RESPONSIBLE FINANCE, EMERGING MARKETS AND

PORTFOLIO DIVERSIFICATION. AN APPLICATION TO THE BRAZILIAN MARKET

Área temática: Ética e Responsabilidade Social

Raphael Acosta

raphael.acosta@yahoo.com.br

Resumo: A associação entre finanças corporativas e sustentabilidade tem vindo a ganhar relevância nos últimos anos, e

sua extensão através dos mercados financeiros desenvolvidos tem sido objeto de inúmeras publicações. No entanto, poucos

estudos lidar com o investimento socialmente responsável em mercados emergentes, e estes estudos são geralmente

limitados a medição de desempenho.

Neste trabalho, propomos a analisar o potencial de diversificação de fundos socialmente responsáveis em relação a

índices tradicionais em uma referência de mercado emergente no mercado brasileiro. Para atingir este objetivo, utilizamos

a metodologia de Engle e Granger (1987) para analisar seus colegas de movimentos em relação ao mercado local, bem

como provas de europeus, americanos e britânicos de referência para determinar se se há ou não há, o potencial para a

diversificação, combinando esses ativos financeiros.

Em vista dos testes, podemos concluir que o rosto de financiamento convencional, SRI no Brasil representa uma

oportunidade para diversificação da carteira para os investidores locais e internacionais.

Palavras-chaves: Investimento de Responsabilidade Social; Mercados Emergentes; A diversificação da

carteira; Bovespa; Cointegração.

ISSN 1984-9354

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

2

1. INTRODUCTION

Research on the performance of socially responsible investment (SRI) has emerged with the

development of socially responsible funds (SR). Two supported debates were then imposed on the

question of the link between ethical performance and financial performance on one side, and on the

comparison of the performance of ethical funds face conventional finance on the other side. To address

these issues, the major developed markets (mainly Western) have been meticulously analyzed.

Surprisingly, if the issue of performance galvanized so many researchers, the integration of SRI

in an optical portfolio diversification has not generated much enthusiasm. However, the integration of

financial markets is a central theme of the international financial analysis. Indeed, consider the

performance of financial assets without contextualize their reciprocal movements seems inconsistent

with modern portfolio theory.

Similarly, research on SRI in emerging markets is still in its infancy. To date, there is a limited

number of similar studies on emerging markets, mainly focused on Asian markets (Aras et al., 2010;

Saleh et al., 2011; Kolk and Muller 2011). This finding is so more regrettable that the development of

SRI in emerging markets is quite singular.

Given this situation, it seemed appropriate to extend the analysis of the movements of funds

and SR index traded on an emerging market benchmark, the Bovespa. This study aims to answer two

questions:

1 / Is there a potential for portfolio diversification between SRI and conventional finance in

Brazil?

2 / Is there a potential for portfolio diversification between the Brazilian SRI and the main

international financial centers?

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

3

2. INSTITUTIONAL BACKGROUND

Portfolio Divertification

Harry Markowitz developed the first major theory of portfolio selection in 1952, under the

name of Modern Portfolio Theory (MPT). This theory suggests that the performance of a portfolio can

be expressed as a weighted average of the returns of each asset, while its risk depends on the variance

of returns of each asset and their covariance. A fundamental aspect of MPT is that even riskier assets

can reduce overall portfolio risk if they have a low covariance with the returns of other assets that

compose it.

Markowitz (1952) formulated the problem of portfolio choice as the mean and variance of a

portfolio of assets. According to him, for a defined variance, an investor can maximize the expected

return of the portfolio, and vice versa, for an expected return an investor can minimize the risk of its

portfolio. The objective of this theory is to determine a combination of risk and return for an investor

to optimize the risk / return. Therefore, rational investors choose their ideal portfolio along the

efficient frontier based on their financial goals and risk aversion.

At the macroeconomic level, the existence of specific geographic area business cycles should

logically help diversify the risk of a portfolio of domestic securities. According to MPT, the

international portfolio diversification is expected to optimize the risk / return, playing on the low

correlation between selected markets. Indeed, if the developed markets the gains from diversification

are more likely to search in the sectorial disparities operation rather than in the geographical

specificities, significant potential gains are still possible by investing in emerging markets.

The first studies on international diversification (such as Grubel, 1968; Levy and Sarnat 1970

and Lessard, 1973) indicated that international investment could be useful to optimize the risk / reward

ratio compared to a locally diversified portfolio. In addition, Levy and Sarnat (1970) and Grubel and

Fadner (1971) have shown that the tendency of securities to move together in the same economy is an

essential element validating the argument that international diversification reduces the systematic risk

more efficiently than the internal diversification. Moreover, Solnik (1974) puts forward the idea that, if

the primary motivation for the creation of an equity portfolio is to reduce risk, then a strategy to

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

4

diversify your portfolio internationally should be preferred at the expense of simple local

diversification. Moreover, Odier Solnik & (1993) Michaud et al. (1996) show that, regardless of the

nationality of investors, investing in different markets provides benefits in terms of reducing risk and

improving performance.

For twenty years, a considerable literature has been produced on the diversification of

portfolios in emerging financial markets (Harvey, 1995; Bekaert and Harvey, 1997; Bekaert and

Harvey, 2000; Bekaert et al. 2009; Pukthuanthong and Roll, 2009). The growing interest of researchers

and investor's motivations can be explained by various reasons, among which we can highlight the

opportunities for portfolio diversification and the recurrence of financial crises. Indeed, since the early

1980s, emerging markets have often been considered the most promising ones because of their high

yields and potential gains they offer in terms of portfolio diversification (El Hedi Arouri et al, 2010).

