements in the Sri Lankan Share Market

27

the informational content of dividend announcements in Sri Lankan market and to compare and contrast with the previous research findings.

2. Literature Review

Many researchers empirically supported the informational content of different events. Researchers were likely to discuss informational content of dividend announcements as it provides more sophisticated analysis of information asymmetry. Any occurrence that affects the share prices can be defined as an event. If security prices reflect all currently available information, then price changes must reflect new information. Event study describes a technique of empirical financial research that enables an observer to assess the impact of a particular event on a firm’s stock prices.

According to traditional view of dividend policy, it suggests that the price of a share depends upon the mix of dividends, required rate of return and growth prospects. Dividend policy has any relevance for its share values remains an unsolved puzzle in Financial Management. When a corporate accepts the traditional view; it may decide optimal dividend policy because it affects the capital structure and leverage of the firm. In contrast to the traditional view, the dividend irrelevancy theory of Modigliani and Miller (1961) suggests that dividend payments have no impact on the shareholder’s value in the absence of taxes and market imperfections. They argued that if a company with investment opportunities decides to pay a dividend, then retained earnings are insufficient to finance all its investments. Further, in a perfect capital market with rational behavior and perfect certainty and with investment and borrowing decisions given, dividend policy has no effect on the value of the firm. The shortfall in funds will be made up by obtaining additional funds from outside sources. As a result of obtaining outside finance instead of using retained earnings, the loss of value in existing shares equal to amount of dividend paid. Furthermore, the value of a firm is primarily determined by its ability to generate earnings from its investments and by its level of business risk and financial risk. Pettit (1972), suggested that most of the information is reflected in the securities prices as of the end of the announcement period. In addition, he confirmed the proposition that the market makes use of announcements of changes in dividend payments in assessing the value of the security. Further, he concluded two major suggestions regarding the rules and regulations of corporate disclosure. First, the financing alternatives available to the firm would be broader, if the information implicit in the announcement could be conveyed to the market in a different way. Second, the use of dividends as an information disseminating device is inefficient since, it is an imperfect means of describing the firms’ future prospects. Aharony and Swary (1980) empirically investigated information content of dividend announcements in dividend increase category, dividend decrease category and dividend constant category. Their findings supported the semi-strong form of EMH. On the average, the stock market adjusts in an efficient manner to new quarterly dividend information. Further, they strongly supported the hypothesis that changes in quarterly cash dividends provide useful information beyond that provided by corresponding quarterly earnings. They concluded that dividend increase issues positive Abnormal Return (ARs), dividend decrease category was provided negative ARs surrounding the event date (divided announcement date). Asquith and Mullins (1986), found that the event day ARs were positively related to the size of the initial dividend payments. They supported that the informational content of dividend hypothesis and thereby the findings are coincide with the semi-strong form of EMH since, the widely available dividend information is fully, instantaneously and unbiasedly reflected in share prices. John and Lang (1991), studied some informational aspects of insider trading around corporate dividend announcements. They found that informational content of dividend change, say dividend increase, depends crucially on the investment opportunities set of the signaling firm vice versa. Further, the announcement effect of dividend increases may be positive or negative depending on the concurrent insider trading. The evidence shows that on the announcement of dividends, considerable insider trading prior to the announcement day.


28

Uddin (2003), found that when dividend is taxable, paying out more cash would increase the shareholders tax liability. Despite these theoretical arguments for not paying dividends, companies through paying cash dividends to their shareholders possibly try to signal about the future earnings prospects. Finally, he concluded that in Dhaka Stock Exchange, the investors do not gain value from dividend announcements supporting the dividend irrelevancy hypothesis. Koch and Sun (2004), investigated the statistical relation between the market reaction to changes in dividends and recent past earning changes. They found a positive statistical relation between the market reactions around the announcement of the dividend change and the prior earnings change. In addition, the strength of statistical relation varies predictably with both the sign of the past earnings change and the sign of the scaled divided change. Moreover, they found negative statistical relation between the market reaction surrounding the dividend change announcements and the prior earnings change. Fung and Petel (2005), noted that the signaling effect of a company’s dividend policy may be used by the management of a company which faces a possible takeover. In the presence of asymmetric information, dividends can be considered as a communication device that allows insiders to convey important information to outside investors. The contents of that information reflect the presence of different economic mechanisms in which dividends affect firm value and its valuation. Dissa Bandara (2001), analyzed informational content of dividend announcements on the Colombo Stock Exchange (CSE) in Sri Lankan share market. He found that Sri Lankan share market responses to dividend announcements and the market reaction to dividend announcements support informational content of the dividend hypothesis, namely the change in annual cash dividends provide information about the changes in managers’ assessment of future estimates of the firm. He observed that Sri Lankan market participants do not instantly stimulate to the dividend information, as the price adjustment process is slower in the CSE than in the US and other developed markets. Dissa Bandara (2001), found that market response differently with the dividend increase, dividend decrease and no change (constant) of dividend directions. He concluded that the market reacts dramatically to the dividend increases and it is in agreement with the proposition that higher the dividend increase larger the positive response. He found dividend decrease category comes out by providing evidence to accept the proposition that higher the dividend decrease larger the negative response and market reaction to dividend reduction is much more than to dividend increases. Constant dividend category does not show any significance response surrounding the event date. Announcement of constant dividend conveys the manager’s indifference regarding the firm future prospects. He strongly emphasized, some of the market participants reveal their preferences to obtain current income rather than uncertain future capital gains disregarding the direction of dividend announcement. He emphasized that there is a considerable informational content of dividend announcements in Sri Lanka. The market responds differently to dividend increases, dividend decreases and no change in dividend directions. He finally concluded that the results do not support the semi-strong form of EMH. Abayadeera and Senaratne (2001), investigated the impact of company specific factors in setting dividend policies in Sri Lankan share market. They found that profitability; available investment opportunities, financing choices, lagged dividends and the liquidity position of the company are the main factors to influence the dividend policies of Sri Lanka. In addition, companies are highly concerned about investors’ preference for high dividend as well as the market responses to dividend announcements.

Dissa Bandara and Samarakoon (2002), extended that the analysis of dividend announcements, firm size and dividend growth in the Sri Lankan stock market. They found that dividend announcements have significant information content in the Sri Lankan share market. Market reacts positively to dividend announcements; the informational content is stronger for the smaller firms and for firms announcing high dividend growth. Further, they found that considerable anticipatory effect for smaller firms. The market takes a considerable time to adjust the share prices compared to develop markets.


29

3 Data & Sample

3.1 Data Collection

The required data for this study is collected from the CSE. The overall sample selects from 12 listed companies covering 9 consecutive years from year 2000 to 2008 which altogether consists 100 events. It has used specific filtering criteria to narrow down the implication of the sample. Therefore, sample has represented the all material implications and most appropriately explained the market as a whole. It is observed that some companies do not have daily closing share prices evenly throughout the sample period. The missing figures were filled using previous day closing prices. In addition, to eliminate the impact on other elements such as bonus announcements, right issue announcements and abnormal price shocks / movements which resulting from the firm specific factors such as stock splits etc., the stock returns have been adjusted carefully during the period under review. Therefore, following data were collected to adjust stock returns. Bonus issue announcement dates and ratio of allocation. Right issue announcement dates, ratio of allocation and per share price of right. New issue announcements, date of issue, proceeds of the new issue. Abnormal price changes due to stock splits etc. This study attempts to test the informational content of dividend announcements in Sri Lankan listed companies. Therefore, it collects following dividend related data. Type of the dividends (this study considers the cash dividend announcements where it

eliminates the stock dividends.) Rate of dividends (the company should pay dividend throughout the sample period). This

study has considered only the final dividend payments made by the sample companies. Dividend Per Share (DPS) (this ratio is computed by multiplying par value of the share

with dividend rate. According to the New Companies Act No 7 of 2007 it does not recognize par value of the share. It considers stated value of the share. Therefore, for the period of 2007, it takes DPS value directly from the CSE electronic database )

3.2 Adjustments for Missing Prices

The sample contains some companies which were not traded throughout all trading days during the sample period. Therefore, the missing prices are filled based on the closing share price of the last trading date. Following criteria is developed to fill the missing prices during the sample period. If the missing prices are within the 20 days from the event date, those missing prices

should not be consecutively 5 days. Even that company has consecutive dividend announcements from past periods; it is automatically eliminated from the sample.

The missing dates within the estimation period and beyond 20 days from the event date; the missing prices should not be consecutively 20 days. In this case, the company has share prices within the 20 days from event date, is automatically eliminated from the sample based on this missing price gap.

3.3 The Time Period

This paper examines the shareholders response to dividend announcements using a sample taken from the CSE that covers the listed firms which announce dividends over a period from January 2000 to December 2008.


30

3.4 The Sample

A sample was taken from listed firms in the CSE. The sample has been incorporated all the relevant data to measure the impact of informational content of dividend announcements. considering the manageability of the research work, this sample must be satisfied the following selection criteria. This research adopted following selection criteria to formulate the sample. i. The company should be listed on the CSE throughout the sample period from January 2000 to December 2008. This criterion automatically exclude telecommunication sector and power and energy sector because, those were categorized during the sample period. In addition to information technology sector, stores and suppliers sector, motors sector and services sector were eliminated due to insignificant market capitalization. ii. The company must have paid dividends continuously throughout the sample period (from January 2000 to December 2008) by way of final dividends. iii. Liquidity of Stock Liquidity of stock means the ability to sell a share at a price close to the prevailing market price (Samarakoon, 1999). It developed the purpose of this criterion is to narrow down the sample which leads to select a more appropriate sample. Generally, stocks that are traded in the stock exchange are required to quickly convertible into cash at a price closer to their true value without a loss. Based on the liquidity criterion, a company is included the sample (Samarakoon, 1999).

iv. Availability of daily share prices of the companies throughout the sample period. In order to calculate the statistical measures, daily share prices must be readily available throughout the sample period.

3.5 Method of Computation, Measurement and Testing of ARs

Based on the previous works of Fama (1970), Pettit (1972), Aharony and Swary (1980), Dissa Bandara (2001), Dissa Bandara and Samarakoon (2002), Saens and Sandoval (2005): This study utilizes the methodology used by Dissa Bandara (2001) and adopts the steps discussed under the methodology section.

3.6 Identification of Events and Event Dates

This study uses dividend announcements as the “event” where any occurrence that affects the share price or the value of the firm. Johnson (1998), defined “event date” is the date on which the effect of an event is presumed to take place, or the date around which a diffused effect is presumed to be distributed. The “event date” is the first market date on which the market participants can respond to the event. The “event date” is also called as dividend announcement date in this study. It is the market date on which the dividend announcement is publicly announced. This announcement is conveyed to the market participants through the Stock Market Daily (SMD) which is the official publication of the CSE.

3.7 Definition of Analysis Period (Estimation Period and Event Period)

Previous researchers used different lengths of estimation period and event period for their studies. Among them, Brown and Warner (1985), used 250 days (239 days for estimation period and 11 days for event period). Pettit (1972), used 12 month period prior to the dividend announcement date. Dissa Bandara (2001), Dissa Bandara and Samarakoon (2002), used 200 trading days for the estimation period and 121 trading days for event period. This study uses 200 trading days for estimation period and 121 trading days for event period as illustrated in Figure 01.


31

Figure 01 : Estimation Period and Event Period

- 261 - 61 -60 AD (0) + 60

Estimation Period Event Period

(200 Trading Days) (121 Trading Days)

In order to compute the Expected Returns (ERs) for the event period, it is required to estimate test parameters for Beta (β) and Alpha (α). The estimation period starts with -61 day from the

event date and it goes back to 200 trading days. It is identified as -261 day to -61 day. All the days referring to the estimation period must be trading days. The event period starts with the day, immediately before the event date and it goes back to 60 trading dates and immediately after the event date again it goes up to 60 trading dates. Therefore, totally it consists of 121 trading dates. Finally, the total period to be reviewed is 321 (200+121) trading days which is considered as the analysis period. It is important to identify the behavior of ARs to compare with the actual stock returns and expected stock returns surrounding the dividend announcement date. The responses for the events are different from efficient market to other markets. In an efficient capital market, prices instantaneously reflect in all available information. Therefore, a person is unable to earn ARs using publicly available information. Efficient markets imply operational efficiency as well as assets prices that are allocationally efficient. Fama (1970), explained three types of capital market efficiencies, namely, weak-form-efficiency, semi-strong-form efficiency and strong-form-efficiency. In a weak form efficient market, the information on past prices or returns is not useful in achieving excess returns. Semi-strong form efficiency indicates that no investor can earn excess returns from trading on any publicly available information. Strong form efficiency implies that no investor can earn excess returns using any information, whether public or not. This study concentrates on semi-strong form efficiency of the capital market. For the purpose of examining the impact of dividend announcements on the share prices, all the relevant impacts of the other events are adjusted as explained in section 3.8.

3.8 Calculation of Stock Returns (Rit) and Market Returns (Rmt)

In calculating ARs, first stock returns and market returns have to be calculated. Before using

the “Market Model” to estimate ARs during estimation period and event period, time series of daily returns for each event under consideration has to be computed since the return data on the CSE stocks are not readily available. This study calculates daily returns for the sample of companies by using collected daily closing prices, information on dividends, right issues, bonus issues and new issues. Following formula can be used for adjusting capital changes and dividends. The returns are calculated based on Samarakoon (1997), by adjusting capital changes and dividends as follows;

Stock Returns ( Rit ) Rit = [((1+RR+BR) * ( Pt + D)) / (RR * Pr + Pt-1)] -1 (1)

Where, Rit = the rate of return of firm ‘i ’ on day‘t ’ RR = right issue ratio BR = bonus issue ratio Pt = closing share price on day ‘t ’ D = dividend per share (dividend rate * par value of a share)


32

Pr = per share price of right issue Pt-1 = closing share price on day ‘t-1’ (previous trading date) This formula calculates the stock returns, while adjusting for capital changes and it reduces several steps as well as time of data preparation for the study. The adjustments of dividends, right issues, bonus issues are made on the date in which the ex-dividend, new issues, ex-rights and ex-bonus dates fall.

Market Returns (Rmt )

Rmt = (ASPIt –ASPIt-1) / ASPIt-1 (2) Where, Rmt = return of the All Share Price Index (ASPI) for t

th day

ASPIt = ASPI for tth day (current day)

ASPIt-1 = ASPI for the day before tth day (previous day)

4. Methodology

4.1 Event Study Methodology (ESM)

The ESM is a commonly used technique to estimate the stock price reaction to firm specific informational events such as dividend announcements, right issue announcements, merger announcements, stock splits etc. or market related informational events such as terrorist attacks, political changes etc. Fama (1991), expresses that the market model can be used to test the market efficiency when the “firm specific” factors are in existed. This paper basically uses Brown and Warner (1980, 1985) methodology. Mainly, Brown and Warner approach randomly selects event dates and stocks to carry out event studies without assuming a particular distribution of stock returns. This approach considers the normality of ARs. It is important to show the impact of the selected event whereas impact from the contaminated events (other events) should be eliminated. Therefore, it is required to distinguish between the expected and unexpected returns on stocks. For the purpose of calculating ARs, ERs have to be computed using parameters estimated from a period prior to the event. Then it, computes ARs by subtracting ERs from the Actual Returns of the market.

4.2 The Market Model

This study uses the market model to measure the impact of changing dividend on the share price of the firm ‘ i ’. The market model pronounces that the return on a security depends on the return on the market portfolio and the extent of the security's responsiveness as measured by beta. The return also depends on conditions that are unique to the firm. Initially, the market model estimates α (Alpha) and β (Beta) which are referring to the returns from the estimation period. The market model is estimated through Ordinary Least Square (OLS) regression to estimate the parameters of the two-variable regression models. Market returns during the estimation period are treated as independent variables while dependent variable is the returns of the firm ‘ i ’ for the purpose of running the regression. Regression analysis produces estimates of regression intercept α and regression slope β. ERs have been computed as follows; Rit = αi +βi Rmt + εit (3)

Where, Rit = the rate of return on firm ‘i’ on day‘t’ Rmt = the rate of return on a market portfolio of stocks (ASPI) on day‘t’ αi = the intercept term (alpha) of stock ‘i’ βi = the systematic risk of stock ‘i’ (beta) εit = the regression error term


33

4.3 Calculation of Abnormal Returns

In event studies, what one wants to measure is how much of the price reaction to news (in this case the announcements of dividends) is abnormal. That is how much of the reaction is in excess of what was anticipated. Therefore, in order to calculate the ARs one has to compute the returns in excess of normal or ERs around the date that new information of a stock is released to the market. ERs are calculated as follows.

E(Rit ) = Rmt (4)

Where, E (Rit) = expected return on firm ‘i’ on day‘t’

= estimated regression intercept (alpha) of stock ‘i’

= the estimated systematic risk (beta) of stock ‘i’

Then it is required to calculate ARs for the event period. According to previous research findings, this model can be based on simple statistical relationships such as the OLS market model or on more theoretical economic models such as Capital Assets Pricing Model (CAPM) and Arbitrage Pricing Model (APM). The AR can be measured in three ways. Dissa Bandara (2001) and Dissa Bandara and Samarakoon (2002) derived OLS market model by substituting market adjusted return model. This study uses OLS market model by substituting market-adjusted return model. Therefore following formula can be derived; ARit = Rit – E(Rit ) (5) By using OLS market Model with substituting market-adjusted return model, following formula can be derived to calculate ARs;

ARit = Rit – [ + Rmt ] (6)

Where, ARit = abnormal return of the firm ‘i’ during the event period Rit = the rate of return of the firm ‘i’ on the day‘t’ in the event period.

4.4 Organize and Group Abnormal Returns 4.4.1 Calculation of Average Abnormal Returns (AARs)

After computing ARs for each event, it should be calculated some representative measure which can be used the sample of securities under consideration. After calculating of ARs for each day of each event, it is summed up to get AARs. “Σ” is product notion of the AARs. The

measure of AARs normalizes the ARs of the selected sample. It can be computed as follows; N

AARt = 1 Σ ARit (7)

N i =1 Where, AARt = average abnormal return for day‘t’ N = number of events in the sample 4.4.2 Calculation of Cumulative Average Abnormal Returns (CAARs)

Using the Brown and Warner methodology, it is commonly observed leakage of information which occurs when information regarding a relevant event is released to a small group of market participants before official public announcement. ARs on the announcement date are then a poor indicator of the total impact of the information release. Therefore, a better indicator would be the CAARs which are the sum of all ARs over the event period. The CAARs captures the total firm specific stock movement for an entire period when the market might be responding to new information.


