YT YouTube2Text

YouTube2Text → Transcripts → Lecture 5: Capital and Labour Frictions with David McKenzie and Chris Woodruff

Lecture 5: Capital and Labour Frictions with David McKenzie and Chris Woodruff — Transcript

by International Growth Centre · 16,687 words · 2,388 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

1:45all right
1:53okay let me uh kick things off so uh
1:56thank you very much for joining us for
1:58this fifth lecture as part of our igc
2:00and grad firms and development course uh
2:03we're uh uh very uh excited today to uh
2:07uh be uh uh moving on to uh uh topics
2:12related to Capital and labor so to put
2:14this into kind of perspective in in the
2:17uh first a couple of lectures we talked
2:19a lot about the technology side within
2:21the production function
2:23um then we had a excellent lecture on
2:26misallocation uh thinking about how our
2:29factors how inputs and maybe
2:31misallocated across firms uh today we'll
2:34kind of dive into two of those key
2:36inputs uh capital and labor and then on
2:39Friday tomorrow we'll be thinking about
2:41uh intermediate inputs and materials Etc
2:44when uh rocker machiavello will be
2:46talking about contracts and today we're
2:49very delighted to have David McKenzie
2:51and Chris Chris Woodruff after this
2:52lecture I think they probably need no
2:54introduction they've made many seminal
2:56contributions uh to the topics related
2:59to funds and development and not least
3:01related to management to supply chains
3:05to property rights and much much much
3:08more but today they'll be talking about
3:10uh um Capital labor and where they have
3:14also had many path breaking
3:15contributions which I'm sure you'll hear
3:17about today so let me uh hand over to
3:20them and they'll be handling questions
3:22and answers as well and remember for the
3:24people who are live please raise your
3:26hand if you have a question and for
3:28others uh enter on the Q a and
3:32um one of David or Chris I will answer
3:34on that so thank you very much and take
3:36it away uh David and Chris
3:38great so
3:40um yeah welcome everyone
3:42um I'm David McKenzie I'll I'm going to
3:44start with labor and then uh you know
3:47Chris will be able to answer any
3:48questions in the chat
3:51um on the Q a on this and and then if
3:53this hands raised
3:54um let me know on on that
3:57um so yeah let me let me get started so
4:00we're going to talk about the
4:03labor first and then Chris will talk
4:05about Capital if you're a panelist
4:08um then I'll stop at a couple of points
4:09to see if there's questions but
4:11otherwise just
4:12um jump in if if you uh have something
4:15very urgent to say so
4:17yeah let's make sure this is working
4:19great so I'm going to talk about our
4:22firm's labor constrained and so uh you
4:25know within that I want to deconstruct
4:27three words to start with and so the
4:29first is you know what do we mean by
4:31firms you know we've already seen that
4:33there's a lot of heterogeneity in in
4:35firms this you know series is about
4:37firms but there's just this whole range
4:38of different types of firms and so when
4:40we ask of firms are labor constrained it
4:43makes a big difference if we're talking
4:44about the most common type of firm in
4:46developing countries if we're talking
4:48about these self-employed owners who
4:49don't have any paid workers and we're
4:51asking are they constrained in their
4:53growth by a lack of Labor or by the
4:56availability of Labor versus if we talk
4:58about sort of small and growing
5:00Enterprises that have a few workers and
5:02potentially would like to hire a few
5:03more versus if we're talking about large
5:06factories and so
5:08um you know we're going to look at the
5:09evidence on these different groups but
5:11you know there's no single firm we
5:13shouldn't write down a production
5:14function for a firm the the questions
5:16are going to be very different for each
5:18of these
5:19secondly you know what do we mean by
5:21labor you know we write down these macro
5:23models where we have F of K and L but Al
5:26here is is not a homogeneous input you
5:28know some types of Labor may not
5:30necessarily be substitutable you know
5:32even if we write some sort of CES
5:35um function on labor it may be that you
5:37know even if we have a hundred
5:38low-skilled educated workers they can't
5:40do what one person with specialized
5:43skills may be that you know some things
5:45may be that there's there's just no
5:47substitutability whatsoever amongst
5:49types of Labor and so you know when
5:51we're talking about firms being labor
5:53constrained or not we we need to
5:55distinguish the types of Labor and so
5:57you know as a rough category you know
5:58there's a massive relatively low skilled
6:00workers who firms might think of as
6:02interchangeable you know just sort of
6:05you know being this labor input that we
6:07think about in the production function
6:08there might be workers with very
6:10specialized skills that are um you know
6:12firms are looking for and then we might
6:14also think about managers and jinkai's
6:16lectures sort of talked about all the um
6:18work on on management but we might also
6:20think about managers as a as as a source
6:22of Labor as an input in this production
6:24function
6:25okay so we need firms we need labor and
6:27then we need to know what we mean by
6:28constrained and so you know typically we
6:31write down this production function we
6:32write sort of profits as F of a k and l
6:35minus their their um you know costs of
6:39um labor and the cost of capital and
6:41then you solve for the first order
6:42condition and you're going to set the
6:44marginal productivity of Labor equal to
6:45the wage rate and so then we can think
6:47of sort of two definitions of what we
6:49mean by constrained which you know we we
6:52can also think of as analogous to to
6:54financial constraints so the first is
6:57that
6:58for some reason firms cannot set the
7:00marginal productivity of Labor equal to
7:02the wage that they can't hire as much
7:04labor as they'd like at that market wage
7:07w
7:08ow and so you know this might seem
7:10strange to think about in developing
7:12countries we often think developing
7:13countries are abundant with labor
7:15there's lots of people looking for jobs
7:17so why can't firms just find as much
7:20labor as they want so you know then we
7:23have to start thinking about other
7:24information frictions is it that
7:27um you know firms don't know what the
7:28marginal productivity of labor would be
7:30they're not sure if it would be
7:32profitable to hire this worker there
7:34could be a lot of uncertainty about the
7:36cost side of it and so it's you know
7:39maybe that's a reason that firms are not
7:41able to set that marginal productivity
7:43of Labor equal to the wage rate
7:45we we're going to talk about matching
7:47frictions and so maybe you know there's
7:49a worker out there who would be
7:50profitable for you to hire but you just
7:52have trouble locating and finding them
7:53and and not all workers are
7:55substitutable so I can't just get any
7:57worker there's only specific workers
7:59that it's going to be useful for me to
8:00get
8:01you know when we're talking about
8:02smaller firms we might think about
8:04indivisibilities we often think about
8:06that in terms of capital but with labor
8:08you know maybe I need you know you know
8:11workers don't want to come and work for
8:12me for 0.3 hours a day
8:15um you know I can't think of Labor as
8:17this um you know completely continuous
8:20input and and this sort of fixed costs
8:23with hiring each worker and so maybe you
8:25know for certain types of firms that
8:27could be an issue and then you know this
8:30will link to the work that Chris is
8:31going to talk about maybe there's going
8:33to be Financial constraints that make it
8:35hard for me to to hire workers either
8:37because work is a compliments with
8:40certain types of capital I run a small
8:42you know tailor shop I need a sewing
8:44machine for each worker that I bring on
8:46and I can't it sort of doesn't make
8:48sense to sort of hire a lot of workers
8:50without hiring in the capital for them
8:52to use or maybe there's a bunch of
8:54upfront costs these workers need some
8:55training before they become productive
8:57and I am just liquidity constrained and
9:00can't afford to sort of pay those costs
9:02of training so you know there may be
9:04some reasons and we're going to look at
9:05whether these are these constraints
9:09but the second sort of definition of
9:11constraint might be sort of a more macro
9:13view that you know that this developing
9:17country wage W is somehow too high or
9:21too distorted and so you know there's
9:23distortions in the economy that mean
9:25that even if I could set the marginal
9:28product through a labor equal to to that
9:30wage rate you know that wage rate is
9:32somehow wrong for the economy and so
9:34maybe it's wrong because you know
9:36there's minimum wage laws that make it
9:38artificially high and so
9:41um you know there's workers who would
9:42like to work for a lower wage and firms
9:44who would like to hire workers for a
9:45lower wage but but
9:47um you know somehow they're constrained
9:49from doing so from these regulations or
9:51they could be other regulations
9:53um there and if we think about some of
9:55these upfront training costs maybe those
9:57you know there's also this inability to
9:58pay negative wages up front
10:01um to to people who you know are
10:03actually costly for me to have it first
10:04and then uh you know productive down the
10:07line so you know we'll come back to that
10:08towards the end but you know to start
10:11with we'll take the wage rate as given
10:13instead of say given the wage rate and
10:15the economy can firms hire as much labor
10:17as they'd like to to hire or are there
10:19some distortions that prevent them from
10:21doing so
10:22okay so
10:24um now with that overview I want to
10:26start with these sort of different
10:27categories of firms and so I'll start
10:30with the self-employed I'll sort of go
10:32through you know this this is the most
10:34typical type of firm and the you know
10:36we've taught in the in this course
10:38already about you know why are there so
10:39many small firms in the developing
10:41countries and so one question is you
10:43know are these you know tiny firms tiny
10:46because they just can't hire enough
10:47workers to grow is that sort of a
10:49constraint somehow is is a lack of Labor
10:52a constraint on on the growth of these
10:54small firms and so Chris and I worked
10:57with Suresh damel to look at this in in
10:59Sri Lanka in an experiment in which we
11:02tried to help firms take on an
11:05additional worker so we started with
11:06these self-employed workers and we gave
11:09firms a subsidy which was in a place for
11:12eight months which lowered the wage that
11:14they would pay sort of covered about
11:16half the cost to almost all of the costs
11:18of bringing on a worker
11:20and then we track these firms with you
11:22know almost uh
11:24um six years of surveys so that we can
11:27look at the impact and the Dynamics and
11:29see whether if we help firms you know
11:31overcome this these these initial costs
11:33and hire a worker
11:35um is that going to help the these firms
11:37grow
11:39so you know just effects in your mind
11:40what are we talking about when we talk
11:42about a firm like this we focused here
11:44on male owned Enterprises uh in Sri
11:47Lanka we went door-to-door to find a lot
11:50of these these firms and there's sort of
11:51just a mix of the typical types of firms
11:53you might find in developing country
11:55doing sort of retail manufacturing
11:58um services so as a top couple of
12:00examples here you know here's somebody
12:02who's doing shoe repair you know the
12:04question is you know would it help the
12:06growth of this firm if he were to take
12:07on another worker to help him you know
12:09fix more shoes and you know is that the
12:11constraint that he he's he's not able to
12:13get another worker or here's another
12:15person that's doing aquaculture raising
12:18tropical fish and you know when we talk
12:20to them they say well you know the
12:21problem is that if you just bring
12:22someone and off the street or the fish
12:23will die because you know it takes a bit
12:25of training to get
12:26um the the worker up to speed to to be
