Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=

# Intermediaries and Supply Chain Distortions

Authors:   
Discussant: None  
Video: https://www.youtube.com/watch?v=PHYYOyrSItw&t=640s


## Talk (00:10:40 – 01:05:03)

[00:10:47] >> Hey. Uh, I'm gonna lower this for a second. Philip. Um, welcome back everyone. We're very excited to have a a tradition that Duncan Thomas started and and uh has been one of the the great innovations and successes of of Dev,  
[00:11:01] which is to have these invited lectures. So, we're very happy to have uh Dillip.  
[00:11:07] This is a little bit different format. It's going to be an hour session and the last 10 minutes will be for questions.  
[00:11:13] Uh, and uh Dillip has said clarifying questions. usually have no questions for the uh speak, but he said he's happy to take clarifying questions. Um so with no further ado, >> thank you. Thank you Sema and Ben for  
[00:11:28] the invitation. Uh great honor to address this group. Uh I'm going to talk about uh topic I've been working on for off and on for over two decades with uh my colleagues Pushkar Maitra, Shandep  
[00:11:42] Mitra and Sugata Visaria. And the more we work on it, the more it seems we the less we understand. Uh it's kind of a receding boundary as you will as you will see. Uh so it's very much ongoing what I'm going to talk about. Uh so lots  
[00:11:56] of eyes have not been dotted and tees have not been crossed but uh hopefully we'll get there sometime. All right. So the topic is uh uh marketing frictions faced by small farmers in in developing countries.  
[00:12:10] uh when I started working on land and and stuff like that and and I was you know after a while at least in this part of India uh I got the message that land was really not so much an issue anymore because of the land reforms that had been carried out or had not been carried  
[00:12:24] out and will never be carried out. So, so uh and then when I talked to people on the ground and small holder farmers and they they said are the real problems we face are uh more on the credit and the marketing side uh and and some  
[00:12:38] access to to good seeds and so on. So anyway, so I at that stage we started okay thinking about all those issues and um so the marketing frictions uh that I'm going to be focusing on here is  
[00:12:52] where uh farmers sell their produce at the farmgate. Okay. Uh for some reason they're excluded from selling in the wholesale markets neighboring wholesale markets. So they sell to trade intermediaries and they have uh limited access to to  
[00:13:06] credit and downstream price information. So the price at which the traders who buy from them will resell uh in the markets. So the context I'm studying is potato farmers in West Bengal and there's a paper that we published some  
[00:13:20] years ago and I'll connect that you know what we did then and with what I'm going to talk about today but other examples on cashews [clears throat] in Mozambique, coffee in Uganda, Coco in Sierra Leon all involving farmgate uh  
[00:13:33] sales to intermediaries. So the the broad questions first are there distortions uh resulting from these frictions what are the size of these distortions then next what are the underlying determinants and I'm going to  
[00:13:48] be focusing here on competition market power of [clears throat] intermediaries and then farmers access to credit and information and then the effects of policy interventions.  
[00:13:59] So I'm going to describe ongoing work that builds on on on our prior work and then we we actually conducted some additional surveys and field experiments. Uh so uh okay so what what am I talking  
[00:14:13] about? The phenomenon in in particular is farmers [snorts] getting excluded from wholesale markets and this is not just uh not just west Bengal it's sort of mostly eastern India. So there's a there's a a very good sort of  
[00:14:27] ethnographic study of these uh marketing issues in Bihar, Orisa and East and West Bengal and contrast to northern and western India. So they're selling mostly at farmgate. So and the question is why?  
[00:14:42] It's a question that you know it's an important question. It's kind of the institution there why it's there. It's it's hard to uh to really know. There probably should be more research on it.  
[00:14:52] But basically the way it works is that the role of the uh of these traders is that they're aggregators. So they negotiate with individual farmers. They inspect, store, transport, deliver in bulk to the next layer of traders. And there happens to be a vertical hierarchy  
[00:15:06] of of traders. And towards the end I'll just show you even within a sort of very narrow spatial uh region there are multiple layers of intermediaries.  
[00:15:18] Now why why uh is this the case? Um and why what is so special about eastern India? Okay, one probably the most important reason is that uh land is extremely fragmented in eastern India.  
[00:15:29] So there just literally you know the average land holding is very small and in any particular area there just sort of literally thousands and thousands of of small farmers. [clears throat] So there are high transaction costs. So when buyers come you we keep asking buyers in the wholesale market why don't  
[00:15:44] you go directly to the farmer and they say well there just too many of them. We want pick up stuff in bulk and we just don't know them. Uh it's just too we just don't have the time uh to possibly sort of transact with each and every one  
[00:15:58] of them. So we just want to delegate it to the local guys.  
[00:16:02] Second reason is that the farmers uh depend on the local traders for credit inputs and information partly because of lack of formal uh adequate access to to formal credit for instance and then access to the mundi is  
[00:16:16] restricted to uh trader associations. So there are these APMC acts in India uh but in it turns out in West Bengal it's not so much the APMC act it's more a tradition. So there are these traders association that sort of will guard  
[00:16:29] entry into the Mundy and so these small farmers will just not be allowed in and then as I'll explain it's you know there are no auctions or sort of real markets in the and the wholesale markets in in West Bengal. It's really sort of armslength negotiation between  
[00:16:44] representatives of these mundy buyers and sellers.  
[00:16:48] And then this is also an environment where there are no farmer co-ops.  
[00:16:52] There's no um no sort of uh what's it called? uh contracting with retail export conglomerates.  
[00:17:01] So this is an illustration of what it looks like on the ground. So you have these farmers uh the distances involved in the local area. So the farmers are selling to village traders where at the farm gate. So that distance is zero. The  
[00:17:15] one alternative they have to the the the the the trader who shows up at at the farm gate is to take they have to carry the produce to some local uh market. So there's some they're called hearts and  
[00:17:28] these hearts span a bunch of let's say 10 12 villages that's the average distance is about 5 kilometers and they can resell there to some other traders who will in in turn go and sell to the  
[00:17:41] the the the traders sitting in the mundy and the distance between these villages the average distance between these villages and the wholesale markets is about 8 km on average so this is a very narrow sort of geographic space and And  
[00:17:56] I'm going to be focused ex entirely on this on the distortions within this narrow space. So I'm not going to be looking at things like competition across Mundy's or interstate competition which other people have looked at in the context of India.  
