YT YouTube2Text

YouTube2Text → Transcripts → BREAD-IGC virtual PhD-level course 2026 | Lecture 7 Political Selection and Representation

BREAD-IGC virtual PhD-level course 2026 | Lecture 7 Political Selection and Representation — Transcript

by International Growth Centre · 15,720 words · 2,608 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

1:28Well, hello. Welcome everyone back to
1:31the bread lecture series. We're excited
1:33this week to have our seventh lecture on
1:36political selection, voice, inclusion,
1:40and
1:42joining us this week are Rohini Pande,
1:44professor of economics and director of
1:46the Economic Growth Center at Yale,
1:49and Katherine Casey, professor of
1:51political economy at Stanford
1:53University.
1:54I believe both of of you have been bread
1:58lecturers before and both have had
2:01interaction with IGC for a number of
2:03years and so we're really grateful for
2:05your your time especially being based in
2:08the US and waking up
2:10quite a bit earlier than your normal
2:12working hours today
2:14and share some thoughts with all of us
2:16on political selection.
2:19Um just briefly
2:21everyone's probably heard me say this a
2:23dozen times but
2:24uh if you have questions please put them
2:27into the Q&A box and if we could try to
2:29keep these very
2:31related to to the content that would be
2:33great.
2:35The second thing is
2:37if you're interested in being a live
2:38panelist please raise your hand and I'll
2:40randomly select
2:42about 15 of you and then you'll have the
2:45option to ask your questions live at the
2:47end.
2:48And
2:50the last thing is that if you've missed
2:52recent lectures
2:54you're welcome to
2:56look at our website and all the
2:58recordings the slides and the reading
3:01lists are all there. So
3:04I'll put that link in the chat as well.
3:07If there's nothing else then I will hand
3:09over to Kate who's going to kick us off.
3:14>> Great thanks so much and hello everybody
3:16you're very welcome. We're excited to
3:18talk to you today about political
3:20selection.
3:22So let me start by saying that this idea
3:25that how we select our political leaders
3:27is important is not a new one. So James
3:30Madison was writing about this almost
3:31250 years ago and he argued that the
3:34primary objective of any political
3:35constitution should be some sort of
3:37system of leadership selection so you
3:39yield rulers with the wisdom to pursue
3:42the common good.
3:44Now fast forward to 20 years ago and Tim
3:47Besley who you heard from earlier said
3:50you know depi- despite this importance
3:52until only recently the modern political
3:54economy literature has not only
3:56neglected the problem of political
3:57selection, and it's been positively
3:59hostile to the topic. Now, if I had
4:01Tim's posh English accent, that would
4:03sound even even more uh persuasive. Um
4:07but fortunately, the last 20 years of
4:09research has made quite a lot of
4:11progress in correcting that neglect, and
4:13that's what Rohini and I are here to
4:15talk with you about. So, let's just
4:18ground this in in big picture. Why are
4:20we talking political selection when in
4:23four, you know, basically a class on
4:25economic development? And the key idea
4:28is that, you know, governments have
4:30these core functions that undergird
4:32market activity. So, they provide local
4:34public goods, they, you know, define and
4:37protect property rights, correct
4:38externalities. And the key idea is that
4:40the government's ability to do these
4:42things well depends in part on the
4:45political leadership. So, what is the
4:46human talent in governments, and what
4:48are the rules that govern their
4:49behavior?
4:50So, just putting that same idea on a
4:53graph, what we have here is kind of
4:55mapping state capacity, political
4:57institutions, and economic growth. So,
4:59if you look at the horizontal axis, this
5:01has data from the World Bank, government
5:04effectiveness. Think about this as as
5:05state capacity. So, you heard about
5:07Adnan and Asim, their lecture talked a
5:09lot about that. The horizontal axis is
5:12what Rohini and I are here to talk
5:13about. So, voice and accountability,
5:14think about this as more broadly
5:17institutions of politics.
5:20And then, the dots are countries
5:23proportional to their population size,
5:25and the shading of the dots is is about
5:28income. So, so darker dots, higher
5:29income.