12:28able to you know manage the fish and so
12:30maybe it would be profitable down the
12:32line but you know I'm not sure whether
12:34it would be profitable for me to take
12:36someone on and initially until I've
12:38trained them up and and you know so
12:39these firms are owning about 130 a month
12:42in profits and you know the question is
12:44would they get more profits if they were
12:46able to hire some workers
12:49and so you know what would Ferry tell us
12:52well you know our standard complete
12:53markets model sort of a Lucas model says
12:56that uh no firms are going to be setting
13:00their marginal products of Labor and
13:02capital equal to these wage rates and
13:04and the cost of capital and so if we see
13:06these differences in employment size if
13:08we see some firms don't have many
13:09workers it's because this this Theta
13:12this ability or this technology
13:14parameter depending what we've been been
13:16calling it it's just low for these firms
13:18these firms are not very productive so
13:20if I see somebody who's very small it's
13:22because they're not very productive
13:24um and they wouldn't actually benefit
13:26from hiring more workers and so
13:29you know the only way that they would
13:30benefit is if I make the cost of Labor
13:33cheaper and so this temporary wage
13:35somebody's subsidy is going to make the
13:37cost of Labor cheaper when an input's
13:39cheaper I'm going to hire more of it but
13:41as soon as that cost goes back up again
13:42I will hire less of it and so there
13:45should be sort of no long run impact of
13:48a wage subsidy there should be a short
13:50run impact when I make it um you know
13:53cheaper I'm going to you know hire more
13:55and then when I remove that subsidy I'm
13:57going to hire less
14:01so then the question is you know why
14:02might we think a subsidy would would
14:05help these types of firms and so you
14:07know one reason is that there could be
14:09these frictions this these sort of
14:12um search and matching frictions where
14:14it I'm unsure of the types of workers
14:17I'm the sort of person running tropical
14:19fish
14:20um I need to find someone I can trust
14:22who's not going to kill all my fish
14:24um it might be you know hard to sort of
14:27um
14:27you know find that person and if it's
14:29costly to to hire and fire them I might
14:31not ever do it in the first place and so
14:34what the subsidy allows me to do is sort
14:36of take a chance on a new worker makes
14:38it you know less costly to to test out
14:42somebody and then if that turns out to
14:43be a good match I'll keep them on and
14:45even when the subsidy ends
14:47um there could be sort of long-term
14:49unemployment
14:50um impacts so that would be sort of one
14:52reason is if there's these search
14:53frictions and I a subsidy helps me you
14:56know subsidize the cost of search
14:59a second reason is that if I'm a small
15:01firm who's never hired workers before I
15:04might not actually know you know what
15:06the ability of my firm is what the
15:08productivity of me is a business owner
15:10able to manage other workers is until I
15:12try it out and so you know what I'll see
15:15is that there's some people who it would
15:17be profitable for this to do who have
15:20never actually taken it out because they
15:21didn't know about their their type and
15:24so then with the swaged subsidy some of
15:26them get to learn they get to learn that
15:28yeah my business actually can you know
15:30be profitable with an extra worker that
15:32I can manage these workers and that
15:35um you know that I
15:36therefore you know even when the subsidy
15:39is gone I actually now can set that
15:41marginal product of labor equal to the
15:43wage rate
15:44um correctly with with the right
15:46information that I didn't have before
15:50and then you know finally there could be
15:52these these sort of Contracting
15:53constraints because of these training
15:55frictions that I talked about where you
15:57know many workers maybe are not
15:58productive in their first few work weeks
16:00on the job and then they're only
16:02productive down the line and so you know
16:05what we would typically do then would be
16:06to say well you should pay people their
16:09marginal product when people first come
16:11maybe their marginal product is is very
16:13low or even negative and so workers
16:15should be paying you to join your firm
16:17at the start and then you should be
16:19paying them down down the line but if
16:21workers are very poor or if there's
16:23minimum wage laws or are there social
16:25norms or these these constraints here
16:27then maybe you can't
16:29um pay these workers these very low
16:30wages at the start and so then
16:34um you know there's this constraint here
16:35and then the subsidy of of the sewage
16:38subsidy can make it
16:40um help you overcome those constraints
16:42and allow you as a firm honor to uh you
16:45know again pay those training costs get
16:47those workers up to the point where
16:49they're productive and then enable them
16:50when the subsidy is gone then these
16:53these firms will still want to keep
16:55these workers
16:57okay so what do we see in practice we
16:59see
17:01um that
17:02you know when this is the the number of
17:04firms with a paid worker or or the
17:06number of workers and the black is the
17:08the treatment group that gets the
17:10subsidy the dotted line is the control
17:12group that does not and so what we see
17:14is that
17:16um when we give firms this subsidy they
17:19hire more workers uh and so the the two
17:22groups
17:23um see this we see this jump up in the
17:25treatment group in the number of workers
17:26and then once the subsidy ends which is
17:30the second one of these two vertical
17:31lines then the the firms start getting
17:34rid of some of these workers they don't
17:35immediately get rid of all of these
17:37workers but over time they just go back
17:40to where the control group is so you
17:42know part of that is for inspiring
17:44workers part of that is the sort of
17:45normal turnover where workers are
17:46quitting but there's no long run impact
17:49of these wage subsidies so for these
17:51sort of self-employed firms
17:53um you know getting this this sort of
17:55short-term subsidy has no long-run
17:57impact on the The Firm size on the
18:00number of workers that these friends are
18:02holding
18:04and so you know then when we ask firms
18:07directly you know why didn't you hire
18:09another worker or you know what was the
18:11what was going on here they say you know
18:14you know really most of them are saying
18:16I there's there's not enough demand it
18:18wouldn't be profitable I don't need
18:19another worker this is sort of
18:21consistent with just a lot of these
18:22self-employed firms having an optimal
18:24size that's just very small that it
18:27would not be efficient for them to take
18:29on another worker
18:31and then you know if we ask is it is it
18:33that it takes time for them to become
18:35fully productive despite the sort of you
18:37know initial response of that fish owner
18:40um you know when most of them say look
18:42it takes at most four months uh you know
18:45six months or shorter it's not that long
18:47a time for for workers to learn you know
18:50what it takes to be productive in this
18:52firm the subsidy that we offered was
18:54long enough to cover that learning
18:55period And even with that um you know
18:58that that's not overcoming issues for
19:00these firms and then you know when it
19:02comes to these search and matching costs
19:04for a lot of these sort of small
19:05microphones a lot of the workers that
19:08they hire it's it's really a lot of
19:09physical labor it's not sort of complex
19:12tasks where it's very hard to ascertain
19:14those those things and they say you know
19:15what I'm really looking for is you know
19:18worker of a certain gender and a certain
19:20physical strength those are the kind of
19:22key things education is not that
19:24important um you know particular fine
19:26skills are not super important and when
19:29we ask how long would it take you to
19:30find a worker you know the median owner
19:32is saying a week you know if I need to
19:34find a worker I can find them in a week
19:36so this again can you know suggests that
19:38for the self-employed firms it's not
19:40that costly to find workers doesn't take
19:42that long um to train them up
19:45um you know really the the Market's
19:47actually working pretty well for labor
19:49for these types of firms the constraint
19:51to these very small firms is not the
19:54lack of Labor it's not that they can't
19:56hire the workers that they would like
19:58so let me take a you know quick check to
20:00see if there's any questions from anyone
20:02online at the end of the self-employed
20:04before we talk to about different types
20:07of firms and different types of waiver
20:13okay
20:14great so then you know um oh let me uh
20:17sort of add one other thing on on this
20:19which is just that you know really a lot
20:22of these these micro firms these small
20:24firms have a lot of Labor slack in them
20:26that is you know when you look at what
20:28the owners are doing all day there's
20:30there's a they have a bunch of time that
20:33they could actually be working a bit
20:35harder or doing a bit more so even when
20:37you do other interventions
20:39um and Chris is going to talk about
20:40Capital interventions but I'm showing
20:42here you know when we give business
20:43training
20:45um
20:45this is from sort of an overview paper
20:47that um Chris and I and some others um
20:51wrote for for the Vox Dev looking at
20:53business training even when business
20:55training has increased firm profits and
20:57so you know here's a whole range of
20:59studies and the sort of increases in
21:00firm profits and the average is a 12
21:03increase in firm profits the figure on
21:06the right showing what happens to
21:07employment and so you know we've got a
21:10whole lot of business training programs
21:11that have made these firms a little bit
21:12more productive able to earn a bit
21:14higher profits but there's a zero impact
21:17on employment a very precise sort of
21:20zero impact on employment for at least
21:22those firms and you know if you earn 12
21:25more profits you can do that by the
21:27owner working but more productively
21:29they've got a bunch of space in their
21:31own time without them having to hire
21:33workers um there so firms need to sort
21:35of grow quite a bit before they start
21:37hitting the constraint that they need
21:38more labor you know a lot of the time
21:40when you visit these small firms the
21:42owners standing around with no customers
21:44and without a lot to to do and you know
21:47there's other constraints to the growth
21:49that are not just the lack of workers
21:51and so I think you know for a lot of
21:53these firms we we think that labor is
21:56going to bind
21:58um later and and um
22:00you know that there's other constraints
22:01that are hitting these firms first
22:04so you know that's the self-employed
22:06what about you know when we start
22:09looking at slightly larger firms
22:12um and so you know let's first talk
22:13about slightly larger firms with
22:15non-specialized workers and so on one
22:19hand we should think that some of the
22:21frictions that I talked about should be
22:23a lot less of a concern for these larger
22:26firms they should have more experience
22:27hiring workers and so this idea that you
22:30know I'm not sure about my type I don't
22:32know what the marginal productivity of
22:34labor would be should be much less of an
22:37issue if you're a bigger firm and you're
22:38constantly using workers you should have
22:40much a much better sense of sort of how
22:42productive as a worker going to be in my
22:44firm How likely am I going to be to
22:46manage workers so those things should be
22:48less of a constraint
22:50secondly you know the search and maxing
22:53fractions should be lower as well
22:55because of this fixed cost of
22:57advertising for workers or sort of
22:59trying to find workers of making myself
23:01known if I'm a bigger firm and I'm doing
23:03more recruiting then I can spread those
23:05fixed costs over more workers and and
23:07they should be you know less of a
23:09constraint as well so it should be sort
23:10of a little bit easier for me to find