[00:18:10] And so what happens in the wholesale market? You've got these big Monday traders who are buying from the village traders and they are in turn reselling to buyers who are coming to the wholesale market and they're these buyers are coming representing big  
[00:18:24] traders from retail markets in uh so we're looking at two districts one district it's going servicing mainly Kolkata and the other one is servicing mainly Bhaneshar which is the adjacent state of Orisa and some you also have  
[00:18:38] some traders representing other states from SAM and so on who are also coming to buy.  
[00:18:45] Okay, so in 2008 when we studied this for the first time and we uh put a lot of effort into measuring prices at different levels, this gives you an idea of the price caps, the local price caps and you can see the farmers here the average price this is averaged over the  
[00:18:59] entire year of 2008. [clears throat] The were selling at uh a price of about 2 rupees 2.14 uh per kg okay of the leading variety which is joti.  
[00:19:12] uh if they were selling in the local markets they were getting a little bit less which is 2.19.  
[00:19:18] When you see the price at which these potatoes are being transacted in the wholesale market by the wholesale trader to the outside buyers it's 485. So it's roughly double more than double what the farm farmers are getting at the gate.  
[00:19:32] And then when it's ending up in the city market it's in the retail level it's 638. So it's roughly a sort of ratio of 3:1 between the retail and the farmgate and about 2:1 between the Mundy and the and the farmgate.  
[00:19:47] And it turned out 2008 was an extremely low price year but normally the price is higher. I'll just show you some graphs.  
[00:19:54] So typically retail prices are about three to four times farmgate prices.  
[00:20:00] Now part of the gap and again I'm going back to the Mundy to Fonggate gap is accounted obviously by transaction and distribution costs that are incurred by the traders. So the screening inspection transport storage credit price risk etc.  
[00:20:13] The remainder would be the their rents or the profit margins which could owe to their market power and then the distortion that might occur between one adjacent set of layers would be compounded if there are multiple vertical layers. So that's there's a  
[00:20:27] well-known double marginalization problem, right?  
[00:20:31] It's difficult to estimate these middleman margins because of problems in measuring all the relevant trader costs uh the effort, the overheads, the credit and risk and so on. So we'll have to live with this problem of unobserved trader costs.  
[00:20:44] So now related literature uh uh mostly in the context of Africa. So on market power of retail traders it's see uh paper by burquist and dinnerstein which  
[00:20:56] is retail traders visa v consumers um that's maze in Kenya and then Atkin and Donaldson consumer products in Ethiopia and Nigeria and they're interested in sort of spatial  
[00:21:08] variations in in market power. Then there's a bunch of papers by Macham Vel and Moraria on relational contracts between traders and farmers and enforcement frictions in those  
[00:21:22] contracts. So flowers in Kenya, coffee in Rwanda, trade of farmer markets uh sort of farmgate sort of more of spot market  
[00:21:34] nature uh Coco in Sierra Leon uh which Kasaburi and Reed uh and in that context uh the markets seem pretty pretty competitive.  
[00:21:44] So what's distinctive about the setting we're looking at here is that this is a trader farmer setting with significant frictions and as I'll show you we measure the markups and the pass through they do indicate significant distortions but these frictions coexist with low  
[00:21:59] trader concentration and absence of any relational contracts. We did a lot of qualitative surveys with both farmers and traders and uh there's almost no uh instance of of relational contracting.  
[00:22:14] So what we're going to do uh is uh investigate the role of different sources of these frictions, market power, information credit asymmetries and the compounding of distortions and the effectiveness of policies of  
[00:22:28] farmer empowerment uh providing farmers with credit information access in lowering these frictions.  
[00:22:36] Okay, so that's by way of introduction. Now the details the institutional details there are two leading varieties uh accounting for about 90% of potato output. The leading variety is joti  
[00:22:51] which is 70%. And [clears throat] chandra muki which is a sort of slightly fancier variety which is about 20%. And apart from that there are about three or four varieties that account for remaining 10%. And we'll focus on uh to  
[00:23:05] ensure that you know I'm not we're not picking up any variety effects. I'm just going to focus on uh on joti and even within joti there's a good quality and not so good quality and we're going to focus on joti good quality and also  
[00:23:19] we're going to abstract from bonds there are some forward sales and so on I'm just going to focus on ready which are spot market sales now potato is in some ways you know kind of a convenient crop to study because it there's minimal product differentiation  
[00:23:33] apart from what I just told you and there's minimal posth harvest processing needed so no economies scale in processing and hold up and all those kinds of issues.  
[00:23:42] The crop is harvested once a year around March or April. More than 80% of the crop is sold by farmers at the time of the harvest. The rest is stored in cold stores uh or sold to the traders who  
[00:23:55] will also put most of it in cold stores and then it's sold later in the year when prices rise. And it just turns out that there's no shortage of access to storage in neighboring coal stores. no  
[00:24:08] sort of increasing costs of of that and there are also significant outofstate exports.  
[00:24:16] So now uh the sample and the data that we use so there are 72 randomly selected villages in two leading potato growing districts Huggli and Westm Nepur and the villages are selected at least 10 kilometers from each other to minimize  
[00:24:30] cross village spillovers. uh sample of 50 potato growers per village and uh with these farmers. So what we did in our earlier paper was was study them for that year of 2008. We're now  
[00:24:44] going to extend those production credit trade surveys with farmers for three years from 2011 to 13. So we survey them three times a year and then we in the year 2013 uh we interviewed traders and  
[00:24:58] we surveyed them. So we sampled two village traders per village based on who farmers were reporting the who they were selling to and then asked those traders in turn who are you selling to and so we went one layer up. Okay. And we could  
[00:25:13] it's very it's hard to sort of get traders to agree to participate in these surveys and disclose all these details.  