34

The CAARs can be calculated as follows; T

CAART = Σ AARt (8)

i =1 Where, CAAR T = cumulative average abnormal return returns up to day ‘T’ T = time (number of days over which ARs are cumulated) AARt = average abnormal return for day‘t’

4.5. Significance Testing

A test of significance is a procedure by which sample results are used to verify the truth or falsity of a null hypothesis. The null hypothesis is rejected, if the values obtained from the calculations are higher than the critical values. The variables AAR% and CAAR% are used to measure the informational content of final dividend announcements and the efficiency within which this information is impounded into the share price. The null hypothesis is that AAR% is drawn from a distribution with zero means; that means announcement of final dividends have no systematic effect on respective share prices on the dividend announcements date. In order to test hypothesis that the AAR% and CAAR% at day i are statistically different from zero, the following t-statistics for AAR % and CAAR% are calculated. 4.5.1 Significance Testing (Parametric) for AAR% Significance of AARs for the event period is tested using T-statistics. The T-statistics are calculated as follows;

Assume; ARit ~ (0 , )

Where,

= Variance of AR in the event period

N

Then, 2 (ARt) = Σ (ARit – AARt)

2 (9)

i =1

2 (AARt)

=

σ

2 ( ) (10)

(AARt) =

σ

2 (ARt) (11)

T (AAR) = AARt

(AARt) (12)

Where, N = number of events in the sample


35

4.5.2 Significance Testing (Parametric) for CAAR%

Significance of CAARs for the event period is tested using the T-statistics and they are calculated in the following manner;

Assume; CAARt ~ (0, σ

2T)

Then, (CAARt ) = T σ

2 (AARt) (13)

T (CAAR) =

CAARt

(14)

σ (CAARt)

Where, T = time: number of days over which AARs are cumulated

5. Empirical Results

Behavior of the Overall Sample

Table 1 presents daily AAR %, daily CAAR % realized by shareholders in different window periods surrounding the dividend announcement date and direction of ARs on dividend announcement date. Figure 2 explains the behavior of CAAR% in the 121-day window period surrounding the event. The potential anticipatory reaction for dividend announcements can be seen in the longer window period. However, to minimize the cumulative effect in the longer window period, it also analyzes the 2-day, 3-day, 5-day and 20-day window periods. Panel-B of Table 1 presents the CAAR% from 20 days before event date to 20 days after the event date. According to Panel - A of Table 1, it shows that the AAR% on the dividend announcement date is 10.43% and it is significant at 1% level. It is the highest value in the (-1, +1) window period; CAAR% value is 11.10%. According to the Panel B of Table 1, the CAAR% for the window period of the (-20,-1) is 3.10% and the CAAR% for the window period of the (0, +20) is 6.90%. The strong and quick market response to the dividend announcement date can be seen on the Figure 2. It shows CAAR% on the Y-axis and 121-day window period on the X-axis. Day 0 is the dividend announcement date (referred as event date). Dissa Bandara (2001), and Dissa Bandara and Samarakoon (2002), found that there is an informational content of dividend announcements in Sri Lankan share market and empirically found that the positive dividend announcements must have a positive announcement effect. Positive dividend announcements indicate positive signals to the market. Market participants respond to the dividend signals accordingly. Positive dividend announcements provide good earning potentials of the market. Market participants respond and may like to adjust share price based on the good news; that is positive dividend announcements have positive impacts on returns. According to the Panel C of Table -1 it summarizes the direction of the ARs on the event date. It provides average number of events which gain positive or negative ARs on the event date. It is clearly indicated that 93% of average events from the sample are having positive ARs on dividend announcement date. On the other hand, 7% of the events have negative ARs. Positive ARs have significant part in connection with the effect of dividend announcements on the event date. Therefore, it is concluded that the overall market respond rate is positive on the dividend announcement date. It is observed that latest 121-day window period, the market steadily adjust to positive ARs. This can be shown in the Figure 2 from day -20 to day 0 window period the CAAR% goes up to 9.79% from -3.76% and the AAR% on -19 day change is significant at 5% level. It is observed that on the dividend announcement day, the CAAR% reaches maximum point. In addition, there is an anticipatory response before the event date. That means the share prices


36

tend to increase before the dividend announcement and shows a gain for some group of market participants in advance. During the 121-day window period, it is shown that the flat effect of CAAR% from day -30 to day -20. It implies that market does not strongly response to the dividend announcement before 20 day period. The anticipatory response is starting from day -20. It is observed that from day -10, steady anticipation is created. Furthermore, it is shown that on day -7, it earned -1.43% of CAAR% and AAR% is at 5% significant level. It shows that the anticipatory reaction is starting at least one week before the event date. Figure 2 depicts that the significance of positive ARs on the dividend announcement date and subsequently prices adjusts slowly after the event. It shows that CAAR% has been increased from -3.76% to -1.42% and are at 5% significant level on -20 day to -7 day. Slightly increased of CAAR% can be seen before 3-4 weeks times and it is not a so strong response compared to one week before the announcement. It is clear that significant anticipatory response is arising one week before the dividend announcement date. It is observed that during the period of -20 day to -1 day, CAAR% has been increased from -3.76% to -0.64%. It implies that the market price adjusts for the new information in advance and it is ready to absorb the gain of the dividend announcement and the event day returns are highly significant. That is almost two weeks before the information on the new event is realized to the market. It is accepted that before taking investment decisions, the price sensitive information should be readily available to the market participants in an efficient manner which describes that the information dissemination process would be efficient in the market. The market participants need price related information before taking the investment decision. It is important to provide accurate, timely and relevant information in an efficient manner. When analyzing the semi-strong form of EMH relating the dividend announcement as an event, the share prices seem to be instantaneously and unbiasedly reflect the informational content of dividend announcements. It is interesting to note that the impact of information dissemination process in the market which describes that the price related information convey to the market participants is a significant aspect in the capital market. The evidence supports that price adjustments would be compressed within 1-2 weeks time and it would be caused due to the recent development of the official publications of the CSE which is providing information through internet and other official publications that the market participants could response accordingly. Occasionally there would be an impact of insider abuse i.e. price sensitive information release to the market by the related insiders, before the official announcement is released and other misconducts of some of the market participants. According to the results given in Table 1 and Figure 2 which show an anticipatory response in the 121-day window period considering the sample, it does not consistent with the semi-strong form of EMH. This conclusion is based on the previous return patterns, significant anticipatory responses on the window period and price adjustments after the dividend announcement date.

Figure 2 : Cumulative Average Abnormal Returns (CAARs %) of Overall Sample (N=100)


37

*** Significant at 1% ** Significant at 5% * Significant at 10%

TABLE-01

AVERAGE ABNORMAL RETURNS FOR THE OVERALL SAMPLE

PANEL-A

AVERAGE ABNORMAL RETURNS (AARs%)

EVENT DATE AAR% T(AAR)

-10 0.12% 0.516

-9 0.15% 0.512

-8 -0.08% -0.321

-7 0.61% 2.173**

-6 0.39% 1.435

-5 -0.29% -0.887

-4 -0.17% -0.735

-3 0.24% 0.626

-2 0.29% 1.132

-1 0.32% 1.353

0 10.43% 6.117***

1 0.31% 0.879

2 -0.87% -0.866

3 0.20% 0.677

4 -0.01% -0.040

5 0.15% 0.675

6 -0.55% -1.827

7 -0.23% -0.921

8 -0.34% -0.899

9 -0.32% -1.347

10 0.33% 1.238

PANEL-B

CUMMULATIVE AVERAGE ABNORMAL RETURNS (CAARs%)

WINDOWS CAAR%

(-20,-1) 3.10%

(0,+20) 6.90%

(-2,+2) 10.50%

(-1,+1) 11.10%

(0,+1) 10.70%

PANEL-C

DIRECTION OF ABNORMAL RETURNS (ARs) ON DAY 0

DIRECTION NUMBER %

Positive 93 93%

Negative 7 7%

Total 100 100%


38

It can be seen that after the event date, CAAR% has been reduced from 10.10% on day +1 to 8.80% on day +7 of after the official dividend announcement. On average from day +10 to day +20 there is a slight decline of CAAR%. It is depicted that anticipation can be shown almost 2 weeks prior to the dividend announcement and price adjustment can be seen almost after the 2 weeks from the dividend announcement date.

6. Conclusion

This evidence supports that the informational content of dividend announcements in Sri Lankan share market and consistent with the previous works of Dissa Bandara (2001) and Dissa Bandara and Samarakoon (2002). It is found that 93% of average events of the sample are having positive ARs on the dividend announcement date where 7% of the events have negative ARs. Therefore, the overall market response rate is positive on the announcement date. There is significant positive ARs reported on the dividend announcement date at 1% significant level and prices adjust even after the event date. Using a sample of 100 events this study finds that a potential anticipatory reaction for dividend announcement as well as a delayed reaction for dividend announcement in the longer window period. It is clearly reported that anticipation is visible two weeks before the dividend announcement date.

A capital market is considered as efficient if the share prices fully, instantaneously and unbiasedly reflect all relevant information. This paper examined the semi-strong form of EMH which asserts that current stock prices reflect not only historical price information but also all publicly available information. The results suggest that on average; the share prices instantaneously and unbiasedly reflect the publicly available information. Further the evidence finds that considerable anticipation before the event date. This anticipation would limit the information dissemination process of the market or insider abuse or any other misconduct of some of the market participants. This may lead some participants to predict the future prospects of the companies and earns ARs over the others. Therefore, it can be concluded that the results do not support the semi-strong form of EMH. However the results are consistent with the previous research findings on the Sri Lankan share market.


39

References Abayedeera, N., and S. Senarathne, 2001, “An Empirical Study of the Factors Influencing

Dividend Policies of Sri Lankan Companies”, Proceedings of the Annual Research Sessions- 2000/2001 of the Faculty of Management Studies and Commerce. University of Sri Jayewardenepura. pp 1-12.

Adelegan, O. J., 2003 “Capital Market Efficiency and the Effect of Dividend Announcements

on Share Prices in Nigeria”, African Development Bank Report, Blackwell Publishing Ltd. pp 218-236.

Aharony, J., and I. Swary, 1980, “Quarterly Dividend and Earnings Announcements and

Stockholders’ Returns: An Empirical Analysis”, Journal of Finance, Vol. XXXV, No.1, pp 1-12.

Asquith, P., and D.W. Mullins, 1986, “The Impact of Signaling with Dividends, Stock

Repurchases, and Equity Issues”, Financial Management, pp 27-44. Bhattacharya, S., 1979, “Imperfect Information, Dividend Policy, and ‘The Bird in the Hand’

Fallacy”, Bell Journal of Economics, Vol. 10, pp 259-270. Below, S. D., and K. H. Johnson, 1996, “An Analysis of Shareholder Reaction to Dividend

Announcements in Bull and Bear Markets”, Journal of Financial and Strategic Decision, Vol 9, No: 3, pp 15-26.

Brown, S. J., and J. B. Warner, 1980, “Measuring Securities Price Performance”, Journal of

Financial Economics, Vol. 8, pp 205-258. Brown, S.J., and J. B. Warner, 1985, “Using Daily Stock Returns: The Case of Event

Studies”, Journal of Financial Economics, Vol. 14, pp 3-31. Cheng, L. T. W., H. G. Fung, and T. Y. Leung, 2004, “Information Effects of Dividends:

Evidence from the Hong Kong Market”, Journal Review of Quantitative Finance and Accounting. Vol 28. No 1, pp 23-54.

Colombo Stock Exchange, 2000-2008, Annual Reports, Fact Books and Data Library

(Electronic Form of Publications) Dissa Bandara, D. B. P. H., 2001, “Dividend Signaling, Market Responses to Dividend

Change and Capital Market Efficiency: An Empirical Analysis”, Proceedings of the Annual Research Sessions- 2000/2001 of the Faculty of Management Studies and Commerce, University of Sri Jayewardenepura, Sri Lanka, pp 23-51.

Dissa Bandara, D. B. P. H., and L. P. Samarakoon, 2002, “Dividend Announcements, Firm

Size and Dividend Growth in the Sri Lankan Stock Market”, Sri Lankan Journal of Management, Vol 7. Nos. 3 & 4, pp 228- 245.

Elfakhani, S., 1995, “An Empirical Examination of the Information Content of Balance Sheet

and Dividend Announcements: A Signaling Approach”, Journal of Financial and Strategic Decision, Vol 8, No: 2, pp 65-76.

Fama, E.F., 1970, “Efficient Capital Markets: A Review of Theory and Empirical

Works”, Journal of Finance, Vol XXV, pp 383-417. Fama, E.F., 1991, “Reply to Efficient Capital Markets: II”, Journal of Finance, Vol XLVI, No.

5, pp 1575-1617. Fung, S., and J. Patel, 2005, “Information Content of Dividend Policy: Evidence from

Relations between Dividend Announcements, Analysts' Earnings Expectations, and Equity Prices, http://www.fma.org

http://www.fma.org/Chicago/Papers/Information_Content_of_Dividend_Policy_ScottFung.pdf


40

John, K., and J. Williams, 1985, “Dividends, Dilution, and Taxes: A Signaling Equilibrium”, Journal of Finance, Vol. XL, pp. 1053-1070.

John, K., and L. H. P. Lang, 1991, “Insider Trading around Dividend Announcements: Theory

and Evidence”, Journal of Finance, Vol. XLVI, No. 4, pp. 1361-1389. Johnson, K.H., 1998, “Graphical Analysis for Event Study Design”, Journal of Financial and

strategic Decisions, Vol 11, No 1, pp 61-71. Kalay, A., and U. Loewenstein, 1985, “Predictable Events and Excess Returns: the Case of

Dividend Announcements,” Journal of Financial Economics, Vol 14, pp 423-449.

Kalay, A., and U. Loewenstein, 1986, “The Information Content of the Timing of Dividend Announcements,” Journal of Financial Economics, Vol 16, pp 373-388.

Koch, A.S., and A. X. Sun, 2004, “Dividend Changes and the Persistence of the Past Earning

Changes”, Journal of Finance, Vol LIX, No 5, pp 2093-2116. Miller, M., and K. Rock, 1985. “Dividend Policy under Asymmetric Information”, Journal of

Finance, Vol. XL, pp.1031-1051. Modigliani, F., and M.H. Miller, 1961, “Dividend Policy, Growth and the Valuation of Shares”,

Journal of Business, Vol 34, pp 411 -433. Pettit, R.R., 1972, “Dividend Announcements, Security Performance, and Capital Market

Efficiency”, Journal of Finance, Vol XXVII, No5, pp 993-1007. Pettit, R.R., 1976, “The Impact of Dividend and Earnings Announcements: A Reconciliation”,

Journal of Business, Vol 49, No1, pp 86-96. Ryan, P.A., S. Besley, and H.W. Lee, 2000, “An Empirical Analysis of Reactions to Dividend

Policy Changes for Nasdaq Firms, Journal of Financial and Strategic Decisions, Vol 13, No 1, pp 35-44.

Saens, R., and E. Sandoval, 2005, “Measuring Price Performance using Chilean Daily Stock

Return: The Event Study Method”, Cuadernos de Economia, Vol. 42, pp 307-328. Samarakoon, L.P., 1997, “Efficiency of the Sri Lankan Stock Market”, Sri Lankan Journal of

Business Economics, pp 1-22. Samarakoon, L.P., 1999 , “A New Measure of Market Liquidity of Shares; the Case of

Colombo Stock Exchange”, Proceedings of the Annual Research Sessions of the Faculty of Graduate Studies, University of Sri Jayewardenepura, Sri Lanka, pp 165-182.

Uddin, M.H., 2003, “Effects of Dividend Announcement on Shareholders’ Value: Evidence

from Dhaka Stock Exchange” http://www.fee.uva.nl/fm/conference/hyderabad Valipor, H., V. Rostami, and M. Salehi, 2009, “Asymmetric Information and Dividend Policy in

Emerging Markets: Empirical Evidences from Iran”, International Journal of Economics and Finance, Vol 1, No 1, pp 203-211.

http://www.fee.uva.nl/fm/conference/hyderabad

41

Interest Rate Behaviour:

The Sri Lankan Case

W.M.Hemachandra*1

Abstract This paper defines the interest rate followed by identification of determinants of interest rates in Sri Lanka. The Supply and demand conditions, monetary policy of the Central Bank, inflation, loan type and several other factors have been identified as determinants of interest rates. After conducting a theoretical analysis, an attempt has been made to confirm the determinants of interest rates by using data and information on interest rate behavior in Sri Lanka during the recent past.

Keywords: Credit, borrowings, loans, monetary policy, policy rates, Open Market Operations (OMO), market rates, inflation, nominal interest rate, Bank rate, repo and reverse repo rates, call money rates, risk, collateral, government securities, duration, fixed deposits, savings deposits, savings, investment, primary market, secondary market, secondary window, financial crisis, term structure, competition, informal market, SRR, market liquidity, inflationary expectations.

1 Dr. W M Hemachandra is an Assistant Governor of the Central Bank of Sri Lanka. The writer extends sincere thanks to late Dr. P W R B A U Herat, former Deputy Governor of the Central Bank of Sri Lanka for his editorial corrections/comments. The author bears the responsibility for all views expressed, and does not represent the views that of Central Bank of Sri Lanka.


42

1. Introduction An interest is paid as compensation for temporarily giving up the ability to spend money. Therefore, it is a price paid for using others’ funds. It may also be defined as the compensation for the service and risk of lending money. Lenders also charge interest to recover their processing cost. Hence, lenders charge interest in order to compensate for inflation and bearing risk, for postponement of consumption, and to recover processing costs. Borrowers are prepared to pay interest because they are able to spend money to afford large purchases presently. Therefore, interest is the cost of borrowing money. Accordingly, interest can be considered as a cost to one party and income to another. Businesses are willing to pay interest to borrow for investments. Similarly, banks are willing to pay interest on deposits because they can lend those at a higher rate.

Interest, computed as an interest rate is usually expressed as a percentage per annum and therefore, can be compared. In a competitive market, interest rates vary from day to day; tomorrow’s interest rate will be different from today’s interest rate.

Central Banks use interest rate as a policy instrument to achieve monetary policy objectives. They expect to manage market liquidity by changing the policy interest rate, depending on the monetary policy requirements. In the long run, the interest rate is an important variable in determining savings and investment. On these grounds, interest rates can be considered as one of the major instruments/variables available in the modern monetary world. The purpose of this paper is to identify the factors that determine the interest rates in the context of Sri Lanka. Initially, it explains the theoretical background of the interest rate and determinants of interest rates. In order to confirm the theory, the paper uses data and policies pertaining to the recent history of Sri Lanka. Section 2 of the paper identifies the determinants of interest rates that are common to any market. Section 3 is devoted to explain the interest rate regimes that could be identified from the theories and the practices of Sri Lanka. Section 4 rationalizes the determinants of interest rates using the data and information from Sri Lanka.

2. Determinants of Interest Rates

According to the established knowledge, interest rates are determined based on a number of factors which include supply and demand for funds, policy makers’ decisions, inflation, types of loans and several others. The following is a brief explanation about these factors.

2.1 Supply and Demand

Similar to determining prices in the commodity market based on the demand and supply conditions, the level of interest rate is also determined by the supply and demand for funds. An increase in the demand for funds will increase interest rates, while a decrease in the demand for funds will decrease interest rates. Conversely, an increase in the supply of funds will reduce interest rates while a decrease in the supply of funds will increase interest rates. The supply of credit is increased when the amount of money made available to borrowers or loanable funds are increased. When a deposit is made into a bank account, the bank uses that money for its business and investment. In other words, bank can utilize deposits to lend money to other customers. Credit is created when money is lent. In this process, when the supply of credit is increased, the price of borrowing (interest) decreases. Credit available to the economy may decrease when the lenders defer the repayment of their loans. As a result of declining in availability of funds the interest rate will increase in the economy. Hence, demand and supply would decide interest rates of an economy.

Interest Rate Behaviour: The Sri Lankan Case

43

2.2 Monetary Policy of the Central Bank

The monetary authority influences the interest rate when it is used as a monetary policy instrument. Therefore, the policy interest rates can vary from time to time depending on the monetary policy requirements. Policy rates decided by the Central Bank can influence the “open market operations” which is basically the buying or selling of securities from or to the market. When the Central Bank buys securities from the market, money is injected, following which the market interest rate decreases. When the Central Bank sells securities, money from the market is drained to the Central Bank resulting in a decline in funds available in the market, forcing a rise in the market interest rate. In this way, Central Banks can influence interest rates of the market. In a tight monetary policy situation interest rates are rising, while the reverse happens with loose monetary policy.

2.3 Inflation

Inflation is another factor that determines the interest rates. The nominal interest rate (i) is expressed as expected inflation (π) plus real interest rate (r) i.e., ( π + r = i). Therefore, the higher the rate of inflation, the higher the nominal interest rate. There will be a difference between the nominal and the real interest rate. This occurs because lenders will demand a higher interest rate as a compensation for the decrease in the purchasing power of the money that they will be repaid in the future.

2.4 Types of loans

The interest rate charged usually varies depending on the amount of loan or deposit. Normally, when the amount of loans to be borrowed is higher, the interest rate charged will also be higher since there is a higher risk. Risk is an important determinant of the level of interest rates. Higher risk leads to a higher interest rate. The bigger the chance of the loan not being repaid, the higher the interest rate. However, if the loan is secured by way of a collateral, the lender has less risk and the interest rate will probably be lower. A larger amount of loans carry a higher risk and therefore the interest rate charged will also be higher. Since Government Securities have a lower risk, the interest rate on Government Securities is expected to be lower than on other instruments, in general.

The duration of a loan is also a factor determining the interest rate. In general, the longer the period that an individual borrows from a financial institution, there is a higher risk and as a result there will be a higher interest rate for the long term compared to the short term.

2.5 Other Factors

There will be different interest rates depending on the nature of the instrument. For instance, fixed deposits which remain for a long period in an institution have a higher interest compared to savings deposits which may remain small or be variable. Similarly, from the recipients’ point of view, the repurchase (repo) rate which is normally the borrowing rate will be lower than the reverse repo rate which is used for lending.

2 In the Government Securities market,

the primary market rate may differ from secondary market rates. There will be a difference between the nominal and the real interest rate. At the same time, the international market rate will be different from the domestic market interest rates. For instance, the interest rate for concessional lending will be lower compared to commercial type loans.

In the Government Securities market, the rate would vary depending on the fund requirement of the Government (the size of borrowing). Cost of services will be a factor that determines the lending rates. The margin added to the interest rate as profit varies and will be a factor that determines the interest rate.

2 Repo is a mechanism to receive funds on temporary basis while transferring the ownership of securities. Reverse

repo is lending funds temporarily on receipt of securities. The period under transaction varies from one hour or so to a longer depending on the life time of the security.


44

3. Interest Rate Regimes

In the Classical theory, interest rates were assumed to be determined by the interaction of savings and investment. Savings increase when the interest rate increases and the investments increase when the interest rate decreases. The equilibrium interest rate is decided when the savings are equal to investment. In the classical theory, it was also assumed that the quantity of money demanded does not affect the interest rate since all economic agents hold a minimum money balance for their requirements. When the savings exceed investment, interest rate declines. Therefore, the interest rate is an equilibrium price in the Classical theory whereas in the later theories interest rate determines money demand. In the Keynesian theory, the interest rate may not determine the equilibrium level of savings and investment. In the Keynesian’s version the process of interest rate determination was assumed as a monetary phenomenon. The nominal rate of interest is determined by the market where the demand for money is equal to the supply of money. Further, investment is inversely related to the interest rate. Therefore, a lower interest rate stimulates investment in a country. The interest rate is a variable which determines the money demand where the behaviour of bond market is involved. It was assumed that when the expected interest rate is higher there will be a higher demand for bonds, reducing the demand for money. Therefore, the interest rate had an inverse relationship with demand for money. In the Keynesian version it was required to maintain a lower interest rate since a lower interest rate will increase investment. Governments were required to maintain a low level of interest rate by way of ceilings, imposing some controls. The Keynesian version prevailed in many countries until such time that the economic liberalization policies were implemented during the 1970s. In contrast to the Keynesian version where an inverse relationship between interest rate and investment was assumed, in the market repressionerists’ theories, the interest rate was assumed to be positively related to investment. The proponents of this theory, such as McKinnon and Shaw, pointed out that the lack of savings in the developing countries and the ceilings on the interest rate were the major problems to increasing investment. Accordingly, they proposed that rather than having ceilings and controls on the interest rates, they should be allowed to be determined liberally. In this way they invited market reforms while removing ceilings on interest rates. As they expected, high interest rates would lead to higher amount of savings and higher savings will promote investment. Therefore, they identified a positive relationship between the interest rate and savings and the interest rate and investment. It was assumed that higher level of savings will increase investment, thereby increasing economic growth. Accordingly, market reforms including removal of interest rate ceilings were proposed by this group of economists.

