BUSINESS GROUP PERFORMANCE IN PERU

BUSINESS GROUP PERFORMANCE IN PERU

JAN GÜNTHER MEIST (1184210)

Rijksuniversiteit Groningen Faculty of Economics

Landleven 5

9747 AD Groningen, Netherlands Tel: +31 (050) 363 7185 Fax: + 31 (050) 363 3720

Abstract

In this paper I test the existence of differences in return between group-affiliated and stand-alone companies in Peru. I find that on average group-affiliated firms listed on the Lima Stock Exchange (BVL) are associated with higher returns instead of my initial hypothesis of a business group discount.

Designing a CAPM model enhanced by a group factor and based on industry portfolios, I determine that business groups’ ability to diversify into different industries can be ruled out as reason for this difference in returns. And that indeed there is no further indication that stock market returns differ for Peruvian group-affiliated and non-group companies.

STOCK MARKETS IN DEVELOPING COUNTRIES - BUSINESS GROUP PERFORMANCE IN PERU

SECTION I... 4

INTRODUCTION... 4

SECTION II... 8

BUSINESS GROUP DEFINITION... 8

BUSINESS GROUPS AND PERFORMANCE... 9

SECTION III... 13

DATABASE AND SAMPLE SELECTION... 13

DATA RESTRICTIONS... 13

VARIABLES,FACTORS AND CONCEPTS USED... 14

METHODOLOGY FOR GROUP VERSUS NON-GROUP ANALYSIS... 17

THE GROUP FACTOR METHODOLOGY BASED ON INDUSTRY GROUPS... 22

SECTION IV ... 27

SUMMARY AND CONCLUSIONS... 27

REFERENCES ... 30

DATA APPENDIX ... 34

Introduction

Business groups are wondrous things. On the one hand, one is tempted to condemn them for the accumulation of power in the hands of just a few controlling shareholders and the degree of influence these few people have on the entire economy; on the other hand, cooperation and forming alliances is doubtlessly a smart thing to do and sometimes even imperative when entrepreneurs are unable to overcome imperfections, in labor or capital markets.

It is no surprise that business groups or conglomerates are one of the most popular forms of business entities, even more so in emerging or developing markets. According to Ungson, Steers and Park (1997) 40% of total output was produced by the largest 30 business Korean business groups in 1996.

Topping this, Claessens, Fan and Lang (2002) show an average group affiliation of almost 70% for a sample of 9 Asian nations, among them Korea and Japan. As far as Peru is concerned, the fact that the then president Alan Garcia called upon a group of only 12 managers – the so called “12 apostles”

– in 1988 to lead Peru out of the economic crises of that time, underlines that few business groups indeed may exert decisive influence on the total economy in emerging country markets.

Business groups have a long-standing tradition in all developed markets, may that be North America, Europe or Japan. Strategic alliances, cross-shareholdings, joint-ventures, hostile or friendly take-over of competitors – all these are forms of cooperation which allow businesses to share information and technologies in order to grow more efficient, more powerful and more profitable. Academic literature has argued that business groups or conglomerates evolve out of the necessity to overcome imperfections in product, capital or labor markets. So find Claessens, Djankov, Fan and Lang (1999) evidence in nine East Asian economies that business groups exist to overcome imperfections in external capital markets, and that diversification takes place as part of vertical integration complementing the creation of internal markets. Kim, Hoskisson and Hong (2004) attribute the existence of the Korean chaebol business groups to market imperfections. But more importantly than

the issue WHY business groups exist – Claessen, Djankov, Fan and Lang (1999) also find evidence for the theory that business group exist to expropriate minority shareholders – is HOW they work and to what end. The Japanese keiretsu and Korean chaebol, as well as Indian business groups have been object of numerous studies to various aspects of business groups. Gangopadhyay, Lensink and Van der Molen (2002) examine the investment behavior under financial distress, Chang and Choi (1988) strategy structure and performance, Claessens, Djankov and Klapper (1999), Campbell and Keys (2002) and Joh (2003) corporate governance and performance issues, Ferris and Kitsabunnarat (2003) and Claessens, Djankov, Fan and Lang (1999) the effects of diversification, and Hoshi, Kashyap and Scharfstein (1990) the behavior under financial stress – to name just a few works from recent history.

There exist a number of studies on business groups in Chile and numerous studies integrate Chile, Brazil or Argentina to a number of Asian countries, like Claessens, Djankov and Klapper (1999), Claessens, Djankov, Fan and Lang (1999). Khanna and Yafeh (2002) use data from Argentina, Brazil, Chile, Colombia, Mexico, Peru and Venezuela among other non-Latin American countries for their article on risk sharing, La Porta, Silanes, Shleifer and Vishny (1998) include Argentina and Mexico in his study, Khanna and Rivkin (2001) examine the performance of group firms in emerging markets among them those of Peru, but data has been so limited and was drawn from several patch work sources. Furthermore, empirical studies to investigate the stock market performance of business groups relative to stand-alone companies are scarce. In this paper I will analyze the performance of group-affiliated companies in Peru relative to their stand-alone counterparts using empirical data of the past decade, hoping to close the gap of information and shed some additional light on this emerging country stock market.

Why the Peruvian business groups would be of interest, one might ask. The answer is quite simple.

During the early nineties the Fujimori government privatized state companies, which encouraged a stream of foreign investors. His successor Toledo kept the country on a low inflation and stable growth path during the past 5 years of his turn. With the Free Trade Area of the Americas (FTAA) on the way, and a free trade agreement between the US and Peru under negotiation, a number of international firms are on the brink of entering the Peruvian market. Some international companies like Telefónica,

Goodyear or Bellsouth are already engaged. However, taking over entire companies by hostile or friendly takeovers may be impossible due to the degree of which business groups dominate the domestic market. This was possible and has been done during the nineties when the state property was privatized – as mentioned – and henceforth made profitable¹. Peru has positioned itself to compete in the foreseeable future with its bigger neighbors Argentina and Brazil and is certainly one of the most attractive markets for foreign investments. This paper deals with whether it is more profitable for this very foreign stock market investor to focus on groups or stand-alone companies. Or more precisely, with the matter whether over the past decade group companies have performed better than their stand-alone counterparts and whether diversification has anything to with this performance.