23:12workers if I'm a bit bigger
23:15and then similarly these
23:16indivisibilities are much less of an
23:18issue if I'm
23:19um you know not the micro firm and I'm
23:21making this big jump from a one-person
23:24firm to a two-person firm you know going
23:26from a nine person firm to a ten person
23:28firm is much less of a a jump these
23:31indivisibilities shouldn't be there but
23:33on the other hand maybe we think that
23:34some of the sort of these regular
23:36relations these frictions around the
23:38minimum wage the sort of issues of
23:40hiring and firing and things may be more
23:42severe as you get bigger and you get
23:43into the um sort of legal
23:46um eye
23:47um of of the government where some of
23:49these things may bind a bit more so I'm
23:51going to go through four cases that give
23:52us sort of a sense of
23:54um in different contexts you know where
23:57this type of Labor and these types of
23:59firms may or may not face different
24:01frictions
24:04so the first is sort of one of the
24:06things that we're you know a lot of the
24:08research is has been on a sort of a big
24:10type of large Factory in developing
24:12countries are these large textile
24:13factories which have been sort of a
24:15playground for a lot of the research
24:16that um you might see through this this
24:18course you know they have their these
24:21firms uh sort of big employers they have
24:24um you know production lines with lots
24:26of data and you know researchers have
24:28been looking at them a lot and one of
24:30the things that you know I've taken from
24:32these studies is just how much turnover
24:35and labor there is in these these
24:37factories these these factories sort of
24:38seem to have no trouble finding
24:41production line workers but there's a
24:43lot of issues in holding on to them that
24:44a lot of these workers sort of come into
24:46these firms and they leave very quickly
24:48um and you know within a couple of years
24:50a lot of these workers have turned over
24:52and so I want to point here to one study
24:54in Ethiopia which uh Chris Blackman and
24:57Stephen Duncan did where they worked in
25:00in not just uh textiles bid in sort of
25:03five light manufacturing firms including
25:05you know flower farm shoe manufacturer
25:07or water bottle plant
25:09and they have this great description in
25:10their paper where they sort of say
25:12workers are effectively disposable to
25:14firms we can just think of them as an
25:15input that we can you know use like uh
25:19any other intermediate input and you
25:22know firms really aren't that picky
25:24about these workers and so in fact
25:26they're so unpicky about these workers
25:28that they allowed these researchers to
25:30randomize which workers would come and
25:31join these firms so this is sort of
25:34suggesting that firms are not
25:35particularly fussy about you know the
25:37match quality that they're
25:40um you know they say there's no shortage
25:42of applicants there's lots of people
25:43applying for these jobs and they're
25:45they're really indifferent as to which
25:47workers come and join these these firms
25:49but on the other hand you know a third
25:51of these workers quit in the first month
25:52and 77 you know three-quarters of his
25:55workers are gone in the first year
25:57and you know the HR people say well you
26:00know because workers are coming going
26:02all the time we don't spend a lot of
26:03time you know on that that match it's
26:06sort of a first come first serve there's
26:08not much of an interview process it
26:09doesn't take them very long to get up to
26:11speed and so you know we bring these
26:13these these workers in they they work
26:16and then they they they go and I think
26:18you know we're seeing this a lot in in
26:21these other sort of large um studies
26:23from from factories in India and
26:24Bangladesh where I don't think there's a
26:26lot of trouble hiring workers but you
26:29know there's things that firms could be
26:30doing that I think relate much more to
26:32these management practices that jinkai
26:34was talking about of you know how do you
26:36have HR systems that are you know
26:38helping retain good workers and get rid
26:40of bad workers how are you sort of
26:42upskilling your workers with training
26:44how are you you know doing things to
26:46hold on to workers that I think are less
26:48of a constraint on the labor side and
26:50more of a constraint on the management
26:51side and so I think again I would point
26:53to you know labor not being The Binding
26:55constraint of of of the growth of these
26:59firms
27:02second sort of case of these large
27:04larger firms that I want to talk about
27:06is sort of and now talking about maybe
27:09trying to hire more educated workers so
27:11this is work that I did with some
27:12colleagues in Jordan where we take a a
27:16big sample again of smes and a whole
27:18range of different Industries and
27:20they're trying to fill sort of admin
27:21assistance sales staff accountants
27:24marketing jobs web development I.T and
27:28so when you talk to these firms up front
27:30they say
27:31you know we have no shortage of
27:33applicants but it's really hard to know
27:35who's good and who's bad
27:37um it's hard to find you know competent
27:39graduates in in reasonable proximity to
27:41the firm and so we offered them uh a
27:44search and matching intervention where
27:47we took a lot of recent college
27:49graduates we did psychometric testing we
27:52did skills testing where they got you
27:54know a computer test and an English test
27:56and an Excel test and things like that
27:58so we could sort of you know measure the
27:59skills for firms we could measure the
28:02the soft skills of firms and provide
28:04firms with this information so you might
28:06think of firms that you know really
28:08struggling to find the right workers and
28:10you come and try and help help them do
28:12so maybe this is going to help these
28:14firms grow
28:15so what happened well you know we made a
28:18whole lot of matches between workers who
28:19said they were looking for jobs and
28:21employers who said they were looking for
28:22workers and very few of those matches
28:25ended up in interviews most of the time
28:27firms said you know actually while we
28:28said we were interested we're doing all
28:30right we're not interested and then when
28:32when we found some and then some of the
28:35workers who said they were looking for
28:36jobs said they weren't interested as
28:37well and then once we got interviews to
28:40take place
28:41um
28:41very few of the firms hired those
28:43workers and then most of those workers
28:45quit in the first month and so you know
28:46we made over a thousand matches and then
28:49you know only nine people out of that
28:51thousand ended up in in jobs and so you
28:55know then when we look exposed to say
28:56what's going on you know while firms
28:59tell you we have difficulty finding good
29:00matches
29:01very few of them would be willing to pay
29:03to learn anything about the quality of
29:05graduates and then when we track in a
29:08panel setting how long does it take them
29:10to fill their their positions you know
29:12have you got a position open today we
29:14come back and ask what what happened
29:15with that position most of those
29:17positions are filled within a month
29:18firms are filling these positions very
29:20quickly
29:21um anyway and
29:24you know basically what's happening is
29:26that firms say well we can fill these
29:28positions we'd like to have you know
29:30better quality workers at the same wages
29:32that we're offering right now but we're
29:34not willing to offer any different
29:35conditions and I think you know we'll
29:37come back to this towards the end I
29:38think this sort of broader distortions
29:40going on here in the macro economy where
29:42the public sector
29:44um means that you know workers only want
29:47to work on certain types of firms that
29:49the wage rate's sort of too high for the
29:51types of firms um willing to you know
29:54pay for these these um workers but
29:57conditional on you know want the supply
30:00of labor is and what the wages you know
30:02firms can can find jobs workers very
30:05quickly and they can fill these
30:06positions and so again it seems like
30:08labor is not the constraint to the
30:09growth of these smes
30:12so let me you know then talk to a third
30:16um you know case which is India you know
30:19which is famous for these industrial
30:21laws that make it really hard to fire
30:23workers particularly once you get to
30:25sort of 100 workers or more and so you
30:27might think okay these firms surely are
30:29labor constrained you know surely now in
30:32a country like India where there's all
30:33these regulations these firms would
30:35really benefit from being able to hire a
30:37lot more workers but these regulations
30:38get in the way and are a distortion and
30:41so you know here there's there's some
30:43very nice work by Shanghai Shay and
30:46co-authors which sort of point out maybe
30:48these regulations don't bind as much as
30:49we might think they do because of firms
30:52finding loopholes with what he calls uh
30:55um Jacquard where you know they're
30:58finding informal solutions to problems
31:00and one of the solutions that's sort of
31:02arisen in India is this rapid
31:03development of a Outsourcing industry
31:06where firms can contract
31:08um workers through this third-party
31:10intermediaries so they don't count for
31:12the purposes of a law and so
31:14you know this is really taken off over
31:16time and so here you know this the
31:19x-axis shows employment and the y-axis
31:22shows the proportion of their workers
31:24who are you know these contracted
31:26workers and once you sort of start
31:27getting to 100 or more you're getting
31:29you know 40 50 of the workers in these
31:32factories
31:33um are getting contracted through these
31:35these outside things and so there's
31:37surely some Distortion here but the
31:39point of view is that when there are
31:41these distortions the economy the the
31:43market is working to try and find
31:45solutions to these distortions and so
31:47these distortions are not as binding as
31:49we might think they are and again you
31:51know maybe this is not the major
31:53constraint on the growth of these firms
31:55because the market has come up with
31:57solutions that that help firms get
32:00around some of these constraints and so
32:02this is not to say it wouldn't be
32:03beneficial to remove these you know
32:05change these laws or to make it a bit
32:07better but conditional on the law you
32:09know firms are still able to hire
32:11workers they're just finding workers
32:12through these workarounds
32:15so then let me you know finish my you
32:18know in terms of talking about these
32:20smes with one final case which I think
32:22is is one that shows where where firms
32:25did seem to be labor constrained which
32:27is from South Africa and you know this
32:29is a very unusual labor market so in
32:31South Africa the unemployment rate is
32:34you know 20 to 25 if we look at youth
32:37that's you know 40 or more it's a market
32:39which is just definitely not clearing
32:41it's a sort of mix of developed and
32:43developing country there's very rigid
32:45labor laws that are very that are
32:48perceived to make it very hard to hire
32:50and fire workers but there's also very
32:52low in formality and so these
32:53workarounds that we see in India and
32:55other countries where you know firms can
32:57get around the law are harder for firms
33:00to do
33:00and so you know this study looks at
33:03firms with 10 to 300 workers that's you
33:06know
33:07um sort of an average of say 50
33:09um for a median of 50
33:12um in a mean of 78 so you know much
33:14larger firms than I've been talking
33:16about with the micro ones and then they
33:18give a treatment group a lot of
33:20information about the labor law and help
33:22these firms understand that the laws are
33:24not nearly as bad as they thought they
33:26were and they find that when that's the
33:28case these treated firms
33:30um increase their hiring by about 11
33:31workers or 13 over a six-month period so
33:35these firms are you know do seem to be
33:37constrained By Their Fear of the law
33:39they're not hiring as many workers as
33:41they want and then when they learn that