[00:25:18] We somehow managed to bribe them. So paid them enough to and uh so we we managed to sort of uh do it for most of 2013 not all of 2013 for 14 weeks but most most of the year was covered.  
[00:25:34] Uh then p market prices. So retail city prices for different varieties are available in the two main urban destinations. So this this information is in the public domain. You can open the newspapers and look at it. But the  
[00:25:46] Monday prices uh is not in the public domain. So [clears throat] the Monday traders uh apparently sell to Monday buyers following closed door negotiations. So what we were told is that you know the potato is kind of got  
[00:26:01] gotten ready and prepared for sale and so on and it's 7:00 in the evening in the Monday the representative of the buyers and the representative the sellers go into a room something happens in that room 10 minutes later they  
[00:26:14] decide on a price okay and in that price information is known only to the traders in that room and their assistants. So what we did in terms of sort of getting the information out is we bribed some of those assistants to release that  
[00:26:27] information on a daily basis at that night. Uh and I'll tell you more about the information experiment.  
[00:26:36] Now these market prices it turns out is high volatility and uncertainty especially across years. So here is this is data from the the uh main mundies in  
[00:26:48] West Bengal of joti potatoes from 2006 to about 2017.  
[00:26:55] So the first thing is you see huge fluctuations across years. So the year 2008 for instance the price was less than five uh most of the year and uh  
[00:27:07] most years the price rises. So uh because the crop is being harvested early in the year and then uh the rest of the year it's just sort of coming out of store provide incentives for people to store and so on the price rises and so that happened in let's say 2009 it  
[00:27:22] happened in 2012 and uh and 13 and so on most years that's what's supposed to happen that's the seasonality in the prices but some years it just doesn't happen. Yeah, >> just a clarifying question. I think many of us have seen this aark data that  
[00:27:36] you're using here. What stage of the process does that correspond to because you said that the Monday prices aren't publicly available. So So what exactly does a mark collect?  
[00:27:44] >> I don't know. >> Okay.  
[00:27:47] >> I I have lots of problems with most of the published u bundy price information.  
[00:27:51] Uh yeah. So we're going to rely on our you know the trader's assistance uh calling in uh the prices uh but sort of using uh  
[00:28:05] I think this is uh yeah but using the AMA data this is an analysis of variance or the sources of variation and you see that the the large the bulk of the variation is being driven by the year dummies okay followed by the seasonality  
[00:28:18] within the year the period and then the interaction between the two and there's relatively little there's some variation across Mondays in a given year but that's relatively uh small.  
[00:28:30] Okay. So now some other important institutional details. Uh so we when we interviewed the traders uh there are large entry setup costs. So the capital requirements about 50,000 uh uh for a  
[00:28:43] trader and that's like the the average uh loan size for a farmer is of the order of 8 to 9,000 a year. So it's about five times what a farmer usually has access to loans for. So that's part of the barrier for entry into traders.  
[00:28:57] So the big farmers sometimes are entering into into into the trading business. And they also have to build up reputation. Uh so experience, reputation for delivery, repayment, reliability and that takes time to build. So there's  
[00:29:10] limited scope for hit and run entry. Trader concentration is relatively low.  
[00:29:16] So the average village has about 150 farmers selling and there are on average about 14 uh traders in each village that are buying. These are the village traders and the median trader market share is 8%.  
[00:29:29] The transactions are spot market. So mostly it's described as you know a trader sitting in a shop and saying this is the price at which I'm buying today or they just run into each other as they go about their business within the  
[00:29:42] village exposed pair-wise meetings uh and in particular a complete absence of any excant risk sharing or relational contracts even if there may be some agreement that okay I you know you will buy from me right uh but they won't  
[00:29:55] agree on a price beforehand. Yeah. [clears throat] Do you know anything about collusion pigeon traders within villages? There are many but they >> that's that's going to be one of the questions but when getting to market power but just in terms of again the  
[00:30:09] descriptive stuff very likely there's some going on because when we ask the traders do you keep track of what prices other traders are selling in do you discuss prices and you know all of that and 50 more than 50% of them said yes we  
[00:30:22] do that uh and there's uh very almost again absence of any exclusive dealing sort of uh of of tying uh of any search or switching costs. So  
[00:30:36] usually you know most farmers sell to and switch between multiple traders. So you ask farmers you know how many people have you sold over the you know past 3 years for instance have you switched and they've switched and no problems in switching.  
[00:30:51] Now uh the other two uh details price information I already told you about the obliqueness and uh absence of uh uh it's not in the public domain the Monday prices. So what where do do the farmers  
[00:31:04] get information is well from the traders that they sell to. So that's the main source of information apart from neighbors and friends. [clears throat] And when we ask them to tell us what do you think is the is the Monday price uh they underestimate by about 40%. And in  
[00:31:19] fact, the predictions they give is closer to the to the the hearts in which they sell.  
[00:31:24] Uh on the other hand, the traders are much better informed. They're all on the mobile phones um and they they know who to call. So the other problem with the farmers when you ask them, why don't you get information about Monday prices?  
[00:31:35] They said we don't know who to ask. Uh but the traders have uh different networks and they're sort of on their phone and they're tracking prices every two or three days.  
[00:31:46] With regard to credit, farmers [clears throat] farmers are borrowing mainly from informal sources at annual interest rates varying between 21 and 29%. The average interest rate at informal interest rate is 25%. Traders  
[00:31:59] have access to more credit as you would expect uh more collateral. They're wealthier. The average interest rate is slightly lower. So there isn't that much of a gap in the cost of credit for traders and farmers. and trade some traders lend to farmers but trade credit  
[00:32:14] is uh is not not very significant. Uh only about 20% farmers receive trade credit and even amongst those that do receive trade credit uh there's no sort  
[00:32:27] of uh connection or interlocking of the trade uh with the with the credit. So uh when you asked if if you have borrowed from uh the the trader that you're selling your output to, does that make  
[00:32:41] any difference to the price that you paid compared to what if you hadn't borrowed from? And they said no, it makes no difference. So it's just two separate transactions. Yeah.  
