The Structure-Conduct-Performance (SCP) paradigm and its

application in South Africa – a review of the empirical

evidence and the implications for competition policy.

S.A. Du Plessis and E.S. Gilbert1

Abstract

Competition policy and its application in South Africa draw on the intellectual tradition of the Structure-
Conduct-Performance hypothesis. The application is neither simplistic nor without qualification, but the
degree of market concentration has remained a central guide to the competition policy in this country.
This hypothesis posits that market structure (measured by the degree of the concentration in, for
example, turnover) is a primary influence on the market power of firms and, hence, their ability to adopt
monopolistic strategies. Simply stated: high concentration leads to monopolistic behaviour which leads
to high mark-ups and (abnormally high) profits. In this paper we provide a critical review of the empirical
support for this paradigm in South Africa in order to test the appropriateness of its application in this
jurisdiction.

A substantial literature has studied the SCP hypothesis in its South African application, a review of which
is provided here. By the late nineties this often polemical literature reached stalemate. More recently a
series of empirical papers (Fedderke and Schaling, 2005, Fedderke et al, 2007 and Aghion, et al. 2006)
have provided further evidence consistent with the SCP’s application, and the authors have used the
evidence to argue for a more interventionist competition policy in South Africa. But here too
controversy has emerged with a recent paper by Du Plessis and Gilbert (2007) which used the
accounting data of a comparative sample of large industrial firms in the US and South Africa for the
period 1980 to 2006. In contrasts with Fedderke and Schaling (2005), Fedderke et al (2007) and Aghion
et al (2006) it casts doubt on the claims advanced by the aforementioned papers, notably the claims
about the level and persistence of profitability of South African firms.

This paper supports a more sceptical view of the recent enthusiasm for interventionist competition
policy in South Africa and suggests a case by case basis for evaluating the extent of competition in
markets.

1
Economics Department, University of Stellenbosch and Graduate School of Business, University of Cape Town
respectively.

1
1. Introduction

"Clearly, such industry concentration makes for unfair competition, inefficient

markets and inappropriate influence over policy ", Thabo Mbeki(2008)2.

While referring to global agricultural markets, Thabo Mbeki's comments reflect a very widely held view
that concentrated markets block the benefits of competition that are necessary for the maximisation of
social welfare. Competitive markets promise allocative efficiency in a static sense, and dynamic
efficiency through product innovation (Motta, 2004). Competition also promises to provide incentives
for efficient production and downward pressure on both costs and prices (Nickell, 1996: 724-725).
Concentrated markets, if not competitive, deny society these benefits. If concentration in an industry
necessarily thwarts competition then there are direct implications for competition policy3.

However it is impossible to observe or measure competition directly – so proxies have to be used

instead. One of the most commonly used proxies used is the level of concentration in an industry4. The
Structure – Conduct – Performance (SCP) paradigm5states that the structure of an industry (of the
degree of concentration) determines conduct (collusion and monopolistic pricing) which determines
performance (abnormal profitability/rate of return). The ability to collude is assumed to be inversely
related to the number of firms and their market shares in an industry, and thus is positively correlated
with concentration. Successful collusion leads to abnormal profits and a loss of social welfare and
potential economic growth.

If correct, the implications of this paradigm for competition policy are simple: concentration is bad as it
automatically leads to lower social welfare, and steps should be taken to avoid or even to reverse it.
Experience has been unkind to this crude version of the hypothesis. As will be explained below
alternative explanations for a positive relationship between levels of concentration and abnormally high

2
Mbeki, T. (2008). Address of the President of South Africa, Thabo Mbeki, to the 118th Inter-Parliamentary Union
th
Assembly: Cape Town, 13 April 2008. Accessed on the 28 of April, 2008 from:
http://www.dfa.gov.za/docs/speeches/2008/mbek0414.html
3
It has been argued that there are (or should be) other goals for the implementation of competition policy e.g.
development of small firms, promoting opportunities for employment and black economic empowerment (BEE)
(Theron, 2001). We shall not focus on these goals as we believe that they are not the fundamental drivers of
competition policy in South Africa. Our focus is on the basis for the assumption that high levels of concentration
necessarily represent low levels of competition.
4
More recently, the level of concentration has been substituted by a focus on the level of firms’ market shares
(Theron, 2001). However this is simply a firm level substitute for a concentration measure.
5
Bain (1951) is most commonly associated with the creation of the SCP paradigm (see Weiss, 1979:1104) but De
Jong and Shepherd (2007) attribute the development of this body of thought to Edward Mason. He was the leader
of the ‘Harvard School’ which included Joe Bain and Richard Caves amongst others (De Jong and Shepherd,
2007:209). According to De Jong and Shepherd (2007:224) Bain’s most valuable contribution to industrial
organization theory was his development of the concept of barriers to entry.

2
profits have been identified. The most influential of these are associated with the ‘Chicago School’6
which strongly emphasises the role of efficiency in explaining the seemingly abnormal returns. Some
empirical studies in the South African literature have taken this perspective, with notably different
implications for competition policy.

This paper reviews the attempts made to empirically tease out the effects of market power (through
market structure and collusion) on abnormal returns from those of efficiency in the South African
context. The many and polemical rounds of this debate on collusion vs. efficiency led to stalemate by
the end of the nineties (Fourie and Smith, 1998).

