Last month, Kathleen Diga at IDRC pointed out a very interesting paper by Megumi Muto and T. Yamano at the Japan International Cooperation Agency Research Institute, called “The impact of mobile phone coverage expansion on market participation: panel data evidence from Uganda”. The paper joins earlier work by Jensen and Aker in examining the mobile’s role in transforming agricultural marketplaces via quantitative analysis. Abstract here:
Uganda has recently experienced a rapid increase in the areas covered by mobile phone networks. As the information flow increases due to the mobile phone coverage expansion, the cost of crop marketing is expected to decrease, particularly more so for perishable crops, such as banana, in remote areas because the increased information allows traders to collect perishable products more efficiently. We use panel data of 856 households in 94 communities, where the number of communities covered by mobile phone networks increased from 41 to 87 over a two-year period between the first and second surveys in 2003 and 2005, respectively. We find that the proportion of banana farmers who sold banana increased from 50 to 69 percent in the communities more than 20 miles away from district centers after the expansion of the mobile phone coverage. For maize, which is another staple but less perishable crop, we find that the increased mobile phone coverage did not affect market participation. These results suggest that mobile phone coverage expansion induces market participation of farmers who are located in remote areas and produce perishable crops.
Although, as I’ve mentioned elsewhere, I’m not a sufficiently skilled econometrician to assess the quality of the statistical models employed in the paper, I found both the paper’s underlying rationale and findings interesting for a number of reasons.
1. They “explicitly consider pathways in which better access to information increases income” (p50), distinguishing between changes to prices and market participation. In this case of bananas (plantains I think), they find no clear effects of mobile use on prices, but rather on market participation (the proportion of agricultural households in a given region who sell part of their crops in the market. They suggest that in the absence of competition between traders, households were more likely to participate in the markets (marketing the availability of ripe bananas for sale to traders), but that traders were able to keep prices paid to these homes relatively low.
2. They back up their model with field surveys, learning how, for example, “traders use mobile phones to set up a time and place to trade banana”, whereas in the absence of mobiles they just arrive unannounced and buy what’s available, waiting until their trucks are full (See Overa).
3. By accounting for distance, they tackle the complex interaction between mobiles, transport costs, and the availability of alternate face-to-face channels for information exchange — indeed they find clearer effects on market participation levels among households situated at least 12 miles away, and even more so at 20 miles away
4. by running the model for different products, and finding different results, Muto and Yamano illustrate the degree to which the mobile’s impact on agriculture, and on enterprise in general, remains quite context-specific. They find the participation model predictive in the case of perishable bananas, but not for less-perishable maize. Reconsidering the Jensen results in light of the Muto&Yamano study illustrates how the presence of multiple (competitive) markets and the pressure of a highly perishable product may have made the Keralan fish market particularly receptive to improvement via mediated communication. Thus the Uganda paper is a helpful cautionary note to those who might be tempted to make generalizable claims about the impact of mobiles in agriculture based on a handful of undeniably excellent studies.
And a couple of minor points for mobile phone researchers:
5. Their model actually does not offer much predictive power at the level of individual households; rather, the results are interpreted at the sub-regional level, comparing places with mobile coverage to those without, rather than homes with mobile coverage to those without. One could ‘unpack’ this a little more than the authors do in their article– perhaps once the trader is summoned by one household with a mobile, others nearby share the benefit of that call and can also sell ripe bananas. Or, perhaps there is more explicit sharing of handsets, such that even farmers without phones of their own can access one and use it to call traders to purchase and transport ripe bananas. We’re making strides on handset sharing but there is room for greater attention to it in our quantitative assessments of impact
5. Finally, despite the attention to mobile telephones and network coverage, this is still a paper about connectivity and the compression of distance, rather than mobility per se. While it is possible that the traders at the other end of the calls from the producers would be unreachable by landlines, the model can’t account for that. We have to look to Overa for that assessment.
In summary, Mutu and Yamano identify greater impacts of the mobile on the marketplace for perishable bananas than for not-so-perishable maize, and find larger effects for distant/remote farmers, for whom information exchange via face-to-face channels is less possible, and for whom the appeal of replacing travel with a phone call would be higher. Studies like these help the ICTD field develop a better understanding of which marketplaces are likely to be impacted by a step-change in the accessibility and affordability of telecommunications services, and which ones are not.