Group-affiliated companies are less risky, because they are diversified. Khanna and Yafeh (2002) find that affiliation to a group reduces standard deviation of returns between 20-30% for the three tigers Korea, Taiwan and Thailand, and Brazil. Essentially, this is one of their reasons for existence. Due to the underdeveloped marked the domestic entrepreneur needs to establish companies outside his own industry to smoothen income. Basically, the lack of financial markets forces the domestic investor to diversify personally rather than by the share diversification. Investing as minority shareholder in such a group, however, bears additional risk. Khanna and Palepu (1999) argue that groups are difficult to monitor and minority shareholders are subject to ‘questionable practices to their detriment’ as do Claessens, Djankov and Klapper (1999). In addition the financial markets as well as judicial system in emerging markets are likely to be underdeveloped and inadequate; hence investor protection is at a minimum. All this implies the existence of a group discount, i.e. group-affiliated companies exhibit low(er) returns. This seems to be logical, since group-affiliated companies share effectively risk, income streams are smoothened over time, and due to the larger degree of collaboration, more bureaucracy is to be expected and going with it hand in hand inefficiency while distributing these income cash flows, as George, Kabir and Douma (2004) argue. Business groups are more predictable and overall less risky; from this argumentation a risk-adverse international investor would therefore be advised to invest in Peruvian business groups. In this paper I will be examining the question, whether Peruvian business groups outperform their solitary rivals or not after correcting for general market risk, and whether this performance can be attributed to the matter of intra-group diversification into different

1See Torrero (2002) for more info on that topic.

industries. Business groups have their own way to interact with the market, they have internal capital markets, tend to be larger than single companies and hence experience form economies of scale and scope. Correcting for this general market risk, however, there remains a risk associated with diversification which makes them less profitable, because less risky.

The remainder of this thesis is organized as follows. Section II defines the term ‘business group’, reviews the current state of literature on business groups, their structure and performance in more detail, and restates my hypothesis. Section III elaborates on the data and methodology used to test the hypothesis. Section IV analyzes and interprets the obtained results. References and Data Appendix are provided at the end of the paper.

SECTION II

Business Group Definition

Literature knows several different types of business groups. In corporate groupings like the Japanese keiretsu member firms are bound together through common characteristics – like ethnicity of owners –

interlocking directorates, or personal networks. The managers of each company, however, have a great deal of autonomy, while they coordinate their activities via a central bank as documented by Hoshi and Kashyap (2001). Then there are horizontal financial-industrial groups in Russia, which Perotti and Gelfer (1999) describe ‘are more properly industry alliances’. Yet the groups that occur most frequently in Western Europe, Latin America and East Asia are family business groups. These groups may be of horizontal or pyramidal structure, but are all controlled by members or friends of the same family.

Shabunina (2005) defines business groups as ‘a group of companies having the same major controller: individual or interconnected group of individuals.’ This is similar to the definition used in La Porta, Silanes, Shleifer and Vishny (1998), who defines as affiliate to a business group a company, whose shares are controlled to 20% by the ultimate owner, i.e. is controlled by the ultimate owner. The problem with this definition is firstly, that within business groups frequently cross shareholdings occur in order to mitigate the principal agent problem within group companies. Secondly, the ultimate owner definition may be difficult to apply to horizontal structures. Thirdly, the 20% threshold used by the mentioned author is based on the notion that 20% of voting rights assures effective control of a company. However, threshold is more or less arbitrary, it assumes that the rest of the shares are widely held and hence control of more than 20 % of the shares is equivalent to being the largest shareholder. In underdeveloped and emerging markets, the number of investors is small and hence widely held companies the rarest of exceptions.

The Peruvian “Comisión Nacional Supervisora de Empresas y Valores (CONASEV)” comparable to the US American “Stock Exchange Commission (SEC)” defines business groups² as ‘a conjoint of judicial persons… that are subject to the control of the same natural person or the same conjoint of natural persons. With the exception, that a judicial person is self-controlled if, when through the distribution of shares and voting rights no natural person or conjoint of natural persons has more than 30% voting rights nor the capability to designate more than 50% of the members of the directorate.’

The point of view that a business group is a conjoint of independent companies acting with the common strategic intent and common control is inherent to all definitions. The trick lies in defining formally control. For the reminder of this paper I will be using the CONASEV definition of a business group, which takes into account the higher degree of control needed in emerging markets like Peru, due to weaker investor protection and judicial systems, furthermore it seems sensible to dealing with Peru using the Peruvian definition of business group.

Business Groups and Performance

Having defined what constitutes a group-affiliated company and a business group as a whole, the next question is, why precisely they should be any more profitable or less profitable than their stand-alone counterparts. The whole is more than the sum total of its parts, it is said. But what precisely is the quality that business group companies have that sets them apart from their stand-alone counterparts?

Business groups are made up of legally independent units that coordinate their actions to maximize profits of the entire group. Business groups can efficiently allocate their resources taking advantage of

2http://www.conasev.gob.pe/acercade/ConsultaCiudadana/ProyReglamento_PropIndirecta_Vinculac_GrupoEcon.pdf accessed 17/08/2005.

Original text:”... Grupo Económico es el conjunto de personas jurídicas, cualquiera sea su actividad u objeto social, que están sujetas al control de una misma persona natural o de un mismo conjunto de personas naturales.

Por excepción, se considera que el control lo ejerce una persona jurídica cuando, por la dispersión accionaría y de los

derechos de voto de dicha persona jurídica, ninguna persona natural o conjunto de personas naturales ostente más del 30% de los derechos de voto ni la capacidad para designar a más del 50% de los miembros del directorio. …”

Where CONASEV is unavailable to provide a connection to an economic group, for the reminder of this paper I will consider this company as stand-alone.

fast growing industries, and divesting from slower growing industries as George et al. (2004) argue.

They use diversification to smoothen the income of the ultimate owner over time, making use of internal capital markets to gain resources for investments. This internal capital market is of particular importance, if the external market is not well developed.

Diversification is associated with discounts and lower risk, efficient internal capital market with premiums, two opposing effects hence. Claessens, Djankov, Fan and Lang (1999) and Claessens, Djankov and Klapper (1999) find for East Asia and Chile that diversification due to risk-reduction is a secondary issue for the ultimate owner, the primary being rather the expropriation of smaller shareholders. Joh (2003) finds the same for a study on Korea. This is in line with La Porta, Silanes, Shleifer and Vishny (1999), who show that family business groups occur more frequently in countries with low investor protection. And despite the great opportunities of an efficient internal market Scharfstein and Stein (2000) predict inefficient resource allocation within groups due to conflicting interests and principal agent problems. George, Kabir and Douma (2004) as well show that resource transfers in business groups takes place from high to low-performing firms, and hence are inefficient.

Claessens, Djankov, Fan and Lang (1999) argue that business groups are positively related to performance, if internal capital markets are more efficient than the external. In consequence, in countries with underdeveloped financial market systems group affiliation is to be associated with higher performance whereas in developed markets discounts should prevail.