Moreover, we note with surprise that the theme of portfolio diversification incorporating SRI is

still embryonic. To our knowledge, there is only one study on the topic, published in 2010 by Roca,

Wong, and Tularam in the Accounting Research Jornal. This study concluded the SR indices

integration of the American, Canadian, British and Australian markets. However, it does not consider

the indices of local exchanges but only local indexes DJSI and the period studied includes simulated

indexes , which significantly limits the scope of these results.

Socially ResPónsible Investment

Since year 2000, the number of studies analyzing the SRI performance in developed markets

has increased exponentially, and most of the funds and SR indices have been targets of deep analysis.

Gond (2006) believes that “this is probably academic question most frequently studied empirically in

the areas of management.” Yet there is a limited number of similar studies on emerging markets.

To illustrate this balance, we find that the majority of identified in the meta-analysis of Revelli

and Viviani (2013) studies are devoted to the U.S. market (58%), followed by the UK market (11 %)

and French market (10%). Indeed, the concentration of studies in these markets can be explained by

their pioneering in the development of SRI. However, with the significant development of SRI in

South Africa and Brazil, we could imagine a greater enthusiasm on the part of the scientific

community to these issues. To our knowledge, only a few local authors are interested in this subject,

and publications are very recent.

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

5

In Brazil, as a first step, the authors have invested most of their time on the different

characteristics of companies including funds and indices SR. Beato (2009), da Costa (2009) and Nunes

et al. (2010) analyze the societal balance sheets, and point the methodological limitations of the ISE.

However, the scope of this research is relative considering the many changes that the index has

experienced. Thus, an update of these studies may be relevant. More recently, Machado et al. (2012)

highlight the existence of a significant relationship between social and environmental investments and

the integration of companies in the index.

Concerning the analysis of the performance of SRI, Cangussú et al. (2009) show that the

returns on the ISE are not significantly different from those of its benchmark over the period 2005-

2007. Nossa, et al. (2009) confirm these results at the lack of significant relationships between social

assessments, environmental and financial performance of various listed companies. More recently,

Ortas et al . (2012) assure that the ISE is less risky and presents a similar profitability to that of its

benchmark index during periods of market stability. Maimon (2012) who demonstrates that SRI

provides risk reduction without affecting the profitability of assets confirms these results. Surprisingly,

these studies were limited to analysis of risk factors and absolute profitability, without measuring the

relative performance such as the Sharpe ratio, or other. A priori, these results seem to confirm the

theories claiming that SR strategies are vectors for better risk control.

3. THE BRAZILIAN CONTEXT

The BM&F Bovespa

Inaugurated in 1890, the Bolsa de Valores de São Paulo (Bovespa) was the first capital market

based in Brazil. In 2000, all 27 regional stock exchanges have merged, to be included in the Sao Paulo

stock exchange. The Bovespa has concentrated all negotiations of the Brazilian equities, and the

regional exchanges now only keep just some local activities. Since its merger with BMF (organized

market derivatives), the Bovespa has become the leading Latin American financial center and one of

the largest in the southern hemisphere. Its benchmark is the Índice Bovespa (Ibovespa). It consists of

core assets traded on the Bovespa in terms of values and liquidity (presented in Table 1), and is not

limited in quantity.

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

6

Table 1 – The first 10 Bovespa capitalisations

Name Sector Capitalisation*

Petrobras Oil and Gaz 318.556

Vale Mines and Minerals 250.878

Itau Unibanco Finance 148.444

Ambev Food 142.904

Bradesco Finance 106.420

Transm. Paulistas Energy 86.827

Banco do Brasil Finance 76.753

Santander Finance 63.847

Ultrapar Holding 60.317

Itausa Finance 51.283

Source : Bovespa 2012

*In million of Reais

Like the major emerging places, we can see a very high concentration of titles around some

areas. First, the mining and energy, which by itself accounts for 41 % of the weight of listed

companies. This concentration is even more pronounced than the industry consists of a very small

number of players. The other predominant sector of this market is the financial and banking stocks.

This segment comprises 31% of the capitalization of listed companies, which indicates that non-

financial and non-energy values (broadly defined) together represent less than 30% of the total

capitalization.

The unique element of this market is probably the economic implication of the Federal

government in the management of listed companies. Indeed, despite the economic reforms of the

Cardoso government (1995-2002) who led a wave of privatizations unprecedented in Brazil, the

Federal government continues to control many companies directly or indirectly, under the status of

“mixed economy companies”. Among the most important we can mention: Petrobras, Eletrobras,

Sabesp , Copel , Cesp , All America... What makes the first player in this market .