The debate has since re-opened with a slew of new studies, using new methodologies and data sets.
These results are central to our review. At the same time the focus of the South African literature has
shifted to measuring mark-ups and the performance of manufacturing firms and with striking results: In
particular, Aghion, Braun and Fedderke (2006) (hereafter ABF) conclude that South African (SA)
manufacturing firms operate in relatively uncompetitive product markets based on their empirical
observations that these firms have been able to generate abnormal profits when compared to their
counterparts in other markets. More specifically, ABF claim that SA manufacturing firms have
consistently been more profitable on a comparable basis over a period extending from the mid 1960s
through to 2006. Using various estimates of mark ups and profitability, they suggest that SA firms have
been between 50% and 100% more profitable than their international peers. Finally, ABF found a causal
connection between the low measures of product market competition they report and low productivity
growth in the same industries in South Africa.

ABF is one of the series of papers commissioned by the Presidency and National Treasury from an
international and local panel of economists (colloquially known as the “Harvard group”) to do research
on growth constraints in South African economy in support of ASGISA (government’s Accelerated and
Shared Growth Initiative). Based on the evidence of, inter alia, ABF, this panel included in its final policy
recommendations a move towards “…the adoption of a pro-active approach for competition policy,
rather than a complaints-driven approach in order to reduce barriers to entry” (National Treasury, 2008:
4).

Gilbert and Du Plessis (2007) examine the robustness of this recent literature using a smaller, but more
detailed data set based on the annual financial statements of large listed industrial companies based in
SA and the United States (US) for the period 1980 to 2006. To deal with the potential impact of
survivorship bias we construct a market capitalisation based index of the top 25 industrial firms for SA
for this period. We then compare the profitability of these firms using the same measures applied by
ABF and find that this profitability ‘premium’ does not exist for these firms for this period.

6
While a favourable interpretation of market efficiency was promoted by many notable figures at the University of
Chicago School, Demsetz (1974) is usually identified as the main protagonist of this school of thought in the
industrial organization context.

3
The conflicting conclusions of these studies do not, in themselves, resolve the debate. We believe
therefore that it is important for government and the competition policy authorities to appreciate the
continued unresolved nature of this debate and not rush to policy revision based on the evidence
presented by ABF, amongst others. While market structure can facilitate collusion, it is not safe to
assume that one necessarily implies the other. Every case should be evaluated on its own merits.

The paper is presented as follows: in Section 2 of the paper we review the literature relating to the
development of the SCP and efficiency paradigms and the early empirical tests. In Section 3 we review
the empirical literature in South Africa which led to Fourie and Smith’s (1998) conclusion of a stalemate.
In Section 4 we review the studies since then and in Section we 5 conclude the paper by reviewing the
implications of the empirical evidence for the conduct of competition policy in South Africa.

2. The SCP and Efficiency paradigms and initial empirical tests

Bain’s (1951) seminal paper was based on an analysis of the performance of US firms in 42 industries in
the latter half of the 1930’s. He found that the rates of return of firms in the relatively more
concentrated half of the industries significantly exceeded those of in the relatively un-concentrated
industries. This he interpreted as evidence for the SCP paradigm.

In what has become known as the Chicago school approach to industrial organisation theory, Demsetz
(1973 and 1974) posited an alternative explanation for the abnormal performance identified by Bain
(1951). His argument was that the abnormal profits observed reflected the higher level of efficiency of
firms, not the presence of collusive behaviour and pricing. Success (due to efficiency) is correlated with
size and thus to concentration. To resolve the ambiguity, so he argued, researchers needed to
distinguish between the impacts of efficiency on performance from those of market power.

In other words: are excess profits due to more efficient/successful firms or presence of collusion (and
thus monopolistic pricing)? Simply testing for a relationship between number of firms and level of
profitability is not sufficient. A more appropriate test, he argued, would be to test for a (positive)
relationship between firms of different sizes and their rates of return relative to the industry average. If
collusion is present, then smaller firms should earn similar (if not higher) rates of return than large firms.
However, if efficiency was driving the rates of returns then a positive correlation with the industry rate
of return should only emerge for large firms.

Other refinements to the debate have included the possibility of observed outcomes reflecting
disequilibria (Brozen, 1971) and so changes in the observed relationship over time could be expected. In
particular Brozen (1970, 1971) argued that Bain (1951)’s analysis was static and that Bain’s data (and
conclusions) may have reflected a disequilibrium situation. Consequently he argued for the need to
conduct longitudinal studies to look for patterns in concentration and performance over time. When he
conducted such an analysis he found that the level of abnormal performance declined over time and
interpreted as support for his disequilibrium argument.

Bothwell, Cooley and Hall (1984) introduced a case for risk adjustment to the rates of profitability of
firms in different industries to this debate. Finally, Mancke, (1974) raised the possibility that any
observed relationship could simply be due to randomness (i.e. spurious correlations).

4
In the next section we review the empirical evidence relating to South Africa leading up to Fourie and
Smith's (1998) conclusion of a stalemate.

3. The initial debate in South Africa (pre 2000)

3.1. The opening blows
Du Plessis (1978) was the first attempt at measuring the extent of concentration in manufacturing firms
in South Africa. Reekie (1984), however, represents the first direct test for the validity of the SCP
paradigm in South Africa. He concludes that the “SCP model is partially vindicated in the South African
setting” (Reekie, 1984:154). Leach (1991) is a direct response to this from the point of view of the
efficiency school and comes to the opposite conclusion to Reekie: "[w]e thus reject the monopoly
hypothesis in favour of the efficiency hypothesis" (Leach, 1992:153). Fourie and Smith (1993) explicitly
recognise that the answer is more likely to be something more complicated than either collusion or
efficiency. However they conclude that"[t]he data do not reveal a systematic positive effect of efficiency
on profitability" (Fourie and Smith, 1993:128).