The classic markets for studies on business groups have been Southeast Asia and India and indicate, nonetheless, the existence of discount for business groups. All in all, Lins and Servaes (2002) document lower performance for firms from Hong Kong, India, Indonesia, Malaysia, Singapore, South Korea and Thailand that are diversified and part of a business group, Joh (2003) shows underperformance for Korean chaebol companies compared to stand-alone firms, Campbell and Keys (2001) find the same. Khanna and Palepu (2000a) find that the largest and most diversified groups outperform, smaller and medium business groups, which, however, make up more than 90% of their sample. Khanna and Rivkin (2001) find that in some economies group affiliation is positively correlated with performance while in others the effect is either negative or not significant. For The Latin American

countries in their study I present their results in Table 1.0. As one can see the results are not conclusive at all and in addition vary greatly from those I obtain in Table 1. 2 Most decisively I find a percentage of group affiliation of up to 78%, whereas Khanna and Rivkin (2001) count only 24% of companies to this category. But whereas they draw their data from various independent sources, I am using official government data. These conflicting results underline one essential fact: data for this particular market is difficult to retrieve and results in the past have been far from conclusive.

Authors, like the previously mentioned as well as Van der Molen and Lensink (2005), associate the quality of a group to its ability to diversify. There seems to be agreement that uncorrelated diversified business groups, i.e. groups that diversified into various industries that are not related to the original core business, render lower average returns, but also involves less risk. Business groups behave thus just like investments into less than perfectly correlated diversified investments in shareholdings.

Because business groups are present in different industries, industry-related shocks can be compensated performance by other group members reducing volatility. This comparison is not entirely accurate, since business groups organize themselves around a common strategic objective, coordinate actions and distribute investment resources to the most efficient firm, which gives the group a competitive advantage. Resources are not transferred from high to low performing companies, but from companies where resources are available, but with low potential of future earnings, to potentially lucrative business opportunities within the group. Based on this competitive advantage business groups will be likely to be experiencing higher returns unlike a simple diversified portfolio of shares.

From the perspective of diversification it is then this very characteristic of wider business activities that diminishes overall group returns, but steadies overall income volatility. Although more opportunities may be realized, so are more failures. To use the terms of the capital asset pricing model (CAPM): A business group is in the end nothing more than a portfolio of shares, whose degree of risk and return depends on the risk of the individual companies and their correlation. But other then in a portfolio of shares, profit is distributed between these companies. In consequence, the share price of each individual company in a business group is smoothened over time, volatility and overall shareholder’s earnings reduced. From this argument the business group, experience reduced risk, but at the same time a ‘business group discount’.

Summing up, the great majority of current literature attributes the existence of business groups to their ability to reduce risk by diversification. According to the Capital Asset Pricing Model (CAPM) lower risk is necessarily associated with lower return. Some authors, however, attribute the business group discount to the expropriation of minority shareholders and overall inefficient internal capital markets. In this paper I will empirically test whether stand-alone companies are more profitable than group affiliated firms for the single country of Peru after having corrected for market risk.

Hypothesis: “Even after correction for market risk there is a difference between group- affiliated versus stand-alone companies in share price terms. Shares of group- affiliated Peruvian businesses experience lower income risk because of diversification and hence lower returns. Peruvian stand-alone firms are more risky and more profitable.”

Since current literature attributes the business group discount to diversification and risk-reduction it seems to me furthermore logical to employ the CAPM as base model to test this hypothesis, and test an added group factor to this very model.

SECTION III

Database and Sample Selection

Capital markets in South American markets are underdeveloped in comparison to Asia, Europe, or the United States. Consequently it is difficult to obtain a sufficiently sized sample to run an analysis, even more so for one single country. I am using the ECONOMATICA database, which compiles data of roughly 6000 shares, indices and exchange rates from the United States, Great Britain and seven Latin American countries including Peru. In this paper I analyze specifically the Peruvian business groups and therefore will be using only and only the data on Peruvian companies listed on the domestic stock exchange.

On the Lima Stock Exchange, 200 companies have enlisted shares, in addition 4 companies have emitted shares abroad and 34 have listed bonds. Of all of these companies the database provides data. However, I restrict the sample to shares of active companies and using the most actively traded of the shares (the last available monthly value), if several shares have been published.

Data Restrictions

The data used is limited in more than one way. The data for Latin America is limited in time. The same is particularly valid for Peru, for which only in the beginning of the nineties, the government decided to privatize state companies, only after 1992 there is an influx of data into ECONMATICA, meaning that only at that time private investor demand necessitated public recording of results. Hence I am considering only the relative short time frame of the last decade (July 1995 to April 2005).

The second impairment of the data is the limited amount of companies for which it is available. Peru is a one-country market, and one might ad not the biggest in Latin America. The database

ECONOMATICA contains data of 125 different actively traded Peruvian shares, for which I loose even more due to the fact that I need to construct industry portfolios and not just use an individual company.

For industry groups like “Textil”, the sub-portfolios for “group-affiliated” and “stand-alone” are actually incorporating a number greater than just 2 companies. Since group-affiliation is such a common occurrence in Peru it is difficult to find industries for which the sub-portfolio of stand-alone companies has more than just one member. I am using only industry groups with more than 2 members in each sub portfolio. But this is by no means compensating the individual volatility by large numbers.

Furthermore, because of the small number and the frequent unavailability of data the monthly average cannot be calculated for various months. Although the data is much more extensive than the data used in Khanna and Rivkin (2001), these restrictions have to be kept in mind while interpreting the results.

Variables, Factors and Concepts Used

In a first step, I establish the CAPM as a benchmark model to predict stock market returns for the Peruvian market. The CAPM provides an easy way to determine the theoretically appropriate price on an asset. The model takes into account an assets sensitivity to the market (individual risk), which can be diversified away by diversification into less than perfectly correlated assets, and systematic risk, which is part of the every asset and therefore cannot be diversified away. The great achievement of the model developed by Sharpe and Lintner in the 1960’s lies in the mathematical explanation for the trade-off between risk and return and its implication on asset prices. Nonetheless apply the usual restrictions between theory and real life, that is to say the assumptions of the model, such as risk aversion of investor, rational expectation, no arbitrage opportunities, perfect capital markets (in particular information flow) and risk-free rates with limitless borrowing capacity and universal access are hardly ever met under normal circumstances. The assumptions of the CAPM are simply hardly ever met in real life, much less in developing markets. Perfect capital markets and risk-free returns arguably exist in the developed countries like Europe, the U.S., Japan and some other East Asian nations, but are certainly in emerging countries hardly found and even more so in Latin America. What

is imperative for the establishing of any asset-pricing model, however, is that the capital market is open to a sufficient amount of transaction during the day, i.e. shares are indeed traded. The trading volume for the Peruvian market over the past decade has been from as little of 1 transaction during the trading day for the single share to several thousand trades. Whether rational expectations are and risk aversion are present in developing countries may be doubted, since information flow is restricted, stock market supervision is often lacking and corruption as well as nepotism flourishes in many countries. The issue of whether the South American markets in general are indeed efficient is much more an issue of whether corruption, nepotism and fraud are at hand, than trade volume. For references see any published newspaper from Latin America, for Peru I suggest “Perú 21” and “El Comercio”. As far as this paper is concerned, the existence of a reasonably deregulated and active financial Peruvian market must suffice to apply the model.