33:43the law is not as bad they do grow
33:46um and and so there seems to be a
33:48constraint here
33:49so this is a case with a sort of
33:51particular labor market particular
33:52infrastructure where it seems like firms
33:54are constrained and there's this
33:55misallocation that there's you know
33:57these firms would be better off with
33:58more workers in the in the there's this
34:00legal strength but one thing I wanted to
34:02point out here is is sort of
34:04um that a methods Point here which is
34:09um you know it very much depends on what
34:11you're measuring here and um even in
34:14this case there's just a lot of churn
34:16going on and so over six months this
34:18x-axis shows how much churn and
34:20employment there was in these South
34:22African firms and you can see some of
34:24these firms are you know adding 350
34:26workers and others are getting rid of
34:28150 workers within that six month period
34:31so this idea that you know it's just
34:34impossible that for these firms to
34:35change their number of workers very
34:37easily is is not there in the data these
34:39firms are making huge changes and then
34:42you know what this means for us doing
34:44studies is that if firms are making so
34:47much change in their work as and then
34:50we're doing an intervention it's really
34:52hard to you know detect a two or three
34:54or four worker change in workers from
34:57our intervention if firms are just
34:59making such big changes in their um
35:01Workforce anyway and so you need a
35:03really big sample to be able to detect
35:05this
35:07um
35:07it also sort of says well is it really
35:10that hard for firms to change if they're
35:11changing that much and then if these you
35:13know the so when they look at the the
35:16change measured one way they're seeing
35:18this big change when they when they ask
35:19directly how much have you hired and how
35:21much have you fired the impact that the
35:24study finds is much much lower rather
35:25than you know 13 workers they find one
35:28worker which is not significant and you
35:31know they say there's a lot of
35:32measurement error but you know again
35:34even in this context it's going to be
35:36very dependent on how you're measuring
35:38this and you know there's a question are
35:39you just speeding up over six months
35:41hiring that you're going to do anyway so
35:43even in this case where it's this very
35:45unusual labor market you know I think
35:46there's reasons to to sort of ask how
35:49much of a constraint labor is um here so
35:51let me just check again if there's
35:53there's any questions
35:55um from the the online panelists
35:59Dave one question that I couldn't answer
36:02on the chat is whether the
36:05um the uh results in South Africa were
36:08sustained over time whether the hiring
36:10additional hiring was sustained
36:12yeah and that's they kind of to that
36:14either because they only look six months
36:16down the line and so this is my sort of
36:18last bullet point here is you know is
36:19this just speeding up hiring that firms
36:21were planning on on doing is is this a
36:24sustained impact so I think this is sort
36:25of an interesting
36:27um you know Counterpoint to my my
36:29argument that most of the time this is
36:31not constrained but we still don't know
36:32whether this is sort of having a long
36:34run impact in in this case
36:36um ridwan you had a question
36:37yes uh thanks for letting me in so uh my
36:40name is ritwan I'm from Delhi School of
36:42Economics so I was just wondering that
36:44how does in uh industry Academia
36:47collaboration uh sort of have been
36:49removing the information frictions
36:51regarding the type of Labor to hire so
36:54what sort of what sort of impact does
36:57industry industry industry Academia
36:58collaboration does have in terms of
37:01removing the information symmetries uh
37:04about the matching you are talking about
37:06yeah and so I think you know this is
37:08where you know there's obviously a large
37:11labor matching industry already there
37:13that's trying to help firms do this but
37:15you know researchers have come in with
37:17ideas of other ways to to help
37:20um you know workers signal their skills
37:23and uh
37:24um you know I think there are good
37:26opportunities to work to get together to
37:28try and improve that
37:30um but you know I think it's it's a
37:33little case specific I'm going to talk
37:34about I think one or two um cases I
37:37think we've got two more questions from
37:38um so so silver
37:40um
37:42um silver jumper
37:43um you want to go sorry if I'm
37:45mispronouncing your your name terribly
37:49Salvo
37:53or Umberto
37:55hi thank you uh I would like to ask are
37:59there any study they explore the
38:01relation between HR or and manager
38:05practice and uh Innovation outcome or
38:08effort
38:12um yeah so that's again sort of pointing
38:14back to the the sort of link between
38:16management practices rather than than
38:18sort of Labor and and then Innovation
38:21and and growth and
38:23um I think there's a lot of car some of
38:26the work there that has been done on on
38:28management has has has looked at more in
38:31developed countries that at correlations
38:33between
38:35um better management practices and more
38:36innovation in our and and d i I don't
38:39know if um things in developing
38:41countries that have sort of showed
38:43causally that that link and particularly
38:46with with sort of HR um practices but
38:49you know again I think this is uh and
38:52I'll come come to this hopefully in my
38:53next a couple of slides about the sort
38:55of types of workers
38:57um so let me try one one more time is
38:59there any other panelists there's
39:01someone with their hand raised who I
39:02quote on um do you want to talk
39:08no okay well let me let me move on and
39:11then we can um I'll I'll take on this so
39:14you know my my sort of sort of bottom
39:16line for this is that often I think
39:18labor is not the big constraint for low
39:21skilled entry-level labor into these
39:22large firms or for the self-employed
39:25that you know many firms complain that
39:27it's hard to find workers that it's hard
39:29to get them to stick around but revealed
39:31preference suggests that actually firms
39:34fill most of these jobs very quickly
39:36they choose not to spend a lot of time
39:38in screening they don't employ a lot of
39:40HR practices to help their best workers
39:42and they can often find lots of ways
39:44around regulations and so it suggests
39:46these wedges are not that big that their
39:48firms are not devoting a lot of effort
39:50to to to doing it that they don't seem
39:52to be really constrained for the most
39:54part
39:55so then I want to sort of finish by
39:56saying you know when might labor be a
39:58cons more of an issue more of a
40:00constraint to to firm growth and so I
40:03think one sort of category is with more
40:05unusual labor market institutions and so
40:07we talked about South Africa I think
40:09another system that sort of comes up
40:12um is is this apprenticeship system and
40:14um that's prevalent in some parts of the
40:16world where it's this mix between a type
40:19of employment and also a type of
40:20education and so
40:22um you know this is a system where the
40:24wage paid is actually typically negative
40:26for a while you're training people who
40:28will compete with you often when they've
40:31um finished and so we can think about
40:33you know various frictions that this
40:35type of system
40:37um causes and so you know there's work
40:39by
40:40and Morgan Hardy and Jamie mclasslin who
40:42looks at this government program in
40:44Ghana and they work with small firms and
40:46with poor unemployed people and you know
40:49the constraint here is that these poor
40:50unemployed people can't afford to you
40:53know pay this fee to have negative wages
40:55at the start and so instead of this this
40:58fee the government does a screening
41:00program and it selects people who
41:02um you know are good amongst these these
41:05types of workers who wouldn't not
41:07normally be able to get um selected by
41:09firms and this you know does find that
41:11there's a group of firms who choose to
41:13be in this program who benefit in the
41:15short run from this type of help they're
41:17getting this additional type of
41:19um labor who who otherwise wouldn't be
41:22there on the market or that they
41:24wouldn't be up the screen on the market
41:25does seem to get some increase in
41:27profits but there's a couple of other
41:29recent studies that have sort of looked
41:31at this apprenticeship environment
41:32that's finding while it helps the
41:33apprentices it's not having measurable
41:36impacts on on these these firms in
41:39Uganda so even in this apprenticeship
41:41case it's not you know I think there's a
41:43lot of frictions on the on the um side
41:46of the job Seeker but it's less clear
41:48that you know firms would really benefit
41:51from taking on a lot more of these
41:52apprentices
41:54so then you know another type of
41:56constraint I think you know we've talked
41:57about sort of less skilled less
42:00specialized labor I think with
42:02specialized labor and more growth
42:03oriented smes then these search
42:06frictions could be a lot higher it could
42:08be harder for owners who are sort of
42:10these fast growing firms to know whether
42:12it's it's useful to take on these types
42:14of Labor and so in work that I did with
42:17Steve Anderson in Nigeria we looked at
42:19you know firms who are fast growing and
42:22when should they take on Specialists and
42:24things like marketing or accounting or
42:27other professional functions and you
42:29know do firms know how to find workers
42:31of these these types and so we
42:33subsidized them using HR firms and and
42:36professional consultancies to fund these
42:38tasks and we find that firms you know
42:40grow and benefit a lot more from from
42:43having this help rather than trying to
42:45learn these skills themselves but
42:46there's a lot of search frictions that
42:48they wouldn't know how to find these
42:49workers by the by themselves and they
42:51wouldn't know that it would be
42:52beneficial for them to do so and so I
42:54think you know for these more
42:55specialized types of workers where firms
42:57haven't got a lot of experience doing it
42:59maybe there's you know so that can be a
43:01limit to their their growth and
43:03similarly you might think of you know if
43:05I'm trying to search over space if I
43:07need to hire someone from another region
43:09or another country even I'm trying to
43:10export and I need specific workers that
43:13you know know if I'm enjoyed and how do
43:15I break into Germany like you know
43:17finding that worker the the search
43:19frictions could be quite high and so
43:21they're you know it's not to say that
43:22there's never any constraints on labor
43:24the search frictions for particular
43:26types of Labor could be high but for
43:28most part I think it's not
43:30so then let me just finish um you know
43:32with I've talked a lot on the micro side
43:34but I think you know we wanted to also
43:36take a step back and say on the macro
43:38side
43:39um you know what we see is that given
43:41the current skill set of the labor force
43:43given the sort of human capital of the
43:45country that we see and the wages that
43:47Prevail in that country many firms seem
43:49to be able to hire and fire as many
43:51workers as they like without a lot of
43:53constraint
43:54but that's not to say that these
43:56countries and these firms couldn't do
43:58better if the supply of workers improves
44:00and if the price of workers um was was
44:03lower and so you know one sort of
44:06interesting suggestion on this is is
44:08this work that looks at whether
44:10um Labor's just too expensive in in some
44:14countries and so galb and all have this
44:16this work that sort of says a lot of
44:18African countries have wages that are
44:21much higher than you should expect given
44:22the level of income in these countries
44:25and so it makes it hard for a lot of
44:26African firms to compete in light
44:28manufacturing and you can point to sort
44:31of things like natural resource curses
44:33or sort of large State involvement in in
44:36some cases
44:37um to sort of explain why that might be
44:39the the case and so the idea here is
44:41actually you know the most constrained