3.1.1. Reekie (1984)

Reekie (1984) used data from the 1976 census of manufacturing where he examined the relationship
between profitability and concentration for 26 (out of 30) 3 digit SIC level industries. As a proxy for
concentration he used the number of firms that accounts for 50% of industry sales7. In line with other
international studies he used the (Sales – Cost of Sales – Wages)/Sales ratio as the profitability measure
for the industry and found a significant Spearman correlation coefficient (at the 1% level) reflecting a
positive relationship between concentration and profitability at the industry level.

He then attempted to apply the Demsetz (1973) methodology by testing for a relationship between firm
size and industry profitability. The only data available was on plant (not firm) size. He divided the
industries into three groups – high, medium and low levels of concentration and three groups by size –
number of employees per plant and calculated the average returns for the firms in each combination
along the two dimensions.

The results were mixed: though Reekie did not offer any statistical tests for differences, the results as
presented, did not show a clear relationship between size of plant and the average rate of return. Large
firms in concentrated industries did earn above average returns when compared to firms in less
concentrated industries as did the smaller firms in these industries. This is consistent with the SCP
hypothesis. Within these industries, however, the larger firms made greater profits in these industries
than the smaller firms – a result which is consistent with the efficiency hypothesis of Demsetz. This
result holds partially in the less concentrated industries as well. In summary, Reekie interpreted these
results as (limited) support for the SCP hypothesis.

3.1.2. Leach (1991)

7
His results hold for the more widely used 80% concentration measure as well.

5
Leach extended Reekie's analysis in two ways: he used a different measure of competition and used
data from multiple manufacturing surveys.

Leach used two measures of concentration – 50% occupancy count and the Rosenbluth Index (derived
from the Gini coefficient). The second one is a measure of relative (as opposed to an absolute)
concentration8. His extended data set included the results of the 1972, 1979, 1982 and 1985 Census of
Manufacturing. He also differs from Reekie in that instead of reporting rates of return for the various
size and concentration groupings, he reports correlation coefficients for these factors for the industries
in each grouping.

Leach found that " a moderate correlation between concentration and profitability for the largest size
class (500+ employees) and, in general, little or no correlation for the smaller size classes" (Leach,
1991:151). He concluded that this difference in profitability of firms of different sizes within industries
supports Demsetz’ efficiency arguments rather than the collusion arguments of the SCP.

3.1.3. Fourie and Smith (1993)

Fourie and Smith start by arguing against "looking for simple lines of causation"(Fourie and Smith,
1993:116). Consequently they combine simultaneous regression analysis using additonal control
variables with descriptive analyses to try and unravel the (possible multiple) lines of causation.

Their analysis uses a new measure of competition that adjusted the total amount produced in an
industry for the amount imported. This allows for the effects of import competition to be included in the
measure of concentration – an important factor in a small, open economy with differing industry tariff
protection levels. Further, the effects of changes in tariff protection could now be added as an
additional explanatory variable in their analysis.

Concentration was measured with the C5% ratio (the cumulative percentage market share of the five
largest producers) as well as the Rosenbluth measure used by Leach (1991). They also include the
change in the C5% measure from census to census to test for the impact of changing levels of
concentration. For profitability they use two measures – the 'A-profit' measure which reports profit as a
return on fixed assets; and the 'L-profit' measure which is the same as Reekie and Leach. Finally, two
explicit measures of efficiency entered their regression analysis: capital efficiency which is defined as the
change in gross output per unit of capital and labour efficiency which is similarly defined as the change
in gross output per unit of labour.

Their initial analysis suggested a positive relationship between profitability and concentration, but
importantly, only from a threshold level of concentration. This threshold effect is supported by their
findings that import protection only seems to play a constraining role on profitability up to a point as the
level of concentration increases. The change in concentration levels seems to explain some of the
variation in profitability. Efficiency measures, however, are not correlated with profitability. On the
contrary, they found a negative relationship between the level of concentration and the efficiency ratio.

8
His data suggest that the latter is a more useful measure than the former, but the results do not differ for the
two.

6
Fourie and Smith used a simultaneous equation approach to test for multi-directional causality between
concentration, profitability, protection and efficiency, but this innovation did not yield any additional
insights though. Finally they completed their analysis with an examination of patterns in the various
variables considered. They found that high concentration, while seems to be required for relatively high
levels of profitability, is not always associated with high profitability. The analysis left Fourie and Smith
with a complex conclusion: their econometrics did not support the SCP or efficiency approaches
unequivocally. Rather, it highlights a complex interplay of factors that makes a single factor explanation
of the relationship between market structure and profitability highly unlikely.

3.2. The Cement Cartel debate

The debate between Fourie and Smith (1994, 1995) and Leach (1994) regarding the cement cartel in
South Africa is an excellent example of the attempts to answer the collusion vs. efficiency debate using a
case study approach (as opposed to the general studies reported elsewhere in this paper). It highlights
the very real problems involved in identifying the relative importance of each of these competing
explanations despite the flexibility offered by a case study.

3.2.1. Fourie and Smith (1994)

Fourie and Smith present an economic evaluation of the cement industry and its cartel. They use the
SCP framework to analyse the industry, but assert their agnostic view of this approach: "[n]o simple line
of causation is presupposed between any of these dimensions" (Fourie and Smith, 1994:123).

Looking at structure, they find that the industry is highly concentrated. Production is dominated by
three producers together accounting for 95% of SA’s cement production and who had a formal cartel
agreement. These producers are also vertically integrated: they all owned their own limestone deposits
– the key input in cement production. They also owned a centralised cement distributor which moved
about 75% of total production. The distributor in turn owned the three largest cement retailers in the
country.