The original CAPM is given by the formula:

t t t t

t

t t t

t

RF IGBVL RF

R

RMRF RF

R

ε β

α

ε β

α

+

− +

=

−

⇔

+ +

=

−

) (

*

*

where α is expected to be zero, Rt is monthly return on investment, and RMRFt is the market premium calculated as difference between return on the market portfolio (IGBVL_t) and the risk free rate (RF_t). I will therefore use the following monthly data sets:

Rt Return on investment is calculated as the percentage change of the stock market quote including cash and share dividends. ECOMATICA provides the stock market quote in dollar terms. Percentage change on the other hand is calculated manually.

IGBVLt As market return I use the average monthly return of the market portfolio represented by the General Share Index at the Lima Stock Exchange (IGBVL). Like R it is calculated manually as percentage change of the stock market quote including cash and share dividends of the listed firms.

RFt As risk-free return on the market, I come to choose the three-month interest rate on US Treasury bills with constant maturities, denoted tcm3m. The reason for choosing the US interest rates is the high degree of dollarization of the Latin American nations and hence the great dependence on dollar exchange rates. For the time period 1989 to 1994, just prior to the Mexico Crisis, Soydemir (2002) shows that there is slow and varying influence of the changes in the three month T-bills on the stock markets of Argentina, Brazil, Mexico, Venezuela, although Chile was not significantly affected. The pattern here is the stability of the domestic currency. Taylor and Sarno (1997) cite U.S. interests as well as business cycles as global factors, whereas credit ratings and secondary market prices for foreign debt (in most cases “Brady securities”, predominantly denominated in US dollars, but also in Yen and Euro) relate to domestic opportunity and risk. Harvey and Roper (1999), contribute the Asian crisis in 1997 to dollar denominated company debt. Since the dollar in Latin America the currency of accounting and investing it seems imperative to me to also use dollar interest rates to define the risk-free return. So far the somewhat more global reasons to consider the US Treasury bill rate as the risk free rate in my model. In addition, Peru has over greater part of the past decade maintained a fixed exchange rate regime over the most part of the sampling period at around 3.5 Nuevo Soles (PEN) per US$, and still retains a managed float³. Technically their money growth rates and therefore their basic interest rates must co-move. This makes the risk free rates interchangeable. Adding this to the fact that the US dollar practically functions up to now as a parallel currency in Peru and that particularly company debt is denominated in dollar terms, I believe to have more than strong reasons to use the US Treasury bill rates rather than the domestic Peruvian rates.⁴

In a second step I will be estimating a model of differences in returns between group and non-group companies based on the CAPM, i.e. I estimate the difference between group returns and non-group returns and express this difference by the variables of the CAPM, following the formula:

[ ^R

^{− RF}

] ⁻ [ ^R

^{− RF}

] ⁼ [ ^β

^{* RMRF}

] ⁻ [ ^β

^{* RMRF}

] ^{+ ε}

I then calculate a group factor based on industry portfolios and incorporate it in this model. I am using the very same data for theses steps as described above. In asset pricing terms I test whether there

3 Source: International Monetary Fund http://www.imf.org/external/np/mfd/er/2004/eng/1204.htm, accessed 5th July 2005

4For more on the issue of dollarization view Schuler (2005)

exists a factor in stock returns that reflects group affiliation. A factor here being a simple explanatory variable that helps predict the stock returns in our model or the difference in returns in the case of using a model in differences.

The group factor is constructed as the difference between returns of group companies minus the return for stand-alone companies. Hence, it basically measures the difference in returns that does exist between the two samples. There are two aspects that come into play, one being the average out- or underperformance of group firms versus stand-alone firms, and secondly the inference, that the general difference covariates with the single portfolio, i.e. that the performance pattern is consistent throughout all industries and time varying. I use the term group factor as describing the group inherent quality to generate income to a higher/lower degree than stand-alone firms after correcting for the groups different dealings with general market risk.

The difference and the reason why I correct for market risk, is that group companies do work more efficiently due to economies of scope and scale – at least they do have the opportunity to take advantage of synergies arising from their size and complementary business branches. They thus have a different profitability and hence risk. The group factor, however here refers to the group ability to diversify. This quality has only partly to do with higher overall returns, but rather the distribution of returns within the group. Taking into account market risk I do correct for cash flows into the group or stand-alone firm. Adding a group factor I make a statement over intra-group cash flows.

I construct the group factor as described as the difference between group and non-group income. The group factor as explanatory variable is the degree of out/underperformance of group versus non-group firms caused by intra-group distribution issues.

Methodology for Group Versus Non-Group Analysis

Table 1. 1 shows the basic statistics for market return (IGBVL), market premium (RMRF), and the risk- free rate. As South America and developing countries have been struck by various crises (Asia,

Brazilian and Argentine crisis at the end of the nineties) it is not surprising that the average market premium during our sample period is negative. Table 1. 2 provides a comparison between group- affiliated versus stand-alone companies, and the overall portfolio. The standard deviation for the group affiliation is lower in accordance with our prediction of lesser risk and more stable income cash flows.

Meanwhile the return for group-affiliated companies is slightly higher than the return for stand-alone companies. This contests the in the general literature proposed idea that stand-alone companies outperform group-affiliated companies. Table 1. 2 also shows the number of member in the sub portfolios. The number of group-affiliated companies exceeds stand-alone companies by almost the fourfold (97 group-affiliated companies, 28 stand-alone). 78% of our sample consists of group members, a result that is comparable to the numbers obtained in East Asia by Claessens, Djankov, Fan and Lang (1999). Table 1. 3 returns the results for the tests of equality of means and variances.

As one can see, although there is on average a difference in means, the test shows that this difference is not statistically significant. The variances of the two sub-portfolios, however are significantly different. We can therefore deduct that purely by standard methods that, although stand- alone companies are as previously assumed riskier, they failed to provide higher returns for my sample period.