[00:32:52] And on the on the figure you showed with the the diagram, maybe I misread it, but it seemed like the price gap wasn't that huge between the farmgate and the hoth and and it was much larger between the hoth and the  
[00:33:05] mundi. So like these statistics about these traders and is this are uh is this about that Hoth farmgate?  
[00:33:14] >> No, these are the village. So it turns out most of the sales are to the village trader. The the farmgate sale. The heart sale is not a farmgate sale. The farmers are carrying their potatoes themselves.  
[00:33:25] The farmgate sale, the trader is picking up the crop at the farmgate and incurring all the transport costs and so on. The hot sale typically happens when a farmer decides not to sell to the  
[00:33:38] village trader, but instead decides, I'm going to go somewhere else. Then he's got to cart his potatoes to the heart, and if he doesn't make a sale there, he's got to cart it back. So this is so that's kind of a a backup option and so that's going to define the outside  
[00:33:52] option and uh most years they're only selling about 5% or less in the hut. So mostly they're selling at the farm gate.  
[00:34:00] So all this is about the village trader showing up at the farm gate >> and so and then then I think implicit in that is most of those traders they're selling to are selling directly to the monthly not going through the >> I'll tell you about that later. So there  
[00:34:13] are the different vertical layers actually between the village trader and the Monday trader. So the first part of my talk I'm just going to ignore that distinction. I'm going to assume that that village trader is the guy who's  
[00:34:26] selling to the outside money buyers. Okay. But the later on I'll get into the sort of the uh the the different trader layers.  
[00:34:36] Okay. So this is for then this is based on u our prices uh the average city mundi farmgate price gaps for the year  
[00:34:47] 2008 and then 2011 to 13. So the uh the red is the city price the uh blue I hope it's showing up. Yeah the blue is the Monday price and you see that the city  
[00:35:02] and the Monday price are moving very closely together. uh the gap is relatively small uh but you then you've got the green which is the farmgate price and the big gap between uh between the Mundy and the uh and the and the  
[00:35:17] farmer price. Okay, some co- movement but obvious to to a much smaller extent between the Mundy price and the farm gate.  
[00:35:26] The other thing to observe is that the 2008 price was depressed as I said earlier and uh we therefore realized after working on the previous paper which was just based on the 2008 data that it was an exceptional year and so  
[00:35:40] we wanted to go back into the field and then continue these surveys for three additional years in the hope that the price would go up which it did fortunately.  
[00:35:52] Now in the year 2013 when we surveyed the traders uh we did a detailed survey of all the other sort of transaction costs that they were incurring. So the transport the storage the handling the loading costs and all that. So using  
[00:36:06] that we calculated the margins. So and then all the production surveys that we did with the farmers we also estimated their margins. So this is the breakdown and the contrast of the margins of the farmers and the local traders. And here the local traders I'm aggregating all  
[00:36:20] the local traders up to the Monday level. So basically the uh the farmer's production cost uh in the year 2013 is 295 and the net margin is 182. The local trader transaction cost all those  
[00:36:33] transport storage loading charges etc is 0.58 and the net margin is 787. So it's about almost four times more than four times the farmer's margin. Yeah. just descriptively cover these different  
[00:36:46] guides at different layers of the market are I shouldn't should I think of local traders as guides who are living large >> yes but local traders >> yeah yeah >> and the Monday traders are you know [clears throat and laughter] another  
[00:37:00] another fail yeah altogether yeah >> okay so uh the first question I'm going to ask is sort of what's the role of competition and uh the wide variation in the retail price shops provides a  
[00:37:15] potential source of exogenous variation uh that we can use to gauge the amount of competition. So the standard IO approach is to use pass through to the farmgate uh prices as a measure of competition and the the problem as I mentioned earlier are the unobserved uh  
[00:37:29] trader costs um the investments they made or the effort involved the time in screening and all of that that's difficult to measure. So you've got some unobserved costs but you've got some observed costs which is you know the large part of cost is the procurement  
[00:37:43] cost of of the potatoes itself and so we have data on that. So just uh using pass through uh you can back out. So these models tell you that what really matters for the traders is their total cost but  
[00:37:55] it's it's enough to be able to observe some component of the local costs provided that they're varying over time and that variation is coming from the intermporal variations in the downstream prices.  
[00:38:07] So based on the assumption of independence of the unobserved trader costs and the downstream demand price shocks. So I'm going to illustrate this method. So before I actually show you the pass through estimates, I'll just sketch a simple BR competition model uh  
[00:38:20] based on the Salup uh locational differentiation model.  
[00:38:25] Okay. So uh so let's say there are n u village traders. Um there's a common resale price at which they are going to be able to resell the potatoes and each of them is going to set prices at which they're going to  
[00:38:40] buy. There's a unit mass of farmers. Each has one unit to sell uh and zero value of own consumption. Now the farmer's outside option is to sell in the hut at a price which is co-moving  
[00:38:54] with the price at which the village traders are going to sell. But the hot traders are kind of have less established networks at the mundy. So they sell at a somewhat lower price. Uh so that's RM. So R is the um uh  
[00:39:08] basically is the comparative advantage of the uh of the village traders relative to the heart traders. On top of that the farmer has to incur a transport cost to take the product to the heart.  
[00:39:21] Now the heterogeneity uh within the within the village is that is locationational. So farmers are distributed uniformly on a unit circle and traders are located at equidistant points and the there's also hetrogenity in the  
[00:39:35] transport cost that farmers have to incur if they were to take it to the hut and just for simplicity I'm going to assume that this is also uniform and independent of the of the location within the village.  
[00:39:47] So now you solve this model. I'm going to focus on parameter values that ensure interior market shares for both between the village and the heart traders. So then there's a unique symmetric bertrand equilibrium farmgate price uh p which  
[00:40:00] depends on the resale price for the uh village traders which is capital M and N which is the number of uh village traders.  