The conduct of the firms in this industry was controlled through a (government sanctioned) cartel that
has been in existence (in various forms) since 1992. It allowed the producers to divide the market on a
geographical basis and maintain prices. There is evidence, however, that the cartel was run to minimise
costs, especially from a distribution perspective9. Fourie and Smith (1994:132) believe that the longevity
of the cartel was due to firstly, the homogeneity of the product and fact that the demand for their
product is driven by factors outside of their control (which meant that price was the only basis for
competition and this is easily observable); secondly, the long term nature of the investments in the
production plants (cooperation would reduces the risk due to cyclical changes in demand over the life of
the plant); thirdly, the joint ownership structures and vertical integration meant that the cheaters had
lots to lose.
9
See the discussion on the minimization of distribution costs (Fourie and Smith, 1994:132). While the fact that the
firms colluded to minimize their joint transport costs does reflect their market power, it also can be interpreted as
an arrangement that leave social welfare enhanced. Given the low value/high bulk nature of cement, distribution
costs are very important. Acting to minimize these costs is good for social welfare. It is not obvious that society
would continue to be better off if firms were not allowed to collude in this particular way.

7
Fourie and Smith evaluate the performance of the industry be reference to the price increases (which
are not higher than Producer Price Index (PPI) changes in related industries); and profitability10.
Compared with other industries the profits of cement producers were found to be significantly higher:
"roughly twice that of the building materials industry and roughly four times that of the manufacturing
sector as a whole." Fourie and Smith (1994:135)11. However, when the accounting returns of the big
three producers are examined, there is evidence of significant variance in the returns of these firms. This
suggests that analysing returns at an industry level may not reflect the complexity of the reality at the
firm level. Caution is required, so it seems, when inferring firm level results from industry level data, an
issue to which we return with respect to the latest round of research on the market structure and
performance in South Africa.

In conclusion, Fourie and Smith recognise that their analysis was not sufficient to answer the question
whether South African society was better off under the cartel as compared to a competitive
environment. However, in terms of the collusion vs. efficiency debate, they concluded that the evidence
supports a monopoly abuse over an efficiency argument.

3.2.2. Leach (1994)

Leach responded with a sharp critique of Fourie and Smith's analysis discussed above. He argued that
they reached their conclusions incorrectly because they failed to consider the threat of new competition
(as there are no barriers to entry); they did not account for the risks facing the industry and its
implications for the firms' behaviour correctly; and their analysis of the performance of this industry was
incorrect.

According to Leach, Fourie and Smith make the fundamental error of ignoring the presence and effects
of free entry in their analysis of the industry. He argues strongly for the lack of formal barriers to entry –
particularly via imports. Consequently, he argues, a cartel's existence cannot be for the maintenance of
monopoly rents (otherwise there would be entry). Rather, he argues, the rationale for the cartel was
efficiency and risk management gains.

On Leach’s argument the nature of the cement industry (high fixed costs, variable demand) meant that a
cartel arrangement was an efficient response from society's point of view as it allowed the firms to
maintain a level of productive capacity and product availability over the demand cycle. Price stability

10
They use three different measures of profitability – two (A-profit and L-profit) of which have been used
previously. The third measure (S-profit) is ratio of accounting data: net profit to gross turnover. The results
discussed above refer to L-profit and S-profit. The A-profit results were discounted due to the capital intensity of
the cement industry compared to other manufacturing firms.
11
They go on to say, though, that these have been extremely volatile over this period. This is understandable
given the cyclical demand for concrete and the high fixed cost nature of the business. The implication of this
(something that Fourie and Smith do not consider) is that risk of these firms is probably higher than other
manufacturing firms and so a direct comparison of average rates of return is incorrect (as it implicitly assumes
equivalent risk levels).

8
also aided the planning process of the customers which can lead to higher levels of aggregate demand
than would be possible under a (less stable) non-cartel environment.

Finally, Leach criticised Fourie and Smith's focus on price increases, saying that monopolies lead to
higher prices, not higher price increases. He also criticizes their emphasis of the S-profit measure which,
he points out, is just as open to the problems of capital intensity as the A-profit measure which they
discount because of this fact. Finally, he criticises their use of the L-profit measure as it includes
depreciation in the numerator. This inflates the results for more capital intensive industries (such as
cement) as this cost is built into their gross margin. He proposes the use of an adapted version of the A-
profit measure12 which shows that the cement industry is no more profitable than other manufacturing
firms.

He concludes that there is an alternative, efficiency based argument for cartels that is supported in this
case by the lack of entry barriers to the cement market. He also argues that the conclusion of the
abnormal levels of profitability is incorrect as they are sensitive to the measure of profitability used.

3.2.3. Fourie and Smith (1994)

Fourie and Smith (1994) replied to leach on two fronts: While they parried with Leach on his technical
criticism, they shifted the debate in the direction of a methodological confrontation with Leach (1994).
They charged Leach with adopting what they called the “Chicago School methodology” and blamed this
method for the lack of progress in the debate. Further they blamed Leach for his insistence "… on
building his entire critique as if our paper adopts an extreme structuralist (SCP) position (e.g. p. 256). It is
of little use, and consequently not deserving of serious consideration" (Fourie and Smith, 1994: 93).
Fourie and Smith developed this methodological criticism further in their next set of papers.

3.3. An attempt at integration

In a series of papers Fourie and Smith (1995, 1998 and 1999) attempted to reconcile the contrasting
conclusions reached by Reekie, Leach and themselves in the context of the greater international debate.
These papers raise some very important insights and caveats regarding the debate, insights which
appear to have faded from the present literature in South Africa..

