YT YouTube2Text

YouTube2Text → Transcripts → Lecture 6: Contracts with Rocco Machciavello

Lecture 6: Contracts with Rocco Machciavello — Transcript

by International Growth Centre · 15,944 words · 2,472 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:01approach
0:31right
0:32now
0:33[Music]
1:10wonderful thank you for uh joining us uh
1:12today which on this uh sixth lecture in
1:15this uh course on funds and development
1:17uh just a reminder that all the other
1:19lectures are posted uh on the course
1:22website which will drop a link to uh in
1:25the uh chat
1:27um and uh Rocco's uh slides uh if
1:30they're not already there we'll also be
1:32posted
1:33um so we're uh very lucky today to have
1:35a Rocco machiavello who who are
1:37undoubtedly uh is the world expert on
1:41topics related to our firms contracts
1:43and development which is what he's gonna
1:45be talking about how does this link into
1:46some of our broader discussions uh well
1:49uh you know
1:51contracts are obviously vital in firms
1:54securing uh inputs uh required uh uh to
1:58to produce and as we kind of showed in
2:00the first lecture uh potentially poor
2:02Contracting environments lead to
2:04distorted input use maybe a
2:07misallocation uh
2:09of various inputs in the production
2:12process and so this fits uh very much
2:15along with the labor and capital
2:16lectures yesterday in kind of trying to
2:19dive into those various uh uh uh
2:22distortions uh that Pete uh talked about
2:24so Rocco I'll leave it to you to take
2:27away and uh we're thrilled to have you
2:31well first of all thank you so much
2:32David uh and Dave for the invitation and
2:35the introduction and then also good
2:37morning good afternoon and for some
2:39people also good evening everyone
2:41so yes I'm gonna talk about uh some very
2:44selected topics and maybe try to
2:46highlight some open area uh um in
2:49relationship to firms contracts and and
2:51development yeah
2:53So the plan is that I'll start with some
2:56very brief General remarks on
2:57Contracting problems okay at the very
3:00high level and uh and then all those
3:03have been looked at uh in the
3:05development economics literature in a
3:07very selected way and then I'm gonna
3:09kind of move on and talk specifically
3:11about the relationship between firms
3:13that motivate why I think that's a good
3:15angle to take to sort of think about
3:17some of the themes that have already
3:18emerged in the previous lectures
3:20and then we in particularly focus on
3:23studies that uh
3:25look at the dynamic incentive
3:28compatibility constraints that are kind
3:30of embedded if you like a lot of the uh
3:31the foundation of these long-term
3:33relationships between firms
3:35now if I have time I'll talk a little
3:37bit about relationship and Market
3:39structure and then uh I hope to have
3:41some time to highlight some open areas
3:43and future work I think this is
3:45generally speaking an area that is at
3:47the same time very vast but also still
3:50uh whatever still a room for a lot of
3:52work and a lot of exciting
3:54so and broadly speaking there are many
3:58different types of contractual
3:59imperfections so the economic theory on
4:02uh Contracting problem is uh enormous uh
4:05kind of we can say that maybe started
4:07originating and really taking off in the
4:0870s and 80s and then later actually also
4:11the empirical literature on a contract
4:13as a as a as a sort of exploded uh in
4:17development but also in other field so
4:19if you can think of asymmetric
4:20information like more or less or other
4:21selection you can think of cases in
4:24which actually
4:25there's no problem with asymmetric
4:27information uh there's observability but
4:29non-verifiability so it's a core that
4:31cannot observe and therefore uh enforce
4:33the contract so for instance holding
4:35complete contracts framework is built
4:37around that uh there's also modeling
4:39approaches that take more uh explicit
4:41into account transaction costs for
4:43instance costly State verification model
4:45okay
4:46in all these cases however we can say at
4:49least in a very vague sense that market
4:50markets are incomplete it is not all the
4:53things that matter are inade adequately
4:56priced okay because they cannot be put
4:58in a contract and we cannot have prices
5:00sort of at reflect the opportunity cost
5:02of resources if you like uh adequately
5:04so you know obviously
5:06um most of you will have heard examples
5:08in Insurance Market thinking in terms of
5:10the arrow the broad framework others
5:13will have seen sort of credit markets in
5:14which there is other selection and
5:16therefore the sort of pecuniary
5:18externalities that risky borrowers
5:20imposed on other borrowers in the
5:21markets uh other if you've studied
5:23development there's a long traditional
5:25study Hungarian institution and kind of
5:27trying to understand why share cropping
5:29is is so prevalent in agriculture in
5:32many parts of the world generally
5:35speaking the empirical nature of your
5:37own contracts has the goal of uncovering
5:40the underlying Market imperfection the
5:42underlying contractual imperfection in
5:45order to design a Better Contracting
5:47solution maybe even better policy
5:49Solutions
5:51and indeed so what I've just sold I
5:54think is a general uh about the
5:57literature trading contracts but given
5:58this type of question and these premises
6:00is no surprise that historically uh in
6:03the intellectual tradition in the field
6:05there's been a very tight connection
6:07between development of contract
6:09economics and the economics of
6:11development yeah so for instance here is
6:12a quote this paper presents a struggling
6:15attempt to give structure to the
6:16statement business in underdeveloped
6:18countries is difficult okay and I don't
6:21know uh if we can have a quick round of
6:24people that can say who recognize the
6:26quote otherwise I'll tell you anyone
6:29recognize the quote
6:31this is in the first paragraph of Uncle
6:33of the market for lemon so this is just
6:36a an example say one of the foundational
6:38papers in uh information economics and
6:41uh and contractual economics actually
6:43yeah there's a starting motivation sort
6:45of trying to understand under
6:46development or the fact that it's
6:48difficult to do business in India okay
6:51and So within development uh the
6:53empirical understanding of contracts of
6:55course has done has improved under our
6:57understanding of many different markets
6:59uh for instance all the literature on
7:01insurance and particular informal
7:02Insurance within communities we can
7:04think about the literature from credit
7:05Market uh the very nice paper by Karen
7:09and Zeman on you know kind of designing
7:11an experiment to entangle their
7:13selection and more or less or in the
7:15consumer credit Market comes to mind
7:16there's a fascinating work on
7:19Contracting in markets for water which
7:21is a very uh important resource in
7:23agriculture and you know kind of plaque
7:25by Contracting problems of all types and
7:28then uh already I mentioned
7:29sharecropping and many others and of
7:31course among these there's also been
7:33work and this is where we are going to
7:35discuss a little bit more during this
7:36lecture hopefully on contracts and firms
7:39okay so how Contracting problems affect
7:43firms uh performance and organizations
7:46in developing countries the ninja three
7:48I would say is a little bit smaller
7:50maybe than some of the owner literature
7:52in development that looks at Contracting
7:54problems that I've just mentioned so
7:56just kind of uh mention three kind of uh
7:59I would say by now classic relatively
8:01early work in uh in this trend so the
8:05first one is the banner you do for now
8:06nowadays classic study on reputation
8:09effects in the software uh industry in
8:12India and um that's a paper essentially
8:16that shows how variables that are
8:18arguably correlated with reputation
8:20think about for instance the age of the
8:22firm that is strongly correlate with the
8:24type of contraceptive these software
8:26development firms and up end up getting
8:28okay and um you know one nice thing
8:32about about this paper is that very much
8:34like it's kind of customer or
8:36traditional in development the authors
8:38went out there to collect their own data
8:40they conducted a survey of about if I
8:42remember correctly a bit more than 120
8:44firms maybe about 150 contracts so that
8:47that these approach of sort of
8:50you know we desire to study Contracting
8:52problems by just relying on sort of
8:54off-the-shelf data I think kind of he
8:56goes a little bit some of the themes
8:57that Eric has talked about in maybe
8:59lecture two on upgrading and sort of the
9:02the need to try to measure things uh
9:04directly uh Chris Woodruff uh who
9:07yesterday told us about credit markets
9:09and returns to Capital of course uh
9:11after his PhD was one of the uh he was
9:14doing a lot of work on contracts so for
9:16instance this fascinating paper that he
9:18has on vertical integration uh in the
9:20Mexican Footwear industry where
9:22basically by distinguishing across
9:24different segments uh women's versus Men
9:27shows he tries essentially to test for
9:30and distinguish between the different
9:31theories of the firm and then um you
9:34know we couldn't we could say that maybe
9:36the work also uh on Capital Market
9:39imperfections so for instance the work
9:41by abijit manager and kaiban munchie in
9:44the looking at Miss allocation of
9:45capital in the tirupore industry is also
9:48kind of related uh to Contracting
9:50problems because ultimately uh the
9:53capital is an input one is particularly
9:55fungible maybe as an input but there is
9:58still an input and so the kind of
9:59Contracting problem that Capital markets
10:02often also do plug uh other other input
10:05markets so that's kind of if you like a
10:07special case of a uh most likely the
10:09kind of marketing credit marketing
10:10profession they discuss in that paper
10:12originate from some underlying
10:14Contracting Contracting problem okay now
10:16there's been of course more recent
10:17contribution and an uncoverables later
10:21both a theoretical argument and
10:24empirical observation can suggests that
10:26repeated interaction can help firms to
10:29overcome or ameliorate or a surge if you
10:32like
10:33these Contracting problems okay
10:35so again to mention two early
10:39papers
10:41specifically on relationships and which
10:44is going to be the focus of my of my
10:46lecture and development don't MacMillan
10:48and Chris Woodruff at the really
10:50landmark and now again a classic paper
10:53on relational Contracting and informal
10:55Contracting between firms in Vietnam
10:58Vietnam at the time was a very
11:00interesting case because it was starting
11:01the transition uh and courts in general
11:04were not very well functioning so they
11:07kind of consider a little bit how search
11:09cost to search for alternative trading
11:12partners as well as absence of legal
11:15enforcement of contract affect the
11:18extent of Trade Credit between these
11:21manufacturing firms in in Vietnam again
11:23this is a again a case where uh
11:26development economists go to the field
11:28design a survey and collect original
11:30data from the survey unlike the binary
11:32Loop flow paper they do not focus on a
11:35specific sector so the question are kind
11:37of more General if you like they are
11:38applicable to and relevant to many
11:41different
11:42sectors and what they document are kind
11:45of patterns that are strongly consistent
11:48if you like with the sort of theoretical
11:50models of of relational Contracting in
11:52particular they show that the longer
11:54duration of a trading relationship which
11:56probably correlates with trust is
11:58associated with uh with more Trade
12:00Credit and so is also the cost of of the
12:03partner to find alternative sources of
12:06uh of uh of inputs okay so partners that
12:08have more trouble in finding alternative
12:11suppliers are more likely to receive
12:13Trade Credit from their supplier this
12:14again because these difficulty in
12:16switching presumably is kind of
12:17conducive to some some trust there's a
12:20lot of other work similar work of course
12:22I mean the the paper is very highly
12:24cited this type of regression has been
12:25done in many other contexts there's a
12:28lot of work for instance by Marcel
12:29fashion and and many other at CSA talks
12:33or that look specifically uh the
12:34informal relationship and how they
12:36affect the the behavior of the
12:38organization and the performance of of