44:43firms might be ones that don't even
44:44exist at all that there should be you
44:46know a bunch of large manufacturing
44:48firms that just don't even exist in
44:50these economies because of those the
44:52structure of these economies and the way
44:54that these economies are set up there's
44:56it's just not efficient to do so
44:58um you know another example is work on
45:00the high cost of middle managers that
45:02because middle managers are so expensive
45:04in developing countries relative to the
45:07the income levels in these countries
45:08firms don't uh it's not efficient for
45:11for firms to hire a lot of these
45:12managers and then you know through the
45:14the management cons um work that we've
45:17seen that prevents firms from improving
45:19and so you know I think this is a really
45:21interesting area for researchers to sort
45:22of think about this this micro macro
45:24interaction and how these sort of macro
45:27determinants of the wage structure and
45:29the economy filter through to you know
45:31whether there's constraints and so I'm
45:32pretty much on time
45:34um I think so I'll see if there's like
45:36one last question and then hand over to
45:37Chris
45:42okay so we'll hand over to to Chris and
45:45then I can answer anything in the Q a
45:50foreign
45:59thanks David there you can see my uh
46:03slides I hope uh so thanks David uh uh
46:06that was great and and finished right on
46:08time and I will try to pick up
46:10um uh from where uh from where you've uh
46:13left off I'm going to talk about the
46:15capital side of this and and it's going
46:17to be sort of similar in the sense of
46:19thinking about uh where there are
46:21frictions that might be uh leading to
46:23inefficient allocation to Capital or uh
46:26uh uh too little uh too little capital
46:28and I think when we think about this in
46:30the in in the context of the the
46:32literature that I'll spend most of my
46:34time talking about
46:35um is we want to think about this as as
46:38a constraint on the Dynamics of firms on
46:40the strength on the growth of of of of
46:42firms and uh and and how uh whether
46:45whether capital is allocated
46:47uh efficiently in that in that sense so
46:50let me first let me first sort of step
46:52back and and and and put this sort of in
46:54a broad context there's there's uh you
46:57know we know that production in in lower
46:58income countries is less Capital
47:00intensive
47:01um a lot of this owes to labor being
47:03concentrated in smaller firms as David
47:04Atkins showed in the you know data on in
47:07the introductory lecture there's a
47:10seminal paper I uh uh
47:13Raghu Rajan and Luigi's and Gales that
47:16shows that Manufacturing in in countries
47:19with weaker institutions more generally
47:21is disproportionately concentrated in
47:23low Capital sectors so it's a sector
47:26level analysis not really a not not a
47:28firm level analysis
47:31um and you know there um the we we might
47:35think of that channel as as being
47:37primarily weak information and and they
47:40talk in particular about accounting
47:41standards uh uh weak information and
47:44enforcement
47:46um
47:47weak Capital markets and then low uh uh
47:50uh low Capital intensity and and uh and
47:53and and smaller and smaller firms
47:56um so the theory I'll put back up the
47:59the David's Theory slide to start with
48:01uh I wanna I wanna think about this in
48:04the same in the same way with the sort
48:06of the Lucas model of a of an owner with
48:08uh with with some uh with some uh
48:11ability uh Theta and uh and you know
48:15wanting to set the marginal product of
48:17labor equal to the wage rate and the
48:18marginal product of capital equal to to
48:20uh uh the the interest rate
48:23um and I've changed only the part that's
48:25in Blue from David's slide to say that
48:27for any given Theta constraints to
48:29Capital reduced firm size for a given
48:31ability
48:33um and uh uh and and this may be
48:36especially so for high-end High ability
48:39uh entrepreneurs and why do I say might
48:43be especially so for high ability uh uh
48:46High ability owners or high ability uh
48:49uh entrepreneurs well first the thing to
48:52say is that's not always the case if we
48:54had perfect Capital markets and perfect
48:56Insurance markets interest rates are
48:58constant across entrepreneurs and each
49:00each borrows to the point where the
49:02where the marginal product of capital is
49:03equal to the interest rate there there
49:06wouldn't be differential uh uh
49:08differential constraint on high ability
49:10uh high growth uh high growth potential
49:14uh owners but if either of those markets
49:17is missing or imperfect which we we
49:19certainly would think to be the case
49:21really realistically everywhere but
49:24maybe especially so in in lower income
49:27lower income countries
49:29um then borrowing limits may be fixed or
49:33credit might be limited by collateral uh
49:36and
49:38um and uh and and since often external
49:41capital is limited to debt uh when
49:45owners are risk-averse and insurance
49:46markets are missing that also May limit
49:49uh uh limit limit growth so if it were
49:54the case that there were a very high
49:55correlation of ability and access to
49:59Capital this wouldn't be a problem but
50:02you know the standard kind of thing that
50:04we think of people who have money but no
50:06ideas and people who have ideas but no
50:08money require that we somehow get the
50:10money from the people with with money
50:11but but but less lower quality ideas to
50:15the people with higher quality ideas but
50:17but but having less uh less money and
50:20that requires Capital markets and then
50:22that's that's where we we start to think
50:24that that maybe there there are some
50:26constraints so look I'm gonna I'm gonna
50:28have three slides here with with a bit
50:31of of you know very very basic um theory
50:35on this I I'm not going to spend a lot
50:37of time on these I want really just to
50:39sort of draw out
50:40um you know two points from them one is
50:44um
50:44so you know we have owners of uh
50:47entrepreneurs or owners will choose
50:49Capital stock to uh uh uh to to to
50:52maximize their their utility which a
50:54function of consumption and consumption
50:56is going to be a function of their their
50:57income and uh and and their ability to
51:00and at less what they they they they
51:03invest in the less less what they invest
51:05in the business and the the two key
51:08things to bring out here are what they
51:10invest in the business is going to be a
51:12function of constraints and it's going
51:14to be in particular a function of
51:15borrowing constraints uh as well as
51:18wealth constraints wealth and income
51:19constraints so that's basically four or
51:21five equations four five and six the
51:23borrowing the wealth and uh and and the
51:26income constraints
51:27um the second thing to to to to to bring
51:30out
51:31um is that risk is also going to play a
51:35a role here and uh the risk is gonna you
51:39know is when when the owners are risk
51:42averse we're going to see that that also
51:44reduces is how much Capital they they
51:46put in their uh they they want to invest
51:49in their business and I think that's the
51:51another place where
51:54um when I step back and say let's look
51:56at the literature uh and what we've
51:59learned about
52:00um how to resolve some of the
52:03constraints that we that we see in the
52:06uh how to make these these these these
52:08Capital markets these external Capital
52:10markets work and function better
52:13um one of the places I want to point to
52:15is the idea of how do we bundle uh some
52:19sort of insurance some sort of risk a
52:23way to deal with risk at the same time
52:26that we're dealing with with capital I
52:28think that's an area where we're we're
52:29beginning to see some pretty interesting
52:31work and we want to you know we I think
52:34we want to think about uh that uh uh uh
52:37a bit more uh a bit more carefully so uh
52:40so you know ignoring risk but allowing
52:43for imperfect credit credit markets
52:45we're going to have the marginal product
52:48of capital not equal to the interest
52:50rate itself but interest rate plus the
52:52shadow value of of capital and other in
52:55in other uses uh that's that's a
52:58function of the liquidity constraints
53:00um but perfect credit markets aren't
53:02going to be enough when we have missing
53:03Insurance markets uh we're going to have
53:06uh uh uh we're also going to have risk
53:10rationing or risk constraints for the
53:12for the uh uh for the uh entrepreneurs
53:16and uh under reasonable assumptions
53:18that's going to give us a marginal
53:21product of capital which is uh which is
53:23a greater than uh greater than the
53:26interest which is which is greater than
53:28interest rate so again sort of
53:29suggesting some sort of constraint but
53:31the constraints not coming from uh
53:33differential interest rates across uh
53:36across entrepreneurs but from the
53:39differential uh the differential
53:41interest rates in the market across
53:42entrepreneurs but from the differential
53:44risk adjusted
53:47cost of cost of borrowing
53:51so with with that in mind with those two
53:53things in mind that we want to think
53:55that you know there there there may be
53:57there may be constraints which are uh in
54:00the Capital Market which are varying
54:02with the ability of the entrepreneur and
54:05we want to think that there are uh that
54:08there are
54:09um uh potentially uh there's potentially
54:12risk rationing or uh an additional
54:16constraint which comes from uh the uh
54:19the fact that entrepreneurs have a large
54:22share of their wealth invested in a
54:25single firm and a single uh uh
54:28undiversified uh uh asset that um that
54:32that that that that may affect um
54:35uh the uh that may affect the uh uh
54:39their willingness to make investments in
54:40their willingness to to grow
54:43um and while David kind of talked about
54:46um five papers uh you know several
54:48minutes a piece I'm gonna uh talk about
54:50several papers a minute uh but I'm gonna
54:52try to bring those together around a
54:54couple of topics and papers that are
54:56quite uh quite closely related
54:59um a lot of the work that's gone on in
55:01this in this area uh with uh especially
55:05with small firms and small and medium
55:08firms is
55:09um has been done through rcts the rcts
55:13are often in a single context with a
55:15single sample and you know we want to
55:17learn across the rcts about the patterns
55:20that we see in the data as we look at
55:23the results from several studies at the
55:25same time so I've tried to group some of
55:27the studies I'll dig into a couple of
55:28them and that I think are particularly
55:30interesting and promising directions to
55:32for the for the literature to go in a
55:35bit more detail but but um but but I but
55:38I'll otherwise try to group some of
55:40these uh some of these studies uh to
55:42talk about them so the first two
55:44non-randomly selected examples of of of
55:47returns from Grants
55:50um first the project that David and I
55:51did with uh Suresh damel again uh in uh
55:54in in Sri Lanka where we gave uh Capital
55:57grants of a hundred dollars to two
55:59hundred dollars uh to firms with that
56:02should say that should say less than a
56:04thousand dollars in capital so these are
56:06firms without paid employees who are on
56:10this the small end of the spectrum and
56:12we wanted to understand what's the uh
56:15return to uh what's the return to
56:17Capital uh in these firms
56:20um what we found was returns of of of
56:23around six percent per month on average
56:26across the whole sample uh and that
56:29compares to micro Finance rates in in
56:32Sri Lanka at the time of a percent and a
56:35half to two percent a month so it
56:37suggests that returns are quite you know
56:39the marginal product to Capital is quite
56:41a lot above the market interest rate as
56:45measured by the uh the the the interest
56:48rates of uh of mfis at the time
56:52um similar design similar kind of
56:54project uh work that David and I did
56:56with uh Marcel first Johnson and uh and
56:59Simon Quinn
57:00um an RCT similar size businesses uh in
57:04uh in in Ghana uh uh where uh we found
57:08that returns for cash grants uh were
57:12similar to the level in in Sri Lanka