5:30So, what's the big picture? Well, you
5:32can see these country dots cluster along
5:35the 45-degree line. So, what does that
5:37mean? So, that means to some extent
5:39state capacity and political
5:40institutions move together, and they
5:43both are positively correlated with
5:46economic prosperity. Okay? That's the
5:48kind of idea that, you know, governments
5:49matter for economic performance.
5:52But, second thing you'll notice is that
5:54there there's a bunch of countries that
5:56are on these kind of off-diagonal
5:58quadrants, about a quarter of them,
6:00right? And that suggests that what you
6:02learn from Asim and Adnan about state
6:04capacity is related, but is somewhat
6:06different. It's not the same phenomenon
6:07as what we're going to be talking about
6:09when we talk about political
6:10institutions and political selection in
6:12particular.
6:13And then, the third thing to think about
6:15is if you look at that lower left-hand
6:17corner, that's just a really challenging
6:20set of countries, right? They're very
6:21low income, they have weak weak state
6:24capacity, and weak political
6:25institutions. So, when you're thinking
6:27about how will those lower-income
6:28countries make progress, you know,
6:30diagonal progress toward higher income,
6:33one part of that strategy might be to
6:35improve political institutions and
6:36political selection as one part of that.
6:40But, what's the problem? Well, you may
6:42have heard a lot of, you know, popular
6:43discussion about democratic backsliding.
6:47Right? So, how are these measures doing
6:49in the recent past? What is the trend
6:51that they're on?
6:52Well, you can see the these pretty
6:54precipitous downward-sloping lines.
6:56These are electoral democracy index.
6:58That orange one is for the OECD. So,
7:00this is democratic backsliding even
7:02within the wealthiest countries. The
7:04very steep pink line is the democracy
7:07index in non-OECD countries. So, think
7:09about that as low- and middle-income
7:10countries. So, what we're seeing is a
7:12sectoral downward trend in the strength
7:14of democratic institutions, particularly
7:16for low-income countries. You know,
7:18accompanied by a little bit of a decline
7:19in state capacity as well, but it's this
7:21pink line that that people are worried
7:24about. And people have noticed, right?
7:26People are mad. So, here's just a bunch
7:28of survey data from around the world
7:31showing that people are upset with their
7:33own politicians and the institutions of
7:36representative democracy. So, here's
7:38some data on the US that only 16% of
7:41survey respondents trust the federal
7:42government and a third identify
7:44politicians politicians themselves as
7:47the biggest problem with the political
7:48system, right? So, we're really going to
7:49focus on that.
7:51Uh across the OECD, across you know,
7:53Europe and other wealthy countries,
7:55those who have trust in government are
7:57fewer than those who lack trust by about
7:595% points. And then across 30 different
8:02African nations, there's still a lot of
8:05support for democracy. So, 75% believe
8:07elections are the best way to choose
8:08leaders, but only 37% are satisfied with
8:12how their own democracy is working.
8:15So, the challenge is if politicians are
8:17big part of the problem, how can we get
8:19better people into government, right?
8:22Into elected office, into public
8:23service.
8:24So, solving that problem involves
8:26improving political selection. And to do
8:29so, roughly runs over three stages. We
8:32have to think about who enters politics,
8:34who puts themselves forward to be
8:35willing uh to serve,
8:38who gets picked up by parties. It's
8:40really hard to win elections and as an
8:42independent in most electoral democracy,
8:44and then who wins. So, you see already
8:47these three stages we're talking about,
8:48you got to think about who are potential
8:50candidates,
8:51who are the gatekeepers, right? You can
8:53think about party leaders naturally as
8:54gatekeepers, and and what do voters want
8:57when you think about who wins the
8:58election. So, so we'll think about those
9:00three stages and those three sets of
9:02actors.
9:04Okay, let's ground ourselves in theory.
9:07So, we have classic models um that show
9:11how elections can serve as a pure
9:14accountability device.
9:16They can serve as a pure political
9:19selection device,
9:20or they can serve a combination of both.
9:23And I'm going to quote Jim Fearon here.
