Bangladesh is known for its fast-growing clothing industry. How do its firms trade with international buyers? How do these relationships impact the country’s development strategy? As she details below, those questions are at the core of a paper that Graduate Institute Professor Julia Cajal Grossi wrote with Rocco Macchiavello, from LSE, and Guillermo Noguera, from Yale.
What gave you the idea for this paper?
This is a longstanding project we have been working on for more than five years. The initial idea for this broader research agenda was that many low-income countries leverage exports as a device for development, having one or a few competitive domestic industries joining international supply chains. This is particularly true for Bangladesh, a canonical example of export-led development with a leading industry – the garment sector – driving growth in the country. Understanding the potential of this development strategy involves examining how firms in this sector trade with international buyers, often very large multinational retailers like Zara and H&M.
There are many micro-aspects to this question, one of which is what the local firms are getting out of those interactions with international buyers. In commonly available data, what we observe about these interactions are the volumes and the prices at which parties are trading. This does not tell much about how they split the gains from these transactions. Our paper combines a novel methodological approach and very detailed data, to recover the two unobservables that shed light on this split: the costs and markups in the buyer-seller interaction. Once we recover these, we characterise how they vary across trade relationships and within a trade relationship over time. We show that a given exporter charges different markups to different buyers, even when selling the same product at the same point in time. We also show that part of that variation corresponds to certain buyers offering price premia as a device to induce compliance from their suppliers.
Can you tell us more about your “novel methodological approach”?
In our paper, we study markups and cost formation in the garment industry. The prices of these interactions contain not only the costs of production but also information about bargaining power and pricing rules. Without being able to separate these, we don’t know if firms are pricing just under marginal cost and making losses. Economic intuition suggests this cannot be a long-run equilibrium but can possibly be a short-run one.
Ours is the first paper to break down relationship-specific prices into markups and marginal costs. This is a big jump relative to where the trade and industrial organisation literature stands: to understand pricing in buyer-seller relationships the only option is to infer from the overall price or quantity what the markups and costs breakdown would be, which requires adopting a specific model of competition and a demand structure. This amounts to making many assumptions, observing a price and saying that if the assumptions hold, this price reflects a certain form of heterogeneity in markups or marginal costs. What we do instead is to propose a model that makes no assumptions on the demand or on the competition structure. Instead we impose only a relatively mild assumption on production and technology, well supported by inside knowledge of the industry, and out of that we can recover the decomposition.
What are the implications of this heterogeneity?
The literature suggests that when markups are heterogeneous, the way in which firms respond to trade liberalisations or transmit shocks to local markets is very different from the benchmark setup in which all firms charge the same markup factor. With heterogeneous markups, trade liberalisation may or may not affect consumer welfare or may affect it heterogeneously. It depends on what is adjusting, the markup, the marginal cost, or both. The source of heterogeneity can be firm productivity, quality or the markets in which firms are trading. In our setup, this is also dependent on who the firm is selling to. From a development point of view, this is particularly relevant because it highlights the importance of forming specific relationships with international buyers – this is not about selling an homogeneous good in an anonymous market. We show that there exists an additional layer of markup dispersion: a given exporter, even in a narrowly defined product category selling in a specific country at one point in time, charges different markups to different buyers. This additional layer of heterogeneity gives one more margin for prices to adjust to trade liberalisation and shocks. This is not a statement specific to developing countries but it is critical for them because it is out of these firm-to-firm relationships that these countries grow.
Once we document that this heterogeneity exists, we study why this is so. In the context of the garment industry, there are international buyers whom we call relational. Through frequent interactions, these buyers offer incentives in the form of promises of future trade or price premia to their Bangladeshi suppliers to circumvent the moral hazard issue that could arise in trade. The markups and marginal costs that we observe are compatible with buyers giving these sorts of incentives. At the other end of the spectrum there are buyers whom we call spot buyers, who choose the cheapest supplier in what can be described as a one-off competitive bidding mechanism. The export orders sourced in this manner tend to entail low marginal costs and command low markups.
Do you see any caveats about this study?
The big caveat that we have in mind and make explicit in the paper is about where the heterogeneity is coming from in this particular industry. Our analysis is what we call “reduced form”. When we recover the markups – in the first part of the paper – we do so in a structural way: we offer a particular parametrisation that we take to the data, and we recover the fundamentals of the problem we model. In the second part of the paper, when we characterise how relational buyers perform, we actually identify the sign of an equilibrium relationship between the buyers’ sourcing strategies and the markups and marginal costs we observe. We cannot claim to know the economic model behind that correlation. That equilibrium relationship can arise as the result of different economic models that shape the distribution of markups and costs via selection, treatment, or both mechanisms. In the paper we cannot distinguish between them – nor is it our aim to do so. We just show that in equilibrium this markup dispersion exists, and that it is systematically and conditionally correlated with these buyer characteristics and we advance what we believe is the most plausible interpretation for such correlation.
The second caveat concerns external validity. The first part of the paper is general; the methodology we offer to recover markups and marginal costs should be applicable so long as the characteristics of production function apply. The second part, however, is specific to the Bangladeshi garment sector. We are not making claims on what drives markup heterogeneity in all industries in all countries. We are addressing the likely source of markup heterogeneity in the context that we study.
What are the policy implications of your work?
I think there are two layers of implications. The more immediate ones are connected to recovering for the first time the breakdown of these prices, as this reflects the split of gains from trade at a very micro level. There is a lingering and very pessimistic idea that firms like Walmart, Zara or H&M are extracting the whole of the surplus in these relationships. We need to recognise that these markups are on average not equal to one – meaning that Bangladeshi manufacturers are selling at prices above their marginal costs. The paper is responding to this misconception. The key to this is the heterogeneity across buyers: relationships with some buyers are compatible with the more pessimistic story of surplus squeezing. We think this is a big area of potential policy intervention and relevant to notions of fair trade.
The second layer of implications touches on how we think of trade more broadly. The existing trade literature models firm interaction problems in a particular way. We offer evidence that suggests we should sophisticate these models in one particular direction, by internalising buyer heterogeneity into the pricing problem. Although that doesn’t immediately imply one particular policy recommendation, it impacts the choice of “lenses” that we use for predicting the effects of policy, for example trade liberalisations, in these countries.
What potential do you see for future research on this topic?
The one thing that we don’t address in this paper is the formation of these relationships, and this is what I am working on in other ongoing projects. I am using the same data and explicitly modelling how these relationships are formed. Most of my research agenda is around these buyer-seller relationships with one party in a developing country and the other in a developed country, and I believe this is a very fertile area of work, that is receiving contributions from different fields in economics, from industrial organisation to international trade and development economics.
* * *
Full citation of the paper:
Cajal Grossi, Julia, Rocco Macchiavello, and Guillermo Noguerax. “International Buyers’ Sourcing and Suppliers’ Markups in Bangladeshi Garments.” CEPR Discussion Paper 13482, Centre for Economic Policy Research, London, 2019. https://cepr.org/active/publications/discussion_papers/dp.php?dpno=13482#.
* * *
Interview by Nayantara Sarma, PhD candidate in Development Economics.
Front picture: Sk Hasan Ali / Shutterstock.com.