Despite its many attractions, the Brazilian stock exchange seems far from passionate academic

researchers. Indeed, to our knowledge, there are still very few studies on the integration of Bovespa in

the world market, and these items are relatively recent. Oliveira and Medeiros (2009) show that the

NYSE and the Bovespa are segmented but there are co- movements between the latter and the Dow

Jones. However, the results also show that the practice of arbitration based on the lead-lag effect is not

http://central-do-investidor.exame.abril.com.br/Acoes.aspx?acao=PETR4http://central-do-investidor.exame.abril.com.br/Acoes.aspx?acao=VALE5http://central-do-investidor.exame.abril.com.br/Acoes.aspx?acao=AMBV4http://central-do-investidor.exame.abril.com.br/Acoes.aspx?acao=BBDC4http://central-do-investidor.exame.abril.com.br/Acoes.aspx?acao=TRPL4http://central-do-investidor.exame.abril.com.br/Acoes.aspx?acao=UGPA4http://central-do-investidor.exame.abril.com.br/Acoes.aspx?acao=ITSA4

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

7

economically feasible because of transaction costs. Conversely, Aloy et al (2013) show that the

Brazilian and Argentine markets are not cointegrated in pairs or with the NYSE.

The Socially reposnible investment

The concept of socially responsible finance has emerged in Brazil in the 1990s, with the UN

Conference on Environment and Development in 1992, also known as the “Earth Summit” in Rio de

Janeiro. On this occasion, a number of institutions concerned with social responsibility have emerged,

including the Institute of Ethos1. Singular case among emerging economies by 2000, the publication of

a non-financial report has been imposed on companies listed on Bovespa, under the theme of

governance. In 2001, a new segment called “corporate governance” has been introduced to identify the

companies and the index “Índice de Ações com Governança Corporativa Diferenciada" aggregates

better evaluated.

Thus, at the end of 2001, the Brazilian bank "Banco Real" (owned by Dutch group ABN Amro)

launched two "Ethical Funds", the first funds SR Latin America. In 2004, it was the turn of "Banco

Itau" bank to launch its own background SR named "Fundo Itaú Social Excelência". Until 2006, the

assets of these funds were very low, insignificant. With the commitment of Bovespa and the launch of

the SR index, the number of SR funds and their assets have increased dramatically. Since 2010, about

60% of companies in the Ibovespa publish sustainability reports, and five of them include the Dow

Jones Sustainability Index (DJSI).

Following the launch of the JSE SRI by the Johannesburg Stock Exchange, Bovespa was

inspired by this experience to launch its own index, the "Índice of Sustentabilidade Empresarial" (ISE)

in December 2005. The index was developed by the Centro de Estudos em Sustentabilidade (GVCES)

of the School of Business Administration of São Paulo of the Getúlio Vargas Foundation (FGV -

EAESP) with the financial support of the IFC. Today, the ISE is governed by a multi-stakeholder -

board including representatives from government, NGOs and investor associations collectively

approve the inclusion or exclusion of companies in the portfolio. Only 200 companies representing the

largest stocks in the Brazilian market are eligible for this index. The application for this membership is

voluntary and occurs in response to the GVCES questionnaire sent to the 200 eligible companies. Like

the JSE SRI, the scope of the questionnaire GVCES adopts the concept of Triple Botton Line and

1 Ethos Institute is the organization of reference in the field of CSR in Brazil: www3.ethos.org.br

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

8

evaluates these three aspects in an integrated way, by adding three additional indicators (disclosure,

product and corporate governance) .

The methodology of this index has the distinction of associating a quantitative approach based

on the analysis of questionnaires sent by companies who want to integrate the index and qualitative,

based on public information. Once the questionnaires are completed, statistical analysis is used to

evaluate and classify the most successful companies. In a second step, the board qualitatively analyzes

each company eligible to distinguish companies that make up the index.

Graphic 1 – The ISE Index Composition

Source : Bolsa de Valores de São Paulo

The capitalization of the ISE index is dominated by the financial sector (Graphic 1), with a

quarter of the total capitalization. This concentration is similar to that of the benchmark - Ibovespa.

The mining sector is also well represented, with 16 % of the capitalization, which is higher than the

concentration of the sector in the benchmark (12%). Finally, the areas of water and energy and food

represent 16% and 10% of the capitalization of the ISE, which corresponds to twice the concentration

in the Ibovespa.

We note that three major companies comprising the Ibovespa were not included in the ISE,

Petrobras, Embraer and AmBev. Petrobras is a petroleum company - South American first

capitalization - which was excluded from the index for reasons related to the marketing of a fuel type,

however, authorized by the Federal government. AmBev and Embraer are respectively Latin American

leaders in the markets for alcoholic beverages and aerospace, and non- integration is related to the

nature of their activities.

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

9

4. EMPIRICAL SPECIFICATIONS AND DATA

The Data

Research on emerging capital markets is not easy. Indeed, the foundations of international

financial data are often incomplete with regard to emerging companies and scholarships are not always

willing to cooperate with academic research markets. This problem is even more pronounced for

information about SRI in these markets. Thus, access to these data can sometimes be complicated,

time-consuming , if not impossible in many cases.

For this study, we use two types of data: indices administered by the Bovespa and the SRI

funds managed by independent companies. The indices were all provided by local exchange

companies in a daily frequency. We selected socially responsible index, the benchmark and the

representative of small and mid-cap indices.

As we explained above, access to information has been very difficult as far as SRI investment

funds are concerned. No SR strategy has been communicated to us, and ANBIMA2 Association sent us

over the main SR funds of the Brazilian market. Table 2 presents funds and socially responsible

indexes studied and their benchmarks.