9:26And so, we're thinking about if you
9:28think about the accountability part,
9:29this is what Fred and Claudio were
9:31talking about. So, accountability is an
9:33agency relationship where the principal,
9:36that's us, the public, is empowered by
9:38some formal institutional or perhaps
9:40informal rules to think sanction or
9:42reward the agent, which is the political
9:44leader. And that agent is obliged to
9:47take actions on the public's behalf,
9:48right? So, this is what Fred and Claudio
9:50were talking about. It's really about
9:52incentives, right? We can think about
9:54there's lots of different types of
9:55politicians, good types, bad types. What
9:57we want is to design a system of
9:59incentives incentives so all those types
10:01are actually going to behave in to
10:03forward our the public's interest.
10:06Okay, putting accountability aside,
10:07what's the pure selection part? Again,
10:09quoting Jim, "Selection is about
10:11choosing the type of leader who will act
10:13competently and faithfully in the public
10:15interest." Right? So, you're thinking
10:17about this matters if you know different
10:19types of people are are are better or
10:20worse at governing. And in the extreme,
10:23if you can get the selection part right
10:25and you can get really the best possible
10:26leader, you don't even have to worry
10:28about incentives so much cuz that person
10:30is going to intrinsically behave in a
10:32line with the public interest.
10:34And then we can put these things
10:36together, but a second class of these
10:39models that's related is these citizen
10:41candidate models and these have kind of
10:43limited commitment. So, whatever that
10:45person says or their identity is going
10:47to
10:48guide them to do, once they get in
10:50office, they're going to deliver their
10:52own most preferred policies.
10:54So, this brings in the idea of group
10:56identity. So, think about think about
10:58gender, think about ethnicity, think
11:00about religion. And this is going to
11:02matter. Selection on group identity is
11:04going to matter if public preferences
11:06over policies vary by group or leaders
11:10are more responsive to members of their
11:12own group. So, so when you think about
11:13the citizen candidate, you can model
11:16this as selection on quality. People
11:18often call this valence. Think about
11:20competence, integrity, virtue, uh
11:23motivation, but you can also think about
11:25this as selecting on a group type, a
11:27prominent type that organizes political
11:30and economic life. So, for my half of
11:32the talk, we're going to talk about that
11:33first kind of selection, selection on
11:35quality, and then Rohini's going to talk
11:37about selection on group identity. And
11:40in our halves, we're we're both going to
11:41show you why selection on either
11:43dimension's going to matter for economic
11:45outcomes and public policy. We're going
11:47to show how gatekeepers and voters
11:51affects who can like come up through
11:53these systems when we're thinking about
11:54quality and group. And then we'll each
11:57focus on one specific set of policies
12:00that we see in the real world that many
12:02countries have experimented with.
12:04Okay.
12:04So, here we go.
12:06One thing I want we want as like a big
12:08picture is to think about causal
12:10inference, right? So, how do you know X
12:13causes Y and it's not Z causing Y or Y
12:16causing X, right? And so, why is this
12:18difficult when we talk about selection?
12:20It's difficult because voters have
12:22preferences over both valence and
12:24identity. And both of these dimensions
12:27can contribute to public policy and
12:29economic performance. So, let's let's
12:31just think about uh couple examples. So,
12:34let's think about gender. So, in a world
12:37with bias, these things are going to get
12:38very muddied. Causal inference about
12:40selection over group versus identity is
12:42going to matter. So, think about voters.
12:44Maybe voters start
12:46with a belief, let's say correct or not,
12:49that, you know, men are a little bit
12:50better as political leaders than women,
12:53right? What are voters going to do? On
12:54average, they're going to be electing
12:56more men. They're going to observe men
12:58in leadership. They're going to learn a
12:59lot about men and and how they lead. And
13:02the longer this goes on, they're not
13:04learning about women. Women start to
13:06seem like a very risky choice cuz they
13:07don't have good information. And a
13:09little bit of bias can perpetuate until
13:11you find yourself in a system where you
13:13only have male leaders and you're not
13:15sure, well, are women worse? Are voters
13:16biased? Right? Do Does competence vary
13:19across gender? So, this is going to make
13:21just if you look at a country in a
13:23snapshot, it's going to be difficult to
13:24draw causal inferences about how we got
13:27here. So, think about some stylized
13:29facts from US politics. So, again,
13:31keeping the gender theme,
13:33so it's true, women hold a minority of
13:36political seats. So, think about the US
13:38House of Representatives or the Senate.