3.3.1. Fourie and Smith (1998)

The authors highlight three reasons why there continue to be such diametrically opposed positions in
this debate. The first reason relates to the inability to discriminate empirically between the alternative
explanations; the second, to the possibility of there being endogenous, simultaneous lines of causation;
and finally, the likelihood of their being different patterns of causation in different industries.

12
The amended version includes operating profit as the numerator (calculated as gross profit plus interest paid
less interest received) and fixed assets as the denominator. His justifications for these changes are brief and this
measure's use of fixed assets instead of total assets meant that it still did not address his own concerns with the
appropriate measure returns (Leach, 1994:268).

9
Fourie and Smith (1998) review five competing explanations for the presence of a positive link between
abnormal profits and concentration in an industry:

1. The SCP paradigm which assumes that concentration leads to collusion which leads to abnormal
pricing and ultimately profits (Bain, 1951);
2. The disequilibrium argument presented by Brozen (1971) that most observed relationship are
temporary and that longitudinal studies are required to address this question;
3. The Chicago tradition that the line of causation runs the other way: efficiency leads to abnormal
profits which in turn leads to concentration as more successful firms come to dominate the
market (Demsetz, 1973 and 1974);
4. A revised form of the Chicago school's approach emphasising that abnormal profits reflect risk
premiums. Accordingly, what seem to be abnormal profits in an industry is in fact compensation
for higher risk (Bothwell, Cooley and Hall, 1984). As concentrated industries tend to be more
capital intensive, which, when combined with cyclical demand (e.g. in the cement industry),
would make the returns to shareholder more variable. This would be reflected in a higher
expected rate of return – which is what is observed in these industries; and
5. Finally, they discuss the argument put forward by Mancke (1974) that the observed levels of
concentration and abnormal performance are simply coincidence i.e. they happen randomly and
the observed pattern is not the result of any systematic process or relationship.

Fourie and Smith’s central point is that the empirical tests used in the literature were not sufficiently
powerful to discriminate effectively between the range of possible explanations for the observed
relationship between levels of concentration and levels of profitability at an industry level. These
ambiguous causalities make the problem of identification incredibly difficult.

They go to highlight the fact that there may well be simultaneous and endogenous lines of causality
governing the relationship between levels of concentration and profitability. For example, the level of
concentration is very likely to be affected by the level of profitability of the firms in the industry. They
illustrate this by discussing the example of advertising – its level can affect the success (profitability) of a
firm (and thus the level of concentration in an industry), but the level of profitability will most probably
influence the amount of money available for spending on advertising. If correct, this would violate the
usual assumptions of Ordinary Least Squares (OLS) estimation (which assume a uni-directional line of
causality) invalid.

Finally, Fourie and Smith were concerned with very likely problem of non-homogeneity: namely the
problem of multiple industry development paths and structures affecting the observed outcomes. Both
the SCP and efficiency schools assume structuralist approaches that result in deterministic equilibriums.
However it is likely that both efficiency and market power arguments are complementary with an
ambiguous overall effect. Assuming a single, homogenous explanation over the entire range is exceeding
simplistic and is most likely to be wrong.

In conclusion Fourie and Smith make reference to the New Industrial Organisation (NIO) and New
Empirical Industrial Organisational (NEIO) approaches which had evolved out of the deadlock discussed

10
above. They argue that these approaches' focus on identifying stylized facts (e.g. regarding the specific
conditions which must be true for a particular result to hold) rather than testing for general truths and
Fourie and Smith were hopeful of these approached may be more fruitful approach.

3.3.2. Fourie and Smith (1999)

In this final paper Fourie and Smith examined the methodological reasons behind the observed
stalemate. They start by emphasising the inappropriateness of looking for "universal laws or truths" in
this context. The stalemate in this literature resulted from the (incorrect) application of an "either/or"
approach to competing collusion/efficiency explanations. It is very likely that both these explanations
are complementary and their effects endogenously determined. The key challenge for researchers,
therefore, is to unpack the process of interaction between these (and probably other) explanatory
factors. They then go to point out the role of ideology in this debate, with a particular focus on the
Chicago school.

In summary, Fourie and Smith argued for methodological pluralism and the recognition that, at best, "a
composite picture can be sketched around which carefully derived conclusions can be gleaned and a
fuller understanding of the nature of inter-variable linkages gained." (Fourie and Smith, 1998:91).

4. The new evidence

We first discuss three recent papers which bring sophisticated econometric techniques to bear on the
question of the link between concentration and performance in the context of the macro economy.
These results have also been used to find a role for competition policy in the process of economic
growth, employment and investment. It is, however, the conclusions they reach regarding concentration
and performance that are the most contentious from the point of view of this paper. ABF, in particular,
makes the rather startling claim that South African manufacturing firms have on average been able to
maintain mark-ups between 50% and 100% higher than their international peers, and this performance
could be attributed to the high (and growing) levels of concentration in these markets in South Africa.

The final paper we review directly challenges this claim and finds very different answers when the
annual financial statements of listed companies in South Africa and the United States of America are
compared.

4.1. Econometric evidence linking concentration to performance

4.1.1. Fedderke and Szalontai (2003)

Fedderke and Szalontai (2003) examined the levels of concentration in manufacturing industries in
South Africa over the period 1972 – 1996. They found high, and in most markets rising, levels of
concentration (as measured by the Gini coefficient and Rosenbluth index). Their next step was to test for
bi-variate relationships between the level of concentration and various measures of economic
performance: output, employment, total capital stock, investment, growth in total factor productivity,
relative real unit labour costs, the ratio of gross operating surplus to value added and labour
productivity. To that end they used the Pesaran, Shin and Smith (1996, 2001) F-statistics to identify the
relationship with an optimal lag structure.