[00:40:09] So now the comparative statics of this u so going to be two effects uh if you hold n the number of traders fixed the farmgate price p is increasing in n okay so u so that's that's kind of a very  
[00:40:23] standard effect u so it's more profitable for the traders to uh to sell the the potato so they're going to compete more vigorously with one another moreover the pass through PM uh is constant if there are monopolsy so if  
[00:40:37] they are colluding effectively there operating like a monopsini then the pass through is constant and otherwise if n is bigger than one there's competition then it's increasing in n so the the price is increasing and convex in it  
[00:40:51] okay the other effect is the entry effect so if you hold the money uh the resale price fixed and the number of traders goes up so then in there's increased competition so both price and the farmgate price is going to go up and  
[00:41:05] then you can extend the model to incorporate indogenous entry so The traders have a distribution over uh fixed costs. U so how many of them will enter will depend on the on the on the profits the variable profits they expect  
[00:41:18] to earn and so increasing now the the resale price will raise the per trader profit holding fixed and so that will invite additional entry which will increase competition.  
[00:41:30] There's a supplementary a third effect uh which is going to come back to this question of whether these guys are colluding or not or when are they colluding.  
[00:41:39] There could be tacit collusion. Okay. So the traders could for be forming a cartel which tries to coordinate and enforce a common collusive price. Now collusion would be harder to enforce in high resale price years because  
[00:41:52] deviation payoffs from the collusive price would be higher and if price is following an IID or a near IID process the strength of future punishments would vary less with the current price. So this would imply that enforcable  
[00:42:05] collusion is countery. So this is a theory that goes back to Rottenberg and Salonar in a in 1986 and I quote the IO literature. uh there's relatively little empirical evidence in favor of this. People have tried and I haven't seen any instances.  
[00:42:20] So that's one of the things we wanted to try here.  
[00:42:24] So uh one way of sort of phrasing this uh this hypothesis in this current context u holding the actual number of traders n fixed the effective number of traders represents the extent of  
[00:42:36] enforcable collusion. So ne is equal to one in low m years and bigger than one in high m years which would imply a proyical farm gate price and pass through and this just but this is qualitatively similar to the demand  
[00:42:51] effect. So it's empirically difficult to distinguish from the demand effect. So we're going to make an effort to do it and you'll see uh that we didn't really quite succeed.  
[00:43:01] Okay. So now let me just go through the intraear pass through regressions. So what we're going to do is we've got these three years uh 11 2011 12 and 13 when prices were rising and they were much higher than in 2008. So we've got these variation in retail prices across  
[00:43:15] years. We're going to look at the intraear passroughs and I'm going to report OS and 2SLS uh regressions. So one of the concerns with the OS would be that local supply shocks u would be a source of indogenity concern. So we're  
[00:43:30] going to use city prices elsewhere in India as an instrument for the uh city price in in the West Bengal urban markets.  
[00:43:39] Okay. So here are the results. Uh the so this is the first set of columns is the pass through from the city to the Mundy and as you can see the uh the pass  
[00:43:53] through is sort of close to one. It's centered around one. Okay, which is what you saw in the picture. So they're moving closely together.  
[00:44:00] But the pass through to the farm gate from the city is of the order of 0.2 even in the high price years. In the low price years in 2011 it's actually negative. It's something that I can't really explain. Uh so something odd is  
[00:44:14] happening in 2011 but in 2012 and 2013 the um the u the pass through is about 0.2.  
[00:44:26] So now if we compare this with pass through estimates uh for people who studied u sort of concentration on trader power in Africa. So in u in the Kenyan maze retail trade you know the it  
[00:44:39] was 2 and you know I think that paper sort of pretty persuasive that the traders were colluding in that context.  
[00:44:46] Atkin and Donaldson it varies between 3 to 7 in Ethiopia and Nigerian retail trade and for farmgate sales uh in Mozambique cashews and Uganda coffee is  
[00:44:58] between 04 to.5 and Sierra Leon was 0.9 so this looks like this is extremely u uh extremely high the the the frictions here  
[00:45:10] in terms of the in interyear variations it tends to it's not that much higher in the high price years 2012 and 2013. So the implied Monday to farmgate pass through from the estimates before this  
[00:45:24] 23 for those two years uh 17 in 2008 and negative in 2011.  
[00:45:32] So one question is well to what extent could this be accounted for? You see some procyclicality here of the pass through to what extent is it accounted for uh by entry.  
[00:45:42] So if you look at entry, the number of traders buying in each village, you see it's substantially higher in 2012 when the price was the highest. Uh but for the other two years, it was pretty much the same despite the fact that the  
[00:45:55] retail price in 2013 was much higher than in 2011. So there's some action you're getting from entry, but not that much.  
[00:46:04] And this is the corresponding uh dot index. And you see this drop in between but uh but not that much variation between the first and the third.  
[00:46:15] So now so it suggests that you know demand and rotten salon effects were also at play if you were trying to explain what was different 2013 from 2011 and testing the Rottenberg salon channel is difficult but I'll describe a test uh  
[00:46:29] in in in the next section. Okay. So let me now move on to uh price information.  
[00:46:40] >> [clears throat] >> So to what extent is lack of information about the price at which their potatoes are going to be resold by the trader?  
[00:46:50] To what extent does this constitute a bargaining disadvantage?  
[00:46:54] Uh if it does then providing Monday price information to farmers might reduce marketing distortion. And when we piloted uh our our the project, we kept asking farmers, would you really like to know the what's the prices in the Monday  
[00:47:08] and the resale prices and they said we'd love to have that information because we keep asking for this kind of information. The tra traders tell us something not sure if that's correct or not. It would really be useful for us to know that. Okay. So um so this is what  
[00:47:22] we studied in a prior paper for the year 2008 and we found negligible insignificant average treatment effects on the farmgate prices providing information and we explained this with a model where traders collude so effectively n is  
[00:47:36] equal to one. Uh now this raises a question well maybe 2008 was uh an exception exceptional year low demand and associated absence of competition among traders prevented farmers from gaining uh from the information. So some  
[00:47:50] of the feedback we got from that paper is that oh well if there had been competition uh then information provisioned to farmers would have been u more beneficial. So that was one of the motivations for us to repeat the experiment uh from 11 to 13 and  
[00:48:05] fortunately the prices did rise. So that gives us an opportunity to test this particular hypothesis.  