As against the McKinnon-Shaw version, the market structuralist group came up with a new version saying that since there is an unorganized market (curb market) in the developing countries, the expectations of McKinnon-Shaw cannot be achieved. According to them, increasing the interest rate would not be a solution in the developing countries, since the increase in interest rates will increase the organized sector interest rates, thus encouraging unorganized sector funds to be transferred to the organized sector. The increase in interest rates in the formal sector will lead to an increase in the interest rates in the informal sector, which is substantial in the underdeveloped countries. If the funds are transferred from the informal sector to the formal sector, due to the increase in the interest rate there will be a scarcity of funds in the informal sector. Therefore, increased interest rates as proposed by McKinnon-Shaw will not promote total savings of a country due to an increase in the interest rates in the informal sector as well.

As Chang and Jung (1984) indicated, substitutability between time deposits and curb market loans may not be high for large savers because these assets have very different risk attributes. Shea Jia-Dong (1992) has indicated that an increase in deposit rates raises savings, increases bank credit, reduces the curb market rate and so stimulates capital by increasing output and welfare. So, curb market behaviour can be incorporated into the McKinnon-Shaw model without producing the neo-structuralist policy conclusions.

According to the thoughts of market failure school, (Grossman and Stiglitz (1980), Stiglitz (1989)) the existence of market failure is due to information asymmetry and therefore, the


45

emphasis is on the need for monitoring of financial institutions. Hence, the collection of information and monitoring of financial institutions have to be done by the Government rather than the private sector. According to them, as the Banks have to use credit rationing in order to reduce portfolio risk, the existing interest rate does not reflect the market clearing rate. Therefore, an interest rate, which is efficient, cannot be found in the market and hence, the market failure school rejects the repressionerists’ view that the financial resources are efficiently channeled by the banking system unless there is an intervention by the Government. It is only when there is no information asymmetry, that the equilibrium rate of interest will be determined according to the market forces. Therefore, Government has to intervene and decide on interest rates. In Islamic banking there is no interest rate. The cost of borrowing can consist only of services cost, overhead cost, risk premium, profit and compensation for inflation. These elements other than interest can influence the cost of borrowing. Interest, which is called “Riba” in Islamic terms, is prohibited and therefore can exist only as a profit margin for lenders.

4. Experience of Sri Lanka

From the very inception of the establishment of the Central Bank, the Bank Rate was used as a policy rate and there has been Treasury Bills rate, savings and lending rates and long term interest rates for Government debt instruments such as rupee securities. Until 1977 there was an administrative procedure to determine interest rates. As and when agents wanted to revise the rates, approval from the authorities had to be obtained. In order to implement the monetary policy, the Bank Rate was used extensively. In the long term capital market there were only rupee securities until 1997.

The establishment of the secondary window for Treasury bills was another initiative that the Central Bank has taken in 1981, which provided room for open market operations. Deviating from selective credit controls, Central Bank switched to controlling reserve money by taking into account Government fiscal operations and the flow of external assets. Open Market Operation (OMO) had a significant advantage over other monetary policy instruments due to its flexible nature. The OMO decided, among other things, the interest rate to be charged for Central Bank securities. Before 1977, interest rate had been administered through changes in Bank Rate and administrative measures. Administered rates were based on the rates in concessional credit schemes for refinance and rural credit schemes. Changes in the Bank Rate had an impact on commercial bank interest rates. The interest rates for concessional credit under refinance schemes and for various developmental credit schemes were decided by the Central Bank after considering the effects of these schemes on developmental activities. Hence, the entire interest rate structure prior to 1977 was an administered one. Along with the liberalization policies introduced in 1977, as the first step of a series of changes, the Bank Rate was raised upward and following this the National Savings Bank and the commercial banks increased the rates they paid to depositors. The changes in deposit rates of the banks were followed by upward movements in the rates on inter-bank call loans. As a result of these changes, the interest rate structure in the economy moved to a higher level. After 1981, Central Bank interest rate policy was linked to OMO. With the opening of the secondary window for Treasury bills and the repurchase facility, the interest rates on these instruments in the secondary market were determined on a daily basis taking into account the monetary management objectives. The interest rates so decided had a bearing on the interest rate structure of the market. Until 1997, interest rates on long-term debt instruments were decided administratively since there was no long-term debt instrument for which the interest rates were market determined. It was only after 1997 when the Treasury bonds were introduced as a long-term Government debt instrument that a ‘market determined interest rate’ for long term debt instruments appeared.

After the active OMO was started by the Central Bank, the interest rate of that market began to be used to achieve the monetary policy objectives of the Central Bank. The Repo and Reverse Repo rates corridor was decided based on the monetary policy objectives which provided lower-end and upper-end margin for policy rates. With OMO becoming active, the importance of the Bank Rate as a policy interest rate disappeared. Among other things, the


46

liquidity management was expected to be conducted through the interest rate changes. A tight monetary policy instance was pursued by increasing interest rates while loose monetary policy was conducted by reducing rates of repo and reverse repo markets. Since 2004, the Central Bank of Sri Lanka has been following a tight monetary policy for which the policy rates were increased. There is a close relationship between the changes of policy rates and the market rates where the market interest rates fluctuate based on these policy rates.

4.1 Policy Rates and the Market Rates

As stated above, policy rate of the Central Bank is one of the major determinants of the interest rates. The Central Bank from time to time changes interest rates based on the monetary policy objectives. The Repo and Reverse Repo Rates are the policy rates of the Central Bank. Market rates change according to the changes in the policy rates. Information given in the Table I shows the relationship between the policy rates and market rates in Sri Lanka.

Table I

Policy Rates and the Market Rates (per cent p.a.)

Year/

Month

Policy

Rate

(Repo

rate)

Fixed Deposit Rates of

Commercial Banks

AWDR* Treasury Bill Market

(364 days)

Primary Secondary Max Min

2002/01 12.00 14.50 9.50 10.53 13.71 13.63

2002/03 11.50 14.00 10.00 10.11 13.61 13.23

2002/07 10.50 14.00 8.25 9.49 12.96 13.07

2003/05 8.25 10.00 6.00 6.14 8.55 8.55

2003/08 7.50 8.30 5.50 5.66 7.52 7.53

2003/10 7.00 8.30 5.00 5.43 6.98 6.96

2005/06 8.25 9.80 5.50 5.45 9.25 9.16

2005/12 8.75 11.50 5.50 6.24 10.37 10.31

2006/07 9.13 12.50 5.50 6.59 10.50 10.52

2007/12 10.50 18.00 5.50 8.18 14.20 14.26

Source: Central Bank of Sri Lanka *Weighted Average Deposit Rates of commercial banks.

Figure 1 indicates the movements of market rates and the policy rates. It is clear from the Figure that when the policy rate has declined, the fixed deposit rates and AWDR of the commercial banks have followed the trend. Similarly, when the policy rates increased, the market rates also increased.

Figure 1


47

Market rates differ from policy rates. Normally, the policy rates are lower than the market rates. For instance, as at end of December, when the Reverse Repo rate was 12% the Weighted Average Prime Lending Rate (AWPLR) was 15.35% and call money rate was 14.87% (CBSL data). The difference between market rates and policy rates is due to various other factors. These factors among other things include cost of the funds, risk premium, type of instrument, duration, transaction type, characteristics of loans or deposits, inflationary expectation, market liquidity, competition, seasonal behaviour, nature of the market (primary or secondary), maturity period, market preference and the technology used. The relationship among these elements can be seen clearly from the behaviour of interest rates in the Sri Lankan market.

4.2 Cost of Services/Funds

The major cost of services of financial institutions includes cost of overhead, labour, documentation, postage and legal expenses etc. The higher the cost for these services the higher the interest rate charged from the borrowers. For instance, as given in Table 2 when the operational cost of the commercial banks is high the interest spread maintained by banks is also high.

Table 2

Cost of Banks and Interest Spread – (% of Assets)

2005

Country Operational Cost Interest Spread

Bangladesh 1.9 2.5

India 2.3 2.6

Pakistan 2.4 3.0

Sri Lanka 3.6 4.2

Source: Central Bank of Sri Lanka

Interest paid to depositors is a cost to financial institutions. Interest charged from borrowers would increase if the deposit rats are increased. For instance, the maximum fixed deposit rate of commercial banks in October 2003 was 8.3% and the corresponding Weighted Average Prime Lending Rate (AWPLR) was 9.14%. When the deposit rate of commercial banks increased to 13.5% in September 2006 the AWPLR increased to 13.58%. As indicated in Table 3, lending rates are higher than deposit rates showing the influence of cost of funds on lending rates.

4.3 Type of Instrument/Market

The type of instrument is also a factor determining the structure of interest rates. What is meant by the type of instrument is whether the instrument is a deposit or loan, whether it is a primary market or secondary market, whether it is a Government security or corporate instrument, or whether it is short term or long-term instrument etc., Normally, lending rates are higher than the savings rates. This is clear from the statistics given in the Table 3.

Table 3

Deposit Rates and Lending Rates (percent, p.a.)

Year/Month AWDR* AWPLR**

2003/01 2004/01 2005/01 2006/01 2007/01 2007/12

7.18 5.21 5.35 6.34 7.60

10.31

11.42 9.01 9.80 12.30 15.52 17.00

Source: Central Bank of Sri Lanka * Weighted Average Deposit Rate ** Weighted Average Prime Lending Rate


48

The following primary and secondary market rates for t-bills show the behaviour of interest rates according to the markets, based on which rates can vary.

Table 4

Primary vs Secondary Market Rates – T-bills 364 days (per cent, p.a.)

Year/Month Primary Market Secondary Market

2003/01 8.85 8.83 2004/01 7.55 7.57 2005/01 7.65 7.62 2006/01 10.37 10.23 2007/01 13.58 13.26 2007/12 19.96 20.11

Source: Central Bank of Sri Lanka

The interest rate will vary depending on whether the market is an international market or a domestic market. In the international market the rates applicable are normally London Interbank Offer Rate (LIBOR) + a margin. Therefore, depending on the liquidity situation in the international market rates charged for international bonds will vary. For instance, because of the financial crisis in USA in 2008, there was a tight liquidity situation in the market and the interest rates for international bonds increased. The LIBOR rates are applicable to international markets whereas in the domestic market rupee rates are applicable. The LIBOR rates are different from rupee market rates.

Depending on the nature of the borrowing in the international market, rates can be changed. For instance, for concessional foreign loans, the applicable interest rate would be 1% or 1.5% whereas for commercial borrowings the rate will be LIBOR + a margin, which may amount to 8 – 10-% in the current context.

4.4 Term Structure

The term structure of the instrument is also a factor in determining the level of interest rates. Longer-term deposits or loans carry higher rates. Loans for a longer term will have higher interest rates compared to those with shorter term. So do the deposits. Longer term deposits will carry a higher interest rate compared to those of shorter term. For instance, financial institutions pay a higher interest rate for 24 months deposits (14.25%) compared to three month savings deposits (5%). Similarly, 364 day Treasury bills usually have higher interest rate than 91 day Treasury bills and long-term Treasury bonds will have higher rates than Treasury bill rates. Table 5 indicates this relationship. Further, if the yield curve is upward sloping, interest rates for longer term maturity of debt will have higher interest rates. At present, this is the situation in Sri Lanka’s debt securities market. (However, with a declining yield curve, the reverse can happen).

Table 5

Term Structure of Instruments and Interest Rates (per cent p.a.)

(December 2006)

Instrument Yield Rate

91 days Treasury bills* 12.76

182 days Treasury Bills* 12.78

364 days Treasury bills* 12.96

02 year Treasury bonds* 12.92

12 month fixed deposit rates of commercial banks

14.00

24 month fixed deposit rates of commercial banks

14.25

*Primary market Source: Central Bank of Sri Lanka


49

4.5 Size of Transaction

The size of the transactions (loans or deposits) is also a factor that determines the interest rates. Normally, a larger loan/deposit will have a higher interest rate compared to a small loan. Similarly if there is a larger deposit from a customer, commercial banks are willing to pay a higher interest rate. Risk is another factor in determining interest rates. If the borrowed amount is larger, the risk to the lender is also higher. The higher the risk for a loan, the higher the interest charged from the borrower. For instance, good collateral and a shorter maturity period will have a lower risk and lenders prefer to lend in those markets than in a market where there is a high risk with less collateral and longer maturity period. It is assumed that Government securities are less risky and therefore, the Government securities carry low interest rates.

4.6 Characteristics of the Instrument

Characteristics of loans and deposits will change the interest rates. Very often, loans have to be secured by collateral. The interest rate charged for loans depends on the type of collateral. Unsecured loans have higher interest rates than loans secured by collateral. For instance, at the end of 2007, unsecured loans of commercial banks carried an average interest rate of 37.43% p.a. while loans secured by stock in trade had an average of 30% p.a. Similarly, the deposits in foreign currency will have a different rate compared to those of local currency. For instance, the US $ Non Resident Foreign Currency (NRFC) fixed deposits of banks carried a rate ranging from 3.5 to 5.75% p.a. whereas the rate for local currency 12 month fixed deposits of commercial banks carried 18% p.a. as at the end of 2007.

4.7 Regulations/Limits

Regulations also matter for interest rate determination. For instance, when there is a higher Statutory Reserve Requirement (SRR), the banks would pay lower interest rates for deposits compared to a regime whether there is a lower SRR. Similarly, taxes also influence interest rates. Higher the taxes on interest, higher the interest rate charged from the borrowers or lower the interest rates paid to the depositors. For instance, when the Government imposed a tax on primary market Treasury bills in May 2002, Treasury Bill rates increased accordingly. CBSL has imposed a ceiling on interest rates paid to depositors in finance companies. For instance, at present, these companies cannot pay more than 4% over the weighted average Treasury bill rate for deposits over 12 months.

4.8 Liquidity in the Market The liquidity situation in the market is a major factor determining the interest rate. When there is high liquidity in the market, the call money rates tend to decline. Similarly, when there is a tight liquidity situation in the market, rates tend to increase. The call money rates increased to a level of 102% p.a. in 1995 due to a shortage of liquidity, among other things. Figure 2 shows the relationship between the call money rates and the liquidity situation in the market.

Figure 2

0

5

10

15

20

25

-10

-5

0

5

10

15

20

2003

/01

2003

/04

2003

/08

2003

/12

2004

/04

2004

/08

2004

/12

2005

/04

2005

/08

2005

/12

2006

/04

2006

/08

2006

/12

2007

/04

2007

/08

2007

/12

% p

.a.

(Rs

Bn

)

Excess Liquidity & Call Money Rates (W.A)

Excess Liquidity (EL ) Call Rates (W.A)


50

The level of Government borrowing is another important factor in determining interest rates. When the Government needs to borrow a large amount from the market, the interest rate of the Government securities market tends to increase. The higher amount of Government borrowing will crowd out the private sector and therefore the funds availability in the market for other activities will decline (tight liquidity), causing interest rates to increase. Tight liquidity situation in the market may happen due to seasonal factors as well. For instance, there are two seasonal incidents taking place in Sri Lanka: one is Sinhala – Hindu New year in April and Christmas period in December. It is clear that during these two seasons, the money demand increases and market liquidity declines. During these periods there is a peak in Treasury bill rates following high money demand.

4.9 Inflationary Expectations

Inflationary expectation is another factor that determines the level of interest rates. If the market expects a higher inflation in the future then the interest rates for some instruments will increase. For instance, Figure 3 indicates the relationship between the inflation and the Treasury bill rates. Towards the end of 2007, inflation has been higher and accordingly the rates of Treasury bills have also been higher compared to 2003.

Figure 3

Inflation affects the real rate structure. The lower inflation leads to a higher real rate and higher inflation leads to a lower real rate. This relationship can be seen from Figure 4.

Figure 4


51

4.10 Type of Market

This is a clear relationship between secondary market rates and primary market rates. When the primary market rates are higher, the secondary market rates also tend to be higher. For instance, Figure 5 indicates the trends in primary market Treasury bill rates vs secondary market Treasury bill rates. Therefore, rates vary depending on the primary or secondary market.

Figure 5

4.11 Competition

Participation of a large number of institutions/instruments in the market would increase competition and supply of funds and thereby interest rates would decline. Higher competition will reduce the cost of lending and increase the rates paid for depositors. Therefore, the nature of competition would decide the rate structure. If there is one lender or very few competitive in the market there is no competition and therefore he may tend to charge high interest rates at his discretion, by maintaining a monopolistic/oligopolistic market conditions. When there are more competing institutions, rates will be a favourable condition for depositors and the borrowers.

4.12 Technology

It is accepted fact that the modern technology produces low cost mechanisms for loan/deposit processing and greater access to finance. This may produce cost efficiency and tend to reduce interest rates compared to a situation where there is a high involvement of labour at higher cost. The relationship between the lender and the borrower, cultural factors, convenience to borrow and computational convenience may also determine the level of operational cost and thereby interest rate paid/charged in the market.

There are interest rate variations among the lenders/depositors in the organized and unorganized/informal markets. Informal market rates are normally higher corresponding to the risk factor. There are instances that money lenders in informal markets charge high interest rates based on daily calculations. Sometimes, if the money lender is a person known to the borrower he may go to that lender irrespective of high interest rates charged with consideration of the personal service. Interest rates may also change due to factors such as natural disasters, changes in Government policies and political atmosphere.

5. Impact of the Determinants

There is clear evidence that the determinants explained above have a greater impact in determining the behavior of interest rate structure in Sri Lanka. During 2004-2008 policy rates of the Central Bank were higher following a tight monetary policy stance. Accordingly, the policy rates (repo rates) were increased from 7% at the end of 2003 to 10.5% in February


52

2007. This resulted in a higher interest rate structure. All deposit rates and lending rates were increased by financial institutions following the higher policy rates. Similarly, when the policy rates were low prior to 2004 and after February, 2009, the market interest rates also declined. Hence, there is an influence from the policy rates in determining the interest rate structure of the country. There is evidence from Sri Lanka to show that cost is also a factor affecting the interest rates. Interest rates paid to depositors is a cost to financial institutions and has to be recovered from the lending rates. When the deposit rates are increased, lending rates are also increased. For instance, when Weighted Average Deposit Rates (AWDR) was 5.21% in January 2004 Weighted Average Lending Rate (AWPLR) was 9.01%. This has increased to 17% when AWDR was 10.31% in December, 2007. The behavior of interest rates, such as LIBOR, Singapore Inter Bank Offer rate (SIBOR) have impacted in deciding the interest rates of debt instruments. When the LIBOR was higher, Sri Lanka’s international bonds had higher rates and vice versa. For instance, LIBOR rates declined due to global financial crisis in 2007-2008 following which the bond rates too declined. The effective rate of Sri Lanka Development Bonds in September 2009 was 4.93% compared to that of 7.04% in June 2006 which was mainly due to higher LIBOR.

3

As discussed earlier, level of liquidity in the market is one of the major determinants of interest rates. As depicted in Figure 2 when there was a tight liquidity in the market during 2006 and 2007 interest rates increased to more than 20% compared to the highly liquid market during 2003-2004 when there was a market rate below 10%. This tight or loose liquidity situation has an impact on the level of interest rates structure. As identified earlier, inflation is one of the other factors that determines the level of interest rates. During 2006 and 2007 inflation was high following which interest rates quoted at primary auctions of Treasury Bills were also higher. One of the major factors that resulted in reducing policy rates in 2

nd quarter 2009 was the low inflation followed by which market rates

declined. Decline in inflation was one of the reasons which resulted in declining T. Bill rates in September, 2009.

The secondary market rates vary based on the primary market rate changes. When there is a lower rate in the primary market, the rates in the secondary market also decline. This is clearly evidenced from the depiction in Figure 5. When Treasury bills rate for 364 day maturity in primary market was 7.24% (average) in 2007, the secondary market rate for the same was 7.19% where as when primary rate for same maturity in 2007 was 19.96% (average) the secondary market rate for the same became 20.11%. Therefore, the primary market rates usually affect the secondary market rates.

6. Summary

This paper examined some of the common factors affecting interest rates and theoretical perspectives of interest rates by examining different interest rate regimes. It highlighted the Classical theory, Keynesian theory along with McKinnon-Shaw’s, neo-structuralists’ and market failure school’s versions as a background to interest rate regimes. Determinants of interest rates were identified later while examining the data from Sri Lanka. Policy rates, cost of funds, type of instruments, term structure, size of transaction, characteristics of the instrument, regulations/limits, liquidity situation in the market, inflation, competition, technology are the factors identified as determinants of interest rates in Sri Lanka. Finally, paper carries a brief technical note to explain how some of these determinants have impacted in establishing interest rates in Sri Lanka.

3.

LIBOR in September 2009 was 0.678% compared to that of 5.64% in June 2006.