The results from the pre-analysis seem to negate our theory straight away. Firstly, group companies are indeed less risky than stand-alone companies. They show a 1.6 % lower standard deviation. But secondly the very same group companies receive a 0.45% premium on their stock returns in comparison to stand-alone firms. This implies that the theoretical inference, higher risk equals higher return, seems to be invalid for my sample. However, the difference in means is not statistically significant and it remains to be seen, if a model that corrects for correcting for risk may explain the difference in return a little be better, once we take into account market performance.

In order to investigate whether there exists a difference in returns between group-affiliated and stand- alone companies I will establish a base model for the returns, namely the Capital Asset Pricing Model (CAPM). The CAPM by Lintner and Sharpe (1964) explains the returns of a portfolio of shares by the covariance of the portfolio with the market portfolio. Its quality lies in relating risk adjusted returns of single portfolios to the overall market risk. Empirically however it fails to explains the entire returns and

fails to incorporate firm specific risk factors. Indeed, there have been several attempts to correct the model by incorporating more specific risk factors to it, one of them being the Fama French model as described by Fama and French (1995), which incorporates in addition a size factor and a book-to- market factor to the model, arguing that these two variables somehow express different risk types the firms shares are experiencing. Fama and French, however, have been criticized for their explanations of the risk and why it should manifest itself in correlations between size and book to market value to returns. In a prior investigation to this paper I established the Fama-French asset-pricing model, including size and book-to-market factors to the 7 available Latin American countries. The additional factors turn out to be of no significance. For a similar procedure the Peruvian market is not big enough, not only is data lacking, but also the stock market is too small as that making splitting making the shares into different portfolios on size and book-to-market factors is possible. But since the factors of size and book-to-market ratio are not significant in the whole of Latin America, it seems reasonable to believe that in Peru they are not significant either. I will therefore concentrate on the original CAPM.

As previously shown the original CAP model is given by the formula:

) 1 (

*

t

t

RF RMRF

R − = α + β + ε

where Rt is the return on the portfolio in time period t. RMRFt is the market premium, which is calculated as market return on the market portfolio (IGBVL, the return on the general index of shares of the Lima stock exchange) minus the risk-free rate (RF, namely the not seasonally adjusted three- month constant return on US Fed Treasury Bills⁵). Following the assumptions of the model, there exists no fixed difference between individual return and risk free return, hence α equals zero.

With these data sets in mind the model can be expressed as:

R

− RF

= α + β * ( IGBVL

− RF

) + ε

( 1 . a )

5 Source: US Federal Reserve. Data was accessed via the Fed’s website on the 6th June 2005

therefore use equation (1.b) as a benchmark model. I estimate equation (1) for the entire portfolio of companies in my sample, for which the results are given in Table 1. 4. The model here shows weak signs of autocorrelation with a Durbin Watson test statistic of 1,746, which is confirmed by a following LM test., whereas heteroscedasticity can be ruled out⁶. Adding an autoregressive factor to the model improves the explanatory value only by roughly 3% and as before, all variables remain significant.

Because of the limited improvement of the model and the theoretical background I decide to use the model, not corrected for autocorrelation as given in equation 1.b for further investigation. Table 1. 4 shows the estimation of the entire Peruvian sample and reveals that

α

level, meaning that the model the model can be improved by an additional factor to explain the monthly returns on Peruvian share prices. Nonetheless, as

α

β

the explanatory value of roughly 80% is high for this model. For the estimation of the two subsets – results are given in Table 1. 5 – the constant is not statistically significant, indicating that a group factor will indeed complete the model. Interestingly enough the model also exposes highly different explanatory values for the two sub-portfolios. Model explains 83% percent of excess returns for group- affiliated companies but only 41% for stand-alone companies. Furthermore, the constant turns out to be not significant for the two sub-portfolios. This implies that the CAPM is a valid and good model to estimate the return for Peruvian groups, although the fact that the explanatory value for group companies is almost twice than the one for stand-alone companies shows that market forces have a much higher significance for groups.

I now turn to investigating the difference in returns between group and non-group companies, which implies taking the difference in the CAPM model as well. I do so, because the pre-analysis has shown that the problem is indeed with the difference in returns in terms of their explanation and their validity, which goes further than just the average.

Estimating a CAPM with a qualitative dummy variable, would assume that the difference is constant over the sample period. Taking the model in differences relates the differences in return between group and non-group companies to market performance.

6The results of these tests are not added due to space restriction, but will be provided upon request.

I am examining whether group firms perform differently than stand-alone firms, it is consequently more direct approach examining differences:

[ ] [ ]

) 3 (

* )

( ) (

* )

( ) (

t t Combined

combined t

t

t t andAlone

St Group andAlone

St Group t

t

RMRF andAlone

St R Group R

RMRF andAlone

St R Group R

ε β

α

ε β

β α

α

+ +

=

−

<=>

+

− +

−

=

−

where

α

= α

− α

β

β

− β

risk between group and no-group companies with relation to market risk. The new

β

the significance of returns for differences in risk. If the difference in returns on average is not statistically significant, as the pre-analysis has shown, it is very well possible that corrected for risk there does exist a difference in returns as well.

The importance in this estimation again lies in the statistic significance of the alpha

α

α

indeed significant there exists a constant difference between group and non-group members in returns even after adjustment for risk. If alpha is indeed negative, than group-affiliated companies underperform their stand-alone rivals. A positive sign on the other hand would implicate that group companies reveal higher share price returns. Our pre-analysis has shown that there is no statistically significant difference between group and stand-alone returns, but there exists a difference in variances. From these results I can already reject my hypothesis that group firms underperform, but correcting for risk and keeping in mind that there is a trade-off between risk and return, it is a possibility that the groups perform better, and that the group affiliation is indeed positively correlated with return. Table 1.6 shows the results for the OLS regression estimation of equation (3). Both coefficients are not statistically significant; implicating that there is neither a constant nor a market dependent difference between returns on group-affiliated and stand-alone companies. Therefore the right hand side of equation (3) is equal to zero, and consequently there is no difference in returns between group and non-group firms. Furthermore, the explanatory value of the regression is with 1.5% adjusted R-squared extremely low, indicating, that if there is a difference it is not based on the market factor. In spite of these results, group affiliation may yet be of substantial influence on returns.

During the sample period several crisis hit Latin America, affecting intra-Latino exports and the stability of currencies. One possible explication of the results may be that industries dominated by stand-alone firms were hit harder by these crises. And that group firms, because of their intra-group financial market were able to compensate for the turbulence in a more effective way than stand-alone firms. The group inherent quality on which to base an additional ‘group factor’ in the model is in my understanding the notion that groups are able to diversify into other industries and transfer resources.

The shocks during our sample period may have favored business groups in overall performance patterns. To verify the obtained results in the following step I therefore construct a group factor based on the returns of different industry portfolios, i.e. taking into account the ability of business groups to diversify.