[00:48:12] So now stop for a minute and think well what do we expect about the interactions between competition and information?  
[00:48:19] Should we expect information provision to be more effective?  
[00:48:24] Well think about it a little bit more and say well not necessarily. So you extend the model that I described earlier to incorporate asymmetric information. The following result u tells you well it may not be useful. So  
[00:48:37] if n is bigger than or equal to true the same equilibrium outcome that occurs when farmers are perfectly informed constitutes a perfect basian equilibrium when farmers are uninformed coupled with off equilibrium path beliefs that seem  
[00:48:52] pretty natural and the particular of equilibrium path beliefs is that when you see dispersion in prices offered by different traders you believe the trader who's offering the higher price. you think the guys that are deviating at are the ones that trying to short change you  
[00:49:05] and uh so you you ignore that and you just look at the price offered by the highest uh offer as you're trying to glean what the resale price must be from you're trying to infer that from the observed price. So the intuition here is  
[00:49:19] that price competition itself reveals Monday price to the farmers and if that were the case information provision would have no effect at all. Okay, so this is what we can test.  
[00:49:30] Uh, so now there's a potential test of the rottenber salon effect. The hypothesis is that traders are more successful maintaining collusion in the low price years but not in the high price years. So now so in that special  
[00:49:44] case where traders collude perfectly operate like a monopsony then there would be an information effect but the information effect would be on the pass through uh because uh so the theory we  
[00:49:57] developed in that paper showed that basically it's an informed principle problem. The principal knows the price at which he's going to resell. The the farmer doesn't. And so it's optimal for the trader to not disclose the price to offer a flat price. So when the farmers  
[00:50:12] are informed, it's optimal for the traders to not disclose that information. But if the farmers are informed, then of course that they know what their outside options are. It's varying. So you would get this this effect on the pass through when they're colluding. But with the equilibrium selection theory that I just told you  
[00:50:27] with the case where they're competing, it would have no effect. So potentially we can test the Rotten Tom solon uh hypothesis by checking the the impact on the pass through. So the pass through effect has to be positive in the low price years and vanish in the high price  
[00:50:41] years. Okay. So uh I have 10 minutes left. So let me tell you the results of the uh information RCT. So 36 villages received publicly posted information. So we had  
[00:50:54] our the the Mundy trade assistants sort of calling in the numbers and then we would call it in to a particular some provenence citizen in the village who would post it in the high school or you know some city you know the local government uh local government office  
[00:51:09] outside that there would be a chart which would give these are the prices uh in the in the in the Mundy for uh for Joti and for other varieties over the last two weeks and that chart was being updated. The remaining 36 villages  
[00:51:22] served as control and there's a minimum 10 km separation uh between the treated and the control villages.  
[00:51:31] And in the 2008 version of this experiment, we had another information treatment which is the private information treatment where instead of the information being posted publicly, it was sent to uh four farmers in the  
[00:51:44] village through mobile phones. So that was a private information treatment as against a public village wise information treatment but the interest of simplicity we couldn't we couldn't have too many treatments. So in the 2011  
[00:51:56] to3 we just kept the public information at that time expecting that the public information uh would have probably more impact because it was going to be accessible to the entire village.  
[00:52:09] So here are the u RCT results of the pro effects of providing public information in 2008 the old 2008 data along with the  
[00:52:21] the uh 2011 to3 data and basically we see nothing null effects both the average treatment effects as well as the heterogeneous treatment effects.  
[00:52:35] So even the pass through uh estimate so we we got a significant pass through effect from the private information treatment in the previous 2008 but with the public information we not see any effect even in 2008.  
[00:52:50] So we failed to find evidence of the rottenberg salon effect. Of course there's a qualification maybe the test is is not powerful enough. Uh so maybe we should have chosen the private information treatment where we did see an effect on pass through in 2008. Uh  
[00:53:04] but that's you know with the benefit of hindsight if we were to do it again maybe we'll try it with the private information treatment I don't see us doing it again but okay all right so information is providing information to farmers is  
[00:53:17] doing zilch okay so he said okay what else can we do well credit maybe credit will help so what are the possible channels that providing credit to farmers might help well it would lower  
[00:53:29] the costs of farmers delaying sales from the harvest to the posth harvest period.  
[00:53:34] So they sell 80% of the crop at the harvest time. Why are they doing that and why are they not holding on? And so there's a nice paper by Burke Burkeist and Miguel uh in Kenya where actually farmers that had better access to credit  
[00:53:48] were delaying and then getting much higher prices overall for the year. The other channel would be if trade credit is important, it would reduce the dependence of farmers on traders for trade credit through interlinkage, lock  
[00:54:01] in, loan repayment charges and so on. Of course, the survey data indicated low prevalence of trade credit. Um uh anyway, so we do we so we're expecting the delayed sales uh channel  
[00:54:14] to be uh the more operative one, but who knows? 20% of farmers are still getting trade credit. That may still account for something.  
[00:54:22] Now uh so we had a bunch of papers on these credit treatments which were orthogonal to the information treatments for these uh 2011 to 13. So there were three micro credit treatments and these loans were designed to facilitate uh  
[00:54:36] financing of uh working capital for potato. Uh uh so um loans were given uh at the time of harvest and we um the repayment period was 4 months later longer than most micro credit. uh so  
[00:54:51] that we didn't have to you know the farmers could actually wait until they had sold the crop to repay the loan and then they could actually renew the loan if they wanted to hold on um they they they didn't want to sell right away. So they wanted to delay sales so they could  
[00:55:05] uh apply for another loan and these another four-month loan and so forth. So they kept we kept the lines open uh for access to these fourmonth loans throughout the year. So if they wanted  
[00:55:17] to uh delay u uh they could and then the selection mechanism was different. There were three different treatments. One was u these were individual liability loans and the selection of who would be offered these loans was delegated to a  
[00:55:32] village trader. Grail was the same as trail except that it was delegated to somebody appointed by the local government and GBL was the traditional group loan. So self- forming  
[00:55:43] five person um uh farmer groups. Okay. And in couple of papers that we wrote, we evaluated the impacts of these different credit treatments on farmer output incomes. But here we're going to focus only on the impact on farmgate  
[00:55:57] prices. Okay. On the marketing friction and each treatment was applied to 24 villages um which was independent from the assignment to information.  