57:15returns to in-kind Grants in kind
57:17meaning we purchased inventories or
57:20assets for the business owners that were
57:23the assets and inventories that they
57:26said they wanted to buy with the with
57:28with the funds those returns were quite
57:31substantially higher uh in uh in in in
57:34Ghana than the cash returns but but
57:36again quite High returns to you know
57:38suggesting quite High returns to Capital
57:40maybe especially if the capital is
57:43actually makes it fully into the
57:46business as it was more likely to do in
57:48the in-kind grants in in Ghana than with
57:50the cash with the cash grants
57:53um then finally uh the Banerjee at all
57:56uh uh uh a project looking at at loans
58:00in Hyderabad India they have a fairly on
58:03the Enterprise side on the on how much
58:06the do the loans affect uh the uh
58:08profits of businesses they have a fairly
58:11noisy zero so it's so so it's
58:13insignificant but you know we wouldn't
58:16be able to rule out fairly large uh
58:18fairly large returns but fairly large a
58:20fairly noisy zero on average
58:24um but High returns among those with
58:26Enterprises uh that had Enterprises
58:28before the loans were were given and I
58:31think that's the point of what the next
58:33point I want to make is that all three
58:35of these these studies and a lot of uh
58:38the other literature that we see on on
58:42on Capital on Capital grants shows a
58:46tremendous amount of heterogeneity not
58:48only in the Enterprises themselves of
58:50course if we look at at the Enterprises
58:53in lower income countries we see
58:54heterogeneity in the cross-section but
58:56we see a tremendous amount of
58:57heterogeneity in the returns to Capital
59:00on this from the shocks that that were
59:02that that the experiments uh provide and
59:05I'll show this in three different ways
59:06in ways that are calculated in three
59:08different in that are that are
59:10constructed in three different ways
59:11first from our Sri Lanka paper where we
59:14have predicted returns based on
59:16characteristics that come out of uh of a
59:20of a regression allowing for
59:22heterogeneity and return based on those
59:24characteristics and those
59:25characteristics include
59:27um whether the the owner the the
59:30measures of the ability of the owner not
59:33quite Theta because we don't we didn't
59:35we don't have a perfect measure of theta
59:37but but
59:39um fluid intelligence digit span
59:41numeracy see these these kinds of those
59:44kinds of ability measures uh and
59:46liquidity constraints in the household
59:49uh and uh in particular wage income in
59:52the household relative to the size of of
59:54the business we we can see quite a large
59:59range of uh of uh uh of heterogeneity
1:00:03across uh across returns in that uh uh
1:00:07returns to the to the to the capital
1:00:09grants uh in that project similarly in
1:00:13Ghana this is for uh for males
1:00:16um I think the the point that that that
1:00:18that I'd make here is if we look at the
1:00:20green line which is the in-kind
1:00:23equipment you know purchased directly
1:00:26for uh the Enterprise is what they what
1:00:28they want to buy
1:00:30um I think the the thing that's striking
1:00:32about this figure is how far out the
1:00:35green line goes that is at the top of
1:00:37the distribution there's really quite a
1:00:39lot of separation between the control
1:00:41group and the cash group while at the
1:00:43bottom of the distribution there's not
1:00:44much separation at all again suggesting
1:00:46a range of of returns to Capital that at
1:00:51the high end was was was quite uh is is
1:00:54quite high and at the low end is uh is
1:00:57is close to zero
1:00:59uh Banner you'd all find the same the
1:01:02same thing with their uh Indian data uh
1:01:05this again with loans uh what they term
1:01:07gung-ho entrepreneurs but you know are
1:01:10are the sub-sample of pre-existing uh
1:01:13entrepreneurs uh in quantile regressions
1:01:16had a No Effects up to about the 75th
1:01:19percentile and then then positive
1:01:21effects and really quite big effects
1:01:25from about the 85th or 90th percentile
1:01:28up so in the upper tail of the
1:01:30distribution uh you know quite quite
1:01:33substantial uh quite substantial returns
1:01:35even though uh not so much uh the
1:01:37returns are not so high in the lower end
1:01:40of the uh end of the distribution so
1:01:43this shouldn't be a total surprise
1:01:46heterogeneity has long been a Hallmark
1:01:48of the micro Enterprise sector Lisa PD
1:01:50has has worked on this back from the
1:01:521980s talking 1980s talking about the
1:01:55heterogeneity in the sector itself and
1:01:58similarly if we look at Adventure
1:02:00Capital firms they typically make all
1:02:02their Returns on a few Superstar
1:02:03performers again quite a lot of variants
1:02:05a lot of firms lose money but a few
1:02:07firms make make a lot of money and in
1:02:10some senses what we see from the from
1:02:12the capital drops kinds of or or and
1:02:15even some of the loan kinds of papers is
1:02:17that sort of uh that sort of uh
1:02:19heterogeneity uh uh as as as well
1:02:24um and uh uh you know one one of the
1:02:28most consistent results where it's
1:02:29reported is is what I've sort of just
1:02:31shown you in those graphs is that the
1:02:32effects are concentrated in the top part
1:02:34of the distribution
1:02:36um and so I'll say this and then I'll
1:02:39and then I'll take barbers barbers
1:02:41question here and obvious challenge is
1:02:43that if they're concentrated in the top
1:02:46end of the distribution we we we may
1:02:49lack enough power in in the in the
1:02:51typical RCT to be able to to sort of
1:02:53really uh uh sort of characterize that
1:02:57that part of the distribution let me let
1:02:59me pause there and and Barbara take your
1:03:01take your question
1:03:05thank you very much so um you mentioned
1:03:08that the
1:03:10specific qualities could determine
1:03:13return to Capital what of a specific
1:03:17country business environment how does
1:03:20that also affect
1:03:22yeah so there's so uh
1:03:26in the so certainly that's going to
1:03:29affect returns to Capital across the
1:03:31whole sector
1:03:32um
1:03:33uh in the
1:03:36the very small Enterprises that we're
1:03:40looking at uh in most of these
1:03:43experiments and I would say the the work
1:03:45that Dave and I have done in both uh in
1:03:47in well in Mexico and in in Ghana in in
1:03:50Sri Lanka the work that I've I've
1:03:52referenced there in in India all very
1:03:56small kind of almost local neighbor
1:03:58mostly local neighborhood businesses and
1:04:01maybe the overall business environment
1:04:04in the country is not going to have as
1:04:06big an effect on these Enterprises as it
1:04:09would on in a minute I'll talk about you
1:04:13know about about work with banks and and
1:04:16and loans that go up to much larger size
1:04:18businesses as it would with those larger
1:04:21size businesses where their growth is
1:04:24going to start to bring them into a
1:04:26regulatory structure and into a uh and
1:04:28into uh uh uh uh you know the the need
1:04:33to sell uh to a much more varied number
1:04:36of of customers and to work outside of
1:04:39their own more narrow networks and
1:04:40things things of things of that sort
1:04:42where the business environment is going
1:04:44to be is going to play is likely to play
1:04:46a much larger role
1:04:49okay so let me let me um uh push on so I
1:04:54want to I want to think about this this
1:04:56uh I want to think a bit about this idea
1:04:57of heterogeneity and selection I think
1:04:59there's been a tremendous amount of very
1:05:01interesting work that's gone on in this
1:05:04uh uh you know in this area and I'll and
1:05:06I'll and I'll talk a bit about about
1:05:08some of that in uh in that that has gone
1:05:11on and is is going on
1:05:13um first question we might ask ourselves
1:05:15is do entrepreneurs self-select into
1:05:17sources of of capital that is do they do
1:05:20they self do the ones who have high
1:05:23returns to Capital are they the ones
1:05:25that that that apply for uh uh that are
1:05:28more likely to apply for loans and more
1:05:30likely to to push to get loans and the
1:05:33second question which uh which is uh
1:05:36which which is is related to this as
1:05:39well is can we identify those
1:05:41entrepreneurs with higher returns and if
1:05:43so how and this is a place where I think
1:05:45you know there's been there's been a
1:05:46there's been quite a lot of work that's
1:05:48uh that that I find really quite
1:05:50interesting that that that is uh and
1:05:52interesting and promising and I think
1:05:54it's an area where where there's more
1:05:55that that that that that that can be
1:05:57that can be done so on the first
1:05:59question the literature on standard
1:06:01credit contracts is somewhat is somewhat
1:06:02mixed um and you know jinkai and others
1:06:05have a boxed up lid on on on this on the
1:06:08on microcredit that that sort of reviews
1:06:10this literature
1:06:11um Rachel meager uses a uh you know sort
1:06:14of a Bayesian hierarchical analysis to
1:06:16look at at a collection of uh rcts on on
1:06:20microcredit finds what people interpret
1:06:23as fairly weak effects of of uh of uh of
1:06:28of the credit on firms on on firms it
1:06:32may have effects in other areas but on
1:06:34but on firms and firm growth relatively
1:06:36relatively uh weak effects but there are
1:06:39several papers that show you know
1:06:41somewhat larger effects on uh uh larger
1:06:45uh firms uh with with bank loans and and
1:06:50I guess one of the first in this area uh
1:06:52the managing to float paper from 2004
1:06:54but that paper had been around as a
1:06:56working paper for quite a long time so
1:06:58it was it really a predate stat it
1:07:01really is one of the early pieces on
1:07:03this uses a bank lending policy change
1:07:06uses bank lending policy changes more
1:07:09there was more than one change where uh
1:07:11the law changed the the Indian
1:07:13government had set aside programs for
1:07:16firms below a certain size and that
1:07:19threshold changed at different points in
1:07:22time and so firms of a given size were
1:07:25uh intermittently eligible and not
1:07:28eligible for these loans these are much
1:07:31bigger firms than the than the than the
1:07:33rcts that have just that I've just
1:07:34talked about
1:07:36um uh on the Range that the thresholds
1:07:38were set at the ranges of 150 000
1:07:41roughly at the given the rupee dollar
1:07:44exchange rate at the time to about seven
1:07:46hundred thousand dollars so so much much
1:07:48much larger than the ones that we that
1:07:51we talked about and they find annual
1:07:52returns of of about a hundred percent a
1:07:54year uh and they spend quite a lot of
1:07:56time convincing us that these are
1:07:59actually this is actually a marginal
1:08:02Capital that this is not sort of
1:08:03substituting for other other Capital
1:08:06that this is that this is uh that this
1:08:08is uh that this is access to New Capital
1:08:10there's there's a a a another uh current
1:08:13working paper by jinkai and Adam Seidel
1:08:16that is quite quite an interesting
1:08:18experiment uh with an RCT and a banking
1:08:21partner in China again quite larger
1:08:24loans quite larger comp companies loans
1:08:26of up to eighty thousand dollars
1:08:29um and the the main point of their paper
1:08:32is to kind of look at the indirect
1:08:34effect of these loans so they work in 78
1:08:37retailer wholesaler markets in a single
1:08:40city in China and they they offer the
1:08:43loans at a slightly below Market
1:08:46interest rate so unfavorable terms but
1:08:48not ridiculously favorable terms
1:08:51um to either eighty percent fifty
1:08:53percent or zero percent of the firms in
1:08:55a given market and the variation between
1:08:5780 and 50 in particular gives them the
1:09:00ability to to look at what's the effect
1:09:02of a competitor of yours getting loans
1:09:05when you're in the control group in the
1:09:07same Market uh where they were I want to
1:09:10focus on the fact that they find quite a
1:09:12substantial of you know
1:09:15um firm specific effects uh now it
1:09:18almost gets completely offset by the
1:09:21negative effects on competitors so
1:09:22there's no Market level effect of this
1:09:24credit or there's not a very large