12:40firms in African context for instance
12:43and the other point I want to uh maybe
12:46make here is that in a companion paper
12:48to this Vietnam paper with Simon Johnson
12:50uh as well as John McMillan and Chris
12:53Woodruff also a little bit again through
12:55survey questioners I sort of try to
12:58consider what is the role of Court
13:00um and there what they show is that um
13:03believes in the effectiveness of course
13:05in transition economy is actually very
13:07strongly correlate with the firm's
13:10willingness at least self-reported
13:12willingness to try out new partners so
13:15that might suggest this sort of earlier
13:17survey evidence that well although
13:20relationships are indeed uh potentially
13:23important to help firm sustained trade
13:25at the same time it's not that courts
13:27they fully substitute for courts courts
13:29are still a very important maybe in
13:31order to generate an environment in
13:34which trading partners with comfortable
13:36experimenting with uh with with other uh
13:39clients and suppliers okay so now what I
13:42will uh you know having kind of wrapped
13:45up is a very broad uh introduction to
13:48some of the themes and and if you like
13:50more kind of classic work in in this
13:52field again highly selected of course I
13:55will now focus on a recent work
13:56specifically on firm relationship okay
14:00and I will try to uh sort of Island as
14:04much as possible and a little bit more
14:05precise way the connection between this
14:08topic with the things that have already
14:10been discussed in the in the previous
14:12lectures okay but maybe this is a good
14:16moment to take a quick pause to see if
14:18there are any questions uh before I sort
14:21of move on
14:23hi Jonathan yes
14:27hi thanks rocker obviously interested
14:31you you mentioned trust and I saw you
14:33published recently on informal contracts
14:35I wonder whether United comment between
14:38formal and informal contracts and the
14:40relationship then of social networks to
14:43that for example the
14:45uh Indian networks in in East Africa and
14:49in southern Africa and other places that
14:52seem to have contributed to uh trust in
14:56Contracting and whether this has an
14:58effect on the larger economy because
15:01it's just totally um anecdotally it gets
15:03I get the sense that paranoid Nigeria's
15:06missing middle is the lack of trust to
15:09create larger organizations other small
15:11ones
15:14uh
15:15oh not yet in a feels like I agree uh is
15:18a is a topic which I'll uh maybe return
15:20very vaguely uh in a few slides so I was
15:23I think I thought it was already in the
15:25next slide
15:26um yeah so there's um there are several
15:29interesting issues here with regard to
15:31trust uh the first one is uh what is the
15:34role so first of all we use the word the
15:37trust in two very different ways right
15:38and uh and I'll come to that in a second
15:40one is within the context of a
15:43relationship maybe we have in mind some
15:44sort of repeated game and under certain
15:46condition if some conditions are
15:48satisfied I trust that you will behave
15:50in a certain way and that's one notion
15:52of trust we also use the word trust
15:55however in the sense of generalized
15:57trust like historical norms and there's
15:59a lot of work that shows that there is
16:01enormous cross-country or cross Regional
16:03variation in the extent which these
16:05Norms are adhered to and to the extent
16:07which uh people trust each other so in
16:10the lecture I'm gonna specifically talk
16:12about the firmer notion of trust as it
16:14would become clear in a second
16:17so thinking specifically about the pair
16:19however I also think that the
16:21generalized Trust of the societal Norms
16:23are very important to think about our
16:26trust in a bilateral pair emerge and
16:28under which condition uh parties will
16:30end up trusting each other
16:32um there's there's a lot of descriptive
16:34work but I don't think that there is a
16:36at least I'm not aware of a lot of work
16:38that establishes a connection between
16:41the variation in uh Norms of trust and
16:46actual contractual outcomes in terms of
16:49uh um either the form or the structure
16:51of relationship
16:53two papers come to mind one is a paper
16:55by a former student of mine Arthur
16:57Blowin uh he has a paper in the AJ where
17:00he showed us that uh sort of the
17:02colonial history in the context of
17:04Rwanda Burundi that is strongly
17:06correlated with trusts in the attitudes
17:07today seems to affect also the tape and
17:11the form of uh interaction or contracts
17:13between between parties and then more
17:15recently we see gadan and quarters uh
17:18they use uh uh v80 so business to
17:21business data in uh in in Bangladesh
17:24sorry in India I think where they also
17:26have information on the cast to which uh
17:29the firms belong to kind of check
17:31whether
17:32this community networks that presumably
17:35can are us a little bit more trust and
17:37if we were just in an anonymous Market
17:39whether that kind of correlates with the
17:41prevalence of uh of uh of uh and and
17:43shape of firms to firm transaction but I
17:46agree that uh more work would be needed
17:49and I think this is a maybe point one or
17:51two in my uh in my concluding slides on
17:53open areas for for future work partly
17:55because it could be that in a partial
17:57equilibrium sense having a relatively
18:00close community that Foster ties but
18:02remember is very good and it allows uh
18:05firms to overcome Contracting problems
18:07I'm thinking here for instance about the
18:08work of time and motion the diamond
18:10industry but at the same time from a
18:12micro perspective right this kind of
18:14type Community might generate Distortion
18:16might prevent uh better matches uh uh
18:19and so forth and I think that we don't
18:21have very good evidence on sort of our
18:23veteranoff is resolved
18:27so just a very long answer to say I this
18:30is a fascinating topic it won't be
18:32square and Center here but
18:35but yeah which I think is very important
18:38all right so uh unless there are other
18:40questions I'll move on and uh kind of
18:42start zooming in a little bit on
18:43relationships between firms and I plan
18:46to First motivate why do a little bit
18:49more in detail why it might be important
18:51as a development Economist to think
18:52about this relationship between firms
18:54and then I'll talk about uh some
18:57empirical work uh some more recent
18:59empirical work
19:00nope
19:02um dealing on Williamson won the his
19:05Nobel Prize for sort of introducing a
19:07transaction cost economics and the study
19:09of governance forms uh uh into
19:11mainstream economics we can think
19:13roughly speaking that there are three
19:15broad governance forms to organize
19:18transactions right a given transaction
19:20can be organized in the spot Market
19:22that's kind of what we teach in our EC
19:24101 you could also the same transaction
19:27of course could be organized within a
19:29firm so instead of using the price
19:32mechanism and the centralized sort of
19:34information to allocate resources we use
19:36Authority okay that's typically in ec101
19:39what we put uh into a black box is a
19:41production function but of course is
19:43what uh instead is the object of study
19:46of of other fields and then in between
19:48these two extreme so the market that
19:50uses prices to allocate resources and
19:52firms that use hierarchies or
19:54authorities or allocate resources there
19:57are long-term relationship between firms
19:58or what the Williamson referred to as a
20:01as e Breeze okay and they are their
20:04leader firms nor Market really because
20:06certainly they are not Market in the
20:08sense that the price is not the only
20:09thing that is used to allocate resources
20:12there's a lot of bargaining there's a
20:13lot of discussion there's a lot of kind
20:15of coordination but at the same time
20:17because the two parties are separate
20:19there's no formal authority of one party
20:21over the other so there's no sense in
20:23which a supplier can tell that their
20:26their customer what the customer is
20:28supposed to do or vice versa I mean at
20:30least outside the scope of whatever is
20:32in the enforceable contract
20:34so I'm prevalent our long-term
20:36relationship I think this asking this
20:38question matters because
20:40um it matters whether you know whether
20:43in a way we use a description of the
20:45economies that are some vaguely accurate
20:48in terms of what's the relative
20:49importance of this governance part and
20:51if you look at the US economy this this
20:54graph is not Based on data it's an
20:56illustration but if you look at the US
20:58economy so the market economy par
21:00Excellence you know at about 40 or 30 to
21:0440 percent of uh uh income essentially a
21:08value added actually he's done by the
21:10government so that's really something
21:11that is not done by markets
21:14within the part that is done by the
21:16private sector if you look at the value
21:18of the data in the census it looks like
21:20about 50 percent of the value on that
21:22takes place kind of within firms okay so
21:25that's also another well known fact by
21:27industrial economy so about half of the
21:29activities of the of the eclipse off of
21:31the activity of the private sector in
21:33the market economy by Excellence doesn't
21:35take place in in markets
21:38what about the remaining between firms
21:40okay how much of that is actually
21:42markets and no markets relationship we
21:43don't know because typically census data
21:46do not uh contain information on the
21:49identity of the trading partner so they
21:50don't allow us to really understand
21:52whether this is a market or whether the
21:54transaction is organized in some other
21:55ways however we can look at Customs data
21:58and Customs data in the US tell us that
22:00about 80 percent of all U.S Imports
22:03occur in pre-existing relationships okay
22:06so it looks like if you look at another
22:07place custom so internet transaction
22:09that cross-borders of goods that come
22:12into the US the vast majority of the
22:14transaction that do not occur within
22:16firms about 30 percent even there occur
22:18with inference actually do not seem to
22:20occur in stock markets they seem to
22:22occur in some between people that trade
22:24repeatedly uh this fact has been
22:27confirmed in uh in many other Customs
22:30there but you could kind of argue that
22:32you know International transaction
22:34longer distance higher asymmetric
22:37information costlier enforcement of
22:40course if we use relationships to
22:42mitigate this problem looking at Customs
22:44doesn't really give us an accurate
22:46description of our prevalent
22:48relationships are it turns out that in
22:50recent er uh
22:52um researchers have gained access to uh
22:54uh typically firms to firms transaction
22:57level data this would almost always
23:00comes from vat type of Records
23:03um with various degree of detail so both
23:06for instance in developing countries uh
23:08uh you know they are available for as
23:11far as I know at least Chile Costa Rica
23:13Ecuador or many other countries and so
23:15one could imagine using kind of this
23:17data set to sort of create some
23:19systematic facts about how prevalent our
23:22repeated transaction between in this
23:25case relatively larger firms at least
23:26since these are creative data that
23:28comply with the taxation I haven't done
23:30that exercise uh it has been on my mind
23:32for a long time to put an email to
23:34people that have access to this data to
23:36ask them to compute a few basic
23:37statistics I wonder what we the other
23:40another way to go about this is to look
23:42at specific industry where we have good
23:44data and there I I just want to mention
23:46a couple of uh contexts where uh I I
23:48have the data so I can sort of compute
23:50these numbers and one is the context of
23:54uh answer this trades in Peru okay so uh
23:57imagine Wars that go out at Sea they
24:00they they catch fish then they bring it
24:02back to process in plan to produce fish
24:04meal this is actually an industry that
24:05has been studied in other papers with
24:08different with different focus and then
24:11the other industry I I can look a little
24:13bit that was number is uh in Costa Rica
24:15where again the regulator
24:18um typically those are regulated
24:19industry where we are we can access a
24:21good data
24:22um there because the regulator
24:24essentially looks after the the coffee
24:26sector in a way that generates a lot of
24:28data we can also look at what's the