[00:56:09] Uh siba I have five minutes for mic part of it. Yeah. Okay. Fine.  
[00:56:15] >> And the sample includes 10 treated farmers, 10 randomly chosen farmers from those who are recommended uh but not treated. So we can select uh we can separate the selection from the treatment effects and then 30 randomly  
[00:56:29] chosen from those who are not offered. And if credit frictions um raise marketing frictions then we expect credit treatment effects on the farmgate  
[00:56:42] price to be positive. And to the extent that you know this interlinkage dependence on traders for trade credit is important we expect higher benefits  
[00:56:54] if the agent in question who's intermediating the credit is not the village trader if it's the local government representative. So we expect the grail effect to be larger.  
[00:57:05] What do we see for these three years?  
[00:57:10] Nothing. So if anything, if you average across the three years, there's a negative effect in grail. Again, that's something we cannot explain. We don't know what's going on there. But for trail and GBL,  
[00:57:22] there's nothing. Um, all right. So, um, that's that. uh in the so the last few minutes I'm going to  
[00:57:36] talk a little bit about the trader hierarchy.  
[00:57:42] So it turns out there are three layers between the farmer and the and the money buyer. There's a bottom layer the village traders that are buying at farmgate. Then there's a middle layer who's buying from the bottom layer and  
[00:57:56] they're selling to mundy traders. And then the top the creamy layer are the Monday traders. Okay, those are the fat cats here.  
[00:58:05] And based on the trader survey uh so we had we interviewed people at both the middle and the bottom but we did not dare even approach the fat cats. So we have the bottom two layers but not the  
[00:58:19] top layer. So between the bottom and the and the top uh the mi sorry the bottom and the middle uh you can see the bottom is selling.  
[00:58:29] So some of them are skipping the middle layer. So some in the bottom layer 42% are selling directly to the Monday trader. 45% of the sales are to the middle layer. But if you look at the middle layer they're selling mostly to the to the top layer. Then we look at  
[00:58:43] the prices that they receive for their sales and there isn't that much of a difference between the prices that the middle and the and the bottom are receiving for their sales at the at the Monday. So just for the sake of simplicity just going to collapse the  
[00:58:57] middle and the bottom into one layer. Okay. I'm going to contrast the top layer with this sort of consolidated middle uh and bottom layer and just sort of get the gaps and then the passroughs.  
[00:59:12] Turns out most of the market power is concentrated at the upper layer.  
[00:59:17] So the margins uh if you remember for 2013 the margin for uh farmers was 182 and then the rest was about 7 something and the breakdown of that 7 something 171 for the uh for the local traders and  
[00:59:32] 616 for the Monday traders. So you know the transport storage everything is being incurred by the by by the by the local traders. So these fat cats just sit there. The potatoes just arrive uh you know sometime in the afternoon they  
[00:59:46] just resell it that evening. Okay. Okay. And they collect about three to four times uh what what everybody below is is earning. Then we estimated passroughs based on the weekly data and the pass  
[01:00:00] through from the city price to the local trader is.3.  
[01:00:05] So the implied pass through across the layers in 2013 67 from city to Mundy 44 from Mundy to local 0.55 from local to farmers. So what this tells us is that most of the action is actually between  
[01:00:20] the top layer of the Mundy and the and the middle and the bottom layers. So it's all within the traders. Okay. I mean there is some friction between the village trader and the farmer. Okay. But you know I mean the pass through is  
[01:00:32] 0.55. It's substantially higher. uh than the 0 23 we estimated. So most of that of the low uh pass through was accounted for by the low pass through from the top to the middle.  
[01:00:46] Okay. So let me end here. What have we learned? Um large frictions u no detectable effects of information of credit provision. So in terms of  
[01:00:57] policy sorry no nothing here. uh what it tells you the elephant in the room is competition. How do you get more effective competition amongst the traders?  
[01:01:09] Uh and really how do you deal with the fat cats? Uh so uh so we go back to sort of possible deregulation of the Mundy's interstate trade. So there are restrictions on interstate trade.  
[01:01:23] Shomito Charaji has a nice paper on that. Uh and then contract farming. So retail conglomerates want to go and buy directly from farmers which they're not allowed to in most states. Some states they are maybe trading online platforms.  
[01:01:36] So I think this is the if if there's [laughter] [clears throat] going to budge it's going to be these sort of proco competitive initiatives and I think number of people are studying the impacts of these these [snorts] marketing innovations. Uh that's what we  
[01:01:50] should hope for. Thanks. can manage manage the questions. We have eight to nine minutes or nine minutes for questions.  
[01:02:04] >> Um yeah, so it's super interesting. I guess I I'm trying to understand just building on your followup points like obviously it seems like competition is the is the key thing going back to the setting part what are the institutional  
[01:02:18] features sorry uh going back to the setting part what are the institutional features that you think are sort of like creating this this uh lack of competition at the top like is it something about sort of the fact that there's like a law that  
[01:02:33] basically says you have to use these mundies or like and if you were to I know If you were to speculate about sort of external validity in othera cases, what what do you think? Is there something in the setting that sort of makes this like you know lack of  
[01:02:45] competition sort of so so crucial there? >> So I think it's a combination of two two sources.  