13:40But, there's this fact that conditional
13:42running, women win as frequently as men.
13:45And if you look at their performance in
13:47office, women actually perform better.
13:49So, how do we make sense of that
13:51constellation of facts? Well,
13:54uh a decade or two ago, there's a whole
13:55lot of literature that was saying, "Ah,
13:58fantastic. Bias is a thing of the past.
13:59We don't need to worry about this
14:01anymore."
14:02But,
14:02then a bunch of influential people
14:04started to come through and say, "Well,
14:05actually,
14:07I think you do." Right? So, so think
14:08about this this Anzia Berry paper looks
14:11at that performance in the US House and
14:13they say, "If voters are biased against
14:15women or parties are biased against
14:17women candidates,
14:18only the best possible woman on
14:20competence
14:22can win. Right? Cuz you have to be
14:23better than the male group identity to
14:26win. And knowing this, only the best
14:28possible women are going to run in the
14:30first place. So, they're going to get
14:32elected and on average, once they're in,
14:35they're going to be higher quality than
14:37men. Right? So, so that's kind of a
14:40shows you how parsing the selection on
14:42identity versus valence can be difficult
14:44to parse.
14:46So, what we want is these unconditional
14:48probabilities. We want the whole chain.
14:49Who enters? How successful are they in
14:52getting through gatekeepers? And how
14:54appealing they are to voters in getting
14:56election.
14:57The last thing I'll say about causal
14:58inference is when we go back to parsing
15:01what Fred and Claudia were talking about
15:03incentives from what Rohini and I are
15:05talking about selection, that's also
15:06tricky because these things can run in
15:08parallel at the same time. So, think
15:11about, okay, if I want to improve
15:13political selection, maybe I should
15:15offer politicians higher wages, right?
15:17Cuz higher competence people have better
15:19outside options, so let me increase
15:20their pay.
15:21So, there are studies that show, and
15:23we'll get to them in a second, yes, that
15:24works. That strengthens selection on
15:27quality and competence cuz you get
15:29better people in.
15:30But now once I'm in office, think about,
15:32okay, but now because the wage in the
15:34next period is also higher, I now have
15:37stronger incentives to stay in office
15:39and to get reelected. And so, I'm going
15:41to exert more effort, right? So, so
15:43that's the accountability channel,
15:44right? So, the same exact policy change,
15:47I'm changing the wages, is going to up
15:49activate both the selection channel and
15:51the accountability channel. So, it's
15:53just, you know, I mean, it's interesting
15:54to think about, but it's also when
15:55you're designing, particularly empirical
15:57studies, you got to be carefully set
15:59them up so that you can parse what's
16:01going on clearly.
16:02Okay.
16:04So, we have this
16:06recent Nobel laureate agenda that making
16:09the point that inclusive political and
16:10economic institutions matter for
16:12long-run growth. This was coming out of
16:14a time when there was a lot of kind of
16:17macroeconomic debates about these very
16:19long-run deep determinants of economic
16:21growth. So, institutions often proxied
16:24by long-term colonial engagements and
16:26and kind of the long arc of of
16:28consequences.
16:30Also looking at things like geography,
16:32like do you have bad access to trade
16:34routes, do you have a bad disease
16:36environment, are you landlocked?
16:38And they're trying to see, you know, how
16:39do these really kind of deep
16:40determinants shape growth?