11
For present purposes we are specifically interested in relationship between concentration and
performance as measured by the ratio of gross operating surplus to value added. Fedderke and
Szalontai’s (2003) results were mixed: they found long-run relationships in only nine of the 23 industries
under investigation when using the Gini coefficient measure of concentration. Of these nine, causality
went from concentration to performance in only five, with the opposite direction of influence in the
remaining four. Using the Rosenbluth index, they found evidence of long run relationships between
concentration and performance in twelve, with the causality running from concentration to
performance in eight and the opposite direction in the remaining four.

Fedderke and Szalontai (2003)also found a positive relationship between concentration and real unit
labour costs while concentration is negatively related to labour productivity (output/unit of labour). On
their interpretation "…the impact of increased concentration in South African manufacturing is not as
such to raise the efficiency of production. Instead it is more likely that concentration leads to the pursuit
of managerial objectives in firms in more concentrated industries, at the expense of efficiency and
profitability."(Fedderke and Szalontai, 2002:12).

Finally, Fedderke and Szalontai (2003) estimated a dynamic heterogeneous panel linking concentration,
growth, and employment creation. While they found links between concentration and growth and
concentration and employment, they did not explicitly consider the impact of concentration and
profitability.

4.1.2. Fedderke, Kularatne and Mariotti (2007)

In addition to the problems of method and interpretation that has held back progress in this literature
ABF (2006: 9-10) have also argued that industry level data has become less reliable since the last
manufacturing census (1996). They warned that “…the reliability of all results based on industry data are
likely to decline substantially after 1996”.

Despite these warnings on data quality Fedderke, Kularatne and Mariotti (2007) recently opened a new
line of research on the profitability of South African manufacturing firms following the methodological
innovations of, particularly, Roeger (1995). While growth accounting at the industry level could be used
to calculate mark-up of price to marginal-cost based on changes in factor quantities, it possible to derive
a dual of the Solow Residual (and hence an expression for the same mark-up) based on changes in factor
prices. Roeger (1995) exploited this dual of the Solow residual to find an expression for the mark-up that
avoided some of the endogeneity problems present in the calculation of mark-ups using the standard
Solow residual.

Fedderke et al. (2007) implemented and extended Roeger’s (1995) method with a panel data set for
South African manufacturing industries at the three-digit SIC level for the period 1970 to 1997 at an
annual frequency using the Pooled Mean Group Estimator (PMGE) as proposed by Pesaran, Shin and
Smith (1999) which allows for heterogeneity across industries in terms of intercepts, short run
coefficients and error variances, while imposing homogenous long-run coefficients across industries.

12
They reported a series of results suggesting that mark-ups were considerably higher in South African
industries than in the United States. Firstly, their estimate of the unadjusted long-run value for the
mark-up across industries was around 80% compared with an average result of 45% reported in the
literature on industries in the USA (Fedderke et al., 2007: 45-46). These results were then extended by
controlling for various factors associated with the size of the mark-up, following inter alia Oliveira
Martins and Scarpetta (1999). The second result of Fedderke et al. (2007: 47) is evidence that the mark-
up in South African industries can sensibly be split into a constant component and a statistically
significant counter-cyclical components.

Thirdly, Fedderke et al (2007: 49-51) controlled for the discipline that international trade might impose
on the size of mark-ups in domestic industry. They find evidence supporting the intuitively plausible
effect that greater exposure to import and export competition is associated with lower mark-ups.
However, the subtlety of their result is to distinguish within-industry and between-industry effects for
international competition and they found that the between-industry effect is the more powerful.

Fourthly, Fedderke et al (2007: 52) returned to the long-standing debate on market structure and mark-
ups and found a powerful association between industry concentration and mark-ups, i.e. greater
concentration is associated with higher mark-ups. Consequently, competition policy aimed at lowering
the degree of concentration in industry could, theoretically, lower pricing power of firms and in that way
improve the competitiveness of South African industry, or so Fedderke et al (2007: 54) argued. Further,
they explored the association between mark-ups and unit-labour cost as a measure of industry
competitiveness. Again the results are subtle: an increase in within-industry cost-competitiveness
lowers the mark-up, while an increase of cost competitiveness for an entire industry relative to the
average for the manufacturing sector enjoys a higher mark-up, i.e. the increased cost competitiveness is
not passed on to customers (Fedderke et al., 2007: 56-57).

The results from Fedderke et al (2007) mentioned up to this point suggests that South African industry
enjoys high mark-ups on an international comparison and that the mark-ups are related to
characteristics of the South African industrials landscape such as high concentration ratios. This
evidence from Fedderke et al (2007) confirms earlier evidence based on a multivariate co-integration
model for inflation in South Africa by Fedderke and Schaling (2005). They estimated an average mark-up
of 30% over unit labour cost for the South African economy (Fedderke and Schaling, 2005: 91), which
was three times as large as comparable estimates for the United States by Ghali(1999).

But Fedderke et al (2007: 58) also report one final result that casts some ambiguity on these findings:
adjusting their mark-up estimate to allow for intermediate goods they find a considerably reduced
mark-up for South African industry. Not only are estimates sharply lower (in a range of 6 to 9%) but it is
also lower than comparable estimates for the USA (13%). Fedderke et al (2007: 59-60) offer two
explanations for this result: firstly, they speculate that there might be errors-in-variables problems with
the data on intermediate inputs, and secondly, that there might be an omitted variable bias. They offer

13
evidence that concentration ratios are incorporated as the potentially omitted variable yields an
estimate of the local mark-up that is higher than the comparable figure for the USA13.