53

References

Abdul Gafoor, A.L., (1996), “Internet Free Commercial Banking”, published by A.S. Noordeen Chang, D. and Jung, W.S., (1984), “Unorganized Money Markets in LDCs: The McKinnon –

Shaw Hypothesis verses the Van Wijnbergen Hypothesis”, Nashville: Vanderbilt University, Department of Economics, Working Paper 84-121, June.

Central Bank of Sri Lanka, Annual Reports, Various issues Central Bank of Sri Lanka, (1998), “Economic Progress of Independent Sri Lanka”: Dornbusch, R. and Fischder, S., (1978), “Macro Economics, Mcgraw-Hill Book Company”. Federal Reserve Bank of New York, (2005), “Interest Rates: An Introduction”. Ghatak, S., (1995), “Monetay Economics in Developing Countries, (2

nd Edition)”, St.Martin’s

Press. Grossman, S.J., and Joesph E.Stiglitz, (1980), “On the impossibility of Informationally Efficient

Markets”, American Economic Review, Vol. 70, No 3, pp. 393-408. Hemachandra, W.M., (2005), “Financial Deepening and its Implications for the Sri Lanka

Economy”, Central Bank of Sri Lanka. McKinnon, Ronald, I., (1973), “Money and Capital in Economic Development”, Brooking

Institute. Shea,J., (1992), “The Welfare Effects of Economic Liberalization under Financial Market”,

Academia Economic Papers 20 (2, September), pp 697-716. Shaw, Edward S., (1973), “Financial deepening in Economic Development”, Oxford University

Press. Stiglitz, J.E., (1989), “Financial Markets and Development”, Oxford Review of Economic

Policy, Vol 5, No4, pp. 55-68. Vance Roley, V. and Gorden H. Sellon, Jr. (1995), “Monetary Policy Actions and Long-term

Interest Rates”, Federal Reserve Bank of Kansas City Economic Review (4th Quarter). Wikipedia, the free encyclopedia

54

Multi-Elements of Exchange

Rate Exposure: Evidence from

Japanese Sectoral Returns

Prabhath Jayasinghe1 and Albert K. Tsui

2

Abstract

This paper examines the adequacy of the exposure coefficient/beta in measuring the entire impact of exchange rate changes on firms’ future operating cash flows. To this end, we investigate the presence of four elements of exchange rate exposure: (a) sensitivity of stock returns to exchange rate changes; (b) sensitivity of stock returns to the volatility of exchange rate changes; (c) sensitivity of conditional variance of returns to exchange rate volatility; and (d) dynamic conditional correlation between returns and exchange rate changes. A bivariate GJR-GARCH-M model is employed to investigate all such elements of exchange rate exposure. We uncover significant evidence for the presence of multi-elements of exchange rate exposure, some of which are not captured by the conventional measure of exposure. The paper contributes to the literature by suggesting a theoretical improvement in the area of measuring exchange rate exposure and citing evidence in support of it.

Key Words: exchange rate exposure; volatility spillovers; multivariate GARCH-M

models; time-varying correlation

1 Department of Business Economics, Faculty of Management and Finance University of Colombo, Sri Lanka

2 Department of Economics, National University of Singapore, Singapore

Multi-Elements of Exchange Rate Exposure: Evidence from Japanese Sectoral Returns

55

Returns Changes in

exchange rate

Volatility of

returns Volatility of

exchange rate changes

(a)

(c)

(b)

(d)

Exchange rate Firm value

Mean

Conditional

variance

(e)

Conditional

correlation

Other firms with which it has linkages

1. Introduction Exchange rate exposure is often defined as the change in a firm’s future operating cash-flows in response to changes in exchange rates. Based on the assumption that a firm’s value is a reasonable proxy to the firm’s future operating cash flows, Adler and Dumas (1984) provides a useful empirical definition for exchange rate exposure: the sensitivity of firms’ market value to exchange rate changes. More importantly, they show that exposure can be measured as a linear regression coefficient of firm value on exchange rate. They also argue that such an exposure coefficient is “a single comprehensive measure that summarizes the sensitivity of whole firm … to all the various ways in which exchange rate changes can affect it” (emphasis added). Their argument is based on the questionable assumption of the time-invariant variances of firms’ returns and exchange rate changes. However, if the variances of returns and exchanges rates are allowed to be time-varying, whether that argument holds or not is ambiguous. The verification of such a claim is yet to be done through research. In this paper, we attempt to look into this matter.

Figure 1 shows a number of routes along which the returns on a certain stock can be exposed to exchange rates, if the variances are assumed to be time-varying. First, as represented by (a), the returns of a certain stock may be directly or indirectly exposed to the exchange rate changes. Second, as indicated by (b), returns may also be exposed to the volatility of exchange rate changes (i.e. in addition to its responsiveness to exchange rate changes, profits of a firm may be sensitive to the degree of fluctuations of exchange rates as well). If the degree of fluctuations is time-varying, firms may respond to it by changing markets for both inputs and outputs, locations of production and hedging strategies. Third, the conditional variance of returns can also be exposed to the volatility of exchange rate changes and this is indicated by (c). Even if the relationship (b) is absent, as long as the conditional variances of returns are exposed to volatility of exchange rate changes and the returns are sensitive to its own volatility changes as represented by (e), there may be an indirect impact of exchange rate volatility on returns (i.e. the degree of fluctuations of exchange rates may have an impact on the degree of fluctuations of a firm’s profits forcing the firm to rethink and change its current strategies, which may in turn lead to changes in its profitability). Finally, (d) indicates the time-varying conditional correlation between returns and exchange rate changes, which implies that the intensity of exchange rate exposure of firms is likely to vary over time. In this paper, we use a bivariate GARCH-type model to investigate the aforementioned four elements of exchange rate exposure. We found evidence for the presence of these multi-elements of exchange rate exposure in a sample of Japanese industrial sectors. Interestingly, there are also cases which are not exposed in terms of the conventional measure (exposure coefficient), but significantly exposed to currency risk through the “other” avenues indicated in Figure 1.

Figure 1: Multi-elements of exchange rate exposure


56

The rest of this paper is organized as follows. Section 2 reviews the theoretical and empirical evidence for the multi-elements of exchange rate exposure. Section 3 briefly describes the datasets used in the study. Section 4 highlights the main features of the model used to capture the multi-elements of exchange rate exposure. Section 5 reports that the empirical findings consists of two parts. The first part contains estimation results and the relevant discussion. The second part examines some elements of exposure through simulations. Some concluding remarks are given in Section 6.

2. Multi-Elements of Exchange Rate Exposure: Some Related Work In this section, we look into the literature on international trade and finance in search of theoretical and empirical evidence for the suggested multi-elements of exchange rate exposure. One can find a vast volume of literature on the first moment exchange rate exposure. Though investigating the exchange rate exposure of conditional variance is somewhat common, second moment exchange rate exposure and the dynamic conditional correlation between the stock returns and exchange rate changes are paid far less attention by the researchers.

2.1 First moment exchange rate exposure of returns The first moment exchange rate exposure is thoroughly discussed in the literature during the last two decades. This element of exposure is mainly represented by the direct impact of exchange rate changes on the profits of the firms that are directly engaged in foreign currency denominated transactions. In addition, it also includes the indirect impact of exchange rate changes on firms’ profits that can occur through their linkages with directly exposed firms. Such indirect impacts may come into being when a firm provides inputs to directly exposed firms, acts as an import competitor in the domestic market or uses internationally priced inputs, even if it does not have foreign currency denominated transactions in its accounts (Adler and Dumas, 1984; Bodnar and Gentry, 1993). Among many others, Allayannis (1997), Bartov and Bodnar (1993), Bodnar and Wong (2003), Dominguez and Tesar (2006) and Jorion (1990) also investigate this element of exchange rate exposure at either firm or industry level.

2.2 Second moment exchange rate exposure of returns Koutmos and Martin (2003) was the first (and only) study to examine the second moment exchange rate exposure. Exchange rate volatility can mainly affect the profits of firms through its impact on firms’ international trade activities. However, in the literature on international trade, there is no consensus on the direction of this impact. Clark (1973) and Hooper and Kohlhagen (1978) present theoretical models in which risk averse producers reduce trade in the face of high exchange rate volatility periods. De Grauwe (1989) argues that this adverse effect of exchange risk is a direct result of the assumption of constant absolute risk aversion which usually eliminates the income effect of risk. If a certain producer is slightly risk averse, then one can expect the results suggested by Clarke (1973) and Hooper and Kolhagen (1978). However, if the producer is highly risk averse, then he may worry about the worst possible outcome and the income effect is most likely to dominate the substitution effect. As a result, the producer may export more in order to avoid heavy revenue losses that he expects high exchange rate risk to bring about.

Hysteretic models of trade, in which exporting firms are viewed as holders of options to exit or enter export markets, also have implications towards the relationship between the exports

(a) Sensitivity of stock returns to exchange rate changes (first moment exchange rate exposure of returns)

(b) Sensitivity of stock returns to the volatility of exchange rate changes (second moment exchange rate exposure of returns)

(c) Sensitivity of the conditional variance of returns to the volatility of exchange rate changes (exchange rate exposure of conditional variance)

(d) Dynamic conditional correlation between stock returns and exchange rate changes


57

and exchange rate volatility. According to these models, the decision to enter or exit international markets is based not only on the relevant explicit fixed and variable costs, but also on the cost of exercising the option. The higher the volatility, the higher the possibility that the exchange rates will be favorable and hence the higher the value of keeping the option unexercised. This widens the “range of no change” within which the firm adopts a “wait and see” policy irrespective of the status of their profits. The resulting inertia in entry and exit decisions suggests that profitability of firms is affected by the volatility in foreign exchange markets. Various versions of hysteretic models of trade are developed in Dixit (1989), Krugman (1989), Baldwin and Krugman (1989), Franke (1991) and Sercu and Vanhulle (1992).

In addition to these theoretical attempts

4, there are several studies that analyze the empirical

relationship between exchange rate risk and exports. For instance, Fang and Thompson (2004) report a negative relationship between real exchange rate risk and export revenue in Taiwan. Fang, Lai and Miller (2005) cite evidence for either negative or positive impact of real exchange rate risk on the export performance for a sample of eight East Asian economies. In addition, they find that the impact of exchange rate risk on export performance in all economies is asymmetric between appreciations and depreciations. Meanwhile, for a sample of five developed economies, Kroner and Lastrapes (1993) find that nominal exchange rate risk exert significant impact on both volume and prices of exports.

Moreover, high volatility in foreign exchange markets may motivate the firms to hedge against currency risk, thus affecting the profits through increased hedging costs (Koutmos and Martin, 2003). The impact of currency risk on profits may also be due to the positive relationship between the currency risk and the prices of hedging instruments. For instance, due to the inherent leverage effects, the price of an option increases convexly with the expectation for a currency’s volatility (Entorf and Jamin, 2003).

2.3 Exchange rate exposure of conditional volatility of returns Engle et al. (1990) suggest the existence of “meteor shower” effect of volatility between two geographically separated foreign exchange markets and cite empirical evidence in support of the same against the null hypothesis of “heat wave” effect. Several studies (Apergis and Rezitis (2001) and Yang and Doong (2004), inter alia) show that meteor shower effect can occur between markets for different assets too. The presence of such effects from foreign exchange markets to equity markets may be due to the fact that “exchange rate news may reflect macroeconomic news, such as trade balance news, real interest rate news, and inflation or expected inflation news” (Apergis and Resitis, 2001).

Bodart and Redding (1999) includes some indirect implications of this relationship. They evaluate how the type of exchange rate regime and hence the variability of exchange rate may affect the volatility spillovers and correlations among stock markets. Bartov et al. (1996) examine the relationship between the variability in exchange rate and stock return volatility. Decomposing this relation into systematic and diversifiable risk components, they argue that variability in exchange rates may influence both systematic and diversifiable component of risk. Apergis and Rezitis (2001) investigate the volatility spillovers between London (New York) foreign exchange market and New York (London) equity market. They conclude that conditional variances of returns in the sample are significantly exposed to the volatility in exchange rates, though the reverse is not true. Kanas (2000) examines the volatility spillovers and constant correlations between six equity and foreign exchange markets. No significantly exposed cases are reported. Yang and Doong (2004) examine both mean and volatility spillovers and constant correlations between stock and foreign exchange markets in a sample of seven countries. They also find weak evidence for exposed conditional variances of returns to exchange rate volatility

5.

4 Strictly speaking, some of these theoretical studies also look into the empirical relationship between trade and

exchange rate volatility. 5 Since Apergis and Rezitis (2001), Kanas (2000) and Yang and Doong (2004) are basically interested in estimating

the volatility spillovers between stock and foreign exchange markets, they do not explicitly use the term exchange rate exposure.


58

Furthermore, if a certain stock’s returns are also sensitive to its own volatility as shown by (e) in Figure 1, exposure of the conditional variance of returns to the exchange rate volatility is likely to have an indirect impact on returns. Elyasiani and Mansur (1998) suggest a similar type of indirect relationship between the stock returns and interest rate risk. In their study, interest rate volatility is assumed to affect the volatility of stock returns which, in turn, influences the returns.

2.4 Dynamic conditional correlation between returns and exchange

rate changes Kanas (2000) and Yang and Doong (2004), whose main objective was to analyze the volatility and/or mean spillovers, estimate the constant conditional correlation between equity and foreign exchange markets in a sample of countries. The former study cites somewhat strong evidence for the correlation between the two markets in almost all the countries in the sample. Jayasinghe and Tsui (2008) find evidence for constant conditional correlation between return on seven Japanese sectoral indexes and exchange rate changes. They also report that the conditional correlations between return on some sectors and exchange rate changes are time-varying.

There exist some theoretical work in other areas which has implications towards the dynamic conditional correlation between returns and exchange rate changes. For instance, hysteretic models of trade suggest that firms that hold an “option” to enter or quit the relevant international market come to a status of inertia during high exchange rate volatility periods. This also implies that the correlation between profitability and exchange rate changes may be weak during such periods. On the other hand, the correlation may improve during low volatility periods. The inevitable result is that the intensity of exposure of such firms may vary over time depending on the degree of volatility in foreign exchange markets.

3. Data Our dataset consists of ten industrial sectors of the Japanese economy during the period from 01.06.1992 to 29.12.2000. As such, the sample contains 2240 observations. The choice is based on availability of data

6. In selecting the sectoral returns, we focus on level 04 industrial

classification, which is based on FTSE actuaries system, available in Datastream7. It

comprises 39 sectors. However, we deliberately selected 10 sectors which are more towards the production of manufacturing goods. The selected sectors include automobile and parts (A&P), construction and building materials (C&BM), diversified industries (DI), electrical and electronic equipment (E&EE), engineering and machinery (E&M), information technology and hardware (IT&H), oil and gas (O&G), pharmaceuticals and biotechnology (P&B), software and computer services (S&CS) and steel and other metals (S&OM), respectively. Market portfolio is assumed to be represented by Nikkei 225, the overall stock index in Japan. All sectoral returns and market returns are expressed in local currency.

Though Adler and Dumas (1983) suggest using a set of bilateral rates to measure exchange rate exposure, this act may lead to multicolinearity among the regressors because most currencies are related to one another and move in the same direction (Jorion, 1990). A parsimonious rectification is represented by collapsing a large number of bilateral exchange rates to a single trade-weighted exchange rate. However, a common problem associated with the use of a trade-weighted basket of currencies is that the nature of a sector’s (or a firm’s) exposure may not correspond to the exchange rates and the relevant weights included in the basket (Dominguez and Tesar, 2006). Alternatively, a single bilateral exchange rate with currency that is important in terms of trade and capital flows with respect to the country in question can also be selected. As such, we measure multi-elements of exchange rate exposure with respect to two exchange rates: (a) a trade-weighted yen exchange rate

6 Our initial attempt to extend the study to two more Asia-Pacific countries, namely Australia and Taiwan, was

hindered by the limited availability of sectoral data. 7 Sectoral indices included in both level 02 that comprises only 4 sectors and level 03 that comprises only 9 sectors

were assumed to be too aggregated in nature to test for possible exposure to exchange rate changes. On the other hand, in order to keep the study within a manageable range, we did not use further disaggregated indices in level 05 which comprises more than 100 sectors.


59

compiled by the Bank of England (BOE) and; (b) a bilateral exchange rate between yen and the US dollar compiled by Morgan Stanley Capital International (MSCI). Weights of the BOE exchange rates are designed to represent the relative importance of each of the other countries as a competitor to the relevant country’s manufacturing sector. Data for both rates are extracted from Datastream and are expressed as local currency price of foreign currency: i.e. an increase in the index indicates depreciation of local currency. Following most of the previous studies, changes in nominal exchange rate have been used. Continuously compounded daily returns and exchange rate changes are calculated as follows:

100*ln1,

,,

tu

tutu R

Rr mxiu ,,

where tuR , and

1, tuR are the closing values of stock prices/exchange rates for the trading

days t and t-1 respectively.

4. Measuring Multi-Elements of Exchange Rate Exposure The GARCH model pioneered by Bollerslev (1986) and its subsequent extensions are well-documented in the literature on modeling conditional volatility in empirical economics and finance. The “plain-vanilla” GARCH model has been extended to accommodate several other stylized facts of financial time series. A few such extensions are noteworthy as we are going to capture those stylized facts in this study. First, many variants of GARCH-type models that are capable of capturing volatility asymmetry have been developed. A widely accepted variant of such models is the GJR-GARCH model of Glosten et al. (1993). Second, as Bollerslev (1987) and others emphasize, financial return series tend to be largely leptokurtic. Accordingly, many empirical studies employ GARCH-type models with t-distributed residuals to capture leptokurtosis. Third, Tse (2000) and Tsui and Yu (1999) point out that the constant conditional correlation between volatilities of two financial time-series, suggested by Bollerslev (1992) may be at odds with empirical evidence. As a remedy, Tse and Tsui (2002) and Engle (2002) suggest time-varying conditional correlation GARCH models. Finally, according to Engle et al. (1987), as the degree of risk associated with the return on assets is likely to vary over time, the compensation required by risk averse investors for holding such assets must also be time-varying. They incorporate this feature into asset pricing models by using a GARCH-in-mean term.

In this paper, we adopt a time-varying conditional correlation bivariate GJR-GARCH(1,1)-M model to capture the four elements of exchange rate exposure of sectoral returns. Residuals are assumed to be t-distributed. The mean, variance and correlation structures are specified as follows:

Mean equation for sectoral returns:

ti

q

k

txgktikitxxtmmti harararaar ,

1

21

1,,1,1,0,

; ni ,...2,1 (1)

Mean equation for exchange rate changes:

txltx

s

l

lxtx rbbr ,,

1

0,

(2)

Variance and co-variance equations:

21

ttt Hz (3)

)|(~|)(| 11,,1 ttttxtitt IfII

2

1,1,

2

1,1,

2

1,1,

2

1,, txtxixtxixtiititiitiiiti dhdh (4)

1,

2

1,1,

2

1,, txxtxtxxtxxxtx hdh (5)

21

,,,, txtitixtix hhh (6)


60

Correlation equation:

121,121, 1 ttixixtix (7)

where tir , is return on sectoral index i at time t; tmr , is return on market index at time t; txr , is

the log difference in exchange rate at time t. In addition, t is a 2 x 1 vector of the daily

shocks of ( ti , tx, ) at time t pair-wise with each sector i. And, 1| tt I denotes the 2 x 1

vector of random shocks at time t given all available information at time (t – 1). We assume that it follows a bivariate t-distribution with degrees of freedom, 0 mean and conditional

variance tH , which is a 2 x 2 variance co-variance matrix. For each sector, the main

diagonal elements of tH are the conditional variances of sectoral returns and changes in

exchange rate, represented by tih , and txh , , respectively. The two conditional variances are

assumed to follow GJR-GARCH(1,1) structure given by the equations (4) and (5). 11, tud if

01, tu and zero otherwise for xiu , . Finally, tz denotes the standardized errors which

are assumed to be independently and identically distributed with mean 0 and variance 1. Some discussions on the model setup are in order. As regards the mean equation in (1) for sectoral returns, we follow Bartov and Bodnar (1994) and others to include lagged variables of exchange rate changes to capture the possible impact on stock returns. This is supported by Bartov et al. (1996) when they argue that “characterizing the exposures of firms on a timely basis may be difficult for the investors due to complexities associated with their determination. Without extensive knowledge of international pricing policies, strategic responses to exchange rate changes, foreign currency positions, or firm operations, investors may wait for the firm to release information about its actual performance before they adjust firm value in response to past exchange rate changes, resulting in a delayed rather than a

contemporaneous relation”. As such, the exposure coefficient 1xa measures the sensitivity

of sectoral returns at time t to the exchange rate changes at time (t – 1). This phenomenon is called the first moment exchange rate exposure

8. Given that the exchange rate is expressed

as local currency price of foreign currency, a positive coefficient implies that sectoral returns increase with a depreciation of the yen. This should be the case for those industrial sectors dominated by exporting firms. Following Bodner and Wong (2003) and others, market returns

( tmr , ) are included in the equation to avoid the spurious correlation component between the

sectoral returns and exchange rate changes. In order to capture the exposure of sectoral returns to the volatility changes in foreign exchange market, we include a cross GARCH-M term. In parallel with the use of the lagged exchange rate changes, we include a lagged

volatility term 21

1, txh in the equation. As such, the coefficient ga measures the second

moment exchange rate exposure. A negative sign would imply that trade flows are adversely affected by exchange rate volatility which, in turn, may have an adverse impact on firms’ profits (for instance, exchange rate volatility may reduce the trade volume and profits of an exporter). A negative sign may also imply that increase in exchange rate volatility will reduce firm’s profits through higher hedging costs. However, as mentioned in Section 2, the sign of

ga can be either negative or positive.