16:42And then some studies came along and
16:44said, all of that is clearly important,
16:46but does the identity of an individual
16:48leader, like what can one person make a
16:50difference for economic outcomes? So,
16:52this Jones and Olken paper came along,
16:54which had a very clever identification
16:56strategy, cuz of course, who's in office
16:58and how long they're in office is all
17:00incredibly endogenous to their
17:01performance. And they said, well, what
17:03if we look at people that exit
17:05unexpectedly? So, people that die during
17:08their elected term, right? And we're
17:10just going to compare GDP growth in the
17:12period right before and right after
17:14their passage and see, do we see an an
17:17abrupt change in GDP growth? And the
17:19answer is yes, they do and it's pretty
17:21sizable. So, if you think like a one
17:23standard deviation unit change in leader
17:25quality is going to be linked to this
17:271.5% point change in GDP growth per
17:30year, right? So, even an individual lot
17:32leader can have a substantial impact on
17:36performance of the overall economy. So,
17:38their study they find that this result
17:39really only holds in autocracies, which
17:41kind of comes back to the broader
17:42institutional point that autocratic
17:45leaders are not particularly guarded by
17:47checks and balances from other branches,
17:48so they can have a lot of latitude to
17:50make decisions that could have, you
17:52know, far-reaching consequences.
17:54Another study came along and took this
17:56time series way back, so they had even
17:58more leaders and they replicate, you
18:00know, the same basic relationship
18:02between leadership changes and growth,
18:04but they hone in on a specific
18:06characteristic that relates to
18:07selectional competence, so education.
18:10And they show that when you have a very
18:11highly educated leader who exits, you
18:14have a much steeper drop in GDP growth
18:17in the period immediately after,
18:19suggesting that there is returns to
18:21competence being a good leader in
18:23managing the macroeconomy.
18:25Okay, so leaders are important for big
18:27economic indicators, but they're also
18:29important for more localized measures of
18:32how well is a government working, right?
18:34So, think about Martinez Bravo, she has
18:36this nice paper looking at Indonesia.
18:39And so, they had a massive school
18:40expansion, which really raised
18:42everybody's education and she looked at
18:44these first elections when these newly
18:46educated people have gotten old enough
18:49to qualify to run for public office and
18:52she sees that then you elect better
18:53educated village heads and they provide
18:56more public goods, right? So, think
18:57about health clinics, water taps, right?
18:59So, they're they're delivering public
19:01services better.
19:02Uh
19:03Rohini has a paper
19:05that's looking at, you know, do local
19:07politicians in India, she shows that
19:09better educated individuals, they're
19:11less opportunistic in the way that they
19:13target poverty cards, right? So, less
19:14for themselves and and more for their
19:16deserving.
19:18And so, all of this is like as if they
19:20Leaders matter for macroeconomy, they
19:22matter for public goods, they matter for
19:23targeting.
19:25How do you get better people into
19:26office? And this comes back to what we
19:28were talking about earlier is this if
19:29you think about like an efficiency wages
19:31argument, well, if you want better
19:33people in public office, you should pay
19:35them more, right? And and this is like
19:38they have higher opportunity costs, they
19:40can make a lot of money in the private
19:41sector. So, if you want really competent
19:42people in the public sector, you need to
19:44pay them more. So, what do we know about
19:46this? Well, that seems to be true. So,
19:48this is a study from Italy that's
19:50leveraging, you know, mayors have these
19:53like population thresholds for their
19:55town where you just click over to like
19:565,000 people, you get paid like 30%
19:59more. So, they look at people on either
20:01side of that threshold and they say,
20:02"Yes, higher wages for mayor does
20:05attract more educated candidates and
20:07they kind of improve the way they manage
20:09the government budgets, right? A similar
20:12study shows uh
20:13>> [snorts]
20:13>> the same result in similar strategy in
20:15Brazil where you're going over these
20:16thresholds and that higher pay it
20:18attracts more candidates, you have more
20:19competition, you get better educated one
20:22and then if you look at how productive
20:23they are in their legislative work, they
20:25they seem to be doing better. I would
20:27say that that Italy paper comes back to
20:29this idea that there's selection and
20:31accountability cuz remember the wage at
20:32the next period also is going to
20:34incentivize me to work harder. And they
20:36can they can parse it and sort of say,
20:38"Yes, that is also what worked, but the
20:40big bulk of these results are due to the
20:42selection channel in that particular
20:44context."