2 Associação Brasileira das Entidades dos Mercados Financeiro e de Capitais

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

10

Table 2 – List of funds and indices studied

Indices Investments Funds

Name Symbol Name Symbol

Ibovespa BovR Banco do Brasil Ações ISE Jovem FIC Bov1

Índice de Sustentabilidade Empresarial BovI ITAU Perso Ações Excelencia Social Bov2

Small Cap BovS ITAU Ações Excelencia Social FICFI Bov3

Mid-Large Cap BovM Bradesco Prime FIC FIA ISE Bov4

FTSE 100 Ftse HSBC FICFI de Ações Sustentabilidade Bov5

Euro Stoxx 50 Euro Santander FIC FI Ethical Ações Bov6

S&P 500 S&P

The period runs from 01/01/2006 to 31/12/2012. The courses will be studied in local currency

for internal comparisons. To test the relations with foreign markets we have selected 3 benchmarks and

tests will be in U.S. dollars3.

The stationarity

In the late 1970s and early 1980s, time series econometrics has shown that the estimation of

standard econometric models from non-stationary time series can lead to "spurious regressions" in the

words of Granger and Newbold (1974).

A stationary time series is considered under three conditions. First, the average must remain

stable over time. Then, the variance must also remain stable over time. Finally, the covariance between

two values of x.t must depend only on the length of the interval between the two sets of values and no

availability. A time series is defined as stationary if its stochastic characteristics remain stable and

continuous over time as:

There are several tests to detect non-stationary time series. We present below the Dickey -

Fuller and the Phillips-Perron tests which are two tests commonly used in the literature.

3 Converted with exchange rates from the World Bank.

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

11

The Augmented Dickey–Fuller test

The statistical test most widely used for the analysis of the stationarity of a time series is the

unit root test proposed by Dickey and Fuller in 1981. Strictly, it is not a stationarity test, but actually a

unit root test, to invalidate the hypothesis of stationarity of the time series. The null hypothesis of this

test shows the presence of a unit root, which is to invalidate the assumption that the time series is

stationary. In other words, this test is similar to validating the non-stationary stochastic type by testing

the following assumptions:

H0 : δ = 0 presence of a unit root (non-stationary process)

H1 : δ < 0 no unit root (stationary process)

The test is then broken down according to three models, estimating the coefficients δ, β and c

by regression. The parameter « β » characterizes a deterministic trend and « c », a constant qualified as

« drift »:

Under the alternative hypothesis, the series follows a stationary process.

Under the alternative hypothesis, the series follows a stationary autoregressive process with drift.

Under the alternative hypothesis, the series follows a stationary autoregressive process of

deterministic drift with.

The main problem of the application of this test comes from determining the number of phase

shifts into the equation to be tested. Several methods exist to adjust this parameter. In practice most of

the work select models that minimize the Akaike information criteria (1974) and Schwarz (1978).

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

12

It is then to test the null hypothesis 𝛿 = 0 against the null hypothesis 𝛿 < 0 using the Student's

statistic t. If the statistic t is greater than the critical value of the specific table of MacKinnon (1996),

the null hypothesis H0 is rejected in favor of the alternative hypothesis, the series has no unit root and

follows a stationary process.

The Phillips-Perron test

Phillips and Perron (1987, 1988) propose another test for non-stationarity of time series. This

test is a non-parametric adaptation of the Dickey and Fuller test, corrected of the presence of

autocorrelation, without adding endogenous delayed as in the method of ADF. The null hypothesis of

the test is, as the DF test, the presence of a unit root. The test procedure involves five steps:

1 / Estimated by OLS three models Dickey-Fuller;

2 / Variance Estimation of short-term :

3/ Estimate of the variance of long term also called « corrective factor » :

To estimate this variance it is necessary to define a number of the delays called "truncation Newey-

West." It is estimated based on the number n of observations:

4/ Calculating the statistical Phillips-Perron

With

5/ Statistical comparison of Philipps-Perron critical values table MacKinnon (1996).

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

13

THE COINTEGRATION

Studies on the co-movement between emerging stock markets and developing countries have

recently gained ground in the literature of financial economics. Engle and Granger (1987) were the

first authors to formalize the concept of cointegration through the development of bivariate

cointegration tests to test the relationship between two variables in the long term.

The idea behind the concept of cointegration is that of a link in the independent time short-term

movements. There is no requirement that these two variables move in a coordinated manner in the

short term. However, there is a long-term stable relationship between the two variables and their

respective movements tend to compensate for a stationary series (Bourbonnais, 2011).

The presence of a cointegrating relationship illustrates the existence of a strong relationship

between the background and SRI benchmark and the lack of diversification potential risk of long-term

market. On the contrary, the absence of cointegration is synonymous with the presence of

opportunities for risk diversification. To test the convergence funds / indexes SR and conventional

long-term, we apply the methodology proposed by Engle and Granger (1987).

Estimating long-run relationships

To assess the long-term relations between the two series, we estimate the following regression

model:

Then, we test the stationarity of zt residue such as:

To test the stationarity of zt residue, you should use the Dickey-Fuller and Augmented Dickey-

Fuller, using the critical values tabulated by Engle and Yoo (1987) and McKinnon (1991). If zt is

stationary, the linear combination of the two is also a stationary cointegrating relationship may exist

between the two series. If zt is not stationary, we can conclude that the two series are not cointegrated.

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

14

The correction error model

The concept states that cointegration two series move jointly in the long term. In other words,

the gap between xt and yt converge average despite temporary fluctuations, such as unusual differences

between the two series - even if they are excessive - will be corrected over time.