1:09:27Market level effect of this credit but
1:09:29there's a there's a large firm specific
1:09:31effect that that suggests that the an
1:09:34individual firm has quite a high return
1:09:36uh to to uh an injection of capital uh
1:09:41in the in the firm so the it the
1:09:43intention to treat sales increased by
1:09:45about 10 log points with with uh uh 28
1:09:48take up for example so so you know quite
1:09:51a quite a substantial uh uh quite a
1:09:53substantial increase in in sales of the
1:09:56treated firms
1:09:58um
1:09:58which in the markets where other firms
1:10:01are not treated are offset uh to some
1:10:05extent by by losses and sales of the
1:10:08other firms in those uh in those markets
1:10:11um a second project in this area which I
1:10:13which I think is is always also think is
1:10:15is is is very interesting and promising
1:10:18in terms of
1:10:20thinking about how do we how do we
1:10:22identify who the potential uh who who
1:10:26the firms are that can potentially uh
1:10:28benefit from larger uh loans uh Garrett
1:10:32Dean Carlin Adam Osmond have a a project
1:10:36which ultimately landed in Egypt after
1:10:38trying uh trying it trying out in
1:10:40several different countries over a large
1:10:43number of of years but but uh they work
1:10:46with the Alexandria Business Association
1:10:48to offer loans that are four times as
1:10:52large as the as the uh uh uh as the
1:10:56previous limits uh borrowing limits of
1:10:59of the of the association so all at once
1:11:02overnight your your offered a loan which
1:11:05is four times larger than you could
1:11:07possibly have gotten uh prior to prior
1:11:10to that
1:11:11um they find modest average effects for
1:11:14the larger loans but they but they find
1:11:16quite an interesting pattern in
1:11:17heterogeneity and I want to spend a
1:11:19little bit of time thinking about how
1:11:20they do this because I think this is
1:11:22this is a paper and a and a technique
1:11:25and a and an approach which is
1:11:27um interesting and promising and and um
1:11:30uh one that uh uh where I think
1:11:34additional work would be uh would be
1:11:36merited and and helpful so first they
1:11:39start with a sample of what what is the
1:11:42experiment what what do they do they
1:11:43start with a sample of previous
1:11:44borrowers who were selected by the loan
1:11:47officers of the ABA of the Alexander
1:11:50Business Association for their potential
1:11:53uh as as being able to absorb a much
1:11:57larger loan so this is they're using
1:11:59local knowledge of the of the loan
1:12:01officers to to to select these uh to
1:12:03select these uh these uh uh candidates
1:12:07for the larger loan within this subset
1:12:10they asked can the officers the loan
1:12:12officers predict who will succeed
1:12:15well it turns out not so well
1:12:18um in fact there's no positive
1:12:20correlation between borrower performance
1:12:22with the standard loan and with the four
1:12:24times loan so the experience I have you
1:12:28know having lent to David you know with
1:12:31with the standard loan over the years
1:12:33and see what see seeing what he does
1:12:35does with this with the with the
1:12:37standard uh loan
1:12:40um that doesn't help me predict what
1:12:42he's going to do with a loan that's four
1:12:44times as as as big it turns out that
1:12:46that those that those performances are
1:12:48somewhat are somewhat different
1:12:50um instead what they find that is that
1:12:54the top performers with the larger loans
1:12:56are better predicted with psychometric
1:12:58and cognitive data that the that the ABA
1:13:01that they collect as researchers but
1:13:03that the ABA does not collect and in
1:13:06fact most lenders uh don't uh you know
1:13:08don't don't typically uh don't typically
1:13:11uh collect although Azam kwaje has some
1:13:14work in this area that has done things
1:13:15with the psychometric uh uh measures and
1:13:19and these measures uh with with with
1:13:21lenders uh so some sometimes I think
1:13:24people are using these these data
1:13:27um instead the type of firms doing best
1:13:29with loans are among the worst
1:13:30performers in the control group
1:13:31suggesting that past performance is a
1:13:33poor predictor and why well the authors
1:13:36argue that all entrepreneurs are
1:13:39optimists but some are really incredibly
1:13:42excessively optimistic and the over
1:13:45optimism pays for small loans because I
1:13:48can't get myself into too much trouble
1:13:50making these smaller Investments but
1:13:52once you give me a loan that's four
1:13:54times as big I can get myself in trouble
1:13:57with making the you know the wrong
1:13:59investment and so what are the what are
1:14:01they they find well first what are they
1:14:03going to do they'll take their data with
1:14:06the psychometric and cognitive data
1:14:08they'll take the standard lender lending
1:14:10data that comes out of the loan files uh
1:14:13past history repayment rates things
1:14:15things of that sort and then they're
1:14:17going to also do a version of this
1:14:19exercise where they take both of the all
1:14:21the data they have both the standard
1:14:23data and the psychometric data and
1:14:25they're going to run a machine learning
1:14:27uh prediction uh exercise where they
1:14:31split the sample 50 50 into a training
1:14:33sample and a testing sample and in the
1:14:37training sample they're going to ask
1:14:39who what are the characteristics what
1:14:42are the types of entrepreneurs who do
1:14:45well with the loan right so it's so it's
1:14:49a it's a it's it's a bit like the the
1:14:52graph I showed you from our Sri Lanka
1:14:54paper where I've got the heterogeneity
1:14:56and returns based on uh in that case it
1:14:59was it was a standard regression but
1:15:01regression with certain characteristics
1:15:03right this is going to be machine
1:15:04learning that's going to give me a
1:15:06prediction of what what variables what
1:15:08types predict High returns they do that
1:15:11in the training set and then they're
1:15:13going to run regressions and they're
1:15:15going to show us the regressions on the
1:15:16testing set so a completely different
1:15:18group of of entrepreneurs if what they
1:15:22were getting out of the training set was
1:15:24there by chance we should expect it not
1:15:26to show up in the in the testing data
1:15:29and so uh what they find is is is
1:15:32summarized on this on this uh on this
1:15:34this graph and I think let me point to
1:15:37the the one on the the one on the left
1:15:39which uses only the the psychometric
1:15:41cognitive data it's
1:15:44um
1:15:45it's the only one where these results
1:15:46are statistically significant but what
1:15:49they find is those who do the worst that
1:15:53work um entrepreneurs with the with
1:15:57characteristics that predict
1:15:59underperformance with the big loans
1:16:01actually lose quite substantial uh
1:16:04amounts on those on those Investments
1:16:07the loans make them worse off their
1:16:09profits are lower there they make risky
1:16:12Investments that that that blow up and
1:16:15and they and they and they and they lose
1:16:17money on them the ones at the top end uh
1:16:21uh do phenomenally well and make uh and
1:16:25make very high returns
1:16:28um notice that in the middle graph which
1:16:31is the the predictions that they would
1:16:33the same exercise that they would do 50
1:16:3750 split a training sample a testing
1:16:39sample uh uh trying to predict
1:16:42characteristics types of of
1:16:45entrepreneurs
1:16:48using the data that the ABA has in its
1:16:52own files they don't find that you still
1:16:55get this pattern of of upward movement
1:16:58but nothing that's but it's highly
1:17:00insignificant so nothing that is
1:17:02statistically significantly different
1:17:04about what I can predict out of the the
1:17:07loan files that tells me about who's
1:17:10gonna who's gonna do well in this uh
1:17:12with these loans and who's not going to
1:17:13do well so I the this kind of
1:17:16heterogeneity and this sort of exercise
1:17:18of how is it that we can uh that we can
1:17:21try to predict who's going to do uh well
1:17:24with things now
1:17:25um you know that they're they're not the
1:17:28first to try to they're not the first to
1:17:29try to do this and let me jump down to
1:17:31the bottom of the slide
1:17:32um uh uh they're they're um there's
1:17:36there's work from business plan
1:17:38competitions where we see panelists
1:17:41trying to predict uh where uh uh where
1:17:46uh which which types of firms will do
1:17:48will will grow uh will grow the most and
1:17:51the experience there is uh is more mixed
1:17:54in terms of what what people are able to
1:17:56to predict but
1:17:58the the broader area of kind of thinking
1:18:02about what is what's what's the
1:18:04heterogeneity and how can we how can we
1:18:06uh uh uh reach that x uh that that
1:18:09heterogeneity how can we figure out what
1:18:10the heterogeneity is I think is
1:18:12something that uh uh that has uh that
1:18:14has a lot of uh a lot of uh uh uh a lot
1:18:17of potential and and it's and it's an
1:18:19and it's a you know still quite a young
1:18:21uh young research uh young research area
1:18:24okay let me pause there for a second and
1:18:26see if there are any uh any questions
1:18:32okay then I'll jump on to the to the
1:18:34next part of this which is uh thinking
1:18:37about the other side of what I said so
1:18:39this is that first bit was on the
1:18:41heterogeneity and how do we think about
1:18:43the heterogeneity of the owners and
1:18:45borrowers and who's going to gain the
1:18:46most from these uh from these loans the
1:18:49second part I want to think about is you
1:18:51know risk rationing and thinking about
1:18:53flexibility of credit and uh um of
1:18:56course business Fortunes in any context
1:18:59vary quite a lot over time we might
1:19:01think that's especially the case in
1:19:03lower income countries where where
1:19:05things are uh often somewhat more uh uh
1:19:09where markets are are are are are
1:19:11sometimes more uh uh sometimes more
1:19:14turbulent than and more and and more
1:19:16seasonal and things things of things of
1:19:18that sort so in the absence of the
1:19:21ability to ensure against this type of
1:19:23variance
1:19:24um is a rigid debt contract optimal
1:19:27where we have lots of variants you know
1:19:29I think Finance Theory would tell us
1:19:30this is a place where we should have
1:19:32Equity not that we should have that
1:19:34that's you know more flexible uh that
1:19:36allows flexibility
1:19:38um so I'm going to start here with
1:19:40thinking about about that sort of more
1:19:43or less standard debt contracts that are
1:19:45tweaked in a way that makes them
1:19:47somewhat more flexible and I think the
1:19:49first work in this area that you know uh
1:19:51arguably is the work that Erica field
1:19:54and others did uh in uh in India where
1:19:58they uh did it conducted an RCT where
1:20:01the treatment group the control group
1:20:03got a standard loan and the treatment
1:20:05group got a loan that allowed them to
1:20:07make no payments for the first two
1:20:09months a grace period these were all
1:20:12second time borrowers it was group
1:20:13lending and it was it was it was India
1:20:16uh uh barbonian uh and Iowa have a a a a
1:20:21a a a a paper in India where they also
1:20:24offer flexibility uh to delay uh the uh
1:20:28payments for up to three months that is
1:20:30the to stop making payments on your loan
1:20:32for three months during uh during the
1:20:35payment cycle at any point not just at
1:20:38the beginning but at any point
1:20:39individual loans experienced borrowers
1:20:41their idea was lots of these borrowers
1:20:44are in places where there are say
1:20:46festivals that have shocks to demand at
1:20:49a certain point in the year and allowing
1:20:52them to delay payments during that time
1:20:54of the year gives them Capital at a time
1:20:56when demand has kind of shifted out and
1:20:58the and the returns are quite High
1:21:01um third paper in Bangladesh again
1:21:03offered uh
1:21:05flexibility to defer the loan payments
1:21:09twice during a a 12-month cycle again
1:21:12individual loans again experienced
1:21:13borrowers and then a final paper uh by
1:21:17Bruno doll in uh in uh in Colombia uh
1:21:21first-time borrowers individual loans
1:21:24three months where they were allowed to