24:30relative prevalence of these governance
24:32forms in that context and in both
24:33contexts the vast majority of food or
24:38coffee is intermediate either within
24:40vertical integrated Shades so processor
24:43that owns uh fishing boards or coffee
24:46exporters coffee meals or between
24:49players that have very long-term
24:51relationship I take like at least three
24:52years uh in the data as a very kind of
24:55conservative bomb
24:56and what is interesting is that if you
24:58think about this
25:00or if you think about coffee
25:02you might think that these are
25:03relatively simple good and in fact kind
25:05of homogeneous may be good or not very
25:07highly differentiated now for coffee we
25:10could distinct maybe there's an argument
25:11for fish as well but certainly less than
25:13many more specialized and customized
25:16inputs that we can think for more
25:18sophisticated production process
25:21and so what transaction cost fear you
25:23would suggest is that what should be
25:25places where if anything vertical
25:26integration and relationships should be
25:28relatively less prevalent okay so it
25:30might be in fact that this detailed
25:31studies give us lower bound of the
25:34prevalence of relationship
25:35in other words I would say with the
25:37bottom line
25:38um is that it seems to be that ec101
25:41market uh oops sorry this is a call from
25:46the iPad sorry about
25:48um it seems that can you see my slides
25:51um
25:52no no
25:53but you can now right but let's see very
25:56soon yes
25:58um so it seems that uh sort of the ec101
26:00market essentially the decentralized the
26:02stock market really is the exception
26:05it's not the typical way in which goods
26:08are intermediated
26:11and okay so I think that this is true
26:14not just in developing countries but
26:15also in reach in a high income countries
26:17there's a student from MIT this year on
26:20the market and going with a wonderful
26:22paper or long-term relationship in the
26:24U.S tracking industry and I think she
26:26argues that about 80 of the industries
26:28actually uh the parking lots is actually
26:30shipped through long-term relationship
26:32so tracking us maybe contract is not
26:35there uh so
26:37I believe that this fact that I
26:39mentioned in these slides is kind of uh
26:41a quiet purpose now why it might be
26:43particularly important if that's the
26:45case to study relationships in
26:47developing count
26:49what David and Dave and uh and and I
26:52think Pete that sort of argued uh or one
26:55one takeaway at least uh uh of uh of
26:58what I've earned during their election
26:59is that markets maybe don't function
27:01very well yeah
27:03then we have had I think uh Eric and
27:05Jing and uh
27:07uh to a certain extent also David and
27:10Chris at least in the part in which we
27:12focus on smaller firms also arguing that
27:14firms in developing countries are uh
27:17unproductive we are not very well
27:19managed they struggle to grow so they
27:21struggle to integrative that's what they
27:23need to do okay so crimes also do not
27:25function very well so if we go back to
27:27the free possibilities essentially to
27:29organize transaction markets
27:31relationship and firms it looks like the
27:33two Alternatives don't work very well
27:34they seems to be constrained it's not
27:36easy to grow large vertical integrated
27:38firms and are relatively few of those
27:40and then and then of course also the
27:43market seems to be plugged by all sorts
27:45of uh of imperfections okay and so that
27:47at least is the sense in which
27:48qualitatively studying relationship
27:50might be particularly important in
27:52developing countries now of course this
27:54this statement should be coveted in a in
27:56a in many different ways uh one way uh
28:01for lack of time which I think is
28:03important to cover this is that
28:04essentially so far I've made an argument
28:07based on the demand for relations right
28:09broadly speaking we have said well
28:10developing countries has a weaker
28:12institutional environment uh therefore
28:15even simple transaction maybe even
28:17simple transaction agricultural markets
28:19a simple trade credit loan uh the
28:22transaction of milk or maybe the
28:24transaction of fish even both simple
28:25transactions they cannot rely on formal
28:27contract enforcement so that increases
28:30the demand for relationships
28:33but what about the supply because it
28:35could be that the same factors that make
28:38a country underdeveloped also reduce the
28:41supply of relationships yeah so for
28:43instance it could be that generalized
28:45trust is very low in developing
28:46countries and there are Polaris are
28:48there to try out or even establish or uh
28:51while functioning relationship
28:53um chords might not be functioning very
28:55well and that again might restrict sort
28:57of the supply of uh of uh of
28:59relationship at the same time that um
29:02that it increases the demand for for
29:05long-term risk
29:07and then another aspect that I won't be
29:09able to emphasize a lot is that the
29:11choice of governance forms of course
29:13interact in equilibrium so Rachel
29:15Crampton has a very nice theoretical
29:17paper in the in the AR in 96 where she
29:21kind of tries to model the relationship
29:22between
29:24long-term relationship
29:26so forms of exchange that are supported
29:29by bilateral Trust
29:31and the emergence of spot Market
29:33if one advantage of spot Market is the
29:37liquidity so the fact that I easily find
29:38a match when I need in the spot Market
29:41if many trading partners are engaged in
29:44trading in bilateral relationships that
29:46will generate only SQL a very uh thin
29:49marketing which is difficult to find
29:51partners
29:52but then from the point of view of a
29:55firm that is choosing whether to trade
29:57in the market or try to establish a
29:58relationship that configuration the fact
30:01that many others are already active in
30:02relationship will make the market
30:03particularly less attractive and so
30:06there's maybe a strategic
30:07complementarity in the choice of
30:09governance forms across firms and so we
30:11shouldn't really be studying which is
30:13what I'm gonna do in a second these
30:15organizational forms choices in
30:17isolation yeah
30:18there can also be Arguments for why the
30:20organizational forms are actually
30:21strategic substitute the general point I
30:24want to do I like is that in a moment
30:26we're gonna study uh we're gonna review
30:28studies that study uh relationships
30:31specifically and kind of start your
30:33relationships so much in isolation but
30:35in fact we should also be thinking about
30:37how the choice of the organizational
30:39forms interact
30:42okay
30:44so this is probably another good moment
30:47to uh I hope I have motivated why I'm
30:50gonna focus on uh relationship between
30:52firms in the reminder of these uh this
30:55lecture and now maybe it's again a good
30:57time to to um
31:00take a break and see whatever our
31:01questions
31:09I don't see any questions so I'm gonna
31:12uh move on okay so now what I'm gonna do
31:16is that uh I think this is kind of the
31:18the main part of what I was hoping to
31:21cover today having motivated why might
31:23be useful to study relationship I'm
31:25gonna try to be a little bit more formal
31:27okay in thinking about relationship and
31:31uh
31:31will provide an approach to study
31:34long-term relationship between firms uh
31:37empirically okay and then I'll try to
31:39connect this work with some of the films
31:42that have come up in the in the earlier
31:44lectures
31:47so uh as Jonathan has sort of uh already
31:51I know that before we um we think that
31:54um we are talking about essentially a
31:56relationship based on trust there are
31:58two different Notions of trust it is
32:00another very nice quote from kanataro
32:02again kind of seemingly relating the
32:05problem of underdevelopment to a problem
32:07of of lack of trust
32:09there's an economist think about trust
32:11in two different ways either as a norm
32:13say generalized trust reciprocity that
32:15is enormous literature and labia
32:18evidence supporting that notion or
32:20within the context of repeated game
32:22framework from now on I'm gonna go down
32:25the repeated games framework group
32:31[Music]
32:34perform trust and uh overcome
32:37Contracting problems however all these
32:40different models some emphasizing all
32:42enforcement problems or their
32:44emphasizing Insurance consideration some
32:46emphasizing asymmetric information and
32:48so forth
32:49all these models rely on a common
32:51Insight namely the idea that the future
32:54rents are necessary to deter short-term
32:57opportunities okay so the simplest model
32:59that capture is inclusion of course is
33:01the repeated prisoner dilemma yeah
33:03suppose you have two players I'm playing
33:06a repeat a prisoner of the RAM and
33:07infinitely repeated prisoner dilemma
33:09with David if we both cooperate I get P
33:12of C
33:13if I deviate any cooperates I get paid
33:16off D but then maybe after he sees it I
33:19have deviated while he was expecting me
33:20to cooperate then maybe you know he's
33:22also going to be fact and then if we
33:24both the fact I get paid off P okay
33:27so of course I'll be willing to
33:29cooperate uh in this framework only if
33:32the payoff of cooperating today and then
33:34keep on doing so in the future is bigger
33:37than the short run temptation to defect
33:40okay which gives me pay off D but then
33:43of course uh uh being uh
33:47being followed by a period of of
33:49punishment okay so we can rearrange this
33:51expression and we see that the parties
33:53will be able to cooperate if the value
33:55of the relationship that stuff that is
33:57in the future okay
33:59is bigger than the current temptation to
34:01the vi
34:03and again this is a very simple model
34:05certainly a very simplistic model even
34:08but this tension between current
34:10Temptation that arise from the fact that
34:12the actions are non-contractable and
34:14future rents is what underpins all model
34:17of uh of long-term uh Contracting and
34:20repeated interactions okay and the
34:22problem if you're interested in doing
34:23empirics of course is that neither the
34:26future value of the relationship nor the
34:30right hand side the temptation to the VA
34:31are directly observing the data and
34:34there are various reasons why they are
34:36not directly observing the data one
34:38simple reason typically said on the
34:40equilibrium path we are cooperating if
34:42we expected not to cooperate there
34:43wouldn't be a relationship in the first
34:45place but then if we are on the
34:46equilibrium path
34:48punishment and the factions are not
34:50supposed to arise so you're not even
34:51supposed to see those in the data yeah
34:54at least if you take the standard uh
34:56framework so we need to find kind of to
34:58think a little bit more carefully about
35:00how we're going to study relationship
35:01empirical
35:03so one approach to measurement of course
35:06has been to say well you know let's give
35:08up on this and let's simply try to proxy
35:10for the uh this future value of the
35:13relationship for instance using
35:14relationship page uh so either measured
35:17in calendar time or a past transaction I
35:20would say 90 maybe 95 percent of the
35:23literature should actually follows this
35:24approach and it's an approach that is
35:26very intuitive because
35:27if two parties have traded repeatedly of
35:30course presumably they have more trust I
35:32mean you know these future value that is
35:35what allows them to overcome a current
35:37temptation is kind of loud however I
35:39have to say that uh the the the the key
35:43of course in the model
35:44is not the past stage per se right is
35:47there is this future is the promise of
35:49having this future rents uh in the
35:51relationship and so in a way we might
35:53say that relational
35:55um right it is different from repeated
35:58failure I always buy my sandwich in the
36:01same cafeteria at the lse but there's
36:03not really no relational trade so
36:04there's a lot of repeated trade there if
36:06you get access to virtil records you see
36:08that Rocco keeps keeps purchasing from
36:10them but there's no relational trade
36:12there's no future expectation of what
36:14they are supposed to do and and what I'm
36:16supposed to do other than you know what
36:18is kind of implied by by sort of
36:20um you know the social norms if you like
36:23of of buying croissant uh uh in London
36:25okay
36:27so there are other approaches one is to