This dynamic aspect is introduced by the error correction model (ECM) defined by Granger

(1981). It is thus a model for defining the adjustments that lead to a situation of long-term equilibrium

between two variables integrated of the same order which is written as follows:

The γ coefficient associated with represents the return force to the target given by the long

term cointegration. It must be negative and significantly different from zero so that the error correction

mechanism is validated. Otherwise, the ECM must be rejected because it would be in the opposite

direction by separating the two sets of the convergence objective.

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

15

5. EMPIRICAL RESEARCH

Stationarity Tests

The tests performed in level give very expressive results. Indeed, none rejects H0 presence of a

unit root. In other words, the assumption of stationarity series is systematically rejected, regardless of

the test used. It appears from these tests that the use of methods of simple regressions between sets

studied would result in a high risk of obtaining « spurious » regressions. This situation is common in

the study of financial time series in levels. Table 3 presents the results of the ADF test and Phillips-

Perron applied to all courses in first difference.

Table 3 – Results of the stationarity tests in first difference

Model 1

Model 2

Model 3

ADF

Prob

a PP

Prob

a ADF

Prob

a PP

Prob

a ADF

Prob

a PP

Prob

a

Bov1 -7,6 0,00 -43,4 0,00 -7,6 0,00 -43,4 0,00 -7,6 0,00 -43,4 0,00

Bov2 -7,6 0,00 -44,0 0,00 -7,6 0,00 -44,0 0,00 -7,6 0,00 -44,0 0,00

Bov3 -42,7 0,00 -42,7 0,00 -42,7 0,00 -42,7 0,00 -42,7 0,00 -42,7 0,00

Bov4 -7,0 0,00 -43,9 0,00 -7,0 0,00 -44,0 0,00 -7,0 0,00 -44,0 0,00

Bov5 -8,4 0,00 -43,8 0,00 -8,4 0,00 -43,8 0,00 -8,4 0,00 -43,8 0,00

Bov6 -9,3 0,00 -43,6 0,00 -9,3 0,00 -43,6 0,00 -9,4 0,00 -43,6 0,00

BovI -6,9 0,00 -41,1 0,00 -6,9 0,00 -41,1 0,00 -6,9 0,00 -41,1 0,00

BovM -32,8 0,00 -32,8 0,00 -32,8 0,00 -42,3 0,00 -32,8 0,00 -32,8 0,00

BovR -7,9 0,00 -42,3 0,00 -7,9 0,00 -42,3 0,00 -7,9 0,00 -42,3 0,00

BovS -5,7 0,00 -36,0 0,00 -5,7 0,00 -36,0 0,00 -5,7 0,00 -36,0 0,00

Euro -9,5 0,00 -44,3 0,00 -9,5 0,00 -44,3 0,00 -9,5 0,00 -44,3 0,00

Ftse -7,6 0,00 -44,8 0,00 -7,6 0,00 -44,8 0,00 -7,6 0,00 -44,8 0,00

S&P -10,1 0,00 -48,0 0,00 -10,1 0,00 -48,0 0,00 -10,1 0,00 -48,0 0,00

Test results converge very pronounced, and the null hypothesis of unit root is always rejected in

favor of the alternative hypothesis. None of the analyzed series possesses a unit root, which leads us to

conclude that they are all stationary in view of their significant p-value threshold of 1%.

Any series of our sample is stationary in level, but they are all in first differences. In other

words, they are all integrated in the first order, denoted I (1). These results lead us to analyze the

presence of cointegration relationships between funds and indices SR and benchmarks.

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

16

Cointégrations analysis

Inter-markets relationship

Table 4 presents the results of regressions between the European, British, American indices and

the Brazilian ones. The table is divided into two main parts. The first presents the estimated regression4

coefficients and the Students associated statistics. The second part reports the results of unit root tests

ADF and Phillips-Perron applied to residues from these estimates.

Table 4 – Estimation of long-term inter-markets relationships

Estimation of equation Residus stationarity test

Coeff t stat C t stat R2

ADF

1

ADF

2

ADF

3 PP1 PP2 PP3

Euro 0,025 4,298 1056 41,272 0,01 -1,88* -1,879 -1,883 -1,90* -1,89 -1,93

BovI Ftse 0,006 1,735* 896 28,246 0,002

S&P 0,449 13,835 390 9,514 0,095 -1,87* -1,877 -1,864 -1,871 -1,868 -1,803

Euro 0,22 1,178* 29044 35,762 0,001

BovR Ftse 0,182 1,792* 28220 28,167 0,002

S&P 7,82 7,367 20215 15,087 0,029 -1,89* -1,898 -1,398 -1,911 -1,911 -1,41

* : result in significant threshold of 10%

** : result in significant threshold of 5%

*** : result in significant threshold of 1%

Overall, the coefficients of determination associated with these regressions are extremely low,

and the coefficients of regressions between indices BovR-Euro, BovI-Ftse and BovR-Ftse are not

significant. In addition, no unit root test applied to residus not proved significant at the 5% level. We

cannot estimate error correction model from these relationships.