The Group Factor Methodology Based on Industry Groups

The argument that group affiliation generally leads to lower risk is faulty as Claessens, Djankov and Klapper (2000) show. They find that whereas business groups in Chile mitigate risk, business group in East Asia are associated with higher risk. Van der Molen and Lensink (2005) attribute the discount in returns associated with group-affiliated firms, to their ability to diversify – to spread the risk over several industries, or countries. Khanna and Palepu (2000a) find that only after a certain threshold is reached, diversification is beneficial to stock market performance of Indian firms. In Khanna and Palepu (2000b) the findings for Chile are the same, with the addition that additional benefits can be retrieved from being of a group, while external market are still developing. This non-diversification benefits, however, decline the further developed the external market becomes.

The Fama and French three factor model associates stock market performance with market exposure (as the CAPM), company size and book to market value. Although they have a hard time linking these descriptive factors to underlining risks, the fact remains that these characteristics can be efficiently used to predict share performance. According to Fama and French (1995) the factors used by them can be associated with common factors in earning shocks. I argue that group affiliation can be sued in

a similar way. And unlike size and book-to-market ratio the connection with risk is clear-cut: group affiliation spreads risk over different industries hence lowering risk and group affiliation makes use of economies of scope and scale. These two effects however are working into opposite direction, one making the group affiliation more risky and arguably less profitable and the second one making group affiliation more profitable and following the risk/return trade-off theory more risky.

Based on the presented literature I come to believe that one of the defining qualities of business groups is the diversification of their business and hence the lower degree of volatility of group return. A diversified business group can compensate shocks to isolated industries by transferring resources to the companies of the group in crisis from their better performing industry sections. The financial backup allows the individual company to undertake necessary investments during crisis times and emerge from crisis stronger than their competitors without financial backup. In the CAPM framework this reduced risk then requires less return. In terms of stock returns thus group-affiliation reduces risk and shields group firms from shocks to single industries. Since group companies are able to work more efficiently because of economies of scale and scope, I will need to correct for market risk, to which the two company types react distinctly different. Even after the correction for this risk, I expect a difference in returns. This difference in returns I then can deduct is the effect of diversification.

For the construction of the group factor I therefore use an industry split. Stand-alone firms are by definition only engaged in one single industry and hence cannot compensate industry-related shocks by income transfer. Industry specific, negative shocks will be experienced to a much harder degree by stand-alone companies resulting in underperformance in comparison to business group firms. In case of industry-specific booms group companies are inclined to transfer resources to lower performing branches of the group, making them less profitable in boom times. Overall group companies are expected to be less volatile when it comes to industry specific shocks. Following the reasoning that less risk materializes into lower returns I expect to see a difference returns between the sub-portfolios of group and non-group firms.

Using an industry split therefore takes into account that industry-related shocks affect business groups and stand-alone firms differently. If there exists a difference between group and non-group returns

based on diversification issues, an industry split will detect it most accurately. The alternative is classifying business groups according to the degree of their diversification, as done in Van der Molen and Lensink (2005), but this cannot be done due to data restrictions.

The shocks that have hit Latin America and Peru over the past decade were mostly related to debt issues that concerned government, exerted inflationary pressure, and impinged on country risk measures. They affected all firms equally, minus business groups that had diversified geographically.

Note that the transfer argument, which makes business groups less volatile, is only valid when not all parts of the business group are affected by a shock in income, i.e. experience financial difficulties - this was the case in the past decade. Again, data to differentiate international and national groups is not available to sufficient degree to allow testing. However, this effect affects only groups and not the difference between group, and non-group firms, thereby not interfering with an industry split.

Diversification is the key to why business groups are theoretically are less volatile. An industry split allows us the construction of a group factor taking exactly this quality into account, when the data to the degree of group diversification is not readily available.

For splitting the sample based on industry I am using the industry categorization ECO with 20 distinct categories, which is ECOMATICA internal. It roughly equals the NAIC categorization, but separates manufacturing industries in greater detail. Each firm is therefore categorized by the fact whether it belongs to a group or not, and to the industry it is in, the portfolio for example“ Agro y Pesca Group”

hence contains all stock market listed companies that are in the Agricultural and Fishing Industries and are part of a business group. Table 1. 7 documents the number of group and stand-alone companies for each industry portfolio for the sample period January 1995 to April 2005. For further analysis I am forced to consider only those industries, for which both – group and stand-alone – portfolios contain at least 2 members. Considering industries for which either one of the sub-portfolios has 1 member creates a volatility that will distort the results. I therefore select only 8 of the 20 industries for further analysis, for which basic statistics are provided in Table1.8. Interestingly enough the basic statistics confirms for 5 of the selected portfolios that group-affiliated companies perform better than stand-alone companies, for 3 industries the reverse is the case. The volatility expressed as

standard deviation of monthly returns is, on the other hand, in all but one case lower for business groups. This data confirms the previous results, that group-affiliated companies most definitely are less volatile in their returns, but that they do not necessarily earn more or less.

Now I turn to constructing a group factor. The aim of a capital asset pricing model incorporating a group factor is explaining the difference in monthly returns between group-affiliated and stand-alone firms, i.e. the percentage in monthly returns that group-affiliated companies earn more than their counterparts. I therefore construct the group factor in exactly this way: taking the average difference between returns on group-affiliates and stand-alone companies, and then adding this factor to the CAPM estimating the returns on the portfolio return. If I simply take the overall average of the differences and then estimate the regression for the individual industry portfolios, their monthly return show up as estimated variable on the left hand side of the regression and on the right hand side, as part of the calculated parameter mimicking the group effect. This would render our estimation useless.

I therefore construct the group factor for each industry portfolio as the unweighted average of differences between group and stand-alone firms for the not estimated part of the industry portfolios.

In other words for the industry group “Agro y Pesca” the group factor is constructed as the unweighted average of differences between group and stand-alone in the rest of the industry groups, namely:

“Alimentos y Bebidas”, “Energía Eléctrica”, “Fondos”, “Minería”, “Otros”, “Siderur y Metalur”, “Textil”.

Table 1. 9 shows the calculation procedures for all industry portfolios. Table 1. 10 reports the basic statistics for the constructed group factors. The medium values do not exceed 0.5% and indeed tend to be low, but only in one of eight cases negative. Standard deviation is consistently around the 7%

mark. On average then the group factor tends to be small, if at all existent.

