More on the role of middlemen in development

I had blogged here earlier on the misleading obsession in development with eliminating middlemen.

In this post, I thought of drawing attention to two BREAD working papers which highlight the point that intermediaries perform an important role and efforts to completely eliminate them will only backfire. 

The first paper by Dilip Mookherjee et al examined the impact of a Bangladesh government decision in 2011 to ban financing intermediaries, Delivery Order Traders (DOTs), from the wholesale and retail market for edible oils on concerns that they were responsible for rising consumer prices since 2008. Their findings,

The standard model... predicts that the pass-though of oil price shocks to the downstream consumer price should have risen after the reform, owing to the de-concentration induced by elimination of an intermediate layer. As crude import prices were falling after the reform was implemented, a higher pass-through rate should have resulted in faster decline in downstream price of edible oil... However, we find that the regulatory effort to reduce market power of financing intermediaries ended up raising consumer prices by restricting access to credit of downstream traders... 

We discuss supplementary evidence consistent with the credit rationing hypothesis, based on case-studies, as well as data on aggregate crude import volumes which contracted sharply (at a time when import prices were falling). The lower passthrough and higher prices following the DOT ban capture the effects of higher interest rates on bank loans in the post-reform period in addition to the tighter credit constraints lowering the volume of oil trade. The wholesale traders faced substantially higher marginal costs which lowered the passthrough rate, and the tightening of credit constraints effectively lowered the price elasticity of the derived demand functions faced by upstream refiners, resulting in higher refiner markups. These disruptions overwhelmed whatever reductions in market concentration resulted from the elimination of the DOTs from the market.

It can be argued that this is a partial equilibrium reaction and if the markets are allowed to operate, the formal financial channels will replace the intermediaries. However, I am inclined to reject this. Even in a general equilibrium, the replacement of intermediaries is unlikely to happen due to challenges of credit worthiness assessment, natural of the financial transactions (instalments and receivables), the inconvenience and difficulty of accessing formal finance, the limited base of formal finance and so on.

In another paper, Mookherjee with another set of co-authors used a clever experimental design to compare the effectiveness of two different types of schemes to select local commission agents for an agricultural credit program in West Bengal. The objective of the program was to select local agents so as to leverage their specialised information and connections with local residents, while also avoiding pitfalls of elite capture. The agents recommended village households for individual liability loans design to finance cultivation of the local cash crops and potatoes, and received a commission 75% of the interest payments made by each borrower. 

In the Trade-Agent Intermediated Lending (TRAIL) scheme, the intermediary was randomly chosen from among private traders within the community with a track record of lending to, and selling and buying from farmers in the village. In the Gram Panchayat Agent Intermediated Lending (GRAIL) scheme, the agent was randomly selected from a list provided by the elected village council.  

The authors examined the relative impacts on farmer cultivation decisions, output and incomes, and the reasons for variations across the two groups. They found,

Given their experience lending and trading farm inputs and output within the community, the TRAIL agents may be better informed about borrowers’ farm productivity and reliability. They could also expect to earn middleman profits from trading the farmers’ output, and so may have chosen to direct credit and business advice to their most productive farmer-clients. GRAIL agents, on the other hand, would have likely acted in line with the priorities of their political party, for example, selecting poor beneficiaries to further their party’s pro-poor agenda. They may also have been motivated differently vis-a-vis their beneficiaries’ projects: they were unlikely to benefit directly if their borrowers had large harvests, and instead could have faced the blame if the borrowers’ projects failed and they fell into economic distress...

Our estimates show that TRAIL loans were more likely to be taken up than GRAIL loans. Among the loans taken up, repayment rates were a similarly high 93 percent in both schemes. In both schemes, beneficiaries borrowed more, cultivated more area and had larger potato harvests. However, potato profits and farm incomes only increased for beneficiaries in the TRAIL scheme (by 40 percent and 21 percent, respectively). This is because TRAIL farmers’ expansion of cultivation was accompanied by a reduction in unit production costs, whereas GRAIL farmers continued to produce at previous (high) costs per acre...

We... find that the TRAIL agent selected more productive farmers than the GRAIL agent did. And yet, this selection difference accounts for less than 10 percent of the TRAIL–GRAIL difference in the average treatment effects on farm income... our analysis indicates that most of the difference can be attributed to the TRAIL scheme generating larger treatment effects on farm income, conditional on farmer productivity. Specifically, our findings suggest that a beneficiary in the TRAIL scheme produced more output at significantly lower per-unit cost and earned larger profit, than a GRAIL beneficiary of the same productivity. Since the loan product and hence borrower incentives were identical in the two schemes, this suggests that the key difference was the borrower’s relationship with their agent...

Since TRAIL agents are also middlemen in the potato trade, they are motivated to increase their profits by helping their clients produce and sell more. In our model, they advise farmers on ways to lower unit costs of production, inducing them to cultivate more potatoes, and earn larger profits per kilogram. This advice is most effective for the most able farmers. On the other hand, the GRAIL agent is motivated by the objectives of the political party controlling the village council. His primary goal is to minimize loan defaults, and accordingly he may intensively monitor poor, less able borrowers and induce them undertake actions and expenditures that ensure crop success, but also lower expected profits. As a result GRAIL borrowers would incur higher costs than TRAIL borrower... the TRAIL agents... increased their interactions with treated farmers, and to a greater extent with more productive farmers. GRAIL agents also interacted more with treated farmers, but to a greater extent with less productive farmers; this appears to have caused them to default less than their TRAIL counterparts.

This summary highlighting the importance of intermediaries is important,

Our results indicate the importance of looking beyond the architecture of local networks, and incorporating the effects of policy treatments on relationships within the network. Our work also sheds light on the specific role that intermediaries play... this paper finds that selection differences explain very little of the difference in the outcomes of the TRAIL and GRAIL schemes; accordingly we focus on the role of interactions between the agent and farmers. Interestingly, the large profits that the traders earn as middlemen appear to be the precise reason why their incentives are aligned with increasing farmers’ output... The more general lesson is that the role of the intermediary was not confined to selecting beneficiaries and achieving better targeting. The differences in the nature of network connections with the intermediary mattered significantly for the eventual success of the program. In particular the endogenous change in this relationship resulting from the intervention accounts for most of the differences in impacts on beneficiary incomes. Policy interventions that employ local intermediaries could benefit from considering this mechanism at the time of intervention design.

The point is that intermediaries, for all their problems, are an important player in informal markets and social systems. Any policy that seeks to eliminate them, as against acknowledge, formalise, and regulate them, is likely to be counter productive.