36:29kind of try to write down specific
36:31models and then derive the proxies from
36:33this model for the prevalence of these
36:35rents
36:36another approach that is very related to
36:39uh what Eric has been discussing is to
36:42Simply go out there and try to measure
36:44directly what are these informal
36:46promises that parties make to each other
36:49and then presumably one could imagine
36:52also different types of other ways of
36:54sort of uncovering empirically whether
36:55there are relationships uh for instance
36:58we could simply estimate structural
37:00models uh and then kind of let the model
37:02kind of allow us to back out to this V
37:05or we could imagine try to use natural
37:07experiments okay uh for instance
37:10talks to Future traits talk to
37:12anticipation of future trade if a
37:15relationship is in place you cannot lead
37:16to an immediate change uh NBA
37:19so now we're gonna talk a little bit
37:21about a couple of papers that have uh
37:23that I've written with the with the with
37:25the different quarters with the meat
37:26market area who is now a teller than
37:28with Arthur Bloom who is now in Toronto
37:30that kind of try to take specifically
37:33this Dynamic incentive compatibility to
37:35constraint to the data and trying to
37:37learn from that
37:43so the first paper is the paper it is a
37:46paper in which uh Amit and myself study
37:48the uh Kenya flower export sector okay
37:52we turns out that Kenya exports a lot of
37:54roses
37:56uh would be fascinating to understand
37:57exactly why the industry developed
37:59better and so forth but I'm not going to
38:00talk about that
38:02but this is a market that is
38:03particularly convenient to study
38:05relationship and it is so for advance
38:08for a different reason first of all
38:10while we were able to get access to kind
38:12of a buyer sellers that are relatively
38:14well detailed I would say at least back
38:16back in the days so with transaction
38:18level records
38:20but the second thing that makes this uh
38:22marketing interesting is that no
38:25governance forms kind of coexist you
38:27either export process through long-term
38:29relationship with Buyers so you have a
38:31wholesaler or an important or browsers
38:33in the UK and you're just shipping these
38:36roses two or three times a week from
38:37Nairobi by playing to the wholesalers in
38:40the UK or you sell your roses on the
38:43Dutch option which are essentially the
38:44stock market par Excellence is a
38:46platform in the Netherlands where all
38:48contracts are enforced and then
38:49everybody kind of participate in the
38:51market and then bids for the Roses so
38:53the Dutch options are actually really
38:55kind of our ec101 Mark okay and those
38:58two forms coexist presumably because
39:00they kind of entail different uh uh
39:02transaction costs
39:04I mean the nice idea is that the
39:07presence of the spot market and
39:09therefore the reference price in the
39:11Dutch options gives us a potential
39:13handle on these Temptations to deviate
39:15which are typically not observed as I
39:17mentioned before
39:19and then the second thing we're going to
39:20also rely on a shock to kind of
39:22distinguish between different models of
39:24of transactions of a sort of
39:27relationship there is a couple of pets
39:30so the left figure looks at the over the
39:32week in the season the season the flower
39:35season so to speak starts in August
39:37September okay so these are the prices
39:40the weekly prices for roses are the dark
39:42souls for three different season and as
39:45you can see there's a seasonality that's
39:47not surprising because flowers are
39:49perishable so you can't store them and
39:51but the seasonality is very predictable
39:53so relative to the average price in the
39:55market around Valentine's Day which
39:57happens there or around Mother's Day
39:59which happens around this time depending
40:02on calendars and Country prices in the
40:04markets are always higher but kind of
40:06predictably so okay in contracts and
40:09this is just a raw data price is in
40:11relationship tend to be very very stable
40:13over the course of the season if David
40:15and I have a relationship we'll kind of
40:17agree a price and then we'll maybe
40:18renegotiated next year if we want to
40:20keep working together we don't use a lot
40:22of uh index prices uh in in this
40:25industry and I you know as you will see
40:28in a second or maybe not but you know in
40:30the paper we argue what we have an
40:31explanation for why that's okay and um
40:34um and of course the idea is that uh
40:37you know an index price uh will maybe
40:41um you know by tracking the stock market
40:43price when it goes up it reduces the
40:45temptation to debate for the seller but
40:47it maybe increases the Temptation from
40:48the V8 for the buyer so one implication
40:50of this relational Contracting model is
40:53that the incentive compatibility
40:54constraints of the two parties can be
40:56aggregated under certain circumstances
40:58therefore if that's the case using index
41:01prices actually doesn't help in uh in
41:03alleviating uh um uh enforcement
41:06constraint
41:07so from the perspective of a seller the
41:10seller could be cheating in many
41:11different ways but one way in which the
41:13seller could cheat in this market is by
41:16sending the promised amount of flowers
41:17in the relationship you see
41:20which is supposed to receive a price PC
41:23on the spot Market at Price PW
41:25and of course
41:27um you know
41:28the this temptation will be particularly
41:31strong at the Valentine's Day when the
41:34price in the market is very high
41:38in fact we have a the paper kind of
41:41develops a little bit of a test to check
41:42whether this constraint is binding or
41:45not the idea is that small price
41:47fluctuation and the times of Valentine's
41:49Day but is when these incentive
41:50constraints should be buying if it's
41:52binding at all so basically imply a a
41:55reduction in the volume traded in a
41:57relationship with an elasticity equal to
41:59what minus one which is exactly what we
42:01see in the data so but the point here is
42:04that uh all of a sudden we have
42:06observable proxies for Temptations to
42:09deviate in the data yeah the uh price in
42:13the market is observed and the price in
42:15the relationship and the volumes in the
42:17relationship are directly observed and
42:19so now we have a uh a measure at least a
42:22lower amount and in fact a tight bound
42:24if the incentive constraint is binding
42:25for the value of the relationship for
42:27the seller
42:29we can then compute this value of the
42:32relationship which is a is a
42:34is a summary if you like it is it's a
42:37dollar value that is directly observed
42:39in the data and what do we find
42:42uh just the script maybe this is comes
42:44out of the raw data are a couple of
42:46interesting facts well the first one is
42:48that uh these values are very large okay
42:51so if you for instance for the typical
42:53seller I believe this is about uh 300 if
42:57I remember correctly of the yearly
43:00Revenue in the relationship so if you
43:02think that you make a 10 margins on uh
43:06on on your relationship uh sorry uh uh I
43:09guess I'm getting it from 300 of the
43:11weekly volume in the relationship
43:13right so it would be six percent of the
43:16yearly Revenue in the relationship but
43:17if you make a profit margin of ten
43:19percent that would be sixty percent of
43:21your of the profits that you gained from
43:23that relationship
43:24um that's the the value of the
43:26relationship is equal to 60 of the
43:29profits that you make in a given Year
43:30from a relationship but a very large
43:31number
43:32and the fact that the number is large
43:34suggests right what does it say well it
43:37says that
43:39presumably uh that number is large
43:41because Temptations to deviate are very
43:43large so it means that there are a lot
43:45of trading opportunities that are
43:47probably valuable in this market but
43:49they don't take place because parties
43:51don't have enough relational Capital to
43:53overcome the associated Temptations to
43:55deviate
43:55yeah so the fact that relationship
43:57people are valuable is an indication
43:59that Contracting problems are very
44:01severe the second thing that we realize
44:03is that if you just pull up the
44:05distribution of his relationship value
44:07across groups of firms you see that
44:11um uh you know in a given year but the
44:14patterns is of course all across years
44:16and you know if you put this stuff in
44:17regressions you see that new
44:19relationships have generally speaking
44:21are less valuable than established
44:23relationship
44:25that of course could be due to selection
44:27and it is indeed the case that
44:28relationship that don't survive
44:30look to be less valuable according to
44:32this uh this measure that relationship
44:34that survived but it turns out that even
44:36if you control for selection
44:38relationship value increases over time
44:41so in this industry what happens is that
44:43over time as David and I kind of keep
44:45trading together maybe in the first year
44:47will be able only to stretch by a
44:50certain amount how much we can sell it
44:51to each other at Valentine's Day the
44:54year after he trusts me a little bit
44:56more I trust him a little bit more will
44:58be able to stretch more and more and
44:59more and more okay because that's
45:01essentially our proxy for uh
45:03relationship value
45:05and then finally we hand over use a
45:09sudden Supply shocks that prevented
45:11firms from being able to harvest flowers
45:14um this was due to a post-electional
45:16violence to kind of trying to understand
45:18okay there are these relationship we
45:20appear to be very valuable their value
45:23increases over time what could be the
45:25underlying mechanism that generates
45:27these forces
45:28and in principle there are many uh
45:32models of a relationship uh so many
45:35models of relationship between predict
45:36that the relationship value is
45:38stationary okay so the simplest model
45:40like the repeated prisoner dilemma that
45:42I mentioned before or models of
45:44relational Contracting which set an
45:46assumption essentially transference
45:47neutrality
45:49actually would predict that the
45:50relationship values should kind of be
45:52stationary over time
45:53we consider different models that could
45:55potentially imply an increasing value of
45:58relationship over time and one such
45:59model is a model in which parties learn
46:03about each other over time maybe the
46:05buyer is learning about the seller's
46:07reliability so think of reliability as a
46:10time over time
46:12so the seller keeps sending flowers
46:13there's never been a problem so over
46:15time I become more optimistic about the
46:17seller reliabilities over time I'm
46:19willing to place larger and larger order
46:21or to rely more and more on these
46:23suppliers at a Valentine Day
46:26one way to test whether that's the case
46:29is that one implication of of uh of
46:32reputational incentive of course is that
46:34reputational incentives are very strong
46:37when parties know each other a little
46:40bit but not too much yeah because if uh
46:44if you have already traded for a very
46:47long time with the seller at that point
46:49you're presumably quite confident about
46:51their type okay
46:52so at that point is a is uncertainty
46:55over the seller type in this theory
46:57model that kind of create reputational
47:00incentives and so a prediction of a
47:03model there is no reputation is that
47:04when there's a shortages of flowers and
47:07you need to prioritize among your
47:09different buyers
47:11you need to choose which buyers to
47:13prioritize
47:16very young buyers or buyers that you
47:18haven't traded a lot with
47:21International not very valuable yet
47:24because they don't trust you that much
47:25and right we are at the early stages of
47:27the relationship and the relationship is
47:29not very valid so maybe there's no lives
47:31incentive to try to protect those given
47:33that you have to give up on something
47:35buyers conversely that uh have traded
47:38with you for many years uh in fact
47:41they've had maybe 1 000 transaction
47:42because every year they have like 200
47:44transactions of flour they've all
47:46received already flowers from you one
47:48thousand times
47:49they're not going to Fair much about
47:51your reliability during the time in
47:53which you really struggle to to produce
47:55and so it is really both buyers with
47:58whom you already have a somewhat
48:00valuable relationship
