Saturday, 10 April 2010

Business Groups in Portugal

   Last week I read an interesting paper about business groups. Business groups, or conglomerates, are firms that have divisions operating in very different sectors of economic activity. For example, General Electric in the U.S. operates in several sectors, such as machine manufacturing, finance, and also electric power generation, transmission, and distribution. In Portugal, the most prominent example is Sonae, with businesses in retail, telecommunications, and wood product manufacturing.
   The paper uses Canadian data to document some stylized facts about business group activity. The most salient one is that business group divisions are larger (e.g. supermarkets owned by Sonae tend to be large compared to local supermarkets, say) , and also more productive. This is interesting because many people complain that business groups are too large and inefficient, whereas the evidence seems to point in the opposite direction.
   The next question is of course to understand why business groups form, leading them to be composed of large and highly productive divisions. One possible explanation, explored here, is that business groups emerge to form an internal capital market. That is, in countries where financial markets do not function properly, business groups emerge in order to facilitate the funding of new businesses, and the cross-division allocation of capital.
   I would say the Portuguese experience appears to provide support for this theory of business group formation. Around 50 years ago, the Portuguese economy was dominated by large and far-reaching business conglomerates. Although business groups are still important today (e.g. Sonae), my guess is that they were far more important during the 1960s. A sign that financial markets are more efficient today than they were in the past.


  1. "The most salient one is that business group divisions are larger (e.g. supermarkets owned by Sonae tend to be large compared to local supermarkets, say) , and also more productive. This is interesting because many people complain that business groups are too large and inefficient, whereas the evidence seems to point in the opposite direction."

    Reading the paper, it seems that plants integrated in conglemerates are , indeed, more productive than plants not integrated in conglemerates; but they are less productive than plants of the same size not integrated in conglemerates:

    "I compare the productivity of conglomerated
    plants to the productivity of plants in single-segment firms. I find that conglomerated
    plants are on average more productive than plants in single-segment firms, but this relationship is reversed when controlling for plant size." (pg. 2)

    "In columns (4) to (6) I repeat the analysis
    adding a control for relative production size of the plant. Interestingly, the coefficient
    on conglomeration dummy becomes negative and statistically significant. Plants that
    are in conglomerates are on average one percent less productive than plants of the same
    size in the single-segment firms." (pg. 25)

  2. Miguel, you make a very good observation. You need a model to interpret the regression results. The post provides a link to such a model. Notice first that the unit of observation in the data is a plant (for the purpose of this reply, take this to mean either a standalone firm, or a division in a conglomerate). Now, suppose a world where plant-level TFP is exogenous. This means conglomeration does not impact the TFP of each division. What conglomeration can do is improve the capital allocation across divisions. In an economy with financial frictions, you're going to find lot's of small but highly productive firms. These firms are too small, because they are credit-constrained. You will also find very large highly productive firms in conglomerates, operating at efficient or near-efficient size. Imagine now, in such a world, running a regression of plant-level TFP on a conglomeration dummy, controlling for plant size. You may easily get a positive coefficient on the dummy. Precisely, this would be a sign of inefficient capital allocation, i.e. many highly productive firms operating at suboptimal size. Conglomerates allow such firms to grow, either by providing them with more external funding, or by reallocating capital away from less productive units. This would increase conglomerate-level TFP, which we don't observe. So, the interpretation you give in your comment is not warranted. Let me finish by saying that there might be other possible interpretations of the same data, including some where conglomeration may impact plant-level TFP. However, I am not aware of any paper fully articulating such view. Hope my very long reply addresses your comment!

  3. Business groups appear when they are successful in their core (initial) business and have available income to invest. Usually they expand into sectors where they can better use their extra resources and thus be more efficient. What normally happens is, whenever they find a strategic opportunity, they make it effective either through merging or acquiring any other company or business in the same sector and most often in a different one. The next step is always to find synergies, which in most of the cases means a cut on cost and an increase on productivity. For instance, support functions like IT or HR of one company of the group, can also take over those functions on the acquired company.

    Although in the short, medium term, these companies might build a strong position, they can deliver the results and they can create wealth, in the long run I have serious doubts about it (I would like to understand what’s the evolution of Sonae’s share of market in retail, without the Carrefour acquisition effect). Innovation is nowadays one of the main drivers for growth, if not the main one and these companies, in their seek for efficiency, they lose often flexibility. That flexibility that is needed to disrupt in the market place, that flexibility that is crucial to put at stake the current model. That’s why these groups innovate not organically, but by acquiring small but innovative businesses with a big potential (Coca-cola is acquiring Innocent in UK, Microsoft is acquiring everyday small Hi-Tech businesses all over the place).

    The car industry in US, mainly GM, is an ultimate example of a giant that failed on his vision for the long term and it is now trapped in its business model. Some other groups that have an option for a more flexible approach like Nike, for instance, are in a better shape to face the challenge, I think.

    As innovation, for me, is one of the biggest challenges of the Portuguese economy and mainly, the Portuguese industry, I think we should look careful for which is the right model to grow the economy. As I stated above, I have serious doubts that a business model like Sonae is the right option.

  4. Thanks for your very interesting observations Nuno. A theory of conglomeration based upon financing frictions would also predict that conglomerates should expand by acquiring small but very productive firms - although for reasons different from the ones you put forth. It would be interesting to try to tell the two stories apart from the data. You also point to costs of conglomeration (loss in flexibility), which is important and should be brought into the analysis.

    Let me also apologize since I mistyped in my reply to Miguel above: one can easily get a negative coefficient on the conglomeration dummy, not a positive one.

  5. This is a big question. I always assumed that large conglomerates exist due to the very nature of capitalism.

    In a market economy, firms tend to consolidate, especially in mature industries. There are certain stages of industry and product development which lead to consolidation.

    There are so many reasons for consolidation, we could spend years on this subject. Another reason is that in developped countries like the US or the UK, a number of bankers in financial institutions are constantly encouraging CEOs of large companies to acquire smaller competitors.

    There are also instances where larger firms acquire smaller competitors to gain critical competencies which they lack.

    Finally, there are emotional or power reasons involved such as CEOs constantly craving for more power which they satisfy by acquiring smaller competitors...

    In my view one of the main reasons why Portugal has been in recesion for years is the fact that SMEs account for 86% of employment. SMEs are facing increased competition from larger competitors from Northern Europe, the US and Asia and just cannot compete.