The Inter-markets Relationship

To test the long-term relationships between funds and indices in the Brazilian market, we use a

similar applied to the inter-markets relationship methodology. These tests are applied to the relations

linking the SR index funds and their benchmark indexes:

4 With the method of ordinary least squares

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

17

Table 5 – Estimation of long-term inter-markets relationships

Estimation of equation Residus stationarity test

Coeff t stat C t stat ADF1 ADF2 ADF3 PP1 PP2 PP3

BovI

Bov

M 2,27 98,68 -39,73 -3,58 -0,43 -0,43 -2,20 -0,40 -0,39 -2,17

BovR 0,03 112,57 62,25 7,60 0,37 0,37 -0,75 0,43 0,43 -0,70

BovS 1,18 206,63 297,69 79,72 -4,76*** -4,76*** -4,81*** -5,12*** -5,11*** -5,15***

Bov1

BovI 0,00 636,31 0,00 2,01 -1,62* -1,62 -2,31 -13,47*** -13,46*** -16,18***

Bov

M 0,00 90,27 -0,05 -3,97 -0,42 -0,42 -2,17 -0,76 -0,85 -2,38

BovR 0,00 101,49 0,07 8,46 0,44 0,45 -0,86 0,32 0,34 -1,09

BovS 0,00 189,45 0,27 69,13 -4,25*** -4,24*** -4,46*** -4,99*** -4,99*** -5,23***

Bov2

BovI 0,01 326,50 0,80 19,18 -2,15** -2,15 -2,54 -4,14*** -4,13*** -4,83***

Bov

M 0,03 165,61 -0,40 -4,20 -1,08 -1,08 -2,49 -2,38** -2,37 -4,24***

BovR 0,00 154,44 1,18 13,72 -0,23 -0,22 -1,07 -0,56 -0,55 -1,48

BovS 0,02 151,35 4,93 70,29 -4,00*** -4,00*** -4,17*** -4,17*** -4,16*** -4,57***

Bov3

BovI 0,01 243,44 1,33 25,48 -1,86* -1,86 -2,60 -2,69*** -2,68* -3,41**

Bov

M 0,03 905,91 0,00 0,00 -1,68* -1,68 -3,07 -2,74*** -2,73* -4,48***

BovR 0,00 167,07 1,54 20,58 -1,02 -1,01 -1,47 -1,20 -1,19 -1,64

BovS 0,02 133,48 5,03 67,80 -3,44*** -3,43*** -3,64** -3,76*** -3,76*** -4,08***

Bov4

BovI 0,00 596,17 0,02 11,02 -1,84* -1,84 -1,85 -11,47*** -11,47*** -12,21***

Bov

M 0,00 114,59 -0,03 -2,78 -0,76 -0,76 -2,39 -1,29 -1,27 -3,23*

BovR 0,00 119,07 0,07 9,24 0,16 0,17 -1,12 -0,01 0,01 -1,37

BovS 0,00 191,21 0,31 81,13 -4,06*** -4,06*** -4,05*** -5,16*** -5,16*** -5,15***

Bov5

BovI 0,00 139,38 0,13 22,71 -0,68 -0,68 -1,90 -1,07 -1,07 -2,27

Bov

M 0,00 204,97 0,03 5,91 -3,33*** -3,33*** -3,32* -4,14*** -4,13*** -3,82**

BovR 0,00 186,20 0,11 26,52 -1,99** -1,99 -1,97 -2,31** -2,33 -2,32

BovS 0,00 90,74 0,36 55,99 -2,93*** -2,93** -3,22* -3,01*** -3,00** -3,46**

Bov6

BovI 0,26 242,25 29,07 27,59 -2,55** -2,55* -2,81 -3,25*** -3,24** -3,62**

Bov

M 0,62 133,14 0,00 0,00 -0,75 -0,75 -2,17 -1,38 -1,36 -3,13

BovR 0,01 132,53 37,46 20,14 0,32 0,33 -0,35 -0,34 -0,33 -1,17

BovS 0,31 155,03 99,07 75,58 -3,33*** -3,33** -3,34* -4,03*** -4,02*** -4,05***

* : result in significant threshold of 10%

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

18

** : result in significant threshold of 5%

*** : result in significant threshold of 1%

In contrast to previous results, internally, we find very many opportunities cointégrations.

Indeed, of the 27 connections tested, 15 long-term relationships were found between SR funds and

indices and benchmarks. Given the large number of models estimated error correction, Table 6

presents a reduced presentation of these models.

Table 6 – Error Correction Estimate Models

ECM BovI - BovS ΔBovI = 0.951649 ΔBovS -0.053308 z(t-1)

t-Student

32.52169

-5.8694***

ECM Bov1 - BovI ΔBov1 = 0.000873 ΔBovI -0.116953 z(t-1)

t-Student

76.33922

-10.522***

ECM Bov1 - BovS ΔBov1 = 0.000889 ΔBovS -0.060690 z(t-1)

t-Student

26.38971

-6.0249***

ECM Bov2 - BovI ΔBov2 = 0.014345 ΔBovI -0.035474 z(t-1)

t-Student

77.69989

-5.7513***

ECM Bov2 - BovM ΔBov2 = 0.030312 ΔBovM -0.019310 z(t-1)

t-Student

62.18433

-3.2804***

ECM Bov2 - BovS ΔBov2 = 0.014646 ΔBovS -0.044142 z(t-1)

t-Student

27.69039

-5.0526***

ECM Bov3 - BovI ΔBov3 = 0.013923 ΔBovI -0.021804 z(t-1)

t-Student

77.70071

-4.5416***

ECM Bov3 - BovM ΔBov3 = 0.029499 ΔBovM -0.024795 z(t-1)

t-Student

64.72687

-3.6901***

ECM Bov3 - BovS ΔBov3 = 0.014270 ΔBovS -0.036646 z(t-1)

t-Student

28.09572

-4.6137***

ECM Bov4 - BovI ΔBov4 = 0.000890 ΔBovI -0.101197 z(t-1)

t-Student

76.95736

-9.7683***

ECM Bov4 - BovS ΔBov4 = 0.000905 ΔBovS -0.063672 z(t-1)

t-Student

26.57937

-6.1278***

ECM Bov5 - BovR ΔBov5 = 2.45E-05 ΔBovR -0.014903 z(t-1)

t-Student

68.97150

-3.6820***

ECM Bov5 - BovS ΔBov5 = 0.000864 ΔBovS -0.021332 z(t-1)

t-Student

26.78114

-3.6355***

ECM Bov6 - BovI ΔBov6 = 0.260824 ΔBovI -0.025752 z(t-1)