20:47Okay, so then
20:49is there any downside to kind of
20:51aggressively pursuing selection on
20:53competence? So, some people worry that
20:56if you prioritize competence too much,
20:59you might be crowding out people that
21:00are intrinsically motivated to serve,
21:04right? So, this is called like social
21:05preferences. Basically, you have leaders
21:07that that put the utility of citizens or
21:09the recipients of public services into
21:11their own utility function. And the idea
21:13is that, you know, super competent
21:15people, there's like a higher
21:17productivity to their efforts, so bigger
21:19return for the same effort. But the
21:21really prosocial public service
21:22motivated guys, they will just
21:24intrinsically work harder. They will put
21:26in more effort. So So how that nets out
21:28is something that people think about. Um
21:31theoretically, it could go either way.
21:32Depends on how these things are
21:33correlated with each other. But what do
21:35we know from the data?
21:37So there's four studies I'm going to
21:39show you. I want to caveat the first
21:40three are not actually about elected
21:42politicians. So this is not quite the
21:43same thing, but they're quite
21:45informative. So these are about
21:46frontline public service workers. So
21:49this trade-off doesn't seem to be a big
21:51problem in recruiting
21:54um community development workers in
21:55Mexico. So they randomized wages and
21:57they find higher wages, you know,
21:59attract better quality applicants. They
22:01make it easier to fill really tough
22:03postings. And there's no
22:05positive, not negative, just zero impact
22:07on average on how public service
22:08oriented these people are. So maybe it
22:11doesn't matter there. Zambia also seems
22:13to not be a huge problem. They're
22:15advertising for community health
22:17workers. They were really saying, "Oh,
22:18this is great for your career and your
22:20upward mobility." It does bring in
22:22people with lower prosocial, you know,
22:24measures, but they're so much better,
22:27they're so much higher competence that
22:28their
22:29performance overall nets out positive.
22:32There's some a similar study in Uganda
22:34that's actually saying, "Well, actually
22:36this is more of a problem. If you're
22:37advertising higher expected pay, you're
22:39not going to get these public service
22:41people uh in and there's going to be
22:43lower effort overall." So So that says
22:45maybe there is is more of a trade-off
22:47there. Okay, this last study is is about
22:50elected politics. So this is in
22:52Pakistan.
22:53And this they show that if you
22:55advertise, these are local elected
22:57offices and say, you know, the social,
23:00the prosocial returns as opposed to your
23:02individual private incentives, they kind
23:04of do a RCT where they're comparing
23:06those two things. And they show that if
23:08you advertise the prosocial part, you're
23:10going to increase entry, those types of
23:12people are going to be more likely to
23:14get elected, and when you trace them
23:16over time, they're actually going to
23:17allocate the public budgets in ways that
23:20are more in line with local preferences.
23:22So, so that's not actually showing the
23:24the competence trade-off part, but it is
23:26showing you that public service
23:27motivation matters in in doing a good
23:30job in local government.
23:32Okay. One thing to note about that last
23:35study is those are non-partisan
23:37elections.
23:38So, in most democracies around the
23:40world, parties are playing a very active
23:43role in gatekeeping who can get kind of
23:46jumping into actually being a candidate
23:48that voters can examine.
23:50Which brings up the question, are
23:51parties good or bad for selection on
23:54quality? Right?
23:56So, there's a lot of reasons to be
23:59skeptical. So, there's a bunch of
24:01studies from a bunch of different
24:02countries that are showing that party
24:04gatekeepers limit the entry of women and
24:07minorities regardless of how qualified
24:09they are.
24:10The idea behind this is party
24:12gatekeepers tend to be elite males that
24:15tend to recruit and promote people like
24:17themselves, right? So, so that would be
24:19I'm kind of selecting on group identity
24:21at the expense of selection on quality,
24:23you know, penalizing certain groups.
24:25Um
24:26another set of studies looks at Sweden,
24:28and as you may know, Sweden has just
24:30absolutely phenomenal data on everyone
24:32including their politicians.
24:34And so, two studies are looking at this,
24:36and they don't reach opposing
24:37conclusions, but they come to slightly
24:38different conclusions about what's the
24:41role of party leaders. Uh the first one
24:43says actually party leaders, they're
24:45they're very merit-based in their
24:47approach to promotion, and so by being
24:49merit-based, they're kind of helping
24:51making Sweden an inclusive meritocracy,
24:53right? So, that kind of representation
24:54and inclusion part doesn't have to be at
24:57odds with selection on quality.