Turning to the mean equation in (2) for exchange rate changes, we assume that it follows an autoregressive process of order s. The sectoral returns are not included as explanatory variables in this equation for two reasons. First, each industrial sector is sufficiently small as compared to the whole economy. It is therefore reasonably safe to assume that the

8 Unlike the conventional augmented market model that is widely used to estimate exchange rate exposure,

suggested bivariate GJR-GARCH model does not capture the impact of contemporaneous exchange rate changes in the mean equation. Apparently, there is a trade off here. Though the conventional augmented market model is able to capture the contemporaneous exchange rate changes in the mean equation, it does not capture the other elements of exchange rate exposure of sectoral returns as specified in the conditional variance and correlation equations discussed in this paper. In this sense, this minor weakness of the model is the cost that is paid for it is being able to successfully capture the other elements of exposure. Nevertheless, the use of lagged exchange rate changes may not be unrealistic in the context of daily data.


61

exchange rates are almost entirely dependent on activities in the rest of the economy (see Bodner and Gentry (1993)). Hence, return on a particular sector is assumed to have negligible effect on the exchange rate. Second, we have performed the Granger-causality tests for all sectors with changes in exchange rates and we find that none of the return series Granger-causes exchange rate changes.

The variance equation in (4) for returns of the i

th sector includes the GARCH(1,1) terms

(represented by i and i ) and the GJR term with coefficient i . In order to measure the

exchange rate exposure of the conditional variance of sectoral returns, a cross ARCH term is

also included and its impact on sectoral volatility is estimated by the coefficient ix . A

positive and significant estimate of ix suggests that an increase in volatility in exchange

rates may increase the volatility of sectoral returns. Moreover, a cross GJR term, the

coefficient of which is ix , is added to capture the possibly asymmetric exchange rate

exposure of the second moment. A negative and statistically significant ix implies that the

volatility increase in sectoral returns caused by a depreciation of yen is greater than that caused by an appreciation of the same magnitude. However, in literature, there is no consensus on whether depreciation is good news or bad news in stock markets. Based on their empirical findings, Ma and Kao (1990) argue that stock prices may be adversely affected by news of appreciation in an export-dominant economy whereas the opposite is the case for an import-dominant economy. Nevertheless, there is a view that depreciation may be some bad news for all participants in the stock markets irrespective of whether they invest in the shares of an exporting firm or importing firm. This is because depreciation may bring about an element of uncertainty as well. “However smaller the devaluation may be, it can also signal the imminent arrival of larger and persistent depreciations” (Maghrebi et al., 2004).

Similarly the variance equation in (5) is assumed to follow a GARCH(1,1) process, together

with a GJR term to capture the possibly asymmetric exchange rate volatility by parameter x .

The inclusion of the GJR term can be justified by the fact that exchange rate changes are often negatively skewed. We will find some support for asymmetric volatility associated with

exchange rate changes provided that the estimated values of x are statistically significant.

Finally, the conditional co-variance between sectoral returns and exchange rate changes is represented by equation (6) which is written as the product of conditional correlation coefficient and the square root of the conditional variance of returns and exchange rate changes. Following Tse and Tsui (2002), we assume that the conditional correlation between sectoral returns and exchange rate changes is time-varying. It is assumed to follow the

structure in equation (7). As such, the conditional correlation at time t ( tix, ) is given by the

weighted average of time-invariant component ( ix ), its own lag term in previous period

( 1, tix ) and 1, tix which is assumed to be a function of lagged observations of standardized

error tz . More specifically,

M

h htj

M

h hti

M

h htjhti

tij

zz

zz

1

2

,1

2

,

1 ,,

, . Both 1 and 2 are non-negative

and 1)( 21 .

Assuming that the standardized residuals of the suggested bivariate model are t-distributed,

the conditional log-likelihood of residual vector t at time t can be defined as follows:

2ln2

ln2

2ln

t

2

)(1ln

2

2ln

2

11

ttttt

t

DRDH (8)

where is the vector of parameters to be estimated; (.) is the Gamma function; tD is a 2

x 2 diagonal matrix whose diagonal elements are 21

,tih and 21

,txh ; tR is the 2 x 2 conditional


62

correlation matrix whose diagonal elements consist of ones and off-diagonal elements are

represented by tix, .

The log-likelihood function of the sample is obtained as

T

t tL1 , where T is the

number of observations. The parameter vector of the bivariate GJR-GARCH-M model is

estimated by maximizing L with respect to . All estimates of the parameters in this paper

are obtained by the method of maximum likelihood using the BHHH algorithm and programs coded in GAUSS.

5. Empirical Findings and discussion In this section, we report and discuss estimation results of the bivariate GJR-GARCH-M model. This includes all four elements of exchange rate exposure of sectoral returns and diagnostic checks for adequacy of the proposed model. Then we move on to examine some dynamic properties of exchange rate exposure of sectoral returns and their conditional variances through simulation. The simulated impulse responses of nine sectors will be discussed accordingly. We also demonstrate by simulation that a possible indirect effect of the volatility of exchange rate exposure on sectoral returns could still be possible even if such returns are not directly exposed to changes of the exchange rate in the mean equation.

5.1 Estimation results

Before sectoral returns and changes in exchange rate (both being )0(I processes) are used

in a vector autoregressive form, one has to check whether each sectoral index is cointegrated with the exchange rate. Existence of such a long-run relationship requires the relevant model to be augmented with error correction terms. As the Johansen cointegration test results

9

indicate, no sectoral index is cointegrated with the exchange rate during the sample period considered, suggesting that the error correction terms are not necessary in the mean equations.

Having confirmed the non-necessity of error correction terms, we review the results obtained from the suggested model. First, the exposure associated with the trade-weighted exchange rate is examined in detail. Maximum likelihood estimates of the suggested time-varying conditional correlation GJR-GARCH(1,1)-M model are reported in Table 1. Returns in five out of ten sectors show first moment of exchange rate exposure (see row 1 in Table 1). These sectors include A&P, C&BM, E&EE, IT&H and O&G. The absolute value of estimated

exposure coefficient represented by 1xa of these five cases range from 0.0523 (t-statistic: -

3.18) in C&BM to 0.1522 (t-statistic: -6.19) in O&G. Exposure coefficients of four out of five significant cases are greater than 0.1, suggesting that returns in those sectors are relatively highly sensitive to the changes in exchange rate. As would be expected, returns on the sectors like A&P, E&EE and IT&H are positively related to the exchange rate changes. This implies that returns on these sectors increase with the depreciation of the yen. This is highly agreeable with economic theory as these sectors are mainly engaged in exporting goods or services. The negative relationship between O&G and exchange rate changes can be attributed to Japan’s heavy import reliance in that sector. Although she does not have significant domestic resources of crude oil, natural gas and other energy, Japan is the world’s third largest oil consumer and second largest energy importer (EIA, 2004). These results are also consistent with previous findings of Bodnar and Gentry (1993) and Dominguez (1998) who report that electrical machinery, precision instruments and industrial sectors are positively exposed to exchange rate changes while oil and coal and energy and utilities sectors show the opposite result

10.

9 Results are not shown here to conserve space.

10 However, Bodnar and Gentry (1993), Dominguez (1998) and this study employ three different industry

classification systems.


63

We note that C&BM sector is also negatively exposed. Unlike in the O&G sector, it is difficult to craft a clear-cut explanation for this in terms of imports and exports. The difficulty is partly due to the different classification systems used in sectoral stock indices and import/export data

11. On the other hand, it is not sensible to solely attribute exchange rate exposure of

sectoral returns to the aggregate import/export trade statistics. As many studies argue, first moment exchange rate exposure is determined by a number of other industry characteristics in addition to imports and exports

12 (see Bodnar and Gentry,

1993, for instance).

As the coefficient of the GARCH-M term indicates, second moment exchange rate exposure

is present in three out of ten sectors (see row 2 of Table 1). The coefficient ga is significant

in the three sectors IT&H, S&CS and S&OM. In the third sector, the sign attached to the

estimate of ga is negative. This suggests that exchange rate volatility may increase the

hedging costs and/or adversely affect the exports in these two sectors. The positive sign

11

For instance, the industrial sectors used in this study are due to the FTSE actuaries system whereas the data in Yearbook of International Trade Statistics are based on Standard International Trade Classification (SITC). Although sectors like oil and gas are common to both systems, in many other sectors, such a commonality does not exist. 12

A detailed discussion of the “determinants” of the exposure of sectoral returns in terms of industrial characteristics and firm-specific factors such as hedging activities is beyond the scope of this study.

A&P C&BM DI E&EE E&M IT&H O&G P&B S&CS S&OM§

1.

1xa

0.1405* (6.30)

-0.0523* (-3.18)

0.0161 (0.45)

0.1175* (6.57)

0.0277 (1.84)

0.1500* (5.80)

-0.152* (-6.19)

-0.0239 (-1.26)

-0.0295 (-0.78)

0.0103 (0.39)

2.

ga

-0.0356 (-0.51)

0.0618 (1.26)

0.0957 (0.95)

0.0295 (0.54)

-0.0130 (-0.28)

0.1646* (2.13)

-0.0759 (-1.11)

0.0266 (0.49)

0.2604* (2.35)

-0.163* (-2.10)

3.

i

0.0800* (4.13)

0.0894* (3.91)

0.0449* (3.62)

0.0762* (4.08)

0.0549* (3.25)

0.0827 (3.90)

0.0555* (3.09)

0.0602* (2.90)

0.1668* (6.17)

0.0775* (4.27)

4.

i

0.0568* (2.07)

0.0736* (2.85)

0.0335* (1.99)

0.0196 (0.91)

0.0690* (3.12)

0.0279 (1.31)

0.0458* (2.31)

0.0387* (1.97)

0.0496 (1.40)

0.0555* (2.24)

5.

ix

0.0352 (1.64)

0.0171 (1.78)

0.1597* (3.37)

0.0142 (1.61)

0.0175* (2.30)

0.0197 (0.92)

0.0112 (0.92)

0.0109 (1.12)

0.1112 (1.83)

0.0678* (2.25)

6.

ix

-0.0213 (-0.87)

-0.0189 (-1.67)

-0.1924* (3.51)

-0.0142 (-1.37)

-0.0175* (-2.05)

-0.0124 (-0.46)

0.0013 (0.08)

-0.0068 (-0.54)

-0.0867 (1.17)

-0.0513 (-1.46)

7.

i

0.8808* (36.32)

0.8729* (35.60)

0.9274* (67.64)

0.9019* (42.28)

0.9002* (46.04)

0.8963* (39.89)

0.9244* (60.33)

0.9143* (39.70)

0.8000* (29.08)

0.8705* (42.47)

8.

ix

0.0539 (1.75)

-0.0496 (-1.58)

-0.0050 (-0.16)

0.0848* (3.13)

0.0246 (1.03)

0.0635* (2.44)

-0.058* (-2.09)

-0.0258 (-0.91)

-0.0086 (-0.27)

-0.049* (-2.13)

9.

1

0.9927* (63.76)

0.9784* (35.21)

0.9747* (24.96)

0.9775* (24.44)

0.7156 (1.40)

0.8780* (5.61)

0.9671* (20.35)

0.9770* (47.85)

0.9979* (139.51)

NA

10.

2

0.0012 (0.41)

0.0056 (0.95)

0.0064 (0.88)

0.0026 (0.44)

0.0083 (0.04)

0.0114 (0.80)

0.0049 (0.67)

0.0041 (0.82)

0.0003 (0.10)

NA

11.

4.9612* (12.87)

5.4010* (12.17)

5.2667* (12.27)

5.8365* (11.80)

6.2954* (10.87)

5.4368* (11.83)

5.4290* (12.08)

5.2129* (12.18)

5.8265* (11.33)

5.3226* (12.58)

Table 1: Maximum likelihood estimates for the time-varying conditional correlation GJR GARCH(1,1)–M model (yen trade-weighted exchange rate)

Notes: * indicates at least 5% level of significance; Values mentioned within parentheses are relevant t-statistics; The estimated model consists of the Equations 3b through 9.

§ For S&OM, constant correlation model is used as

the time-varying version of the suggested model did not converge. Except for the time-varying correlation parameters, the results from the two versions of the model are largely similar and available upon request; NA: Not

available.


64

attached to the estimates of ga in the other two sectors may suggest that income effect

dominates and the firms in these two sectors are likely to increase exports in order to avoid possible revenue losses that may stem from volatility of exchange rates. Interestingly, though S&CS and S&OM sectors are exposed to the volatility of exchange rate, they are not exposed to the exchange rate changes.

In addition, as evidenced by the estimated coefficient of own GJR term ( i ), there is support

for asym-metric volatility in returns in seven sectors (see row 4 in Table 1). These sectors include A&P, C&BM, DI, E&M, O&G, P&B and S&OM. All seven estimates are significant at 5% level and bear the expected positive sign, suggesting that the leverage effect is at work when there is a reduction in sectoral returns. The estimated value is at its maximum (0.0736; t-statistic: 2.85) in the C&BM sector, meaning that it is more vulnerable to negative shocks in the stock market. The smallest value among the significant cases can be seen in DI sector (0.0335; t-statistic: 1.99). Volatility associated with exchange rate changes is also found to be asymmetric in five sectors

13.

As suggested by the significance of the coefficient of the cross squared error ( ix ) in

Equation 4, the exchange rate exposure of conditional variance is present in three cases. In other words, “meteoric shower” effect can be observed in the three sectors DI, E&M and S&OM (see row 5 in Table 1). The largest coefficient (0.1597; t-statistic: 3.37) is found in the DI sector, meaning that its volatility is highly sensitive to the volatility changes in foreign exchange market. The smallest value (0.0175; t-statistic: 2.30) is reported in E&M sector. In

addition, the estimates of ix are significant at 10% level in two more sectors (C&BM and

S&CS). The sign of ix is positive in all these five cases, suggesting that an increase in the

volatility in foreign exchange market may spillover as an increase in the volatility of sectoral returns. Furthermore, we find evidence of asymmetric exchange rate exposure of the conditional

variance in several sectors. The coefficient of cross GJR term ( ix ) in Equation 4 is significant

at 5% level in two sectors, namely, DI and E&M (see row 6 in Table 1). Again, in one sector

(C&BM) the estimate of ix is significant at 10% level. The implication is that the returns in

these three sectors are not only highly sensitive to volatility in foreign exchange market but

also vulnerable to the depreciation of yen. The sign attached to the estimates of ix is

negative in all three cases. Given that the exchange rate is expressed as local currency price of foreign currency and an increase in the exchange rate indicates depreciation, this suggests that depreciation of yen will increase the volatility in sectoral returns than an increase in volatility brought about by appreciation of yen of the same magnitude. This is because the depreciation of local currency always contains an element of uncertainty and hence acts as ‘bad’ news, which may result in program trading which has the potential of decreasing of stock prices through increased selling pressures (Maghrebi et al., 2004).

The suggested time-varying conditional correlation model converged only for nine industrial

sectors14

. Out of those nine cases, the time-invariant component of the correlation ( ix ) is

significant only in three cases. Moreover, in eight out of those nine cases 1 is highly

significant while 2 is not significant in any of those cases (see rows 9 and 10 in Table 1).

These results suggest two patterns in the correlation between Japanese industrial sectors and the exchange rate changes. First, although the correlation between the returns and exchange rate changes are time-varying in these sectors, time varying correlation is more likely to be dependent on its own past and less likely to be disturbed by the recent changes reflected in standardized residuals. Second, in many cases, although the time-invariant component is not significant, time-variant component is significant, suggesting that the time-

13

Results are not reported to conserve space. 14

The estimates of the tenth sector, namely S&OM, are based on the constant correlation counterpart of the suggested time-varying correlation GJR-GARCH model.


65

invariant component alone is not a reliable measure of the correlation between the two variables.

In order to evaluate the robustness of the above findings, we also estimate the multi-elements of exchange rate exposure associated with a yen-US dollar bilateral exchange rate

15. With

the chosen bilateral exchange rate, first moment exchange rate is present in six sectors (A&P, C&BM, E&EE, E&M, IT&H and O&G). However, second moment exposure of returns is limited to only two sectors (IT&H and S&CS). The exposure of the conditional variance of returns is also confined to two sectors, namely, DI and S&OM. As for the time-varying

conditional correlation between sectoral returns, ix is significant in three sectors (E&EE,

IT&H and O&G); 1 is significant in all nine sectors

16; and

2 is not significant in any of those

sectors. On average, exposure of sectoral returns to the yen-US dollar bilateral exchange rate is largely similar to their exposure to yen trade-weighted exchange rate. The similarity in terms of the presence of non-conventional elements of exposure with both exchange rates is of particular importance.

Our main findings are summarized in Table 2. There is no sector whose returns and variances are simultaneously exposed through all four routes indicated in Figure 1. The results related to the sectors DI, E&M, P&B and S&CS are all consistent with the main

argument of this paper. The coefficient 1xa in any of these sectors is not significant

17 and,

according to the conventional measure, the four sectors are not exposed to currency risk18

. However, returns and/or conditional variance of some of these sectors are significantly exposed to the exchange rate volatility. In some sectors, correlation between the returns and exchange rate changes are time-varying and statistically significant. These evidences suggest

15

Results are not shown here to conserve space. 16

Suggested time-varying model did not converge for S&CS sector and the estimations are based on the constant correlation counterpart of the suggested model. 17

The coefficient of 1xa is significant in the E&M sector for the yen-US dollar bilateral exchange rate.

18 Here, one may argue that the coefficient of the contemporaneous exchange rate change term might have been

significant (if it was included in regression) though the coefficient of the lag term is not. As the results of univariate regressions reveal, in all these four cases, the coefficients of both contemporaneous and lag exchange rate changes are not significant.

1xa

ix ga ER(tw) ER(bi)

ER(tw) ER(bi) ER(tw) ER(bi) ER(tw) ER(bi) ix 1

2 ix 1

2

A&P *** *** – – – – * *** – * *** –

C&BM *** *** – – * – – *** – * *** –

DI – – – – *** *** – *** – – *** –

E&EE. *** *** – – – – *** *** – *** *** –

E&M * ** – – ** * * – – – *** –

IT&H *** *** ** ** – – ** *** – *** *** –

Oil&G *** *** – – – – ** *** – ** *** –

P&B – – – – – – – *** – – *** –

S&CS – – *** *** * – – *** – – dnc dnc

S&OM – – ** ** ** – ** dnc dnc – *** –

Table 2: Evidence for multi-elements of exchange rate exposure: a summary

Notes: ***, ** and * indicate that the relevant coefficient is significant at least at the 1%, 5% and 10% level

significance, respectively; The relevant coefficients are xa, ga

, ix, ix

, 1 and 2 in Equations 1, 3 and 7 respectively; dnc: time-varying correlation version of the suggested model did not converge; ER(tw): yen trade-weighted exchange rate; ER(bi): yen-US dollar bilateral exchange rate.


66

that the entire currency risk faced by a firm/sector is not fully captured by the conventional “exchange rate exposure coefficient” alone.

Table 3 shows the diagnostics including the summary statistics of the standardized residuals obtained from the regressions based on the trade-weighted exchange rate. Diagnostics for the standardized residuals from the regressions based on the yen-US dollar bilateral exchange rate are almost the same and are not reported here. Although the standardized residual series still show some non-normal features, Kurtosis and Jarque-Bera statistics in all the sectors have remarkably decreased. The non-normality associated with return series is assumed to be taken care of by the estimation of parameters by a t-distribution based model. As row 11 of Table 1 indicates, the degrees of freedom values of t-distribution ( ) for each

sector is very low and highly significant. More specifically, ranges from 4.9612 (A&P: trade

weighted exchange rate) to 6.3251 (E&M: bilateral exchange rate) and the relevant t-statistic is always greater than 10. This justifies the selection of t-distribution as the underlying stochastic structure of the time series of sectoral returns and exchange rate changes. Note

that the Ljung-Box statistics with 20 degrees of freedom for standardized residuals ( )20(iQ )

and for squared standardized residuals ( )20(2

iQ ) are significantly low as compared to those

of the return series. Since we employ a multivariate model, two more tests have been employed to diagnose any remaining nonlinear dependencies in the cross product of the standardized residuals. Box-Pierce type test suggested in Tse and Tsui (1999) is used with

correlation coefficient-adjusted cross product of the standardized residuals ( tixtxti zz ,,, ).