[01:02:53] One is the political economy of of [clears throat] restrictions at the mundy. um the traders form pretty powerful interest group uh no matter which government  
[01:03:07] which you know so we've had changing politics in West Bengal um so the left turns out that the one of the leading coalition partners of the left was the forward block and the forward block is essentially the the traders uh  
[01:03:22] representatives and there are a couple of institutional studies of the rice trade and so on which document that the left pretty much sold out to the trader group. Um, so I think so that's probably a big part of it and that continued with  
[01:03:35] the TMC after the left. Now with the BJP has just taken over West Bengal. Now they're actually the first thing, one of the first things that the BJP government announced, we lift all restrictions on interstate transactions. So, so I think there's a political economy part of it,  
[01:03:48] but the other part of it is this extreme land fragmentation here. So this is specific to eastern India. In most other parts of India, farmers take their produce directly to the Mundy's and sell there and there are auctions. Now, of course, there are problems there because  
[01:04:02] there are stories about how the people who are buying at the auctions are colluding in the auctions and so on. So, it in terms of external validity, no, I'm not going to claim any. Okay, it's this is eastern India, but it's not just West Bengal. It's it's a whole, you  
[01:04:16] know, it's a substantial part of India. I don't know if that helps.  
[01:04:22] So you haven't said anything about the the supply elasticity on the part of the farmers and that seems like an important part of the story, right? It's going to affect the extent to which the the pass through is also going to have important implications for the welfare consequences of any kind of reform. Um  
[01:04:35] is that basically zero or or is it quite substantial? uh we've we've started working on it and the problem is dealing with all the lag you know how do you model price expectations and all the sort of d given the huge fluctuations in prices how are farmers forming price  
[01:04:50] expectations and so on and how do you handle the dynamics so that's something we are working on absolutely yeah but it's it's important >> say like Rachel Gabriel Laura Dean were  

## Q&A (01:05:03 – 01:10:31)

[01:05:03] the four and maybe take a couple questions Yes. So I just want to mention on the external validity. I actually think it's much more common than you're saying. I mean in Africa you have a lot of very  
[01:05:17] small farmers. Um in the casori has a lot of work that isn't published which says there's kind of eight traders between farmgate and market. there's  
[01:05:31] um and I would say the literature in general says that providing information to farmers doesn't lead to any um increase in price in the farm gate. So I think that's a very general finding. Um  
[01:05:46] and also I'd say that the the Kenya example where providing credit for storage affected farmgate is again the exception. Again, Lorenzo has a paper which shows that in Sierra Leone it  
[01:06:01] doesn't have any effect because the traders shut it down because they have market power because in that case it's because they provide credit though that they have market power.  
[01:06:12] >> I see. >> Yeah.  
[01:06:17] Um Gil given where you ended up on this slide, Dillip, I mean the the you know it sounds like you think compet complation related to competition has to be you know um where the where we should  
[01:06:29] explore further um where I'd love to hear what your thoughts about are about sort of future research either by you or others um you know exploring this further like what what sort of mechanisms or policies instruments uh  
[01:06:42] approaches might you want to explore to improve competition um and you know reduce these margins. Gab.  
[01:06:52] >> Yeah. >> Um I wanted to ask what you think about the possibility of um socially inefficient entry. Too much entry like in Manu and Winston 86. And so and the specific question is how busy are the traders? You showed that they have they  
[01:07:06] serve 8% of the village so a dozen farmers. Do they work in multiple farmers or do they actually spend a lot of time doing something else or not doing?  
[01:07:13] >> No, they do lots of other things. So a this is not potato is not the only stuff they do. They they deal with many other commodities. They're also selling farm inputs and the typical trader is operating in about five villages, five  
[01:07:26] different villages. So they're I think typically they they specialize in certain villages but then they will occasionally sort of wander into the neighboring villages if if it's profitable. Uh and so in the high price years you see more of this and I think  
[01:07:39] there's I think Kasaburi's work in the the Kasaburi read paper also documents this. So you get some sort of Yeah. So there's potentially overhead the manq effect could be there. I don't know. Uh  
[01:07:53] but yeah, so uh the answer is yes. Uh right.  
[01:07:57] Um Dean, I think you know competition but it's it's not you know you're not going to just be able to wave the competitive suppose you were to get the politics right and allow competition to happen. It's still not going to be easy  
[01:08:11] for an outofstate state buyer to come and transact with these thousands and thousands of small farmers and typically what happens is some you know Reliance is this conglomerate and uh it has  
[01:08:26] actually started buying directly from farmers in certain states whether they will succeed in eastern India I've talked to some industry experts and they think it's very unlikely because they'll eventually have to again resort to the local trader coalitions And I believe  
[01:08:40] there are reports from China where also retail conglomerates wanted to be able to contract directly with the farmers.  
[01:08:46] They tried for many years and they gave up and eventually now they're back dealing with the intermediaries. So it because this transaction costs are also a significant part of the problem.  
[01:08:56] >> Last question Laura I know there are more questions but I'm sure D will be happy to answer them during the coffee break. I was wondering, you mentioned at the beginning of the talk that there's no farmer cooperatives here and in other places those have been effective ways of kind of counteracting this type of  
[01:09:09] market power and I'm just wondering is it the extreme land fragmentation or what features of this environment prevent farmers I from self-organizing or from coordinating with the local traders to try to gain some of these rents.  
[01:09:23] Yeah, I I I wish I knew. I mean, I've studied farmer cooperatives, you know, in other parts of India, there so much heterogeneity, you know, I think the cooperatives that work have to be organic, bottom-up cooperatives. That's partly the sociology of what it takes to  
[01:09:38] organize uh self-organized. It seems to work quite well in western India. But even within western India, it seems to depend on the particular crop. So, but basically, it's not operational here.  
[01:09:48] The credit cooperatives are there but for some reason they don't they don't try to the cooperatives the farmer cooperatives don't seem to I once asked farmer saying why don't you you know they don't want to talk about this but this guy was actually giving me  
[01:10:03] dinner in his house because I was stranded and so he kind of opened up and I asked him why come on why don't you guys get together and try and approach some of these mundy buyers he just looked away and he said there will be blood in the streets  
[01:10:17] >> on that happy note break and restart at 10:20.