Wednesday, April 7, 2010

Where does the money go?

Was chatting recently with Sitabhra Sinha from the Institute of Mathematical Sciences in Chennai, India about some ideas to study the network structure and dynamics of money and information flow in rural Tamil Nadu and he passed along a bunch of papers to me. One of them was Patterns of dominant flows in the world trade web by Serrano, Boguñá and Vespignani. This mix of people from Europe and the U.S. took international trade data and reconstructed the trade network to look at imbalances between countries. They then asked where one dollar generated in any given country ends up being ‘absorbed’ or accumulated in the world. They called this the ‘dollar game’. For big consumer nations like the U.S. no surprise that a big chunk ends up in Japan and China. Where the money ends up accumulating is not completely informative about which country is better off though, since this doesn’t tell you anything about who can generate more currency to begin with. However, it did get me thinking about what it would look like to reconstruct this for rural and urban India and track where our microfinance loan money is ‘absorbed’. My guess is that the rural borrowers are net consumers and much of it ends up in Chennai or Mumbai or someplace like that rather than actually staying in the rural areas. If anyone has data to the contrary (or supporting my guess), I would love to hear about it.


  1. I'm an engineer and not a scientist, so I naturally gravitate to asking questions related to problem solving.

    How could this information be gathered, and what machinery could be put in place to track it over time?

    Are you proposing that one potential goal of increasing rural borrowing is to reverse the flow and turn the rural borrower into net producers?

    If we could track the flow with any kind of fidelity, we could target your lending in such a way as to test whether or not reversing that flow would result in a true standard of living benefit.

    It's all about the data.

  2. Matt, yes, great comment and I agree, its all about the data. I do think that if rural borrowers become net producers overall that more money will circulate among them at any given time and change the standard of living. To see what the money flow patterns look like though is pretty tricky. One way is to gather self reported data from a sample of people on where they have received/spent money. It will not always be clear though which geographic region the transaction should be attributed to. For example what if they bought something from someone in town who has only come to the town to set up a stall for the day in the market place but will go back to their village at the end of the day? Another way could be to track particular currency notes that we put into the system to see where they are at the end of some finite period of time. That would require people calling in to report that they have the note and that is a challenge. Still, its not impossible and I am thinking about how we could understand the dynamics of money flow at reasonable cost and even track how our other interventions impact the patterns of flow. Suggestions?

  3. Well... My initial reaction is to point out that I have no experience with the situation "on the ground". Are there, generally speaking, permanent elements of economic infrastructure in the villages (i.e. small stores, schools, etc)?

    If so, then I think we'd have a mechanism to at least measure purchases at some fixed locations. That would be a start. There'd be some work around normalizing what it means to buy a root vegetable in one village vs. buying a coil of rope in another village, but we could begin to see how the money flow changes over time with increased lending.

    If there is nothing like that, then I wonder about the self-reporting. It seems that this is possible. Again, I don't know the circumstances at the villages, but there could be campaigns set up for representative villages, educating local people about the initiative and what they can do to help (check in with our people periodically to provide data).

    You'd need a lot of research in either case to decide what counts as representative, to establish a baseline and of course you'll need control villages where no lending interventions take place. You could possible incent the control villages by promising lending opportunities at the end of the trial.

    Can this be accomplished?

  4. Having a control village is tough because there are a number of lenders and someone will lend everywhere. However, assuming we can come up with a good way to collect data, I think we could understand the dynamics for the flow of money by getting data at multiple slices in time and fairly intelligently predict the outcomes of lending within the system as a whole. The data could also be used to model those dynamics and understand what network changes could substantially alter the outcomes. Meantime, we will continue to ponder a good way to gather data...

  5. One indicator could be the uniformity in labour rates and its availability around the year.

  6. Subbu, Not sure I follow your comment entirely. Could you elaborate?

  7. My gut feeling is that you will have a hard time getting people to participate based on it being an innovative initiative to understand rural economics. If you can make it fun, however, make it a game, your chances increase greatly. I'm sure you are familiar with "Where's George", I realise these folks don't have the same kind of technology at home or the office to do something exactly like Where's George, but if you can share the results with them so that they can see where the money *they* spent ends up, you might get their interest.

    I don't know the situation there, but I read about high cell phone adoption in places that lack the traditional infrastructure. Perhaps they could text message the serial number from the currency, and a short description? Or geolocation?

  8. Josh, I think you are absolutely right for an urban educated audience but the rural folk here in Tamil Nadu (India) are just totally different. One big difference is that they are not shy about talking about their income, so presumably they wouldn't be shy to answer questions about where they spend their money. Another difference is that they have a lot of time and don't mind spending an hour or more talking to a surveyor. Also, about 30% of our borrower base is illiterate, which means that they cannot even sign their name. Cell phones are making rapid headway but still have some ways to go to make it all the way into the interiors, technology is virtually nonexistent in many places and for the most part, I hate to say this, but I doubt they would care at all about the results. I'm enjoying the discussion and will post further on this soon.

    On a different note, a game is a fun idea and I'm thinking about it....

  9. I'm about half way through _Linked_, and I can't help starting to think of the individuals or businesses as nodes, and the exchange of currency for goods/services as a link.