Runs test is applied with the cross product of the standardized residuals ( txti zz ,, ). In terms of

Sector A&P C&BM DI E&EE E&M IT&H O&G* P&B S&CS S&OM

*#

Mean 0.0143 0.0329 0.0113 0.0086 0.0181 0.0147 0.0116 0.0334 0.0230 0.0202

Maximum 5.295 7.1825 6.0302 5.7326 5.511 4.4874 5.7243 6.0521 7.4237 5.5578

Minimum -4.2696

-4.3904

-4.1056

-4.4358

-5.4343

-5.0839

-3.7878

-4.6357

-4.5838

-3.9618

SD 0.9776 0.9813 0.9743 0.9774 0.9711 0.9660 0.9604 0.9737 0.9736 0.9737

Skewness 0.3133 0.6586 0.2003 0.0843 0.1274 0.1151 0.2643 0.3603 0.3264 0.3436

Kurtosis 5.4441 6.3093 5.2082 5.2728 4.4954 4.4226 4.9450 5.3726 5.4504 4.9731

J-B statistic

593.69 1183.00

469.68 484.35 214.51 193.66 378.80 573.32 599.66 406.35

)20(iQ 22.90 34.09 12.92 27.24 28.37 39.80 22.22 21.21 27.97 27.58

)20(xQ 30.08 28.31 28.97 28.46 27.86 28.28 29.11 28.08 27.43 28.91

)20(2

iQ 21.95 45.73 27.92 21.51 21.88 20.53 8.49 14.40 19.11 6.60

)20(2

xQ 18.59 19.06 17.97 18.41 18.32 18.43 18.71 18.60 18.61 18.83

)20(ixTTQ 18.00 29.46 14.57 15.31 10.35 23.69 33.29 22.13 16.16 18.10

ixRuns -1.41 -1.87 0.24 -1.19 -0.34 -0.78 -0.36 1.08 -1.92 0.47

Table 3: Diagnostics: sectoral returns (trade-weighted exchange rate) §

Notes: )20(iQ,

)20(xQ,

)20(2

iQ and

)20(2

xQ are Ljung-Box statistics of residuals and squared residuals from

equations 1 (returns) and 2 (exchange rate changes) for 20 lags. )20(ixTTQ

is the Box-Pierce type test for cross

product of the residuals suggested in Tse and Tsui (1999). The test statistics associated with (.)Q , (.)2Q

and

(.)TTQ tests are assumed to follow a χ

2 distribution and the critical value at the 5% level of significance with 20

degrees of freedom is 31.41. ixRuns is the runs test statistic for cross product of the residuals from equations 1

and 2. For samples greater than 30, it is assumed to follow a standard normal distribution. *A remarkable outlier

which has not been filtered is removed to get these summary statistics in sectors O&G and S&OM; # Residuals

are obtained from the constant correlation version of the suggested model.


67

these two tests, non-linear dependencies in almost all the sectoral indices have been adequately captured by the proposed model. Results of the all tests, together with the summary statistics, imply that the proposed bivariate GJR-GARCH(1,1)-M model is adequate for capturing all four elements of exchange rate exposure of sectoral returns

19.

5.2 Simulation results In order to better understand the dynamic properties of some elements of exchange rate exposure, we simulate impulse responses to a unit shock in exchange rate changes through nine selected sectors using estimates of the bivariate GIR-GARCH-M model. The remaining sector is not picked because neither its return nor conditional variance is exposed to exchange rate changes or volatility of exchange rate changes. The selected sectors can be further divided into four groups. Group A comprises sectors where only returns are exposed to exchange rate changes. They include A&P, C&BM, E&EE and O&G. Group B contains two sectors whose returns are exposed to exchange rate changes as well as the volatility of exchange rate changes. However, their conditional variances are not exposed. The two sectors in question are IT&H and S&CS. In the sectors classified under group C, only the conditional variance of returns is exposed to the volatility of exchange rate changes. The sectors in this group are: D&I and E&M. Group D consists of a single sector whose returns and conditional variance are exposed to the volatility of exchange rate changes, namely, S&OM. But the returns in this sector are not exposed to the exchange rate changes.

Figure 2: Impulse response functions: Group A

(i) Automobile and parts

0

0.05

0.1

0.15

0 5 10 15 20

ri

0

0.01

0.02

0.03

0.04

0.05

0 5 10 15 20

hi hx

(ii) Construction and building material

-0.06

-0.05

-0.04

-0.03

-0.02

-0.01

0

0 5 10 15 20

ri

0

0.01

0.02

0.03

0.04

0.05

0 5 10 15 20

hi hx

(iii) Electrical and electronics equipment

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0 5 10 15 20

ri

0

0.01

0.02

0.03

0.04

0.05

0 5 10 15 20

hi hx

(iv) Oil and gas

-0.2

-0.15

-0.1

-0.05

0

0.05

0 5 10 15 20

ri

0

0.01

0.02

0.03

0.04

0.05

0 5 10 15 20

hi hx

19

Any other asymmetric GARCH specification could be chosen and would have been adequate for this task. But the focus of this paper is not on choosing the most suitable asymmetric GARCH model for estimating exchange rate exposure.


68

Figures 2, 3, 4 and 5 display the corresponding impulse responses of each group of the selected sectors, respectively. As can be observed in Figure 2, sectoral returns are exposed to an exchange rate shock and the impact of such a shock is relatively larger in A&P, E&EE and O&G. Nevertheless, it dies down quickly and then becomes negligible in all four sectors

20. Figure 3 shows the impulse responses for sectors in group B. Unlike in group A,

the impact of the exchange rate shock on the returns is persistent and dies down relatively slowly. The reason is that the returns in those sectors are exposed to the volatility of exchange rate changes. However, the exchange rate shock does not exert any impact on the conditional variance of returns. The pattern of impulse responses of sectors in group C is shown in Figure 4. It can be observed that only the conditional variances are exposed while the returns are not. The impact of an exchange rate shock on the conditional variance is relatively more persistent

21. The effect of the exchange rate shock on the conditional variance

of returns in DI sector is even greater than the impact on its own volatility. Figure 5 indicates the pattern of impulse responses of the single sector in group D. As the returns show second moment exchange rate exposure, the impact of the exchange rate risk on returns is relatively persistent and dies down slowly. Initially, the impact of the shock on the conditional variance is even greater than the impact of it on its own volatility. However, after the 10

th period, the

impact becomes weaker than that on the exchange rate volatility.

Figure 3: Impulse response functions: Group B

(i) Information technology and hardware

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0 5 10 15 20

ri

0

0.01

0.02

0.03

0.04

0.05

0 5 10 15 20

hi hx

(ii) Software and computer services

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0 5 10 15 20

ri

0

0.01

0.02

0.03

0.04

0.05

0 5 10 15 20

hi hx

Overall, we observe that the patterns indicated by simulated impulse responses corroborate the findings reported in Sub-section 5.1. Particularly, sectors in group B, C and D support the main argument of this paper that exposure coefficient alone does not adequately capture the entire currency risk faced by firms. To further elaborate on this issue, we perform a simple simulation experiment using the DI sector from group C to demonstrate a possible indirect effect of the volatility of exchange rate changes on returns (as indicated by the dotted line (e) in Figure 1). We first add a GARCH-in-mean (GARCH-M) term in equation (1) to measure the sensitivity of the returns to its own volatility. The estimate of this new parameter is 0.1163 and is statistically insignificant with a t-statistic value of 1.55. However, we use this parameter for demonstration purposes. We then simulate the returns by initiating a unit shock in exchange rate changes for the DI sector with the added GARCH-M term in the mean equation. Results are shown in Figure 6. As can be observed from the left panel, a somewhat persistent indirect impact could still affect the returns via the GARCH-M term, even if the returns are not directly exposed to the exchange rate changes. More importantly, there would be no such impact if the conditional variance of the returns is not exposed to exchange rate volatility or the returns are not sensitive to its own volatility.

20

Changes in exchange rates are not shown in the graph for the purpose of clarity. 21

For comparison, conditional variance of the exchange rate changes is also plotted in the same graph.


69

Figure 4: Impulse response functions: Group C

(i) Diversified industries

0

0.2

0.4

0.6

0.8

1

0 5 10 15 20

ri

0

0.05

0.1

0.15

0.2

0 5 10 15 20

hi hx

(ii) Engineering and machinery

0

0.2

0.4

0.6

0.8

1

0 5 10 15 20

ri

0

0.01

0.02

0.03

0.04

0.05

0 5 10 15 20

hi hx

Figure 5: Impulse response functions: Group D

Steel and other metals

-0.04

-0.03

-0.02

-0.01

0

0 5 10 15 20

ri

0

0.02

0.04

0.06

0.08

0 5 10 15 20

hi hx

Figure 6: An indirect impact of exchange rate volatility on returns (DI sector)

0

0.01

0.02

0.03

0.04

0.05

0.06

0 5 10 15 20

ri

0

0.05

0.1

0.15

0.2

0 5 10 15 20

hi hx

6. Concluding Remarks A bivariate time-varying conditional correlation GJR-GARCH(1,1)-M model is employed to estimate the multi-elements of exchange rate exposure in sixteen Japanese sectoral indices. More specifically, the paper focuses on four important elements of exchange rate exposure: sensitivity of sectoral returns to exchange rate changes and volatility of exchange rate changes; sensitivity of conditional variances of sectoral returns to volatility changes in the foreign exchange market; and the conditional correlation between sectoral returns and exchange rate changes. In general, we find evidence for the existence of all four elements of exchange rate exposure in Japanese sectoral returns. First, returns in five sectors (A&P, C&BM, E&EE, IT&H and O&G) show first moment exposure to exchange rate changes. Second, returns in three sectors (IT&H, S&CS and S&OM) are significantly exposed to the volatility of exchange rate changes. Third, conditional variances of the returns in three sectors (DI, E&M and S&OM) are exposed to the volatility of exchange rate changes. The positive sign attached to the relevant coefficient suggests that conditional volatility in these sectors increases with the increase in volatility in foreign exchange market. Fourth, the exchange rate exposure of the conditional variances of the returns in two sectors (DI and E&M) are asymmetric in the sense that volatility increase caused by depreciation of yen is greater than that caused by appreciation of yen. Fifth, returns in eight sectors are significantly correlated with the exchange rate changes and the correlation is time-varying. In addition, time-varying correlation between sectoral returns and exchange rate changes is more persistent and less


70

likely to be disturbed by recent changes. The fact that the time-varying correlation parameters are significant even though the time-invariant component is not significant in some sectors implies that the time-invariant component is a misleading measure of the correlation. Finally, and more importantly, returns in four sectors (DI, E&M, P&B and S&CS) are significantly exposed to “other” elements of exchange rate changes though they are not exposed in terms of the conventional measure of exchange rate exposure (namely, exchange rate exposure coefficient/beta).

The simulation exercise reveals some interesting patterns of the dynamics of the multi-elements of exchange rate exposure. First, if the returns are exposed only to the changes in exchange rate, the impact of a shock would die down soon. Second, when the returns are exposed to the volatility of exchange rate changes, a shock in foreign exchange markets may have a somewhat persistent effect on returns than the effect that would have been generated if the returns are exposed to only exchange rate changes. Third, even if the returns are not directly exposed to the changes in exchange rate or volatility of it, as long as the returns are sensitive to its own volatility, there may be a somewhat persistent indirect impact on the returns via the exposure of the conditional variance of the returns to the volatility in foreign exchange markets. Finally, if the conditional variance is significantly exposed to the volatility of exchange rate with sufficiently large parameters, the impact of a shock in foreign exchange markets on the conditional variance of the returns may be even higher than its impact on the conditional variance of exchange rate changes.

The empirical evidence gathered through the estimation of the suggested model for ten Japanese industrial sectors, together with the results of the simulation exercise based on the parameters estimated, strengthens our main argument presented in Figure 1. In the presence of those “other” elements of exchange rate exposure, the entire currency risk actually faced by a firm/sector is not fully captured by the “exchange rate exposure coefficient” alone. As such, taking the conventional exposure coefficient as the sole measure of exchange rate exposure of firms/sectors may provide us with misleading results.


71

References Adler, M. and B. Dumas (1984) “Exposure to Currency Risk: Definition and Measurement”,

Financial Management, 13:. 41-50. Allayannis, G. (1997) “The Time-Variation of the Exchange Rate Exposure: An Industry

Analysis”, Working Paper 97-29, University of Virginia. Apergis, N. and A. Rezitis (2001) “Asymmetric Cross-Market Spillovers: Evidence from Daily

Data on Equity and Foreign Exchange Markets”, The Manchester School Supplement: 81-96.

Baldwin, R. and P. Krugman (1989) “Persistent Trade Effects of Large Exchange Rate Shocks”, Quarterly Journal of Economics, 104(4): 635-654.

Bartov, E., and G. M. Bodnar (1994) “Firm Valuation, Earnings Expectations and the Exchange Rate Exposure Effect”, Journal of Finance, 49(5): 1755-1785.

Bartov, E., G. M. Bodnar and A. Kaul (1996) “Exchange Rate Variability and the Riskiness of US Multinational Firms: Evidence from the Breakdown of the Bretton Woods System”, Journal of Financial Economics, 42: 105-132.

Bodart, V. and P. Reding (1999) “Exchange Rate Regime, Volatility and International Correlations on Bond and Stock Markets”, Journal of International Money and Finance, 18: 133-151.

Bodnar, G. M. and W. M. Gentry (1993) “Exchange Rate Exposure and Industry Characteristics: Evidence from Canada, Japan and the USA”, Journal of International Money and Finance, 12: 29-45.

Bodnar, G. M. and M. F. H. Wong (2003) “Estimating Exchange Rate Exposures: Issues in Model Structure”, Financial Management, Spring: 35-67.

Bollerslev, T. (1986) “A Generalized Conditional Autoregressive Heteroskedasticity”, Journal of Econometrics, 31: 307-327.

Bollerslev (1987) “A Conditional Heteroskedastic Time Series Model for Speculative Prices and Rates of Returns”, Review of Economics and Statistics, 69: 542-547.

Bollerslev, T. (1990) “Modeling the Coherence in Short-Run Nominal Exchange Rates: A Multivariate Generalized ARCH Model”, Review of Economics and Statistics, 72: 498-505.

Clarke, P. B. (1973) “Uncertainty, Exchange Rate and the Level of International Trade”, Western Economic Journal, 11: 302-313.

De Garuwe, P. (1988) “Exchange Rate Variability and the Slowdown in the Growth of International Trade”, IMF Staff Papers, 35: 63-84.

Dixit, A. (1989) “Entry and Exit Decisions Under Uncertainty”, Journal of Political Economy, 97: 620-638.

Dominguez, K. M. E. (1998) “The Dollar Exposure of Japanese Companies”, Journal of the Japanese and International Economies, 12: 388-405.

Dominguez, K. M. E. and L. L. Tesar (2006) “Exchange Rate Exposure”, Journal of International Economics, 68: 188 – 218.

EIA (2004) “Country Analysis Briefs: Japan, Energy Information Administration” http://www.eia.doe.gov/emeu/cabs/japan.html.

Elyasiani, E. and I. Mansur (1998) “Sensitivity of Bank Stock Returns Distributions to Changes in the Level and Volatility of Interest Rate: A GARCH-M Model”, Journal of Banking and Finance, 22: 535-563.

Engle, R. F., T. Ito and W. L. Lin (1990) “Meteor showers or Heat Waves? Heteroskedastic Intra-Daily Volatility in the Foreign Exchange Market”, Econometrica, 58(3): 525-542.

Engle, R. F. (2002) “Dynamic Conditional Correlation – A New Simple Class of Multivariate GARCH Models”, Journal of Business and Economic Statistics, 20: 339-350.

Engle, R. F., D. M. Lilien and R. P. Robins (1987) “Estimating Time-Varying Risk Premia in the Time Structure: The ARCH-M Model”, Econometrica, 55(2): 391-407.

Entorf, H. and G. Jamin (2003) “German Exchange Rate Exposure at DAX and Aggregate Level, International Trade, and the Role of Exchange Rate Adjustment Costs”, Darmstadt Discussion Papers in Economics # 126.

Fang, W. S. and H. Thompson (2004) “Exchange Rate Risk and Export Revenue in Taiwan”, Pacific Economic Review, 9(2): 117-129.

Fang, W., Y. Lai and S. M. Miller (2005) “Does Exchange Rate Affect Exports Asymmetrically? Asian Evidence”, Working Paper 2005-9, University of Connecticut .

Franke, G. (1991) “Exchange Rate Volatility and International Trading Strategy”, Journal of International Money and Finance, 10: 292-307.

http://www.eia.doe.gov/emeu/cabs/japan.html


72

Glosten, L., R. Jegannathan and D. Runkle (1993) “On the Relation between the Expected Value and Volatility of Nominal Excess Return on Stock”, Journal of Finance, 46: 1779-1801.

Hooper, P. and S. W. Kohlhagen (1978) “The Effect of Exchange Rate Uncertainty on the Price and Volume of International Trade”, Journal of International Economics, 8: 483-511.

Jayasinghe, P. and A. Tsui (2008) “Exchange Rate Exposure of Sectoral Returns and Volatilities: Evidence from Japanese Industrial Sectors”, Japan and the World Economy, 20(4): 639-660

Jorion, P. (1990) “The Exchange Rate Exposure of U.S. Multinationals”, Journal of Business, 63: 331-345.

Kanas, A. (2000), “Volatility Spillovers between Stock Returns and Exchange Rate Changes”, Journal of Business and Finance and Accounting, 27 (3), April/May: 447-67.

Koutmos, G. and A. D. Martin (2003) “First and Second Moment Exchange Rate Exposure: Evidence from U.S. Stock Returns”, The Financial Review, 38: 455-71.

Kroner, F. and W. D. Lastrapes (1998) “The Impact of Exchange Rate Volatility on International Trade: Reduced Form Estimates Using the GARCH-in-Mean Model”, Journal of International Money and Finance, 12: 298-317.

Krugman, P. (1989) “Exchange Rate Instability”, London: MIT Press. Ma and Kao (1990) “On Exchange Rates and Stock Price Reactions”, Journal of Business,

Finance and Accounting, 17(3): 441-450. Maghrebi, N. M. J. Holms and E. J. Pentecost (2004) “Are there Asymmetries in the

Relationship between Exchange Rate Fluctuations and Stock Market Volatility in Pacific Basin Countries”, Conference Proceedings of The 12

th Annual Conference on Pacific

Basin Finance, Economics, Accounting and Business, Bangkok. Polasek, W. and L. Ren (2001) “Volatility Analysis During the Asia Crisis: A Multivariate

GARCH-M Model for Stock Returns in the US, Germany and Japan”, Applied Stochastic Models in Business and Industry, 17(1): 93-108.

Sercu, P. and C. Vanhulle (1992) “Exchange Rate Volatility, International Trade and the Value of Exporting Firms”, Journal of Banking and Finance, 16: 155-182.

Tse, Y. K. (2000) “A Test for Constant Correlations in a Multivariate GARCH Model”, Journal of Econometrics, 98: 107-127

Tse, Y. K. and A. K. Tsui (1999) “A Note on Diagnosing Multivariate Conditional Heteroscedasticity Models”, Journal of Time Series Analysis, 20: 679-691.

Tse, Y. K. and A. K. Tsui (2002) “A Multivariate GARCH Model with Time-Varying Correlations”, Journal of Business and Economic Statistics, 20: 351-62.

Tsui, A. K. and Q. Yu (1999) “Constant Conditional Correlation in a Bivariate GARCH Model: Evidence from the Stock Markets of China”, Mathematics and Computers in Simulation, 48: 503-509.

Yang, S. Y. and S. C. Doong (2004) “Price and Volatility Spillovers between Stock Prices and Exchange Rates: Empirical Evidence from the G-7 Countries”, International Journal of Business and Economics, 3(2): 139-153.

73

HR Managers

of South Asian Banks:

A Study of

Strategic Orientation

Ajantha S Dharmasiri1

Abstract This paper investigates the level of strategic orientation of HR Managers of Banks in South Asia. The research problem was formulated as: “With the increased emphasis on human capital as a global trend, what is the extent of strategic orientation of Human Resource Managers in South Asia?” The objectives were to identify the personal factors as well as organizational factors that contribute to the strategic orientation of HR Managers and to identify the interactional effects of personal and organizational factors. The scope of the research was limited to selected high-performing Commercial Banks from Bangladesh, India, Pakistan and Sri Lanka. A mixed methodology involving a predominantly qualitative approach supported by a quantitative component was adapted. In-depth interviews were conducted with the Chief Executive Officer, Chief Finance Officer, and the Head of HR of each company. Quantitative questionnaires were also administered to capture additional data. The key personal factors emerged were tacit knowledge, conceptual skills, and business awareness. Leader’s recognition of HR, conducive organizational environment and performance orientation emerged as key organizational factors. Interactional effects of the above factors were also discussed highlighting Competency Deficit and Opportunity Deficit.