24:59The second paper is about zipper quotas
25:01and Rohini will talk more about this,
25:02but I'll just say they're saying, "Yes,
25:04that's true. There is, you know, you
25:06know, promotion based on merit, but
25:08there's also these incentives that
25:10people only want to promote on merit to
25:13the in so far as it doesn't compromise
25:15their own leadership position." So, you
25:16get like kind of mediocre males
25:18um that are are the ones that kind of
25:20resist these merit-based promotions.
25:23And that that latter point resonates
25:25with this theoretical model that's
25:26basically showing party recruitment of
25:29candidates as this like rent max
25:31maximization opportunity that leads to
25:33mediocrity, right? Not meritocracy, not
25:36the best people, not the worst people,
25:37but just kind of these meh kind of
25:39average level people.
25:41Okay. So, if we if there are at least
25:43some reasons to be skeptical of what
25:45party elites are doing, I want to I want
25:47to turn to one specific policy that
25:50different countries have experimented
25:51with, which is primaries.
25:54Right? So, if you're worried about some
25:56downsides of party gatekeeping,
25:59one thing you can do is give more power
26:01to voters in deciding amongst all these
26:04people that are interested in being the
26:05candidate, who do we want to pick to
26:07represent the party in a particular
26:09race.
26:10So, in the US, Orrano and Snyder have
26:12done a lot of work on this. This This
26:14was motivated the introduction of
26:17primaries was motivated by this
26:18skepticism of what these party elites
26:20are actually doing when they choose
26:22candidates, right? So, they think about
26:24basically there's a bunch of white dudes
26:26in a smoke-filled room and they're
26:28self-dealing and they're making trades
26:29and they're basically picking candidates
26:31with very little regard from who's
26:33really going to serve voters best. So,
26:36primaries in their words are a great
26:38reform to limit the ability of political
26:40and economic elites to manipulate and
26:42profit from the nomination process.
26:44So, these are they would be kind of the
26:45primary advocate view and they've done a
26:48bunch of work on different types of
26:50elections in the US. And they show that
26:52where you can really see primaries
26:54helping and promoting selection on
26:56quality is in safe open seats. So, what
26:58does that mean? It means a seat in a
27:00party's stronghold, so they're
27:01guaranteed to win the general election,
27:03and it's open, so there's no incumbent
27:05sitting there that might be looking for
27:07some sort of privileged action from
27:09their party.
27:11Um and, you know, other models talk
27:13about to other related things, you know,
27:15you might get higher quality candidates,
27:17but they might be more ideologically
27:18extreme cuz often times we worry that
27:21voters that turn out in the primary
27:22stage are much more extreme than voters
27:24in the general election. Um but but it
27:27kind of sets the ground that that this
27:28could be a a good
27:30a good reform for selection on quality.
27:33But there's this kind of remaining
27:35skeptical camp that says, "Okay, sure,
27:39you can give voters the choice, and you
27:41will build representation cuz you'll
27:43give them what they want, but most
27:45voters are not very informed about
27:47politics. They may not understand how
27:49government works. They may not know what
27:51are the kind of skills and competencies
27:53that enable you to perform well on
27:54government. So, if voters are really
27:57poorly informed, they'll you'll increase
27:59representation, but you'll get these
28:00really low-quality politicians in. And
28:03if you think about party leaders as the
28:05selectors instead, they definitely know
28:07how government works. They know the
28:09skills you need, and they have really
28:11good information on these politically
28:13ambitious people that want to be
28:15candidates. So, this kind of raises the
28:17question, well, if you give more say to
28:18voters, take it away from party leaders,
28:21we what's going to happen to
28:22representation, and what's going to
28:23happen to selection on quality? And
28:25coming back to these causal inference
28:27challenges,
28:28it's really hard to parse those things
28:30because, you know, political parties are
28:31incredibly strategic organizations,
28:34right? So, they don't change how they
28:35select candidates for anything but like
28:37extremely strategic and endogenous
28:39reasons, which makes parsing cause and
28:41effect of, you know, whose voice matters
28:44difficult to do.