The group factor constructed in the described manner relates the performance of group firms in one particular industry to the performance in other industries. Groups perform consistently differently than stand-alone firms, and then the group factor will capture this difference in performance as a common characteristic to all group firms. As I argue the group factor is associated with common influences in earning shocks, hence I expect the group factor significant in estimating the difference in returns between group and stand-alone firms. If group association after adjustment for market risk is indeed

beneficial to share performance, then he group factor should be positive, if the contrary is true the group factor coefficient is to be negative. Since we define the difference as group returns minus stand-alone returns, positive effects of group affiliation will increase the difference, while negative effects will decrease it. From the first analysis it the group factor is to be expected not to be significant, and if not so then positively.

To provide a benchmark I estimate the original CAPM without the group factor. The results are given in the right hand tables of each page in Table 1. 10. Results for the group “Fondos (Funds)” are not given, because due to lack of observations the model cannot be estimated. None of the estimation exhibits heteroscedasticity and tests for autocorrelation lead to near singular matrices, but only in the case of the two portfolios “Otros (Others)” and “Textil (Textiles)” the Durbin Watson test statistic, implies autocorrelation⁷, the standard errors are adjusted to account for heteroscedasticity and autocorrelation. In none of the estimated models constant (C) or market factor (RMRF) out to be significant, furthermore the explicatory value of the models is with 0.8 to 4%, and in only “Siderur (Siderur)” sub-portfolio 17%, remarkably low in terms of R-squared, adjusted R-squared is lower and even negative. The group factor is in only in two cases significant and this only at a 5% level. Adding the group factor does not improve in all cases the R-squared value of the estimation, and if so only in marginally percentage rates, minus the “Siderur” portfolio. Furthermore, the variable coefficients are changing from positive to negative values, depending on the industry portfolio in question. This is valid for all variables – constant, market- and group factor. Indicating that all coefficients are approximately zero in value. Additionally the group-factor coefficient varies in size between –0.3 to +1.19 when added to the model, negating a clear pattern or even a trend in the estimation. In conclusion this means that, there is no difference in returns between Peruvian group and non-group companies – there exists no group inherent quality affecting Peruvian monthly stock market returns.

7Again, the test results for heteroscedasticity are not given due to special restriction, but will be provided upon request

SECTION IV

Summary and Conclusions

There is no evidence of a difference in returns between group-affiliated and stand-alone companies in Peru over the past decade. Although on the one hand side, on average group companies have outperformed non-group companies, there is no evidence of statistical significance, nor that group affiliation has any influence on monthly stock returns.

I have argued in my introduction that group companies are less risky, because of the support of their sister companies they have resources to fall back on in times of crisis and vice versa are backing up sister companies in times of crisis. Literature, in particular Van der Molen and Lensink (2005), has argued that the group inherent quality to diversify into different industries, allows compensating for industry specific shocks hence reducing risk. This risk reduction would then translate into lower returns for the investor. Hence I hypothesize, that there is a difference in returns between a group-affiliated than their supposedly riskier stand-alone counterpart. Because of this risk issue I expect that business groups in comparison perform worse. I estimate a model for the significance in differences between group and non-group companies. Working with the difference in monthly returns allows me to try to explain the difference in returns correcting for market risk. Still, even after correcting the risk there is no indication of a difference between returns of the two sub portfolios.

In spite of this outcome, I then test the hypothesis that there is an effect from diversification for group companies that makes them superior. I construct groups based on industry belonging and subdivide the groups into group and non-group portfolios. It is noteworthy that indeed non-group firms tend to be riskier than group companies – their standard deviation exceeds in all but one investigated cases the standard deviation of the stand-alone portfolios. Unlike predicted by our assumption on the trade-off risk versus return, returns are only in 5 out of 8 industry groups greater for group-affiliated companies.

I then construct a proxy for group-affiliation by taking the average differences between group and non- group for all but the industry to be estimated. Following, I estimate the model for differences in monthly returns between group and non-group members for each industry. First without the proxy for group affiliation, in a second step adding it. The results are very much against a group factor for the Peruvian stock market. Only in two cases the group factor turns out to be significant and neither one of the other factors is statistically significant for all industry groups. In addition the sign of the group factors coefficient changes between positive and negative values, depending on the industry. I can thence not conclude the existence of a general group discount or premium for Peruvian diversified business groups, but because of the range of the estimated figures that the group factors coefficient is zero. In conclusion group affiliation in Peru does not have a considerable effect on the stock performance of Peruvian firms.

I find no proof that group affiliation in Peru leads to higher or lower returns in terms of share prices.

Neither can I argue that due to lower risk or lower efficiency the returns are lower, nor is it true to claim that business groups have higher efficiency and thus higher returns, also a direct comparison seems to indicate this, contradicting the majority of studies for Asia⁸. Like Claessens, Fan and Lang (2002) I find no correlation between group membership and return. There simply is no difference in monthly returns between stand-alone and group-affiliated companies in Peru. The presented model in differences between group and non-group-affiliated companies shows that essentially stock market returns are equal for the two portfolios, although on average group companies do perform better than stand-alone companies in term of monthly stock market returns.

There are several potential interpretation for this result. The first of which being, that business groups simply do not provide additional benefits to affiliated companies. There are potential benefits from diversification, group formation, and creating an intra-group financial market, making efficient use of superior knowledge to exploit investments. But adverse effects of group affiliation include expropriation of minority shareholders, lack of control, and indeed inefficient allocation of resources.

My results indicate that advantages and disadvantages balance each other out. Diversification as such

8Campbell and Keys (2002), Ferris, Kim and Kitsabunnarat (2003), Joh (2003), Lins and Servaes (2002).

has not provided any discount or premium to Peruvian business groups; neither have economies of scale or scope materialized.

More likely, however, is that we are seeing the effects of the opening of the Peruvian economy.

Shimizu (2004) argues that family business and business groups need to attract foreign capital and management knowledge in order to survive in the now more competitive economic environment.

Khanna and Palepu (1999) have shown for India, that foreign investors tend to prefer transparent groups or stand-alone companies. If indeed foreign investors were to prefer Peruvian stand-alone companies then the additional investment would compensate for non-group membership. It might very well be that not group affiliation, but FDI and foreign business knowledge play the decisive role for Peruvian stock market returns. Since groups have greater market power and benefit from economies of scope this would explain the higher average returns.

To prove this interpretation accurate this paper though is too limited and follow up papers are needed on the topic, for efficiency considerations and to test the FDI interpretation, it would seem to be imperative to use, not stock market data, but actual EBIT (Earnings Before Interest and Taxes) figures.

The simple and result of this paper is that there is no proof of a group inherent quality that provided Peruvian business groups with higher stock market returns than stand-alone companies over the past decade. Neither is there evidence of a business group “discount” for Peru. Peruvian group affiliates tend to be less risky and earn more, but to deduct a statistical significant group factor to incorporate in an asset pricing model the difference in returns is simply too weak, thus my initial hypothesis is to be rejected.