Keywords: HR Managers, Strategic Orientation, Competency Deficit, OpportunityDeficit, Commercial Banks

1 Dr. Ajantha Dharmasiri, Postgraduate Institute of Management, University of Sri Jayewardenepura, Sri Lanka;

Email: [email protected], Web: www.pim.lk


http://www.pim.lk/


74

Introduction

Human Resource Management (HRM) is increasingly gaining attention in business circles across the globe with the recognition of people factor in competitive advantage maintaining. Human Resource (HR) Managers have a major role to play as members of senior management teams striving towards profitable and sustainable growth. A recent survey on emerging trends in HR (Ruona et al., 2003) revealed that there is much concern about HR’s absence in the “boardroom” and in organizational leadership positions. Also, the question of value addition and particularly “impact on the bottom line” has been one of the core areas under discussion (Ulrich, 1998). In this paper, HR Manager is meant to be a senior person engaged in Human Resources activities as an occupation. In other words, a manager with experience and qualifications, handling the responsibilities associated with the Human Resources function of an organization (Dessler, 2003). Wheelen and Hunger (1988) define strategic orientation as thinking and action of a person, reflecting the long term plans of the organization he/she works for. It includes the involvement in the strategy formulation, implementation and evaluation stages of the strategic decision making process. It can also be stated as a structured process of thinking to “see the unseen”, leading to action. Bing et al. (2003) in describing the challenges to the field of HR state that “the challenge to practitioners is to move beyond a silo mentality in which solutions can only be found within HRD and to embrace a perspective that organizational problems are systemic and require systemic solutions”. Considering the region of South Asia, a scenario can be seen where paradoxically, the best brains and worst poverty are both present (Bhushan, 2000). He further states that “… the role of managerial development, as a harbinger of social and economic change can play in getting the South Asian region its rightful place in the committee of nations”. Strategic orientation of HR Managers would be of prime importance in such an endeavour, in aligning HR strategies with organizational direction in order to reap the true benefits of regional synergies. Regarding the state of HRM in India, Budhwar (2001) comments that the pressures imposed by overseas operators on local operators have created a strong need for Indian HR Managers to seriously emphasize issues related to performance. Singh (2003) studied the strategic HR orientation and firm performance in India, and highlighted a significant relationship between strategic orientation and firm performance. In the case of Pakistan, a significant emphasis on the need of HR is evident. According to Khilji (2001), the function of HR holds a promise for employees in Pakistan and it is the way forward. Considering the state of HRM in Bangladesh, Siddiquee (2003) endorses a similar view point. According to him, despite the prevalence of a broad consensus on the strategic importance of human resource factor, the paradigm shift is yet to take place in public bureaucracies of developing countries including Bangladesh. Nanayakkara (2003) points out that “weak work ethic, individualism, feminity and weak uncertainty avoidance, which is not a productive combination of variables for a strong organizational culture could explain some of the underdeveloped characteristics of human resource management in Sri Lanka.” According to him, the overall HRM performance was below the expectation levels and the HR strategy - business strategy link was found to be the weakest link. The above country-specific details of the level of HR highlights the need for an in-depth study incorporating the entire region. This paper presents a part of such an endeavour, in fulfilling a research gap.

HR Managers of South Asian Banks: A Study of Strategic Orientation

75

2. Research Aims Against the background of the above discussion, the research problem was stated as follows:

“With the increased emphasis on human capital as a global trend, what is the extent of

strategic orientation of Human Resource managers in South Asia?”

Research questions raised were converted into objectives of the study. They were:

i. To identify the personal factors that contribute to the strategic orientation of HR Managers in South Asia

ii. To identify the organizational factors that contribute to the strategic orientation of HR Managers in South Asia

iii. To identify the interactional effects of personal and organizational factors by way of exploring the trends emerging in relation to strategic orientation of HR Managers in South Asia

iv. To recommend key strategies to organizations as well as for the HR Managers, to improve the strategic orientation of HR Managers in South Asia

3. Glimpse of the Literature

Attempt was made to present briefly the research findings of the key concepts associated. Strategic Human Resource Management (SHRM), Strategic HR Responses, and Regional Realities of South Asia were the three key components in focus.

3.1 Strategic Human Resource Management

According to Schuler & Jackson (1987) Strategic Human Resource Management (SHRM) is, “Explicitly linking HRM with strategic management processes of the organization and to emphasize coordination and congruence among the various human resource management practices”. Wright et al. (1994) defined SHRM as, “The pattern of planned human resource deployments and activities intended to enable a firm to achieve its goals.” The definition of SHRM linking with business strategy has been given by Truss and Gratton (1994) as, “The linking of HRM with strategic goals and objectives in order to improve business performance and develop organization culture in order to foster innovations and flexibility.”

3.2 Strategic HR Responses

Gratton and Truss (1994) stated that HR’s strategic role is to adapt individual HR practices (recruiting, rewarding etc.) to fit specific corporate and competitive strategies. As they further elaborate, “For any particular organizational strategy, there is purportedly a matching human resources strategy.” In response to the changes at strategic levels, the role of the HR function is also changing dramatically (Schuler and Jackson, 1987; Ulrich, 1998; Armstrong, 2006). As a strategic partner, the HR Manager has to acquire the business mastery, a path that would lead him / her to greater interactions with the top team (Ulrich and Broackbank, 2005). Ulrich (1998) reported that the highest variability in the data in relation to the strategic partner suggesting the diverse nature in strategic partnership among HR Managers. Moving beyond isolation as just HR experts, true leaders combine their expertise of HR policies and practices with knowledge of the business to lead change effectively (Rothwell, 1996). Comparatively, fewer studies on strategic aspects of HRM have been done in South Asia, mainly owing to the “infancy stage of HRM” (Budhwar, 2004). As Chanda and Shen (2007) observe, the literature available on HRM practices is largely conceptual and concludes that HRM practices can help to create sustained competitive advantage, especially when they are


76

aligned with a firm’s competitive strategy. It points out to a gap where more empirical studies are needed, particularly in the Asia-pacific context. This research fulfils such an empirical gap.

3.3 Realities of South Asia

South Asia has a rich heritage of culture and tradition in addition to its vast population. It is on the threshold of economic expansion, amidst variety of challenges. South Asia is a “humanly rich” region, which ranks among the world's most densely-populated areas. Around 1.6 billion people live there – about one-fourth of all the people in the world. The region's population density of 305 persons per square kilometre is more than seven times the world average. South Asia is a region with diversity in abundance. In exploring the human richness of the region, several Human Resource related implications can be seen. Encouraging organizations to contribute more for economic growth demands attracting, nurturing and retaining key talent. Ensuring the poor are not marginalized includes equal opportunity without discrimination. Contributing towards the society in the form of corporate social responsibility may assist the deserving poor. HR has to act in all these fronts in delivering results. Hence, the broad need is a holistic approach taking into consideration multi-faceted realities of the South Asian region. The resulting outcome will be a well-balanced economic growth with sufficient emphasis on poverty alleviation. As Sinha (2002) opined, with human resource at the core of the new economy of South Asia, the gap between what can be imagined and what can be achieved will always get narrowed.” In a similar line, Sheth (1990) observed that South Asian countries need dedicated and competent professional managers in government, agriculture, industry, trade and commerce, science and technology, co-operatives, public services, administration and all other forms of socially productive enterprise, where managerial expertise is also crucial input in formulation and implementation of development plans at all levels, including relevant concepts, techniques, and tools in the various areas of management. The required response from strategically orientated HR Managers in South Asia is aptly highlighted here.

4. Beginning With Basics

In exploring the strategic orientation, a basic model which is technically called the concept – indicator model was developed, based on the literature review. Figure 1 contains the details:

Figure 1: Concept – Indicator Model

Strategic Orientation of HR Manager

Organizational Factors

Personal Factors


77

5. Moving Ahead This research adapted a mix methodology which was predominantly qualitative. Grounded theory approach as proposed by Glaser and Strauss (1967), and subsequently modified by Strauss and Corbin (1990) was selected as the appropriate way to understand the patterns emerging from a relatively smaller sample such as the one selected. Additionally, Strategic Orientation Questionnaire (SOQ) adapted from Ulrich and Brockbank (2005) and Strategic Integration Questionnaire (SIQ) adapted from Chanda and Shen (2007) were used as quantitative data capturing instruments. Purposive sampling was used with emphasis on a sector having high strategic significance in South Asia. Banks are the pillars of modern economic development and they act as a catalyst for economic growth and development of the country (Kamal, 2007). Accordingly, selection of four leading banks, based on the country level business information such as the “top 10 business entity lists” was done. The list of specific banks chosen was further endorsed by the academics of the respective countries whom the researcher met as part of his regional study tour.

5.1 Data Collection The companies involved with the study were code-named on request in order to maintain anonymity. The details are as given in table 1.

Table 1: Sample Selected

Country Commercial Bank

Bangladesh BdBank

India InBank

Pakistan PkBank

Sri Lanka LkBank

There were three respondents from each company, viz. the CEO, head of finance (CFO) and the head of HR (HRM), resulting in 12 in depth interviews, with the personal involvement of the researcher. Quantitative instruments (SOQ and SIQ) were also administered to this sample.

As the sample was relatively small, no attempt was made to use detailed inferential statistics, in association with the quantitative instruments.

5.2 Sample Profile The respondents included 4 HR Managers, 4 Chief Executive Officers and 4 Chief Financial Officers. It was a 100% male sample, highlighting the relatively less representation of women in senior managerial positions in South Asia. Considering the education level of the respondents, the details shown in figure 2 can be seen.


78

Figure 2: Education Level of Respondents

Source: Survey Data In relation to the education level of the respondents, it is evident that all but one of them were graduates with majority possessing a postgraduate qualification, inclusive of one doctorate. This indicates the level of emphasis given by the commercial banks to education as a vital part of professionalism. With regard to the education, further details can be seen, as depicted in figure 3.

Figure 3: Local vs. Foreign Qualifications of Respondents

Local, 9, 76%

South Asian, 1,

8%

American, 1,

8%

European, 1,

8%

Source: Survey Data The proportion of respondents having an overseas qualification is significantly higher compared to other sectors. This indicates the need to equip oneself with rapidly developing technology and advanced management practices (Armstrong, 2006). In relation to the work experience of respondents, the details are as shown in figure 4.

First Degree,

2, 17%

Masters

Degree, 9,

75%

Doctorate, 1,

8%


79

Figure 4: Work Experience of Respondents

Between 20 -

25 yrs, 7, 58%Between 26 -30

yrs, 2, 17%

Above 30 yrs,

3, 25%

Source: Survey Data According to figure 4, a large majority of the respondents are having a work experience between 20-25 years. This highlights the “relatively younger” top management representing higher dynamism in line with industry needs.

6. Analysis of Data Both qualitative as well as quantitative data were involved in the research. Qualitative findings and quantitative findings are discussed separately with subsequent assessment of their convergence.

6.1 Qualitative Findings Based on the qualitative responses collected as a part of the in-depth interview, several aspects of the HRM in focus were captured, as depicted in table 2.

Table 2: Key Strengths, Issues and Achievements of the HRM as Identified by the CEO and the

CFO

Key Strengths Key Issues Key Achievements

Experience in Bangladesh as well as Asia Pacific

Adoptability of a multinational mindset to a local bank

Revamping the performance management system

Systematic approach with specialized knowledge in HR

Less emphasis on educating functional managers on HR

Changing the performance management system incorporating individual and team targets

Well balanced, having the right mix of strategic thinking and dedicated execution

Need to develop the entire HR team

Development of robust HR systems

Has the ability to drive multiple initiatives

Low level of inter-departmental relationships of the HR team

HR Systems development

Exceptional dedication and commitment

Need to significantly improve the HR team

Introducing a performance management system

Highly qualified in the field of HR

Less business awareness in the HR team

Establishing a performance management system

Possession of required competencies

Need to understand the complexities of the industry

Revamping the performance management system

Very competent in handling HR

Need to tackle complex issues promptly and professionally

Improving the performance management system

Source: Survey Data


80

According the above table, all three columns highlight the need of personal factors and organizational factors contributing to strategic orientation of HR Managers. In moving further, coding of the transcriptions was done resulting in more than 80 codes being generated. They were clustered and such clusters were labeled in line with the concept-indicator model. Details of such personal factors and organizational factors emerged are as shown in table 3.

Table 3: Emergence of Personal Factors

Source: Survey Data The above table depicts the emergence of three distinct personal factors through qualitative data.

As the HRM of BdBank stated : “I am satisfied with my involvement in business decisions, but I need more understanding on the finer points of the banking industry. As most of the HR people in local banks perform a transactional type of a job, it is very difficult to transform the mindsets. For example, talent management is absolutely zero. I of course can introduce attracting, nurturing and sustaining key talents, because of my MNC exposure. Whether the industry is ready for such new approaches is a question. But I am very optimistic on one key aspect. When more and more foreign banks enter Bangladesh and offer innovative solutions, we can’t just wait.” (Extracted from transcription related to HRM, BdBank, 03/03/07)

The above statement is a clear indication of the HRM knowing the ground situation and the challenges that lie ahead. Such an appreciation of reality is of vital importance for the strategic orientation of HRMs.

In a similar manner, organizational factors were also captured. Table 4 contains the details.

Cluster Contents (Salient Items) Cluster Label Definition from Literature

- Involvement in regional projects - Influence of past incidents - Demonstration of tacit knowledge at

work - Social networking by HRM - Experience and exposure to diverse

settings

Tacit knowledge (PF1)

Knowledge gathered through experience in, and exposure to, the strategic decision making process. (Adapted from Nonaka and Takeuchi, 1995)

- Assessment of status of HR - Assessment of strategic challenges - Assessment of strategic role of HRM - Assessment of the HR team - Envisioning for future - Innovative solutions for strategic HR

issues

Conceptual skills (PF2)

Envisioning and analytical skills needed for strategic decision making. (Adapted from Katz, 1974)

- Awareness of business needs - Awareness of company practices - Critical assessment of mgt. structure - Awareness of key org. challenges - Awareness of business achievements - Critical assessment of industry

Business awareness (PF3)

The understanding of the nature of business one’s organization is engaged in. (Adapted from Ulrich and Brockbank, 2005)


81

Table 4: Emergence of Organizational Factors

Cluster Contents (Salient items) Cluster Label Definition from Literature

Appreciation of achievements of HRM Appreciation of strategic role of HRM Association of Leader and HRM Leader’s awareness of HRD Leader’s contribution in HR decisions Recognition of HR capability

Leader’s recognition of HR (OF1)

The understanding of the CEO on the importance of HR. (Adapted from Avolio etal., 1991 and Pareek, 1997).

Assessment of HR implementation Assessment of org. performance Collective approach to performance Setting of targets as per the Balanced Scorecard Recognition of strategic HR achievements

Performance orientation (OF2)

The amount of trust and emphasis from the organization toward the achievement of set of objectives. (Adapted from Kaplan and Norton, 2001)

Assessment of friendly org culture Autonomy enjoyed by the HRM Recognition of HRM by the organization Supporting environment for innovation Trust in HRM by the organization

Conducive organizational environment (OF3)

The organizational climate which is influential for HR Manager to be strategically oriented, with openness in sharing information and encouragement for innovation. (Adapted from Rao, 1990 and Pareek, 1997)

Source: Survey Data According the table 4, three distinct organizational factors emerge through the qualitative data captured.

6.2 Quantitative Findings Based on the data collected using the Strategic Integration Questionnaire (SIQ) adapted from Chanda and Shen (2007), the following picture can be seen.

Figure 5: Strategic Integration of HR in Four Commercial Banks

0.0

1.0

2.0

3.0

4.0

5.0HR Strategy Formulation

HR Involvement in Business

Line Manager Involvement in HR

Facilitating Performance Management Fulf illing T & D Needs

Empow ering Line on Remuneration

HR for Employee Relations

BdBank InBank PkBank LkBank

Source: Survey Data According to figure 5, it appears that HR strategy formulation is one dimension where there is a significant variation among the four banks. Depending on the time of commencement of


82

strategic HR initiatives, the current state of progress may occupy varying levels. In other words, it could be a case of early starters vs. late starters. Considering the data captured through the Strategic Orientation Questionnaire (SOQ) adapted from Ulrich and Brockbank (2005), the scenario as depicted in table 5 can be seen.

Table 5: Responses to the SOQ

Statement Mean Std. Dev.

1 He/She is actively involved in developing long-term goals of the organization. 4.25 0.62

2 He/She is consulted by the MD/CEO in making key decisions on people. 4.67 0.49

3 He/She designs HR strategies in line with broad organizational strategies. 4.58 0.51

4 He/She serves in cross-functional teams tackling strategic issues. 4.17 0.39

5 He/She introduces key measurements to assess HR effectiveness. 4.25 0.62

6 He/She is conversant with the use of HR related IT systems. 4.67 0.49

7 He/She is aware of the key competitive challenges. faced by his/her organization. 4.42 0.67

8 He/She strives to optimize the people contribution in overcoming strategic challenges. 4.50 0.52

9 He/She acts as a coach in ensuring the other key senior managers play their people management role.

4.67 0.49

10 He/She strives to improve the internal communication so that the key initiatives and results are cascaded down in an effective manner. 4.58 0.51

Source: Survey Data

The mean scores for the 12 responses indicate a trend towards a higher mean. However, as the standard deviation figures indicate variations across the four banks, the awareness of key competitive challenges faced by the organization (Item 7) has obtained the highest diverse responses. Figure 6 shows the diversity of such responses.

Figure 6. Responses from Banks on HRMs Awareness of Key Competitive Challenges

0

1

2

3

4

5

BdBank - HRM

BdBank - CEO

BdBank - CFO

InBank -

HRM

InBank -

HRM

InBank -

HRM

PkBank -

HRM

PkBank -

CEO

PkBank -

CFO

LkBank - H

RM

LkBank - H

RM

LkBank - C

FO

Source: Survey Data

The above Figure indicates consistency among many organizations with the exception of the relatively lower rated BdBank. Even though, the HRM in focus was a veteran in the field of


83

HR, he was relatively new to the industry. The scenario is expected to change over a period of time. The above discussion indicates a convergence between the qualitative findings and the quantitative findings. Such a scenario is in line with inter-method triangulation advocated by Denzin (1978).

7. The Emergent Model The centerpiece of this research is the emergent model, which depicts the key patterns emerged from both qualitative and quantitative approaches. Expanding on the concept-indicator model, which was the starting point of the exploration, more details could be incorporated. Accordingly, based on the personal and organizational factors emerged through data in relation to the strategic orientation of HR Managers, the emergent model as depicted in figure 7 can be seen. The emergent model depicts the interactional effects of personal factors as well as organizational factors in relation to strategic orientation of HR Managers. The researcher intends to introduce two key concepts, in relation to varying levels of personal and organizational factors, namely, Competency Deficit and Opportunity Deficit. It is worthwhile investigating the characteristics of two such scenarios.

Figure 7: The Emergent Model \\ Source: Survey Data

High

Level of

Pre

sence o

f

Pers

ona

l F

acto

rs

Low High

Level of Presence of Organizational Factors

Strategic

Involvement

Operational

Involvement

Opportunity

Deficit

Competency

Deficit

Strategic Orientation

of HR Manager

Tacit Knowledge

Business Awareness

Conceptual Skills

Lea

der’s

recognitio

n o

f H

R

Perf

orm

ance

Orienta

tio

n

Conduciv

e

Org

aniz

atio

nal

Environm

ent


84

Competency Deficit refers to a scenario where the organization provides an opportunity for the HR Manager to get involved in the strategic decision making process, he /she does not capitalize by moving ahead appropriately, as there are serious competency gaps in him/her. Attention has to be paid to the context, where there is a significant recognition of people as a key asset in the organization, and there is sufficient emphasis on strategic planning process. In such a case, why HR Manager does not play the role of a strategic partner, in one way, could be attributed to issues with his/her competencies, that need to be systematically explored. Opportunity Deficit refers to a scenario where the HR Manager is willing and able to become a strategic partner, the organization does not provide the needed opportunity. It can be caused by not having an enabling leadership, non availability of a supportive climate, and non existence of a performance orientation in the organization. Relevant contextual aspects need to be considered in this case as well. Despite a significant recognition of people and sufficient emphasis on the strategic planning process, why there is a lack of opportunity for the HR Manager to play the role of a strategic partner, demands an in-depth investigation.

Based on the emergent model, the researcher intends to propose several propositions that need further investigation through detailed studies. P1: Greater the presence of tacit knowledge in the HR Manager, lower will be the

competency deficit, resulting in a higher level of strategic orientation of the HR manager.

P2: Greater the presence of conceptual skills in the HR Manager, lower will be the competency deficit, resulting in a higher level of strategic orientation of the HR manager.

P3: Greater the level of business awareness of the HR Manager, lower will be the competency deficit, resulting in a higher level of strategic orientation of the HR manager.

P4: Greater the leader’s recognition of HR, lower will be the opportunity deficit, resulting in a higher level of strategic orientation of the HR Manager.

P5: Greater the presence of a conducive organizational environment, lower will be the opportunity deficit, resulting in a higher level of strategic orientation of the HR Manager.

P6: Greater the emphasis on performance orientation, lower will be the opportunity deficit, resulting in a higher level of strategic orientation of the HR Manager.

The findings discussed so far are in line with research conducted by Hoque and Moon (2001) who established that: “The growing number of specialists using the HR title who are well qualified, are more likely to be involved in strategic decision making processes and are most likely to be found in workplaces within which sophisticated methods and techniques have been adopted”. Further, the findings arrived at a South Asian setting is of significance due to variety of reasons. As Khatri (1999) observes, conducting strategic HRM studies in other parts of the world, for example Asia, would help to meet the shortage of empirical work in the field in those parts of the world and also serve as a vehicle for comparative studies.

8. Limitations of the Study As the research was based on a predominantly qualitative approach with exploratory nature, inability to predict the strategic orientation of HR Managers of other organizations in South Asia, based on the findings of this research, could be stated as a limitation. Accordingly, selection of a larger sample with organizations representing all the major industries will be useful for a better understanding of the strategic orientation of HR Managers in South Asia. Impact of national culture on the firm performance was not considered in this study. Differences emerged from the data in relation to four Commercial Banks could have been


85

analyzed with inferences to cultural dimensions such as those recommended by Hofstede (1991).

9. Policy Implications and Recommendations Based on the findings of the research, following recommendations can be made, for the organizations.

9.1 Strategic Involvement of HR It needs to be ensured that the HR Manager is involved in the strategic decision making process of the organization. This may be translated into several key actions such as: i. Encouraging the HR Managers involvement in cross functional teams tackling strategic

issues ii. Presence in top team in providing people related information on issues and implications

of decisions. iii. Coaching element of HR Manager where he/she acts as the internal consultant in

advising the line managers on HR matters, with the premise that “all managers should be HR Managers”.

iv. Regular meetings between the CEO and the HR Manager, in discussing key people matters affecting the business.