48:02but not you haven't yet established your
48:04own reputation that you'll try to
48:05prioritize during the time of Crisis
48:08okay now of course so it turns out that
48:10this is what we find in the data so
48:12namely that there is embedded u-shaped
48:14relationship between the experience in
48:16the relationship how much we have traded
48:18in the past and whether the buyer gets
48:20prioritized at the time in which the
48:22sellers need to make these choices
48:24because of an exogenous supply shop
48:25there's other evidence that sort of
48:27corroborates this idea that in this
48:29market it seems that one reason why
48:31relationships are important is because
48:33probably trading directly over you know
48:36remove transaction costs you don't have
48:38to send the flowers to the Netherlands
48:39but when once that happened it exposes
48:42party to lack of contract enforcement
48:43and two opportunities and one model that
48:47is consistent with the value of
48:48relationship with the idea that we kind
48:49of grow over time uh confidence in our
48:52party's uh partner reliability okay
48:58now in this uh yes could I just um ask a
49:01couple of Q A's yes thanks for there's a
49:05couple of people who've been asking
49:06about uh the role of intermediaries we
49:09see them as being very common in
49:10developing countries do they play a role
49:13in long-term relationships and a second
49:15question more directly on the flowers
49:17paper is you know what are the sellers
49:19getting out of their long-term
49:20relationships is it a promise to buy at
49:23low season or when margins are bad
49:26yes banks these are uh these are both uh
49:28great uh great question okay on the
49:31first one
49:32yes intermediaries are indeed uh very
49:34important uh anecdotally in many
49:38industry sector in the industry that
49:39I've looked at I think one one
49:42possibility is that uh another
49:44implication of uh of these models of uh
49:47reputation is that there are economies
49:49of scales in reputation in many contexts
49:52um and uh so it could be that uh given
49:55that many firms in developing countries
49:57are small
49:58the more firms or left SQL are at the
50:01disadvantage in trying to uh create a
50:03reputation is that something that I may
50:04be mentioned again in the last slides as
50:06an open area for for future work
50:08but if you grant me that this
50:10speculation is correct uh this kind of
50:12fear informed speculation is correct
50:15then there is a natural role for
50:16intermediaries because to the extent
50:18that intermediate is aggregate business
50:21uh they might sort of have a slightly
50:23larger scale and they might have an
50:25advantage in uh sort of a sustaining
50:29relationship and sustaining reputations
50:30in relationship and in fact there's some
50:32really work for instance soprana Bardon
50:34and dilip Mukherjee and the third
50:36quarter but unfortunately I don't
50:38remember the model of this uh in the um
50:41AJ
50:43um a few years back
50:45um so yes intermediaries are very very
50:47important and uh you know one
50:49possibility there might be many other
50:51reasons of course right so repetition is
50:53not the you know Contracting problems
50:55that make reputation valuable I'm not
50:56the only thing that is going on
50:58um but that could be a reason and then
50:59there was the second question instead
51:01which was about yes what do these uh
51:03exporters get from relationships
51:06yeah so um
51:09the pair kind of benefits by Saving on
51:12one way to think about the the Kenya
51:14context that the pair uh benefits by uh
51:18saving on transaction costs
51:19that is what generates the events in the
51:22relationship then how the rents kind of
51:24materialize for the business right how
51:26we split both those those gains that
51:29will vary across relationships uh one
51:31way would be to Simply share some of the
51:33gains in the forms of higher prices I
51:36believe my court or I mean as a more
51:38recent paper looking at the Ethiopian
51:40data in the flower industry and it does
51:41find that prices inside relationship are
51:45slightly higher than the average spot
51:46market price that wasn't the case During
51:49the period That We examined with the
51:51Kenya data if I remember correctly so it
51:54could be that the value comes in the
51:55form of an assured Demand right or a
51:57very stable prices for instance so it
52:00might come in the form of some kind of
52:01insurance either demand insurance or
52:04more stability in cash flows for the
52:06seller
52:07yeah
52:08any other
52:10and actually this is an important
52:12question because in in different context
52:15of course the value of The Sword and the
52:18lying source of the value of the
52:19relationship will be different but for
52:21instance in the flower paper right it's
52:23insurance
52:25against shock was the main source of
52:27value to be in a relationship for the
52:29seller we shouldn't see these
52:31relationship we should see probably a
52:33downward sloping relationship between
52:34reliability and experience because an
52:36insurance model will say well you know
52:38when the relationship is more valuable
52:39that's where you're gonna get more
52:40Insurance yeah so in a way the value of
52:43the test here is to distinguish the
52:44reputation the acquisition of a
52:46reputation for reliability
52:48now you have people that trust you and
52:49then that wouldn't otherwise unless they
52:51were built from instead another possible
52:54source of uh of volume is relationship
52:56which could be insurance which doesn't
52:57seem to be the the primary force
53:00in this context
53:04I need to depend on what am I doing on
53:06time
53:07um
53:08so
53:09um
53:10with the without improving we have
53:12another paper that uh um it's brought a
53:15similar intuition so the idea of using
53:17shock to reference prices to uh ask a
53:20couple of different questions relative
53:22to the flower paper
53:25um in most theoretical models cheating
53:27actually does not happen along the
53:28equilibrium path there is no cheating in
53:30the flower bed maybe there's a lot of
53:31cheating in the industry but we were not
53:33equipped to study whether cheating
53:34happens however it's important to uh to
53:38know whether cheating happens in
53:39practice first of all it's important to
53:41know whether our models are accurate
53:43given that the majority of models
53:45implies that cheating does not occur in
53:46equilibrium
53:48but second and more importantly
53:51if cheating is a problem really what
53:53theory tells you that you can't just
53:56detect cheating and unfair what's the
53:58cost of this Contracting problem from
54:00The observed amount of cheating because
54:01parties will presumably anticipate the
54:04possibility of cheating and they will
54:06take precautionary actions and so the
54:08big cost of cheating comes probably in
54:11the form of its precautionary actions
54:12that are taken to mitigate the exposure
54:15to cheating
54:16well of course it's difficult to study
54:18cheating empirically for a couple of
54:19reasons the first one is that typically
54:23data set records the trade that takes
54:25place not the trade that was supposed to
54:27take place so we don't really know
54:29whether uh there was if you like a
54:32contractual default okay and then second
54:34even if we don't observe contractual
54:36defaults we wouldn't know whether the
54:38default occurs for forced mature reasons
54:41as opposed to instead opportunistic
54:43Behavior by one of the two parties
54:46so with Arthur we study the coffee
54:48Market the international coffee markets
54:50we have access to nice data on uh um
54:53International uh forward contracts in
54:57the international coffee Market
54:59these are essentially the contracts for
55:01which very large firms I'll come back to
55:03to this point below so at least for
55:05developing countries we are talking
55:06about large firms
55:09um in our sample the average firms has a
55:12revenues of one million dollar per year
55:13okay so these are not the micro firms
55:16that uh that that we have been talking
55:18maybe about most of the time yesterday
55:19or or earlier
55:22um so these are large firms that sell
55:24coffee through forward counters the
55:26former country is essentially a promise
55:27that I'll deliver coffee to David I can
55:30say in six months okay
55:32and we're going to exploit this
55:33intuition that the default will happen
55:34when market conditions change
55:36sufficiently to place a relationship
55:38outside its self-enforcing trade so
55:42essentially we go back very much to the
55:44same type of incentive compatibility
55:46constraint that we had before this is
55:48the quantity that I've promised to David
55:50this is the price in the contract with
55:52David this is now the price the world
55:54market price for coffee okay and the
55:57nice feature of coffee of course is that
55:59uh these are forward contracts so uh
56:02with a certain delivery date so what
56:05matters is the spot market price and the
56:07delivery date relative of course to our
56:09expectation which are observable because
56:12at the time in which we wrote the
56:13contract we know what is the future
56:15market price quoted for the delivery day
56:17and so the seller roughly speaking will
56:20default on the contract okay if market
56:24prices vary in such a way that this
56:26incentive compatibility constraint can
56:27no longer be satisfied
56:29so you see that the prediction here uh
56:32well there are two different type of
56:34contracts in the market
56:35there are fixed right contracts in which
56:37the price in the contract is uh oops
56:40sorry I moved
56:43all right
56:47uh there are fixed price contracts in
56:49which the price in the contract is fixed
56:52at the time of Contracting
56:55and then there are so-called price index
56:57store contracts on differentials in
56:59which David and I agree that at the time
57:01of delivery he will pay me a price that
57:04is equal to the word Price Plus a fixed
57:06differential
57:07immediately clear trade-off right the
57:10fixed price contract is good because it
57:12gives a lot of stability I know how much
57:14I'm gonna get paid I'm not exposed to uh
57:16to the this world market price
57:18fluctuation which is something that uh
57:20firms in developing countries can't
57:22really ensure against
57:24at the same time the problem is that uh
57:27the fixed price contracts potentially
57:30creates this problem that if market
57:32prices changes a lot then the seller
57:36might want to default on the forward
57:37contract
57:41so we have this prediction and if you
57:43just look at the data that's exactly
57:44what we find in the data again this is
57:46raw data
57:47so what we plot is the percentage of
57:49default the likelihood of a contractual
57:52default
57:53okay
57:54against these price surprises which is
57:57the ratio between the spot market price
57:59of delivery divided by the future
58:02markets for the delivery date quoted at
58:05the Contracting date
58:07so a ratio of well above one means that
58:09relative to what we were expecting the
58:11spot market price will be at the time in
58:13which uh I'm supposed to deliver the
58:15copy the spot market price is increased
58:17a lot and as you can see
58:19there are it's few contracts that have
58:21this very large uh increase in uh in
58:24Market uh in market prices of these were
58:26contracts that were assigned at the
58:27times in which there was a coffee Market
58:30World prices kind of increased a lot
58:32bones are the contracts in which
58:34defaults are concentrated and this by
58:38the way only happens as the theory
58:40predicts if the contract is a fixed rise
58:42contract as opposed to a differential
58:44price contract of course this is the raw
58:46data we need to put this into
58:47regressions and so forth
58:50it looks like uh we have a couple of
58:52different exercises that our estimate
58:54suggested probably about 50 of the
58:57observed defaults in this market are
58:59strategic in nature okay so cheating is
59:02a real possibility it looks like sellers
59:04can't really be fully trusted
59:06and so one prediction of of
59:10implication of this is that if you think
59:12about which kind of contracts will form
59:15end up receiving very much like in the
59:18spirit of water of energy and do flooded
59:20in the software industry
59:22if you have a lot of relationship value
59:24with their with your buyer right your
59:27buyer trust a lot that even if there's a
59:30big change in the price you're not gonna
59:32default then you're gonna be given a
59:35fixed price contract
59:37If instead the relationship value is