References

Campbell II, T., & Keys, P. 2001. Corporate governance in South Korea: The chaebol experience.

Journal of Corporate Finance 8: 373-391.

Chang, S. J & Choi, U. 1988. Strategy, structure and performance of Korean business groups: A transactions cost approach. Journal of Industrial Economics 37 (December): 141-58.

Claessens, S., Djankov, S., Fan, J. P. H., & Lang, L. H. P. 1999. Corporate Diversification in East Asia: The Role of Ultimate Ownership and Group Affiliation. Policy research working paper no.

2089, World Bank.

Claessens, S., Djankov, S. & Klapper, L. 1999. The Role and Functioning of Business Groups in East Asia and Chile. ABANTE 3 (1): 91-107.

Claessens, S., Fan, J. & Lang, L. 2002. The benefits and costs of group affiliation: evidence from East Asia. Discussion paper no. 3364, Centre for Economic Policy Research.

Fama, E. and French, K. 1995. Size and book to market factors in earnings and returns. Journal of Finance 50: 131–155.

Ferris, S., Kim, K. & Kitsabunnarat, P. 2003. The costs (and benefits?) of diversified business groups:

The case of Korean chaebols. Journal of Banking and Finance 27: 251-273.

Gangopadhyay, S., Lensink, R. & Van der Molen, R. 2003. Business Groups, Financing Constraints, and Investment: The case of India. SOM Paper. University of Groningen.

George, R., Kabir, R. & Douma, S. 2004. Business groups and profit redistribution a boon or bane for firms? Discussion paper no. 2004-124. Tilburg University.

Harvey, C. & Roper, R. 1999. The Asian Bet. Mimeo. Duke University.

Hoshi, T. & Kashyap, T. 2001. Corporate Financing and Governance in Japan: The Road to the Future. Cambridge MA. MIT Press.

Hoshi, T., Kashyap A., & Scharfstein, D. 1990. The role of banks in reducing the costs of financial distress in Japan. Journal of Financial Economics 27 (September): 67-88.

Joh, S. 2003. Corporate governance and firm profitability: Evidence from Korea before the economic crisis. Journal of Financial Economics 68: 287-322.

Khanna, T. & Palepu, K. 1999. Emerging markets business groups, foreign investors, and corporate governance. Working paper no. 6955. National Bureau of Economic Research.

Khanna, T. & Palepu K. 2000a. Is group affiliation profitable in emerging markets? An analysis of diversified Indian business groups. Journal of Finance 55:867-891.

Khanna, T. & Palepu, K. 2000b. The future of business groups in emerging markets: Long run evidence from Chile. Academy of Management Journal 43:268 – 285.

Khanna, T. & Rivkin J. 2001. Estimating the performance effects of business groups in emerging markets. Strategic Management Journal 22:45-74.

Khanna, T. & Yafeh, Y. 2002. Business Groups and Risk Sharing around the World. Working paper no. 2002-8. Center for Economic Institutions.

Kim, H., Hoskisson, R. & Hong, J. 2004. The Evolution and Restructuring of Diversified Business Groups in Emerging Markets: The Lessons from Chaebols in Korea. Asia Pacific Journal of Management 21:25–48.

La Porta, R., Lopez de Silanes, F., Shleifer, A., & Vishny, R. 1998. Corporate Ownership around the World. Journal of Finance, forthcoming.

Lins, K. & Servaes, H. 2002. Is Corporate Diversification Beneficial in Emerging Markets? Financial Management, Summer 2002: 5-31

Perotti, E.& Gelfer, S. 1999. Red Barons or Robber Barons? Governance and Financing in Russian Financial-Industrial Groups. European Economic Review, forthcoming.

Scharfstein, D. & Stein, J. 2000. The dark side of internal capital markets: divisional rent-seeking and inefficient investment. Journal of Finance 55:2537-2564.

Shabunina, A. 2005. Consolidation of Ownership and the Role of Business Groups in the Russian Corporate Sector. Heriot Watt University, Edinburgh. Augustin Cournot Doctoral Days.

Sharpe, W. 1964 Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risk.

Journal of Finance: 425-442.

Shimizu, T. 2004. Family Business in Peru: Survival and Expansion under the Liberalization.

I.D.E Discussion paper nr 7. Institute of Developing Economies.

Soydemir, G. 2002. The impact of the movements in US three-month Treasury bill yields on the equity markets in Latin America. Applied Financial Economics, 12: 77-84.

Taylor, M. A. & Sarno, L. 1997.Capital flows to developing countries: long and short-term determinants. The World Bank Economic Review, 11(3): 451-470.

Torrero, M. 2002. Impacto de la privatización de sobre el desempeño de las empresas en el Perú. Documentos de Trabajo no. 41, Grupo de Análisis para el Desarrollo.

Ungson G. R., Steers, R.M. & Park, S. 1997. Korean Enterprise. Harvard Business School Press:

Boston, MA.

Van der Molen, R. & Lensink R. 2005. Group Affiliation and Firm Risk: Evidence from Indian Stock Returns. Working paper. University of Groningen.

DATA APPENDIX

Table 1.0 Results from Khanna and Rivkin

Peru Argentina Brazil Chile Mexico

Number of Observations 99 129 629 1780 344

Group 26% 51% 48% 36% 32%

Number of Firms 29 27 122 445 70

Group 24% 48% 47% 36% 27%

Number of Groups 5 9 39 45 12

Firms per Group

Average 1.4 1.4 1.5 3.5 1.6

Minimum 1 1 1 1 1

Maximum 3 4 3 19 5

Return on Assets (ROA)

Average 6.2 6.8 4.3 6.5 7.5

Standard deviation 12.5 6.9 8.6 12.6 6.5

Median 6 6.1 3.7 6.1 7.2

Minimum -31.2 -9.1 -57.7 -101.6 -47.1

Maximum 39.7 29.7 43.8 118.4 25.8

ROA among group firms

Average 3.1 4.9 5 8 8.1

Standard deviation 12.3 5.5 7.7 11.5 4.8

Median 5.1 5.2 4.4 7.3 8

ROA among non-group firms

Average 7.3 8.7 3.7 5.7 7.3

Standard deviation 12.4 7.6 9.4 13.1 7.2

Median 6 8.5 3.3 5 6.7

Industry representation

Agriculture / extractive 30% 22% 14% 16% 18%

Manufacturing 44% 43% 45% 31% 36%

Transport 3% 12% 15% 24% 11%

Retail / wholesale 0% 11% 6% 4% 18%

Services 22% 12% 21% 25% 17%

Time Period of Data 91-97 90-97 90-97 88-96 88-97