9.2 Strategic Alignment of HR It also needs to be ensured that that HR policies and practices are aligned to the organizational strategic direction, and are reflective of organization’s long term goals. This should be practically approached as: i. HR Manager is well updated on the strategic direction of the organization, with regular

reference to organizational documents, meeting minutes and discussions. ii. Devise a checklist for HR initiatives to ascertain the strategic alignment, and apply it in the

case of every new HR initiative. iii. Cultivate the practice of connecting the HR actions with the broad organizational

objectives, especially with the HR Manager leading by example in grooming the HR team.

9.3 Strategic Contribution of HR It needs to be ensured that HR function, and particularly the Head of HR is accountable for people- related strategic matters and thus should achieve agreed targets. Quantification of targets with appropriate matrix will be a perquisite for this. In executing the key tasks, the following practical actions need attention: i. Appropriate usage of strategic results tracking tool such as Balanced Score Card, with

emphasis on people related parameters. ii. More focused HR scorecard, representing HR strategies, objectives, initiatives and action

plans should be in place. iii. Regular review of strategic contribution of HR team and communicating the results to all

concerned should be given priority. Such reviews may also act as “course correction” opportunities.

Considering the HR Manager as an individual, the following recommendations can be made based on the findings of the research.


86

9.4 Upgrading Competencies of HR Managers HR Manager cannot survive in an increasingly changing business environment without regularly enhancing his/her knowledge and skills with desired attitude. Some possible actions are: i. Be in touch with current research done on HR, with understanding and applications where

relevant and possible. ii. Embark on development programs such as short courses on diverse aspects, not only in

HR but also in relation to business as a whole. iii. Think and act with more business sense, with greater understanding on business, by way

of being in touch with the rest of the organization as well as the outside world.

9.5 Exploring Opportunities for HR Managers Professional networking by means of active membership in organizations related to HR is becoming increasingly important. Some possible practical actions in this direction are: i. Becoming a member of already existing HR body, such as Institute of Human Resource

Management ii. Form new bodies to establish greater cooperation related to common issues, despite

organizations competing with each other. All HR Managers in banks getting together in one forum to discuss common people issues is one such example.

iii. Facilitating activities involving more than one organization in line with CSR initiatives, with a clear understanding of the community needs.

10. Conclusion As an overall conclusion of this study, it can be stated that the strategic orientation of the HR Managers selected from a purposive sample of four high performing Commercial Banks in Bangladesh, India, Pakistan and Sri Lanka showed a satisfactory level of strategic orientation, confirming the findings of related studies done earlier. It is expected that the findings will contribute to the broadening of the existing knowledge on the HR Managers in South Asia, with academic, managerial, and societal implications.

11. Acknowledgements The author wishes to acknowledge the support rendered by the Association of Management Development Institutions of South Asia (AMDISA) in providing the financial assistance for a South Asian management study, through the Commonwealth AMDISA grant awarded for eligible doctoral candidates.


87

References Armstrong, M. 2006, “A Handbook of human Resource Management Practice”, 10th edition,

Kogan Page, London. Avolio, B. J., Waldman, D. A. and Yammarino F. J. 1991, “Leading in the 1990s: The Four “I”s

of Transformational Leadership”, Journal of European Industrial Training, 15 (4): 9-16. Bing, A., Khan, M. and Short, D. C. 2003, “Challenges to the Field of HRD”, Advances in

Human Resources, 5 (3): 342 – 351. Bushan, R. 2005, “An Introduction to Advantage – South Asia”, Editorial of the AMHIDSA

Publication, Hyderabad: 5. Budhwar, P. S. 2001, “Human Resource Management in Developing Countries”, Rutledge,

London: 75-90. Chanda, A. and Shen, J. 2007, “Measuring Strategic Integration of HRM by Developing the

Instrument HRMSI”, Proceeding of the 9th South Asian Management Forum: 505-514. Denzin, N. K. 1978, “The Research Act: A Theoretical Introduction to Sociological Methods”,

2nd Edition, Prentice Hall, Englewood Cliffs, 1978. Dessler, G. 2003, “Human Resource Management”, 9th Edition, Prentice-Hall of India, New

Delhi: 3-25. Glaser, B. G. and Strauss, A. 1967, “The Discovery of Grounded Theory”, Aldine, New York,

1967. Hofstede, G. 1991, “Cultures and Organizations: Software of the Mind”, McGraw-Hill, New

York. Hoque, K. and Moon, M. 2001, “Counting Angles: A Comparison of Personnel and HR

Specialists”, Human Resource Management Journal, 11 (3): 5 -22. Kamal M. 2007. “Productivity and Profitability of National Commercial Banks of Bangladesh: A

Comparative Analysis”, The Cost and Management, 33 (2): 19-28. Kaplan, R. S. and Norton, D. P. 2001, “The Strategy Focused Organization: How Balanced

Scorecard Companies Thrive in New Business Environment”, Harvard Business School Press, Boston.

Katz, R. L. 1974, “Skills of an Effective Administrator”, Harvard Business Review, Sep-Oct: 90.

Khatri, N. 1999, “Emerging HR Issues in Singapore”, International Journal of Manpower, 20 (8): 516 – 529.

Khilji, S. E. 2001, “human resource management in Pakistan in Budhwar, P. S.(Ed.)”, Human Resource Management in Developing Countries, Routledge, London : 102-120.

Nanayakka, G. 2003, “Patterns of Human Resources Practices in Sri Lanka and a Cultural Explanation”, Proceeding of the Faculty Seminar at the Graduate School of Economics of Nagoya University.

Nonaka, I. and Takeuchi, H. 1995, “The Knowledge Creating Company”, Oxford University Press, New York: 60 -65.

Pareek, U. 1997, “Training Instruments for Human Resource Development”, Tata McGraw-Hill, New Delhi: 75-109.

Rao, T. V. 1990, “The HRD Missionary: Role and Function of HRD Managers and HRD Departments”, Oxford and IBH Publishing, New Delhi: 50 -70.

Rothwell, W. J., Prescott, R.K. and Taylow, M. W. 1998, “Strategic Human Resource Leader”, Davis-Black Publishing, Pao Alto: 26-37.

Ruona , W. E. A., Lynham, S. A. and Chermack, T. J. 2003, “Insights on emerging Trends and the Future of HRD”, Advances in Developing Human Resources, 5 (3): 272 – 282.

Schuler, R. S. and Jackson, S.E. 1987, “Linking competitive strategies with human resource management practices”, Academy of Management Executive, 11 (1): 207–219.

Sheth, N. R. 1990, “Management Development and Education in South Asia”, Interman, Geneva: 27

Siddiquee, N. A. 2003, “Human Resource Management in Bangladesh Civil Service: Constraints and Contradictions”, International Journal of Public Administration, 26 (1): 1 -10.

Singh, K. 2003, “Strategic HRM and Firm Performance in India”, International Journal of Human Resource Management, 14 (2): 540 – 543.

Sinha, D. P. 2002, “South Asian Management: Challenges in the New Millennium”, Association of the Management Development Institutions in South Asia, Hyderabad.

Strauss. A. and Corbin, J. 1990, “Basics of Qualitative Research: Grounded Theory Procedures and Techniques”, Sage, Newbury Park, CA.


88

Truss, C. and Gratton, L. 1994, “Strategic Human Resource Management: A Conceptual Approach”, International Journal of Human Resource Management, 1 (5): 636–86.

Ulrich, D. (1998) “A New Mandate for HR”, Harvard Business Review, Jan-Feb: 124-133. Ulrich. D. and Brockbank, Y. 2005, “The HR Value Proposition”, Harvard Business School

Press, Boston: 200-222. Wheelen, T. and Hunger, D. 1998, “Essentials of Strategic Management”, Longman, New

York: 34-41. Wright, P., McMahan, G. and McWilliams, A. 1994, “Human Resources as a Source of

Sustained Competitive Advantage: A Resource-Based View”, International Journal of Human Resource Management, 1 (5): 299–324.

89

1 P W D N R. Rodrigo is the Additional Director, Policy Review and Monitoring Department of the Central Bank of Sri Lanka

Book Review

The Cost of Capitalism:

Understanding Market Mayhem

and Stabilizing

Our Economic Future

Author : Robert J. Barbera;

Publisher : McGraw-Hill Business Book Publishers (2009).

ISBN :

978-0-07-162844-0

Reviewed By : P W D N R Rodrigo 1


90

This book examines the causes of the financial crises and lessons to be learned, and goes on to question the concepts of infallibility of markets, rationale economic agents, preoccupation of monetary policy with inflation etc. Hence, this is a rare book that should be of great interest to an extremely wide audience from the market practitioners, central bankers to scholared economics and finance. In particular, it is a must read for central bankers since it convincingly argues that preoccupation of relevant central banks only with prices and inflation disregarding the bubbles in the financial markets was a primary cause of the financial crises in the past. Barbera’s remedy for financial crises is that when asset bubbles start to build, the Fed should raise rates ( “lean against the wind”). According to him the notion that the Fed should worry about systemic risk only after a crisis hits is madness.

The author translates the theories of Myman Minsky22

(who formulated a theory called “ the financial instability theory, which says as people take more and more risks in a boom, any small disappointment cause a huge consequences due to the increased amount of debt/ high leverage) into language both accessible and entertaining for non economists. Although Barbera uses Minskey’s theories, he disagrees with Minsky's remedy - government-guided investment, and instead proposes that central and commercial bankers become more aware of the phenomena (boom and bust cycles)and act accordingly.

This book is also an essential reading for any investor large or small. Anyone planning to assume any amount of risk, be that in the financial markets, real estate, business, just about anything, will benefit from its lessons. Using a two fictional characters, Hana and Hal, he illustrates in simple language how increased leverage leads to dramatic loss even when house prices drops a little.

The main theme of the book revolves round the financial crisis of 2008 and it repeatedly asserts that bubbles and financial crisis are an inherent feature of capitalist economic system , and that large scale government interventions are the cost of capitalist economic system which, according to the author is still the best economic system. While agreeing with free market system, the author takes to task mainstream market economists who have embraced the mistaken view that capitalism as infallible. In uncomplicated terms the cause-and-effect relationship between Fed policy and a capitalist economy is well illustrated by the author, and sound recommendations are offered for fundamental changes in Fed policy decision making. It is not only filled with valuable insights on the financial boom and bust cycles and provocatively raises numerous questions on the operation of financial markets and conduct of monetary policy.

The book contains 16 chapters divided into four parts ; Part I deals with the Financial Markets and Monetary Policy in Perspective, Part II covers the Economic Experience -1985-2002, Part III discusses the Emerging Realities -2007 :2008 and finally Part IV is devoted to Recasting Economic Theory for the 21

st Century. Main points contained in the book could be

summarised as follows ;

☼ Stability and economic tranquility lead to over confidence that these temporary conditions were permanent features of the economy. As a result, confidence grows leading to the taking of excessive risks, driven by greed and herd instinct. Because many of the investors become over confident, and leverage their investments too heavily, even a very small economic downturn renders large numbers of investors insolvent. The denial of conventional economic theory, of this inherent instability of the capitalist economy, causes even dangerous bubbles and more severe recessions than necessary.

22 Myman Minsky ,Professor of Economics of Washington University in St Louis (1965-1900). He proposed theories

linking financial market fragility, in the normal life cycle of an economy, with speculative investment bubbles endogenous to financial markets. Minsky claimed that in prosperous times, when corporate cash flow rises beyond what is needed to pay off debt, a speculative euphoria develops, and soon thereafter debts exceed what borrowers can pay off from their incoming revenues, which in turn produces a financial crisis. As a result of such speculative borrowing bubbles, banks and lenders tighten credit availability, even to companies that can afford loans, and the economy subsequently contracts.

http://en.wikipedia.org/wiki/Washington_University_in_St_Louis

http://en.wikipedia.org/wiki/Financial_market

http://en.wikipedia.org/wiki/Speculation

http://en.wikipedia.org/wiki/Investment_bubble

http://en.wikipedia.org/wiki/Endogeneity_(economics)

http://en.wikipedia.org/wiki/Prosper

http://en.wikipedia.org/wiki/Cash_flow

http://en.wikipedia.org/wiki/Irrational_exuberance

http://en.wikipedia.org/wiki/Bank

http://en.wikipedia.org/wiki/Lender

http://en.wikipedia.org/wiki/Loan

http://en.wikipedia.org/wiki/Economy

The Cost of Capitalism: Understanding Market Mayhem and Stabilizing Our Economic Future

91

☼ Over the past 25 years financial market crises occurred with painful regularity due to errors and excesses. Meanwhile, our willingness to pretend that capitalism is infallible has grown. Wild Wall Street swings have included the 1987 Crash, the early 1990s S&L Crisis, the LTCM Meltdown, the dot-com crash of the late 1990s, and now the 2008 crisis in U.S. real estate. In addition, Asia had Japan's early 1990s collapse, followed by a panic several years later in emerging Asian economies.

☼ Mainstream economists are ill suited to forecasting future economic crises. The author considers their economic theories and forecasting tools to be flawed, in their failure to comprehend that capitalism is prone to cycles. Further more the academics created the complex mathematical models underpinned by the assumption that people are well informed and act rationally. He writes “They failed to acknowledge that financial markets periodically go haywire.”

☼ The idea of Alan Greenspan, that central banks should do nothing to burst asset bubbles is wrong. Instead, it is suggested that central banks should build the level of corporate-bond spreads into their models. When spreads are low, risk appetites are high, as they were in 2005-06. That should lead central banks to tighten monetary policy. When spreads are high, they should ease.

☼ The current crisis suggests that monetary and fiscal policy cannot be driven exclusively by economic fundamentals such as inflation and unemployment. When interest rates are low, consumers and businesses do not just borrow money; they borrow money to buy assets, setting up a feedback loop that can eventually lead to a bubble. When such a bubble is inflating, government revenues (in the form of taxes on capital gains, bonuses, corporate profits and property sales) tend to be strong. As governments are now discovering, such revenues collapse very quickly when the bubble bursts.

☼ From 1945 to 1985 there was no recession caused by the instability of investment prompted by financial speculation--and since 1985 there has been no recession that has not been caused by these factors. Yet, as Barbera shows, the Fed behaved as if inflation were the economy's only enemy. He argues persuasively that American capitalism needs "a new paradigm," one that recognizes the wisdom of Minsky.

☼ Adams Smith's idea that self-interest could produce an orderly society benefiting everybody seems to have been carried too far. The author writes “"We sort of morphed from Adam Smith's invisible hand, that markets move things in a very helpful direction, to some notion [that] free markets have an infallible hand,"

☼ The simple capitalism model that Smith described no longer seems to fit the complicated, highly interconnected global economy of today. Therefore “ it's time to update our economic thinking. In conventional industries, the demise of companies leads to Schumpeter's “creative destruction” with capital being reallocated to more productive areas. But in banking and finance, a crisis leads to “deflationary destruction” as capital is eliminated. Businesses, investors and consumers lose confidence; borrowers are unable to repay their lenders, who suffer as well. Therefore, Schumpeter's idea is not good always. For example “ when a great many financial institutions have lent the wrong way and there's this chance for a dominolike default, there's nothing creative about that destruction. You've got to prevent it. That's the cost of capitalism: Periodically you will have to come to the rescue of the financial system." ( example : rescue of AIG)

☼ Investors and policy-makers can reduce the risk of truly gruesome outcomes if they better plan for the violent economic storms, which history confirms are always over the horizon. Investors will learn how to gird themselves for the roller-coaster ride that is free market capitalism; policy makers will find out how to plan for crises they know will occur at some point; and academic economists will rethink their pursuit of ever more elaborate mathematical models that bear no. resemblance to the real world. Stop pretending that people are always rational and that markets are always efficient and be prepared for market mayhem..


92

☼ Minsky divided the peoples’ behaviour during a steadily growing output and low inflation, into three phases. In the first, investors take on little enough debt that they have no trouble meeting their capital and interest payments. In the second, they stretch their finances so they can only afford the interest. In the third, or Ponzi phase, they take on debt levels that require rising prices to be safely financed (the homebuyers who took on 125% mortgages at the peak of the property boom were a classic example). When markets reach this fantasy land, a small change in the fundamentals or in investor attitudes can be enough to cause the system to unravel. Once prices start to drop, borrowers start to default on their loans, or seek to sell their assets, causing prices to fall further.

☼ The Government policy makers must respond to rapidly rising markets with the same urgency they muster in responding to rapidly falling markets. There is also a need for regulations that provide “safety margins for the myriad nonbank financiers” that arose outside 1930s-era banking industry reforms. But the author cautions that policy makers must recognize that “innovation on Wall Street, over time, dulls the applicability of a given set of regulations,” thereby mandating constant vigilance and regular updating of the rules.

The weakness of the book is that it does not describe, in adequate detail, at what time central banks should try to tighten monetary policy in response to asset price booms. Furthermore, it is not clear whether increase in interest rates (by the FED) would have stopped the housing bubble since the Fed was in fact raising interest rates before the crisis.

A practical lesson that is useful is that if we accept Minsky’s idea that financial markets are not always right, then we might be willing occasionally to act against market interests, however loudly bankers would complain.

Finally, the book is recommended as an excellent reading for central bankers, since the analysis is pragmatic, accounting for human nature and its consequences in the context of the financial system.

GGuuiiddeelliinneess ffoorr pprreeppaarraattiioonn ooff ppaappeerrss ffoorr tthhee CCBBSS Journal ooff

MMuullttiiddiisscciipplliinnaarryy SSttuuddiieess

Author A1 , Author B

2

The abstract: It should be a brief (100-200 words) synopsis of your paper. Its purpose is to provide a quick outline of your presentation, giving the reader an overview of the research. This is an important aspect of your paper, as it is this description that may attract the reader to continue and finish your full report. In particular, the main contributions of the paper should be explicitly mentioned in the abstract. No citations or cross-references should be used.

Keywords should be so chosen that they best describe the contents of the paper.

1 Author is with the Department of Department name, University/Institution Name (as a footnote of the title page) 2 Author is the designation of author, name of Department, Central Bank of Sri Lanka

11.. IInnttrroodduuccttiioonn

The CBS Journal of Multidisciplinary Studies

(CBS-JMDS), publishes original research papers on a

vast range of research topics. These instructions provide

you the basic guidelines for final preparation of papers

accepted for publication in the CBS-JMDS.

22.. LLiitteerraattuurree RReevviieeww 2.1.Fonts

Papers should be prepared for review using

Microsoft Word. The main font to be used throughout the

paper is Helvetica whilst Times New Roman has to be

used for the Paper Title and Authors Name. Try to follow

the font sizes specified in Table 1, as best as you can.

Table. 1. Font sizes used in the paper

Paper's Parts Font

Size Italic Bold

Paper Title 24 √ √ Authors' names 12 √ √ Abstract 10 √ -

Keywords 8 √ -

Main text, References 10 - - Headings 12 - √ Sub-headings 12 - √ Footnotes, Authors' affiliations, Sub- and Superscripts 8 √ -

2.2. Formats

We encourage you to limit the paper to a maximum

of 20 pages. In f o r m a t t i n g y o u r A 4 -size p a p e r ,

t h e t o p m a r g i n should be set to 20mm, bottom

margin to 25mm. The text should be justified. Do check

the spelling prior to submission.

33.. IIlllluussttrraattiioonnss The figures and tables should be positioned at the

tops and bottoms of pages whenever possible. Figure

captions should be below the figures; table captions

should be above the tables. Try to place the figures and

tables after their first mention in the text.

Eg-: Table 1 Deposit Rates and Lending Rates (percent, p.a.)

Year/Month AWDR* AWPLR**

2007/01

2007/12

7.60

10.31

15.52

17.00

All half-tone illustrations (pictures/photographs)

should be clear black and white prints. Do not use

photocopies.

44.. RReeffeerreenncceess aanndd

FFoooottnnootteess All references should be listed in alphabetical order at the

end of the paper. The title of the paper should be placed within

quotes. Number footnotes separately in superscript. Place the

actual footnote at the bottom of the page in which it is cited.

55.. EEqquuaattiioonnss Number e q u a t i o n s c o n s e c u t i v e l y w i t h

e q u a t i o n numbers in parentheses align with the right

margin, as in Eq. (1). Be sure that the symbols in your

equation have been defined before the equation appears or

immediately following. When you refer to equations in the

text, refer to " (1)" except at the beginning of a sentence:

"Equation (1) is used...."

k

log Pi

= β0 + ∑ βj χij (1)

1 – Pi j = 1

66.. AAbbbbrreevviiaattiioonnss

aanndd AAccrroonnyymmss Define abbreviations and acronyms the first time

they are used in the text, even after they have already been

defined in the abstract. Do not use abbreviations in the title

unless they are unavoidable.

77.. CCoonncclluussiioonn The conclusion may review the main points of the paper,

do not replicate the abstract in the conclusion. A conclusion

might elaborate on the importance of the work or suggest

applications and extensions.

88.. AAppppeennddiixx Appendixes, if needed, should appear before the

acknowledgment.

99.. RReeffeerreenncceess

Asquith, P., and D.W. Mullins, 1986, “The Impact of

Signaling with Dividends, Stock Repurchases, and Equity

Issues”, Financial Management, pp 27-44.

Shea,J., (1992), “The Welfare Effects of Economic

Liberalization under Financial Market”, Academia Economic

Papers 20 (2, September), pp 697-716

journal of studies - centre for banking studies · college of asia and the pacific ... cbs journal...

Documents