59:40more the buyers say look I can't trust
59:43you Rocco so I better give you a
59:45differential price contract
59:47okay
59:48so that's an implication of the theory
59:50we can test this implication uh uh in
59:53the data and we find support for this
59:55but notice that this has a couple of uh
59:57interesting uh implications the first
1:00:01one of course is that uh the uh it's the
1:00:04better relationship that sign fixed
1:00:06price contracts right so
1:00:09they test so we detect strategic default
1:00:12in a way on those relationships that are
1:00:13very strong and that because they are
1:00:15very strong they trust each other and
1:00:17they border already they're there to
1:00:19sign fixed price contracts but
1:00:22contractual default of course the big
1:00:23cost of contractual default is not the
1:00:26few contracts that gets defaulted is the
1:00:28fact that many forms potentially remain
1:00:31exposed to price risk because they don't
1:00:33even get a fixed price contract to begin
1:00:34with so here are the possibility of
1:00:36contractual default for exposes these
1:00:39very large firms to an insurance a
1:00:41missing Insurance Mark okay
1:00:47so I now want to spend uh maybe uh five
1:00:51minutes to relate a little bit what we
1:00:54have learned from these uh two studies
1:00:57um to what we discussed in uh what was
1:01:00discussed by Dave and David and and Pete
1:01:02particularly and and we'll set an extent
1:01:04to Chris
1:01:06so in both models Super Paper kind of
1:01:09really brings out the importance of
1:01:11thinking about this relationship value
1:01:13yeah
1:01:16and it seems that in both in contact
1:01:19this wheel matters we can quantify it is
1:01:22large and uh it drives both the
1:01:24contracts that people sign the firm sign
1:01:27and the way they responds to shock
1:01:31if you think about the flower paper one
1:01:33implication of the flower paper is that
1:01:35the amount of flowers that the firm sell
1:01:37is constrained by this bill by the Viva
1:01:40the fair Mass with the buyer
1:01:43survivability relay exactly two wedges
1:01:45that they've talked about and that I
1:01:47think are very implicit uh were not so
1:01:50implicitous they were quite explicit in
1:01:52Pete's discussion as well
1:01:53so in other words Contracting problems
1:01:55give us a lens or give us if you like a
1:01:58micro Foundation or a way to think about
1:02:00wedges and in particular to think about
1:02:03heterogeneous wedges across sperms
1:02:06in the context of flowers what do I mean
1:02:09by what of course is the fact that you
1:02:11know when I'm solving the the the the
1:02:12the the the the the the problem of the
1:02:15firm right the firm faces a constraint
1:02:17they are not able to sell as much as
1:02:19they would like so they can't really
1:02:20equate marginal revenues to Maxine of
1:02:22course that's the sense and the gap
1:02:25between marginal revenue and marginal
1:02:27cost will depend on this bill which is
1:02:29presumably heterogeneous across firms
1:02:31and so the weight will also be
1:02:32heterogeneous acrosphere
1:02:34now of course in the context of the
1:02:36flower paper you might say but Rocco you
1:02:38told us that these firms also sell on
1:02:39the spot market so it could be that the
1:02:42marginal flower is always sold to the
1:02:44spot Market in which case everybody face
1:02:46the same price there's no web
1:02:48well if that was the case that would be
1:02:50true however notice that the average
1:02:53revenue right still will depend on the
1:02:55mix of views that you have in your
1:02:57client portfolio and so at the very
1:02:59minimum it would be the case that the
1:03:01connection between the average uh
1:03:04product revenue and the marginal product
1:03:06Revenue will be different that
1:03:08connection essentially will be different
1:03:10across firms okay the two will no longer
1:03:13be the same differentially so across
1:03:15firms
1:03:17as I mentioned uh so this just to
1:03:20emphasize the idea that this type of
1:03:22Contracting problem or starting this
1:03:24type of Contracting problem and kind of
1:03:25trying to understand what is v and what
1:03:27it comes from give us a handle to think
1:03:28about wedges potentially
1:03:31second we mentioned that relationship
1:03:33appear to be very valuable so that that
1:03:35implies that the many valuable trading
1:03:37opportunities do not take place uh and
1:03:40that there are these very large distort
1:03:42potentially large distortion on the
1:03:43demand side even for these very large
1:03:45firms
1:03:46both the flower exporters and the coffee
1:03:49exporters are very large firms for
1:03:51developing country standards
1:03:54and then finally if you think about the
1:03:55flower paper
1:03:57we identified this idea about reputation
1:03:59intended as beliefs about the seller
1:04:01reliability matter is one of the sources
1:04:03of volume is relationship if these
1:04:06beliefs matter then try your beliefs
1:04:08matter
1:04:09but then prior beliefs by necessity do
1:04:12not come from uh interactions that we
1:04:13have had with that particular partner
1:04:15right that must come from perception
1:04:17that you have about the Kenyan exporters
1:04:20more General and so that also suggests
1:04:22that uh Eve reputation is important The
1:04:25Logical implication of that is that
1:04:27there should be informational
1:04:28externalities in the form of potentially
1:04:30Collective reputation that also matter
1:04:32and here there is a GMI with a couple of
1:04:35other quotas have a really really nice
1:04:37paper on restart on uh on a collective
1:04:40reputation in the uh China final Dairy
1:04:43sector but I believe that a lot more
1:04:45work should be done to think a little
1:04:47bit about the policy implication
1:04:48particularly in export markets of uh of
1:04:51uh of this type of reputational models
1:04:54the the paper with our tour the one that
1:04:56detects cheating kind of goes a little
1:04:58bit further in uh in in these in
1:05:00establishing connection with some of the
1:05:02lectures that we talked about before and
1:05:03it also because there is a model that is
1:05:06kind of calibrated okay and so it's a
1:05:08model that reveals that for these very
1:05:10large exporters the possibility of
1:05:12defaulting on these contracts reduces we
1:05:16estimate their output by 10 to 15
1:05:18percent on average okay but what is
1:05:21interesting is that uh what the what the
1:05:25uh what what the what the paper
1:05:27highlights is that within the same
1:05:29industry
1:05:30the type of Market imperfection that the
1:05:34firm face
1:05:35is very heterogeneous yeah so yesterday
1:05:38Chris gave you a framework uh thinking
1:05:41about credit constraint and thinking
1:05:43about it sooner costing both being
1:05:45reason why marginal product of capital
1:05:47remains above the interest rate
1:05:49in this context we have determined that
1:05:51for these very large firms some firms
1:05:54have a lot of relational Capital with
1:05:56their buyers they are trusted a lot
1:05:57there's essentially no problem there
1:05:59they are unconstrained they achieve
1:06:00first best
1:06:01problems that have medium relational
1:06:03Capital being a credit component they'll
1:06:06get price Insurance yeah but they might
1:06:09not be able to get all the funds
1:06:11necessary to to to to to to to exhaust
1:06:14essentially the trade with the buyer
1:06:15because the amount of trade is
1:06:16constrained by this relational value
1:06:19and it will firms are kind of new or
1:06:22they don't have a lot of relational
1:06:23value with their customers then in that
1:06:26case they face an insurance constraint
1:06:28yeah so even if you give them money they
1:06:31would not want to get more of his money
1:06:32because they are remain exposed to to
1:06:35price risk so we have kind of a context
1:06:37in which if we really go deep and try to
1:06:39understand the underlying Contracting
1:06:41problem we see that actually the market
1:06:44that is missing can be heterogeneous
1:06:46across firms and it's kind of endogenous
1:06:48okay so we shouldn't maybe think about
1:06:51uh Missing markets in an exogenous sense
1:06:54or you know the credit Market doesn't
1:06:55work well because within firms within
1:06:58industry firms will face very different
1:07:00constraints
1:07:01I also think that these uh a couple of
1:07:03things I I think it's remarkable that we
1:07:05find this evidence on firms that are so
1:07:07large because most of the evidence for
1:07:09this type of distortion I think comes
1:07:11from from smaller firms uh at least a
1:07:14capital Distortion that Chris talked
1:07:16about even if you think about the banner
1:07:17you do flow
1:07:18uh paper in restart which is a fantastic
1:07:21paper testing for credit constraint
1:07:24um the firmware the typical loan size is
1:07:26about ten thousand dollars the typical
1:07:27loan size of the scope export is like
1:07:29four hundred thousand dollars so these
1:07:31are very large firms
1:07:32the second thing is that again thinking
1:07:34carefully about what's the source of the
1:07:36market failure is bonds implication for
1:07:39example it is policy implication if
1:07:40you're trying to develop uh tools that
1:07:43help these large firms to add risk
1:07:46the firm is borrowing money actually The
1:07:49Firm doesn't have a very large demand
1:07:51for adding this price risk uh because
1:07:53essentially gambling with the with
1:07:55borrowed money okay
1:07:56and in the paper we also I don't have
1:07:59time to cover it in detail here but we
1:08:01also kind of try to do a little bit of
1:08:03reduced form exercise to test for a
1:08:05credit constraint using a regression
1:08:07discontinuity design because we know how
1:08:09much working Capital Loans these firms
1:08:11are given and we do find evidence that
1:08:13some of these firms are credit
1:08:14constrained and when that happens the
1:08:17fact that the firm is credit constraints
1:08:19because of this Contracting problem
1:08:20comes stream actually tends to lower the
1:08:22price that the firm can pay to its
1:08:24suppliers and the suppliers are farmers
1:08:26so what that's suggestive I mean this is
1:08:28very prediminary evidence it's all
1:08:30independent is that Contracting problem
1:08:32up the export gate actually kind of
1:08:35reduce welfare weighing up further up
1:08:38the chain in this context and impose
1:08:40externality on the small older farmers
1:08:43yeah
1:08:45okay so so far we have uh I've tried to
1:08:49argue that relationship in many
1:08:51widespread in many contexts
1:08:54we can learn a lot if we think about
1:08:56relationship there's a lot of work on
1:08:58this we can learn quite a bit I think
1:09:00specifically if we think hard about
1:09:01taking Dynamic incentive compatibility
1:09:03constraint to the data
1:09:05we have discovered at least when we were
1:09:07able to do that that relationships are
1:09:09very vulnerable
1:09:11we are then maybe trying to make this
1:09:13connection that is the relationship
1:09:15values are heterogeneous across firms
1:09:17even across relationship and they might
1:09:20give us a land
1:09:21to microphone and even endogenize wedges
1:09:24uh and and marketing Perfections that
1:09:27have been discussed in in the previous
1:09:29lectures and then I gave you a couple of
1:09:31examples uh for lack of time I don't
1:09:33give you more that kind of really
1:09:34understanding the source of the market
1:09:36imperfection might also heal their
1:09:38specific policy uh implementation I
1:09:41think this is again a good time to take
1:09:43a little bit of a pause uh before I move
1:09:45on to to the next session and and to
1:09:48concluding remarks
1:09:51so let me take a look at the Q a and if
1:09:54you have questions I've got a fantastic
1:09:56day yeah
1:09:58thanks yeah um yeah I'll read out um
1:10:02one question from the Q a ah okay yeah
1:10:06so it comes from Luis
1:10:07um is asking um a common complaint by
1:10:11small producers in developing countries
1:10:13is about intermediaries keeping most of
1:10:15the value generated at their expense
1:10:18usually I would think of this as a
1:10:20consequence of
1:10:22intermediaries Market power
1:10:24can one make a theoretical case for
1:10:26their High share being based on the fact
1:10:29that it is
1:10:31they who invested in a long-term
1:10:33relationship with the final buyers
1:10:36I I so I don't know if Dave I speak this
1:10:39question because he knows I love this
1:10:40question please this is a great question
1:10:43not only a theoretical case can be made
1:10:46I think that we need to try to make