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

19

t-Student

65.81284

-4.8867***

ECM Bov6 - BovS ΔBov6 = 0.273290 ΔBovS -0.044823 z(t-1)

t-Student

27.12455

-5.0316***

* : result in significant threshold of 10% ** : result in significant threshold of 5%

*** : result in significant threshold of 1%

All the estimated relationships are significant at the 1%, reflecting a high degree of integration

of these series. The hypothesis of cointegration relationship between the charge and its benchmark

index is systematically rejected with the notable exception of BovS index.

We note that, with the exception of background Bov5 a significant cointegration relationship

was measured between the SR and SR index funds. Similarly, all funds and Brazilian SR index have a

significant cointegration relationship with the Small Cap Index. These singular results reflect two

trends. On the one hand, there is a significant long-term relationship between SR funds and responsible

index. On the other hand, the composition of the index funds and SR seem to incorporate a significant

proportion of securities from small cap companies.

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

20

6. CONCLUSIONS

Econometric tests applied to SRI Brazilian market given significant results that lead us to four

distinct conclusions:

1 / The Brazilian market is not cointegrated with developed markets selected.

The study of cointegration relationships between the Brazilian market and benchmarks of

European, British and American places demonstrated the potential of diversification offered by the

Brazilian stock exchange as a whole. Indeed, despite the financial collaboration partnerships signed

between Bovespa and the NYSE, no cointegration relationship was detected between these different

markets.

2 / There is a strong potential for internal diversification in this market between conventional

finance and responsible finance.

Cointegration tests on the fund and the index SR did not detect any long-term relationship with

the benchmark and the average capitalization. This reflects the potential for diversification that offers

SRI compared to conventional finance.

3 / Ethical funds are cointegrated to socially responsible index.

Strong relationships cointegrations were highlighted between ethical funds and SR index.

Indeed, with the exception of Bov5 all funds are cointegrated at BovI. This leads us to believe that

these funds are very similar to the ethical index composition.

4 / Ethical funds and index are cointegrated in the Small Cap Index.

A strong relationship was found between the SR index and the Small Cap index, and tests of

SR funds have yielded similar results. Funds and managers indexes seem to incorporate a significant

proportion of securities from small-capitalization companies, which confirms the study of the

composition of the SR index point 3.

Finally, the empirical study was conducted with exploratory manner on a subject and a concept

still minimally processed. Indeed, research on the nature of the relationship between conventional and

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

21

socially responsible long-term finance still seems very developed in view of the conceptual challenges

it poses, like the theme of SRI in emerging markets . This study provides a new perspective on the

analysis of the performance of SRI in Brazil, with a special focus on its capabilities in terms of

portfolio diversification. The results demonstrated the potential for portfolio diversification offered by

this asset class for both domestic and international investors.

To further this study, it would be appropriate to extend this analysis to other emerging markets.

Furthermore, an analysis of the composition of the financial indices may be judicious in providing

additional qualitative component in this exploratory study.

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

22

REFERENCES

Aloy, M., Boutahar, M., Gente, K., & Péguin-Feissolle, A. (2013). Long-run relationships between international

stock prices: further evidence from fractional cointegration tests. Applied Economics, 45(7), 817-828.

Akaike, H. (1974). A new look at the statistical model identification. Automatic Control, IEEE Transactions

on, 19(6), 716-723.

Aras G., Aybars A., & Kutlu O., (2010). Investigating the relationship between corporate social responsibility

and financial performance in emerging markets. International Journal of Productivity and Performance

Management, Vol. 59 Iss: 3

Beato, R., Souza, M., & Parisotto, I. (2009). Rentabilidade dos índices de sustentabilidade empresarial em

Bolsas de Valores: um Estudo do ISE/Bovespa. Revista de Administração e Inovação, 6(3), 108-127.

Bekaert, G., & Harvey, C. R. (1997). Emerging equity market volatility. Journal of Financial economics, 43(1),

29-77.

Bekaert, G., & Harvey, C. R. (2000). Foreign speculators and emerging equity markets. The Journal of

Finance, 55(2), 565-613.

Bekaert, G., Hodrick, R. J., & Zhang, X. (2009). International stock return comovements. The Journal of

Finance, 64(6), 2591-2626.

Bourbonnais, R. (2011). Économétrie-8e édition. Dunod

Cangussú Rezende, I. A., Goldner Nunes, J., & Salles Portela, S. (2009). Um estudo sobre o desempenho

financeiro do índice bovespa de sustentabilidade empresarial. Revista de Educação e Pesquisa em

Contabilidade (REPeC), 2(1), 93-122.

da Costa, F. M. G., Voese, S. B., & Rosa, L. (2009). Custos e investimentos ambientais praticados pelas

empresas do setor de energia elétrica que participam do ISE Bovespa 2008/2009. Revista Contabilidade e

Controladoria, 1(3).