4.1.2. Aghion, Braun and Fedderke (2006)

ABF (2006) extended the work of Fedderke et al (2007) in three directions: firstly, they calculated mark-
ups and other measures of profitability for South African industry using three data sets (an industry level
panel data set from UNIDO, an industry level panel from TIPS and a firm level panel for listed
companies). These data are for South African and a large international cross section. Secondly, they
examined these measures of profitability over time and controlled for measures of product market
concentration and, finally, they studied the association between product market competition (as
measured) and their estimate of productivity growth.

We are particularly interested in the results ABF (2006) obtained from the firm level panel as this study
offers firm level calculations as a robustness check. ABF’s (2006) firm level panel contained data for
listed companies in 56 countries including South Africa for the period 1980 to 2004. They found that
listed South African firms earned profits on average around 50% higher than the international average
when profitability was measured with Net Income/Sales, Net Income/Assets and Net Income/ Equity
though the same South African firms had much lower Gross Margin, Market to Book Ratio and Profit-
Earnings Ratios. They also found no evidence of systematic variation over time for the these measures
of profitability, nor evidence that large firms enjoyed higher profits than smaller firms (Aghion et al.,
2006: 11: 12). These specific claims are investigated in Du Plessis and Gilbert (2007).

4.2. Have SA firms really performed that well?

Gilbert and Du Plessis (2007) responded to ABF by using financial data taken from listed companies'
annual financial statements. This allows for the correction for survivorship bias and the calculation of a
range of different measures of performance. However, it also introduces a size bias as their index
methodology includes data for only the largest 25 industrial firms listed on the exchange (by market
capitalisation).

It has long been recognised that empirical studies based on data of listed companies can be subject to
the effects of survivorship bias. This is an inherent problem that occurs when a sample is drawn only
from currently listed firms, for which a historical series is then constructed in a backward-looking
manner. This naïve selection will lead to biased historical estimates as it systematically ignores the
results or data from firms that used to be listed but no longer form part of the relevant index. By
excluding the firms that are no longer listed, data is collected only for the firms that have survived. In
the context of this case, such a biased sample of firms and their returns would lead to overly high
estimates of mark-ups and profitability, because it is usually the most profitable firms that survive.

Overcoming this problem involves a careful identification and inclusion in the sample frame of firms that
are no longer listed14 over the entire sample. Unfortunately ABF do not clearly specify which firms were

13
Fedderke et al’s (2007: 59) concern that “South African data on intermediate inputs is not fully reliable” is laudable,
but extends also to the other data used in studies of this kind, including especially industry concentration ratios.

14
included in their analysis. According to ABF (2006: 8) they used “(f)irm-level (Worldscope) evidence from
publicly listed companies. The firm-level evidence is based on Worldscope data for publicly-listed
companies in 56 different countries since the early 1980s. The dataset contains yearly balance sheet and
P&L items, and other basic firm characteristics.” Table Four in ABF suggests that the data was sorted by
3 digit (SIC) manufacturing industry. Table Seven then compares the ratios of the listed companies to
those of the industry level as computed from the UNIDO’s International Statistics data base. However
there is no further discussion in the text of the nature of the firm level data set used. This very brief
discussion does not answer the following very important questions:

1. How many companies were included for each country (most importantly, for South Africa)?

2. Has this data set been corrected for survivorship bias?

Given the long period (1980 – 2006) studied by ABF, the chances are very high that their sample is
drawn from an incomplete list of firms. However, the lack of discussion of this bias by ABF and the lack
of information as to the number or identity of the firms included in the analysis suggests that their
conclusions may be subject to this bias.

Gilbert and Du Plessis (2007) used a firm-level data set corrected for survivorship bias to test the
robustness of ABF’s claims. However, this clarity came at a price: the complexity of the data collection
process limited them to the comparison of only two markets – SA and the USA. They analysed relative
performance in terms of both margins: Operating Profit, Profit Before Interest and Tax (PBIT) and Net
Profit After Tax (NPAT); and returns: Return on Assets (ROA) and Return on Equity (ROE).

Gilbert and Du Plessis’ (2007) results directly challenged ABF's conclusions if performance is measured in
terms of operating, PBIT or NPAT margins. There was no evidence to conclude that, on average, SA
industrial firms were more profitable that their US counterparts in terms of these measures for the
period 1980 – 2006, nor was there greater persistence in the profitability of SA firms. However if
performance is measured in terms of ROA or ROE then their analysis provides some (though limited)
support for ABF's conclusions. SA industrial firms have consistently outperformed their US counterparts
over this period by significant relative margins. However these estimates are still only half as large as
those presented by ABF.

5. Conclusion and implications for competition policy

The debate on market structure and performance in the South African literature of the eighties and
nineties ended in stalemate. But it was followed by the adoption of a Competition Act in 1998 which in
its treatment of “dominant” forms and their potential anti-competitive practices (sections 7 and 8) are
dealt with from a perspective close to the SCP hypothesis. Accordingly, dominant firms are defined in
terms of their market shares on the following schedule:

14
Firms may no longer be listed for several reasons. They can go bankrupt or decide to de-list. Alternatively their
market identities change due to specific corporate actions e.g. mergers, acquisitions or restructurings. The effect is
the same from a survivorship bias perspective.

15
A firm is dominant if:

“(a) it has at least 45 per cent of that market;

(b) it has at least 35 per cent, but less than 45 per cent of that market, unless it can show that it
does not have market power;

(c) it has less than 35 per cent of that market, but has market power .” (Competition Act, 1998:
section 7, italics in the original)

While crossing these arbitrary boundaries do not themselves imply anti-competitive behaviour, Reekie
(1999: 269) is correct to argue that “…there is at least an implied presumption of “guilt” based on the
disputed Structure: Conduct: Performance paradigm of industrial economics” built into this legislation.