1:10:47empirical kind of tests for whether
1:10:49that's correct or not there's recent
1:10:52work I'll mention it later on in the
1:10:54slide so this is actually exactly what's
1:10:55coming up not with intermediarism
1:10:57foreign
1:11:02of course interested in uh kind of
1:11:05understanding how Market powers of these
1:11:07intermediaries the process price is paid
1:11:09to Farmers okay this is work that
1:11:11deploys IO tools uh Michael Rubens has a
1:11:14wonderful paper in the Chinese Top
1:11:15Tobacco industry uh also has a paper
1:11:20um well Lauren Falcao I think Lauren
1:11:22will talk a lot about those issues
1:11:24um a lot of these papers which are
1:11:27absolutely fantastic
1:11:29um circles if you like on a specific
1:11:31Market Market affection which is Market
1:11:33power yeah concentrated wire Downstream
1:11:36they have Market Powers markdown farmers
1:11:38get lower price
1:11:41one could make the argument however that
1:11:43is uh in a lot of these agricultural
1:11:45chains uh price is not the only thing
1:11:48that matters right Farmers might receive
1:11:50loans from their buyers uh Farmers might
1:11:53receive insurance they might receive
1:11:55inputs and so forth and it could be that
1:11:57a certain degree of Market power is
1:12:00needed in order for these buyers to
1:12:02provide these services to the to the to
1:12:04the farmers and if that's the case
1:12:06indeed there's going to be a trade-off
1:12:08and that's exactly what we're going to
1:12:09look at now in the next slide okay
1:12:12uh
1:12:14I had a couple of other questions so
1:12:15since Louise already allowed me to talk
1:12:17about the next slide let's see so
1:12:18Michelle can we think about the
1:12:20relationship between firms in developing
1:12:21countries and their suppliers in the
1:12:22same way
1:12:25what is the V2 suppliers in this case uh
1:12:29I think so the the example I I I if it's
1:12:33true that in terms of methodology none
1:12:36of what I've said in the flower for the
1:12:39flower paper or the coffee example kind
1:12:41of specific to developing countries but
1:12:43in both cases the context our developing
1:12:45countries exporters towards richer
1:12:47countries uh so I think conceptually yes
1:12:49we can definitely apply uh I think very
1:12:52similar conceptual framework uh to
1:12:54develop to develop in countries uh I
1:12:56think that if that was the question now
1:12:59of course uh if one was able to do the
1:13:02same study in the same industry rich and
1:13:04poorer country would be very will be
1:13:05fascinating right to compare whether
1:13:07relationships are more prevalent in
1:13:09poorer countries and in richer countries
1:13:10whether
1:13:12um the relationship value is larger in
1:13:14developing countries and in richer
1:13:16countries and so forth but unfortunately
1:13:17I'm not aware of people yet that have
1:13:19attempted that that exercise
1:13:22in the in the coffee paper we had some
1:13:25cross-country variation because the
1:13:26exporters come from different countries
1:13:28but coffee is only produced by
1:13:30developing companies and to be fair
1:13:31although we were pushed to look for
1:13:33correlation between institutional
1:13:35quality of the exporting countries and
1:13:37relationship value
1:13:38I I'm not comfortable kind of
1:13:40extrapolating any of the patterns that
1:13:42maybe appear in figure 15 of the paper
1:13:44yeah but it's a great question
1:13:47um Rocco maybe I can just summarize the
1:13:49other two um yes uh very I mean this
1:13:53maybe will come later but both uh John
1:13:55and guruk are um interested in asking
1:13:59you to explore say a bit more about
1:14:00policy implications John is talking
1:14:02about sort of a
1:14:04efforts to improve courts and Industrial
1:14:06Court systems Google Rook is um
1:14:09wondering what policies would increase
1:14:11incentives for firms to write better
1:14:13contracts
1:14:15okay these are if I can let me go and
1:14:19then there are the next break I'll
1:14:21answer it not the answer the answer is
1:14:23that I don't know but I think those are
1:14:25both great questions I have some
1:14:26thoughts about it and I think that they
1:14:28are kind of priority areas for for for
1:14:30work really and I can think also I mean
1:14:33one could imagine some experiments and I
1:14:35know that some experiments are ongoing
1:14:37um some experiments with the conferences
1:14:40Chris is doing I think work in Mexico
1:14:41that is fascinating with courts a labor
1:14:43contract uh there's also work other type
1:14:46of experimental work with courts not
1:14:48necessarily focusing on relationship uh
1:14:50we could imagine digital platforms
1:14:52sometimes helping with contract
1:14:53enforcement at least kind of with the
1:14:55spread of information and reputational
1:14:56incentives and I think that there is
1:14:58ongoing work also on on that
1:15:01um
1:15:01there are also it would be fascinating
1:15:03to put together the data to exploit
1:15:05either cross-sectional variation in
1:15:07chords performance like in India in
1:15:09Brazil there are nice papers that use
1:15:10that variation with kind of firm to form
1:15:12data to kind of try to understand how
1:15:15the effectiveness of chords impact the
1:15:17relationship I know China has enacted
1:15:19reforms in that sense but I don't have
1:15:21the Chinese debt but if anyone uh has
1:15:24good Chinese data if that might be that
1:15:26the right place kind of to to try to
1:15:28tackle uh tackle this question have
1:15:30inside that as we'll see in a moment I
1:15:33think the implication of the world in
1:15:36which relationship are important
1:15:38the theoretical implication is not that
1:15:40perfect or improving contact enforcement
1:15:42will necessarily improve Market function
1:15:45okay and we'll get back to this in a in
1:15:48a sec
1:15:50Okay so
1:15:52so we have talked about these bees right
1:15:54and so far I've simply said well you
1:15:56know Veronica is this relationship value
1:15:58they take them as given we have tried to
1:16:00learn about them how large they are uh
1:16:03whether they correlate with contract
1:16:04choice or nothing but so far I've kind
1:16:06of taken them as exogenous okay but a
1:16:09nice feature now Reviving in the v is
1:16:12that we can start to think about whether
1:16:13to what extent they are endogenous okay
1:16:15now even when if they weren't exogenous
1:16:18we can write down models stick them in
1:16:21our modern as wedges right that's
1:16:22probably like almost as Iceberg cost or
1:16:24like similar and then you know we can
1:16:27quantify our model taking into account
1:16:28uh these uh this watches and you know it
1:16:31would be useful to know how large and
1:16:33support but we can take treat them as
1:16:34parameters but of course they are not
1:16:36parameters right because they are is the
1:16:39net present value of the relationship so
1:16:40they really depend on many things and
1:16:43among the many things they depend on
1:16:45Market structure okay so that's why now
1:16:47I want to think a little bit about
1:16:48Market structure and relationships
1:16:51for marketing developing countries are
1:16:53often portrayed as this functional okay
1:16:55and so there is this perception that
1:16:57increase competition will be able to
1:17:00improve firm performance and management
1:17:02and there is abundant evidence that's
1:17:04actually the case so I believe
1:17:06please don't go out and say oh I was
1:17:09following these scores uh you see a
1:17:11bread and there was this crazy economies
1:17:13that say that competition is bad I
1:17:14believe competition is good okay
1:17:17and he said that if relationships are
1:17:20important and relationship relies on
1:17:22future rents
1:17:24if competition reduces ranks
1:17:28or increases temptation to deviate in
1:17:31certain contexts competition could be
1:17:33harmful right that's obviously this is a
1:17:35theoretical argument that has been made
1:17:37the theoretical argument that equals a
1:17:39little bit the argument that Lewis was
1:17:41asking in his questions about
1:17:42intermediaries yeah
1:17:45so with the meter we have done a paper
1:17:47where we have kind of tried to test
1:17:49these uh this hypothesis
1:17:51specifically in the context in a context
1:17:54in which relational contracts are
1:17:56important okay so maybe if you are in a
1:17:58market in which the restaurant contracts
1:18:00are not so important uh you know we will
1:18:03expect the usual uh effects of
1:18:05competition to play out
1:18:07so to do that we needed to
1:18:11uh we need to so we basically asked in
1:18:14this paper uh um the following question
1:18:17we asked those competition causes
1:18:18relationships to break down
1:18:21and if so what are the consequences of
1:18:24relationship breakdown for Market
1:18:25functions and for participants yeah and
1:18:28then ideally also what are the
1:18:29mechanisms through which relationship
1:18:30competition causes relationship to break
1:18:33down well
1:18:34um this is a tricky uh this is a very
1:18:37tricky set of questions to go after
1:18:38because well first of all you have
1:18:40measurement problem you need to go out
1:18:41there and measure this relationship that
1:18:43we do a survey fine
1:18:45Beyond of course is that uh competition
1:18:47is an indulging of alcohol right is a is
1:18:51um well his Market structure which is
1:18:52endogenous so
1:18:54so what we tried to do we tried to look
1:18:58for a context which perhaps will happen
1:19:00to be the Rwanda coffee sector in which
1:19:03relational contracts are important and
1:19:05in which we might be able to come up
1:19:07with a relatively good instrument for
1:19:11competition okay so we started the
1:19:13Rwanda coffee sector we're at about 400
1:19:15000 smaller than Farmers scattered
1:19:17around the country
1:19:18and then the government kind of um
1:19:21the secretary essentially is liberalizer
1:19:23one has been doing very well so there's
1:19:24been a lot of kind of coffee washing
1:19:26station or coffee meals this is the
1:19:28first stage processing of coffee that
1:19:30I've been entering the sector and that
1:19:33because coffee needs to be processed
1:19:35within hours of our best this meals are
1:19:39all scattered in the countryside yeah so
1:19:43uh well I'll show you a map uh maybe
1:19:45later on now here is a map of Rwanda I
1:19:47don't know if you can see the the yellow
1:19:49dots are where the meals are located
1:19:51yeah you see the meals need to go very
1:19:54close to where the cough is produced
1:19:55because the coffee needs to be
1:19:56transported to the meal within hours of
1:19:58being harvested so that gives us nice
1:20:00geographical variation there will be
1:20:03some means that face a lot of
1:20:04competition because they are in zone
1:20:06that are very densely populated by means
1:20:08relative to the amount of coffee that
1:20:10can be produced and means instead maybe
1:20:12that are a little bit more isolated and
1:20:14therefore face a little bit less
1:20:16competition
1:20:17so we do a serving of all the means in
1:20:19the sector and of random samples of
1:20:21farmers around these means and to kind
1:20:25of try to measure directly the extent to
1:20:27which we rely on relational contract
1:20:29okay so which relational contracts here
1:20:31means that before harvest the meal gives
1:20:34inputs to the farmers at harvest the
1:20:38farmer sells on credit to the meal which
1:20:40is very important because this meals a
1:20:42very large working capital requirements
1:20:44so they tend to be credit constrained so
1:20:46if the Farmers Supply on uh on credit
1:20:48that's relaxes declarative constraints
1:20:50for me and then finally uh whether after
1:20:53our best meal makes a second payment so
1:20:55we pay the Trade Credit and provide
1:20:57others and basically what you find is
1:21:00that first of all these measures of
1:21:02relational Contracting so the extent to
1:21:04which farmers and meals rely on these uh
1:21:07interlinked transactions that are
1:21:09clearly non-enforceable in this
1:21:10contracts
1:21:12um very strongly correlates with meals
1:21:13performance yeah unit cost capacity
1:21:15utilization
1:21:18and then we kind of try to construct
1:21:19this instrument for competition not
1:21:22lifting that formula
1:21:24conditional on the tweetability for
1:21:27coffee cultivation in Milan catchment
1:21:29area
1:21:31the suitability around the mill
1:21:33catchment area will drive placement of
1:21:36other competing meals but potentially
1:21:39not performance of the meal itself since
1:21:41the meal doesn't get coffee from the
1:21:42from the green area so if the meal only