Dickey, D. A., & Fuller, W. A. (1981). Likelihood ratio statistics for autoregressive time series with a unit

root. Econometrica: Journal of the Econometric Society, 1057-1072.

El Hedi Arouri, M., Bellalah, M., & Nguyen, D. K. (2010). The comovements in international stock markets:

new evidence from Latin American emerging countries. Applied Economics Letters, 17(13), 1323-1328.

Engle, R. F., & Granger, C. W. (1987). Co-integration and error correction: representation, estimation, and

testing. Econometrica: journal of the Econometric Society, 251-276.

Gond, J. P. (2006). Construire la relation (positive) entre performance sociétale et financière sur le marché de

l’ISR : de la performation à l’autoréalisation ? Revue d’Économie Financière, n°85, avril.

Grubel, H. G. (1968). Internationally diversified portfolios: welfare gains and capital flows. The American

Economic Review, 58(5), 1299-1314.

Grubel, H., Fadnar, K., 1971. The interdependence of international equity markets. Journal of Finance 26 (1),

89-94.

Harvey, C. R., 1995. Predictable risk and returns in emerging markets. Review of Financial Studies, Vol. 8,

Issue 3, pp. 773 - 816.

Kolk A. et Muller A. (2011). CSR Performance in Emerging Markets Evidence from Mexico. Journal of

Business Ethics - Volume 104 / 2011

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

23

Lessard, D. R. (1973). International portfolio diversification: a multivariate analysis for a group of Latin

American countries. The Journal of Finance, 28(3), 619-633.

Levy, H., Sarnat, M., 1970. International diversification of investment portfolios. American Economic Review

60, 668-675.

MacKinnon, J. G. (1996). Numerical distribution functions for unit root and cointegration tests. Journal of

applied econometrics, 11(6), 601-618.

Machado, M. A. V., Macedo, M. Á. D. S., Machado, M. R., & de Siqueira, J. R. M. (2012). Análise da relação

entre investimentos socioambientais ea inclusão de empresas no Índice de Sustentabilidade Empresarial-(ISE)

da BM&FBovespa. Revista de Ciências da Administração, 14(32), 141-156.

Maimon, D. (2012). An assessment of Brazilian SRI funds. VIII Congresso nacional de excelência em gestão, 8

e 9 de junho de 2012

Markowitz, H. (1952). Portfolio selection. The journal of finance, 7(1), 77-91.

Michaud, R.O., Bergstrom, G.L., Frashure, R.D. & Wolahan, B.K., 1996. Twenty years of international equity

investing: Still a route to higher returns and lower risks? Journal of Portfolio Management, Vol. 23, Issue 1, pp.

9-22.

Newbold, P., & Granger, C. W. (1974). Experience with forecasting univariate time series and the

combination of forecasts. Journal of the Royal Statistical Society. Series A (General), 131-165.

Nossa, V., Cezar, J., & Nossa, S. (2009). A relação entre o retorno anormal e a responsabilidade social

e ambiental: um estudo empírico na Bovespa no período de 1999 a 2006. Brazilian Business Review,

Vitória, 6(2), 121-136.

Nunes, J. G., Teixeira, A., Nossa, V., & Galdi, F. C. (2010). Análise das Variáveis que Influenciam a

Adesão das Empresas ao Índice BM&F Bovespa de Sustentabilidade Empresarial. BASE–Revista de

Administração e Contabilidade, 328-340.

Odier, P., & Solnik, B. (1993). Lessons for international asset allocation. Financial Analysts Journal, 63-77.

Oliveira, G. R., & Medeiros, O. R. (2009). Testing the Existence of Lead-Lag Effects Between the US and

Brazilian Stock Markets. Brazilian Business Review, 6(1), 1-20.

Ortas E., Moneva J. & Salvador M. (2012). Does socially responsible investment equity indexes in emerging

markets pay off? Evidence from Brazil. Emerging Markets Review, vol. 13, issue 4, pages 581-597

Perron, P., & Phillips, P. C. (1987). Does GNP have a unit root?: A re-evaluation. Economics Letters, 23(2),

139-145.

Phillips, P. C., & Perron, P. (1988). Testing for a unit root in time series regression. Biometrika, 75(2), 335-346.

Pukthuanthong, K., & Roll, R. (2009). Global market integration: An alternative measure and its

application. Journal of Financial Economics, 94(2), 214-232.

Viviani, J. L., & Revelli, C. (2013, May). SRI portfolio management: a meta-analysis. In 30th International

French Finance Association conference (AFFI).

Roca, E., Wong, V. S., & Tularam, G. A. (2010). Are socially responsible investment markets worldwide

integrated?. Accounting Research Journal, 23(3), 281-301.

Saleh M., Zulkifli N., Muhamad R., (2011). Looking for evidence of the relationship between corporate social

responsibility and corporate financial performance in an emerging market. Asia-Pacific Journal of Business

Administration, Vol. 3 Iss: 2

Solnik, B. H. (1974). Why not diversify internationally rather than domestically?. Financial analysts journal,

48-54.

XI CONGRESSO NACIONAL DE EXCELÊNCIA EM GESTÃO 13 e 14 de agosto de 2015

24

Schwarz, G. (1978). Estimating the dimension of a model. The annals of statistics, 6(2), 461-464.