1:21:25make interest payments only not defer
1:21:28the whole loan they still had to pay the
1:21:30principal but interest payments only
1:21:31they find uh so what so what do we find
1:21:34you know what do we find out of this
1:21:35first of all field and all find that the
1:21:38grace period leads to much higher
1:21:40investment rates very high returns of
1:21:43about an 11 monthly return uh compounded
1:21:47and that the capital in the business
1:21:49after a year is 82 percent higher but
1:21:52they also find quite substantially
1:21:54higher default rates so that it looks
1:21:57like it's not profitable for the mfi to
1:22:00to to to to offer that off of that
1:22:02contract
1:22:03but if you look across those even those
1:22:06four studies that I just uh written down
1:22:09there you get a mixture of high risk
1:22:11High return Investments that's true in
1:22:14field and and Battaglia uh you get uh
1:22:17the ability to meet demand searches as I
1:22:20talked about in in barbonian agrawal oh
1:22:23you get no improvement at all you get
1:22:25higher default or you don't get higher
1:22:27default there are differences that maybe
1:22:29differences in the context that made
1:22:31differences whether it's group lending
1:22:33or individual lending it may be
1:22:34differences in uh in the uh uh in the
1:22:38types of borrowers whether the first
1:22:40time or or or or experienced borrowers
1:22:42and things like that again this seems
1:22:44like an area where digging in to try to
1:22:47understand who benefits from the
1:22:49flexibility of the loans and who is it
1:22:51who who who uh who uh may may not
1:22:55benefit is uh is is is a right a ripe
1:22:58area that we might that we might think
1:22:59about kind of doing something more
1:23:01systematic with are there differences in
1:23:03the design of clients is the predictable
1:23:05heterogeneity this paperback
1:23:08Kryptonian Osmond sort of says that look
1:23:11there's more heterogeneity within any of
1:23:13these samples than there is across
1:23:14samples and it's really there that we
1:23:16need to be we need to be looking and
1:23:18that's that's kind of uh what I want to
1:23:20say okay in the last few minutes that I
1:23:22have I'm going to say a bit more about
1:23:24thinking about pushing the debt this
1:23:27this sort of uh flexible debt contract
1:23:30even further to kind of pushing it
1:23:32towards something that looks more like
1:23:34an equity contract and I'll talk about
1:23:36uh one uh one paper that that one one
1:23:40experiment One One Pilot uh of an
1:23:43experiment that David Suresh and I did
1:23:45in uh in Sri Lanka which uh which I
1:23:47think of as kind of a failed experiment
1:23:49uh um that attempting to make uh a leap
1:23:54into micro Equity contracts so we we
1:23:57took a series of uh nine nine
1:24:00Enterprises and invested uh uh five
1:24:02thousand dollars uh on a royalty-based
1:24:05revenue sharing uh contract in lieu of
1:24:07fixed payments so not not share of
1:24:10profits uh we knew better than to try to
1:24:12figure out what the profits of these
1:24:13businesses were but share of of revenues
1:24:16turns out that's also pretty hard to to
1:24:18to to figure out and and and monitor
1:24:21um the principal can be paid partly
1:24:22through through the royalty rate and and
1:24:24we set up sort of the contract where it
1:24:26was flexible that that these these these
1:24:29with with choice of parameters you could
1:24:31make it more equity-like or more
1:24:32debt-like
1:24:34um
1:24:34um the failure actually didn't come as
1:24:37much from uh Under reporting of of
1:24:40royalties as it did just from the
1:24:42inability to to enforce
1:24:44non-non-repayment it turns out that the
1:24:47Sri Lankan legal institutions didn't
1:24:49recognize the contract as
1:24:50collateralizable they didn't recognize
1:24:52it as a debt contract it was an
1:24:54innovation that that required a sort of
1:24:57change in the legal structure that that
1:24:59wasn't uh you know that that made it
1:25:01difficult to uh
1:25:03uh to uh uh to to enforce uh in in in in
1:25:08the in the courts
1:25:10um so that was perhaps a leap too far
1:25:13towards the equity Direction but there
1:25:15are a couple papers out there that I
1:25:17find interesting in this area and I
1:25:18think they're um there again it's a
1:25:21place where uh the the literature is
1:25:23ripe for more more Innovation and more
1:25:26experimentation with with what we might
1:25:28get to work to to bring in sort of more
1:25:31um risk sharing into the external uh
1:25:34into the external uh credit so one paper
1:25:37bar it all uh uh uh works with an mfi uh
1:25:44in uh in Pakistan uh that offers a
1:25:46rental contract
1:25:48um and it it is a for for large lumpy
1:25:50assets so think of a like a
1:25:51three-wheeler or or something like that
1:25:54um where the borrower has the right at
1:25:57any point to return the asset to the to
1:26:00the lender uh and uh and and you know in
1:26:03that in that case they've lost their
1:26:05rental payments that they've made to
1:26:07that point but it's rent to own so after
1:26:09a certain period period of time when
1:26:11they've when they've made a sufficient
1:26:14number of rental payments the asset is
1:26:16uh they they fully own the asset their
1:26:19share of ownership starts at 10 and then
1:26:22each time they make a payment it
1:26:24increases until they've until they've
1:26:26they've bought they've bought the entire
1:26:27asset so there's some risk some
1:26:29insurance here in the sense that they're
1:26:31able to return the asset
1:26:33um kadara at all uh similar sort of
1:26:36thing with bicycles in in Kenya uh where
1:26:39the payments are linked to measurable
1:26:41sales they work with uh with a candy
1:26:43company that that uh food company that
1:26:46that where they have the wholesale data
1:26:48that allows them to uh to say what the
1:26:51what what the uh what the actual uh
1:26:54sales are and and measure that that
1:26:56quite carefully
1:26:57um the these royalty-based contracts are
1:27:00not so uncommon in Venture Capital
1:27:02there's a company Nigerian company
1:27:04called move that makes loans to Uber
1:27:07drivers that also uses a a similar sort
1:27:09of royalty based uh uh uh contract so a
1:27:13few concluding thoughts
1:27:15um first a couple of takeaways there's a
1:27:17lot of interesting work uh scraping
1:27:19information to Aid targeting and and
1:27:21selection I think there's a lot more to
1:27:23be that can be done in this area but
1:27:25that's that's a place uh thinking about
1:27:27this heterogeneity of entrepreneurs and
1:27:29how do we target this uh better
1:27:32um the second is that that's almost
1:27:33certainly not the first best external
1:27:35credit instrument it may be the second
1:27:37best it may be the best we can do in
1:27:39lots of contexts but moving to more
1:27:41Equity like instruments and figuring out
1:27:42how to do that how to push that down the
1:27:44the the size chain I think is is is
1:27:47quite is is is is is an area that's
1:27:50that's that's quite interesting in both
1:27:51cases I think we do well to look at what
1:27:53Venture capitalists are doing and say
1:27:55how can we do that with smaller firms
1:27:56and perhaps with
1:27:59um machine learning and and
1:28:00computational capabilities that these
1:28:02these opportunities will will open up to
1:28:04be able to do this at scale on a smaller
1:28:07a smaller firm the the last thing I want
1:28:09to say is there's not an enough time
1:28:11this was far too focused on small firms
1:28:13and debt to justify the full agenda
1:28:15that's here on on Capital and uh and
1:28:19firms I needed to say more about angel
1:28:22and Venture Venture Capital uh finance
1:28:25and investment Readiness and other other
1:28:27potential uh High Flyers
1:28:30um so I'll just point to the work on
1:28:32accelerators that Juanita Gonzalez or
1:28:34Reba and others are uh and and some of
1:28:36her collaborators are doing which I
1:28:38think is is a promising area with with
1:28:40sort of faster growth firms and then
1:28:42some work on investment Readiness that
1:28:44David and others have done in in the
1:28:46Balkans kind of thinking about how we
1:28:47get firms uh uh ready to scale up and uh
1:28:51I'll stop there I think we've got about
1:28:53two minutes and I see uh
1:28:55um has a has a question so let me let me
1:28:59uh yes hi there so I have a question
1:29:02that in this rcts experiments especially
1:29:05where there are the forms have been
1:29:07given the flexible
1:29:09um uh time of borrowing do you
1:29:11experience like there is a correlation
1:29:13between a higher default rate and their
1:29:16productivity like is it necessary that
1:29:19these these firms if they default more
1:29:21on their loans are they more productive
1:29:23or uh yeah so is it the same
1:29:26productivity or is are there and the
1:29:28other barriers that are affecting these
1:29:30development so the one the one
1:29:32experiment that uh that finds higher
1:29:36default rates is the field that all
1:29:38project in in India the others uh don't
1:29:41find higher default rates than they than
1:29:43than in the standard in the standard
1:29:45contract
1:29:46um and in and in uh the field at all
1:29:50paper as I recall it it
1:29:52it's the people who default are the ones
1:29:56who decide I've grown big enough that
1:29:58I'm never going to have to get loans
1:30:00from you again and therefore the
1:30:02relationship doesn't matter as much to
1:30:04me so that would suggest that they're
1:30:05likely to be the high uh ability uh
1:30:09entrepreneurs I mean the let me say that
1:30:12in general there's one of the
1:30:14constraints and it's the it's the it's
1:30:16the constraint that Garrett Brian and
1:30:18Dean Carlin and Adam Osmond went after
1:30:20is that
1:30:22um there there's a sense that uh that
1:30:26there's a that growing out of a micro
1:30:30Enterprise out of a micro lender is
1:30:32difficult because the best borrowers for
1:30:35micro lenders are are borrowers that the
1:30:37micro lender wants to hold on to those
1:30:40micro lenders typically in most
1:30:42countries are not part of credit bureaus
1:30:44so nobody else knows what your what your
1:30:48uh uh repayment rates are uh in the in
1:30:52except the micro lender and you know
1:30:55part of what they were trying to do is
1:30:57figure out how can we get micro lenders
1:31:00to pass off their clients to to sort of
1:31:03more formal lenders that's initially
1:31:05what they they tried to do and so they
1:31:07found this the mfi that would scale up
1:31:09the loans within within the mfi but um
1:31:12but I think that that does suggest that
1:31:15these these high ability businesses that
1:31:19are bought you know that are borrowing
1:31:21in these in the these uh from these uh
1:31:24these at some point they may they may
1:31:29feel like they're self-sustaining they
1:31:30don't need external credit and then it
1:31:32may be that that's when they say look at
1:31:34this relationship is not important to me
1:31:35anymore and and I it might increase my
1:31:37chances of of defaulting but I think
1:31:39that margin is also one where where
1:31:42there's where there's uh
1:31:44um where there's a sense that they're
1:31:46challenges let's say uh in in in the way
1:31:49the capital markets work going across
1:31:51that that barrier
1:31:54okay I'm out of time we're out of time
1:31:57let me close up here and just remind
1:31:59everybody that uh thank you first of all
1:32:02thank you all for participating for for
1:32:04joining uh remind you that uh the next
1:32:07lecture will be tomorrow at the same
1:32:09time uh Rocco machiavello uh we'll talk
1:32:12about uh Contracting issues and and and
1:32:14uh and uh in uh at at the at the same
1:32:19time so
1:32:20thank you all thank you David and and
1:32:22Dave for for organizing this and uh
1:32:26good evening good afternoon good night
1:32:30Wherever You Are
1:32:32good morning from Boston thanks again
1:32:34Chris yes thank you so much
1:32:37thanks David

About this transcript

This page contains the full transcript of Lecture 5: Capital and Labour Frictions with David McKenzie and Chris Woodruff by International Growth Centre, generated from the public captions YouTube serves with the video. The transcript has 16,687 words across 2,388 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.