1:21:45gets copied and therefore operates only
1:21:47based on the gray area the green Air the
1:21:50suitability for meal placement in the
1:21:52green area will give us potential an
1:21:54instrument for uh competition okay now
1:21:57there's a lot of
1:21:58things that go into this computation of
1:22:00the instrument but let's say that we
1:22:02kind of do come up with an instrument
1:22:04that very strongly correlates with
1:22:06competition
1:22:07and in a review form sense that strongly
1:22:10correlates with the uh with the use of
1:22:12relational Contracting so what we find
1:22:14is that conditional suitability for
1:22:16coffee and milk placement in their
1:22:18catchment areas means that are
1:22:20surrounded by more suitable places
1:22:22that's a that's kind of what help
1:22:24associate a bit with the argument so if
1:22:26you're on a meal that is surrounded by
1:22:27very good places
1:22:29it turns out that you have uh lower and
1:22:32more erratic capacity utilization and
1:22:34therefore higher processing costs so
1:22:36your performance is worse you also
1:22:38produce lower quality coffee we tested
1:22:40the quality in the lab a little bit like
1:22:41Eric was mentioning
1:22:43you you the meal doesn't pay higher
1:22:46prices to Farmers so it's not that
1:22:48competition here translating to Farmers
1:22:49receiving higher prices in fact we find
1:22:52that farmers sell less coffee to all
1:22:55means to one way they are access they
1:22:57process more coffee at home which is
1:23:00very inefficient and is associated with
1:23:01much lower prices yeah
1:23:04the reason why this is interesting is
1:23:06that it provides an example uh an
1:23:09empirical example uh of a situation in
1:23:12which because the two Market
1:23:14imperfections interact right there's a
1:23:18Monopoly power that comes from the fact
1:23:20that these means need to be uh located
1:23:21in the countryside so that's what Luis
1:23:23was saying but at the same time the fact
1:23:25that this Market power might be needed
1:23:28or a certain degree of Market power
1:23:29might be needed in order to sustain
1:23:31these relationships that are then
1:23:33valuable because of all the other
1:23:35services that they produce that they
1:23:37deliver to the farmer yeah
1:23:40this is already used for
1:23:43the paper that kind of tackles Louise
1:23:46question in this context kind of needs
1:23:48to be written so that would need to be a
1:23:50paper right that really takes seriously
1:23:52sort of Market power and Mark Brown as
1:23:54well as potentially the services at this
1:23:56relationship facility
1:24:00there are police implications from the
1:24:01Sun I unfortunately actually they we did
1:24:05this project in partnership with the
1:24:08coffee board in one that they they took
1:24:10the study and decided to try to
1:24:12implement apparently not very
1:24:14successfully zoning regulations yeah
1:24:16because they say oh you know if the
1:24:17problem is competition then it should
1:24:19prevent Farmers to switch across means
1:24:22that I don't think is the policy
1:24:23recommendation that comes out of the
1:24:25study because what the IV is picking up
1:24:27is the effect of an extra meal right is
1:24:29it is the late is the effect of an extra
1:24:31meal well on average means already phase
1:24:33four to five competitors
1:24:35the first best of course would be to
1:24:37have a contact enforcement which is a
1:24:39little bit what the Costa Rica
1:24:40regulation does okay it would be to
1:24:42facilitate farmers and means to rely on
1:24:45formal contract enforcement
1:24:47in sustaining this type of relational
1:24:49companies but the theory tell us that uh
1:24:53imperfect con so in improving contract
1:24:56enforcement will not necessarily improve
1:24:58Market functioning unless it completely
1:25:01eliminates the Contracting problem and
1:25:03the reason is that a formal contract is
1:25:06imperfect right formal contractor could
1:25:09use as the outside option through the
1:25:12relationship breakdown
1:25:13and so should you try to improve formal
1:25:16contract enforcement uh but only
1:25:19partially you might actually end up
1:25:21hurting relationship because what you
1:25:23are doing is that you are improving the
1:25:24outside option of the default option
1:25:26right and therefore preventing the
1:25:28relationship from achieving the service
1:25:30that they need to be uh achieved this is
1:25:33not a character with contract
1:25:35enforcement I don't know of any evidence
1:25:37that has this prediction okay
1:25:39but the logic is similar to the
1:25:42literature that argues that improving
1:25:45for instance information in the market
1:25:46can facilitate collusion and there is
1:25:49evidence for that okay so it's a very
1:25:51similar logic and it's a logic that work
1:25:54through the outside option
1:25:57uh that is used in the relationships
1:25:59okay so I think an implication is that
1:26:02yes we can think about policy policy
1:26:04implication of studies and I think
1:26:05trying to find cheap ways of improving
1:26:07contract enforcement should be a
1:26:09priority but we need to be mindful that
1:26:11fearing doesn't tell us that necessarily
1:26:13improving contract enforcement will
1:26:15improve our very much like improving
1:26:17competition might not improve outcomes
1:26:19or improving information might not
1:26:21improve outcomes yeah
1:26:22because this really comes from the
1:26:25structure of the incentive compatibility
1:26:26constraint and the fact that the outside
1:26:28option as a counter Wheeling Force
1:26:31improving things that improve the
1:26:32outside option reduce rent and makes it
1:26:36harder to sustain relationship okay so
1:26:37if relationships are still needed then
1:26:39you might actually the intervention my
1:26:41backfire
1:26:42Okay so
1:26:45I have to wrap up there's a uh I take
1:26:48two minutes to wrap up the first thing I
1:26:49wanted to say all these slides will be
1:26:51available and of course get in touch if
1:26:54you if you want to discuss any further
1:26:56points yeah first of all I wanted to
1:26:58talk about other works on Market
1:27:00structure and relationship which I think
1:27:03is important nicely Lauren will talk
1:27:06about competition I believe in the uh
1:27:09next week and I will imagine that we
1:27:11should definitely talk about the paper I
1:27:13hope and possibly also about
1:27:17okay so please stay tuned
1:27:22Contracting a free
1:27:27so shocks gets transmitted differently
1:27:29if there are these interlinkages for
1:27:31instance
1:27:32a fascinating paper where the Bangladesh
1:27:36government banned intermediaries in the
1:27:39soil lead chain thinking that these
1:27:41intermediaries at a lot of Market power
1:27:43and uh and we're capturing rents turns
1:27:46out that when you remove this
1:27:47intermediary from the market
1:27:49price transmission decreases
1:27:51and the market becomes less efficient
1:27:53and the reason is because with
1:27:54intermediaries yeah sure they have
1:27:56Market power but they are actually the
1:27:58one that provided Faith credit that is
1:28:00necessary for wholesalers and processor
1:28:03to import seeds so this is a wonderful
1:28:05example of how kind of taking into
1:28:08account Contracting problems really
1:28:10changes the way we think about standard
1:28:12question that uh that are very important
1:28:14in trade and that are you in trading
1:28:16development to to to to understand our
1:28:18Market function for instance possible
1:28:21I wanted to tell you more things but I
1:28:23I'll drop so I also promised that I
1:28:25would say something about open areas I
1:28:28have maybe four or five slides on
1:28:31operators
1:28:32first of all structural models I think
1:28:35that there is a lot of work to try to
1:28:37bring estimate structural model of
1:28:39Contracting problem and relationship
1:28:41there are wonderful examples in IO that
1:28:45are closely related and are mentioned
1:28:47zombie slides there are also some
1:28:49beautiful examples of recent paper
1:28:50mostly from Junior scholars in
1:28:53development that are really really very
1:28:54good and I think I have if I had to pick
1:28:57one it's not about relationship but I
1:28:59think this paper by Nick Ryan I put it
1:29:01as kind of even on the reading list
1:29:03because it's really a wonderful study
1:29:05about taking into account Contracting
1:29:08problem lead us to change our view on on
1:29:11on standard problems and and in this
1:29:14case in particular on efficiency of
1:29:15procrement auction
1:29:17other areas of uh that are open for work
1:29:21I think we need more systematic facts we
1:29:23need to integrate this agenda into macro
1:29:26and trade models uh to enrich them if
1:29:28you like to think a little bit more uh I
1:29:30mean in a more than one's way about
1:29:32wedges if we think that these bees are
1:29:34important
1:29:35I think that if these are important they
1:29:38give V creates a very natural
1:29:42stores of non-convexity we talked
1:29:44already about the fact that there are
1:29:46economies of scale in reputation there
1:29:48are economies of scales and credibility
1:29:50and so it could be that thinking about
1:29:53these give us a handle to understand why
1:29:55firms small firms don't grow why they
1:29:57cannot grow and why even if you boost a
1:30:00lot of intervention not much is going to
1:30:02happen potentially if the problem is
1:30:04essentially create this future rights or
1:30:07the lack of his future rents okay
1:30:09um I have some work that speaks to that
1:30:11but a lot more should be should be done
1:30:13and then this was like a two or three
1:30:17lectures I think we would touch some
1:30:18Advanced topics such as well what's the
1:30:21relationship between uh with the
1:30:23vertical integration and firm boundaries
1:30:25uh we'll talk about sourcing system
1:30:27where do relationships come from so far
1:30:30I've just talked as if we do firms end
1:30:33up having relationships but
1:30:34relationships are managed then they come
1:30:36from somewhere
1:30:37we should talk about relationships
1:30:39within firm because there's not on
1:30:42relationship between firms is the
1:30:43relationship between employers and their
1:30:45employees and those are also maybe very
1:30:48important and will give us a different
1:30:49handle in understanding uh variation in
1:30:52management practices and then of course
1:30:54the collusion so partner talked about
1:30:57relationship that happened to be
1:30:58beneficial for the parties involved in
1:31:00the relationship but there's no there's
1:31:02no reason why this relationship would be
1:31:03beneficial for the market as a whole and
1:31:06of course informal or implicit collusion
1:31:08is a case of colluding firms benefiting
1:31:11from their informal effective and
1:31:13illegal relationship but of course the
1:31:15market as a whole potentially suffering
1:31:17from it and then as I mentioned we need
1:31:19a lot more work kind of trying to
1:31:20evaluate policy concrete policies trying
1:31:24to ameliorate Contracting problems um I
1:31:27think I I'm gonna stop here I'm happy to
1:31:30stay a little bit longer if people have
1:31:31questions and otherwise please do shoot
1:31:34me an email uh if you want to chat more
1:31:40foreign
1:31:44I think uh given our um need for
1:31:48credibility here we will uh have to stop
1:31:50things we've already gone over by a
1:31:52minute despite Rocco's generous offer to
1:31:54uh
1:31:55to chat further but um those who have
1:31:58remaining questions uh you know I I I'm
1:32:01guessing Rocco wouldn't mind answering
1:32:03them on email he's even kind of
1:32:04explicitly offered to do that right here
1:32:06and uh Rocco we're grateful for that but
1:32:08and uh above all else we're grateful for
1:32:11you know everything you've done the last
1:32:1290 minutes this is a I know a huge
1:32:14amount of work for you and it's also a
1:32:16huge value for for us for for students
1:32:19interested in this topic for many years
1:32:20to come so thank you and again thanks
1:32:23everyone for joining us so we will meet
1:32:24again next week on Thursday and Friday
1:32:27at uh 10 a.m EST I hope to see you there
1:32:31and thanks for joining us today bye yeah
1:32:33thank you everybody shoot an email if
1:32:35you want thanks Dave and thanks David as
1:32:37well

About this transcript

This page contains the full transcript of Lecture 6: Contracts with Rocco Machciavello by International Growth Centre, generated from the public captions YouTube serves with the video. The transcript has 15,944 words across 2,472 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.