A new empirical literature has since reinforced the structuralist interpretation of the link between
concentration and performance in South African industry. Indeed, the empirical claims have been
stronger than ever, and have lately culminated in the policy recommendation of the Harvard Group that
Competition Policy itself be revised to create a sharper tool with the aim of pro-actively lowering
concentration in the service of higher economic growth. But we have also argued that the recent results
are no less contestable than their predecessors of the nineties, indeed a careful use of firm level data
failed to confirm the sensational claims about the profitability of South African firms.

Consequently we would suggest that a less interventionist competition policy would be appropriate for
South Africa. The evidence is too mixed to justify the application of the rules proposed by the Harvard
School. The evidence does not unequivocally support the conclusion that market power is necessarily a
sign of monopolistic practices. Each case needs to be treated on its merits.

References

Aghion, P., M. Braun and J. W. Fedderke (2006). Competition and Productivity Growth in South Africa.
Cambridge, Ma., CID Working Paper No. 132, August 2006.

Bain, J. S. (1951). "Relation of Profit to Industry Concentration: American Manufacturing 1936-1940."

Quarterly Journal of Economics, 65(August): 293-324.

Bothwell, J.L., Cooley, T.F. and T.E. Hall (1984). "A New View of the Market Structure-Conduct-
Performance Debate." Journal of Industrial Economics, 32 (June): 397-417.

Brozen, Y. (1971). "Bain: Concentration and Rates of Return Revisited." Journal of Law and Economics,
14 (October): 351-370.

16
De Jong, H. W. and W. G. Shepherd. (2007) Pioneers of Industrial Organisation, Edward Elgar Publishing
Ltd: Cheltenham, UK.

Demsetz, H. (1973) “Industrial structure, market rivalry and public policy”, Journal of Law and
Economics, Vol. 16: 1-9.

--- (1974) "Two systems of belief about monopoly" in Goldschmidt (ed.), Industrial Concentration: The
New Learning, Little, Brown and Company: Boston.

Du Plessis, P. G. (1978). "Concentration of Economic Power in the South African Manufacturing

Industry." South African Journal of Economics, 46(3): 257-271.

Du Plessis, S.A., and E.S. Gilbert (2007) “Mark-ups and competition – a comparison of the profitability of
listed South African and American industrial companies.” Paper presented at the bi-annual conference
of the Economic Society of South Africa, held in Johannesburg, 10-12 September, 2007

Fedderke, J. W., C. Kularatne and M. Mariotti (2007). "Mark-up Pricing in South African Industry."
Journal of African Economies, 16(1): 28-69.

Fedderke, J. W. and E. Schaling (2005). "Modelling inflation in South Africa: a multivariate cointegation
analysis." South African Journal of Economics, 73(1): 79-92.

Fedderke, J. W. and G. Szalontai (2003). “Industry concentration in South African manufacturing

industry: trends and consequences, 1972-96.” Centre for Social Science Research Working Paper 05/105
[Online]. Available: http://www.cssr.uct.ac.za.

Fourie, F. C. v. N. and A. Smith (1993). "Concentration, Tariff Protection and Industrial Performance in
South Africa." South African Journal of Economics, 62(3): 197–207.

--- (1999). "The Concentration-Profits Stalemate II: Ideological Rifts and Methodological Stress." South
African Journal of Economics, 67(1): 66-94.

--- (2001). "Revisiting the concentration-profits relationship in South Africa: moving the debate beyond
deadlock." Journal for Studies in Economics and Econometrics, 25(2): 25-60.

Ghali, K. H. (1999). "Wage Growth and the Inflation Process: A Multivariate Co-integration Framework."
Journal of Money, Credit and Banking, 31(3): 417-431.

Hamilton, L. (1996). Data analysis for social scientists. New York, Duxbury Press.

Leach, D. F. (1992). "Concentration and Profits in South Africa: Monopoly or Efficiency?" South African
Journal of Economics, 60(2): 143-155.

Mancke, R.B. (1974). "Causes of Inter Firm Profitability Differences: A New Interpretation of the
Evidence." Quarterly Journal of Economics, 88(2): 181-193.

17
National Treasury, 2008.Press Statement on the release of research papers on growth in South Africa by
the international panel. 8 May, 2008. Pretoria: National Treasury.

Nickell, S. (1996). "Competition and Corporate Performance." Journal of Political Economy, 104(4): 724-
746.

Oliveira Martins, J. and S. Scarpetta (1999). The Levels and Cyclical Behaviour of Mark-ups Across
Countries and Market Structures. Paris. OECD Economics Department, Working paper No. 213.

Pesaran, M. H., Y. Shin and R. P. Smith (1999). "Pooled Mean Group Estimation of Dynamic
Heterogeneous Panels." Journal of the American Statistical Association, 94: 621-634.

Reekie, W. D. (1984). "The Structure-Conduct-Performance Paradigm in a South African Setting." South

African Journal of Economics, 52(2): 146-155.

--- (1999).“The Competition Act, 1998: An Economic Perspective.” South African Journal of Economics,
67(2): 257-288.

Roeger, W. (1995). "Can Imperfect Competition Explain the Difference between Primal and Dual
Productivity Measures? Estimates for US Manufacturing." Journal of Political Economy, 103: 316-330.

Theron, N. (2001) “The economics of competition policy: merger analysis in South Africa”, South African
Journal of Economics, vol. 69 (4): 614-658.

Weiss, L. W. (1979) “The Structure Conduct Performance Paradigm and Antitrust.” University of
Pennsylvania Law Review, 127(4) April: 1104 – 1140.

18