28:46So, just in these last few minutes, I'm
28:47going to share
28:48uh
28:49a paper that I did with my co-authors in
28:51Kono and ABK in Sierra Leone that was
28:53really trying to target this specific
28:55question of like, what happens when you
28:57give voters more say in selecting
29:01candidates? And so, this is for
29:03parliamentary elections, but you can
29:04think about them like in the US context,
29:06it's basically the same institutions as
29:08like selecting people to be in the US
29:10House of Representatives. Okay. So, we
29:13evaluated a field experiment that did
29:17two things. So, first of all, in the
29:18status quo, what happens? Party leaders
29:21just select candidates.
29:23Comes to the general election, voters
29:25pick between the parties. Why why we
29:27think this is problematic? Well, Sierra
29:29Leone, like the US, has a lot of
29:31geographic sorting with respect to party
29:33preferences. So, there's like deep
29:35strongholds in the north of one party,
29:37deep strongholds in the south. So,
29:39basically, whoever the parties have
29:41picked in their selection process is
29:43almost guaranteed to win the general
29:44election. So, voters, in a way, are a
29:46little bit irrelevant in the
29:47strongholds. So, what could you do
29:49instead?
29:50Well, this experiment was working with
29:52the parties and they held these party
29:54conventions. So, those are That's a
29:56picture of one. Those are two aspirants.
29:58Aspirants mean you're aspiring to be the
30:00candidate. So, these guys are competing
30:01with each other.
30:03They are debating each other on their
30:05merits, on their policy visions, and
30:07they're in front of party leaders, rank
30:09and file members, regular voters. And
30:11these conventions happen. They're
30:13broadcast over local radio many times
30:15over the subsequent weeks. So, voters
30:16that weren't there
30:18uh can see it. So, so that's thing one.
30:19Thing two
30:21is that This is where the research team
30:22comes in. We went around to these
30:24communities and we took a representative
30:26sample of voters and we visited them at
30:28their house and we said, "Okay, this
30:30party, here's an example, they have four
30:33potential candidates in this
30:34constituency, who do you want them to
30:36pick?" Right? Then we just compiled that
30:39data in these one-page reports. You can
30:40see in this example, the party was
30:42looking at four candidates and really
30:44had a very clear front-runner with
30:46voters. And then these reports are
30:48delivered to party leaders.
30:49And the important thing, so this is not
30:51a US primary cuz there's nothing binding
30:53about this. It's really just alleviating
30:56information constraints
30:57for party leaders. So So what do I mean
30:59by that? Well, the conventions give
31:01party leaders really good information on
31:02their own choices, like what do their
31:04own aspirants look like? And then really
31:06importantly, it tells party leaders how
31:08do voters feel about these different
31:10choices. And we're going to find that
31:12that second part is really important
31:13because parties don't actually have very
31:15good information on what voters want.
31:18Okay, so this is a randomized controlled
31:19trial, so this is what it looks like.
31:21You have a big lottery over here on the
31:22left. You put some constituencies in the
31:24treatment group, some constituencies in
31:26the control group. Those big boxes are
31:29just collecting surveys, survey data
31:31from voters, aspirants, party leaders.
31:32So that's exactly the same in treatment
31:34and control.
31:35But you can see in the treatment group,
31:37there's the conventions and the radio
31:38broadcast. That last voter survey,
31:40that's what goes into the voter report
31:42to say, you know, if you were to have a
31:43primary, who who would win?
31:45And the control group is just politics
31:47as usual, but we're still collecting
31:49that voter survey, so we know who would
31:52win a primary, but we're not going to
31:54compile that report, we're not going to
31:55share it with parties, but we're just
31:56going to see, okay, at the end of the
31:58day, um does representation go up or
32:00down? Right? Are voters more or less
32:02likely to get their choice? And this
32:04this will be my last slide because you
32:06can basically summarize this entire
32:08paper in two pictures. So what are the
32:10results?
32:12Well, the first result is this had a
32:14very large impact on on increasing
32:17representation as we've defined it as
32:19the voters get their first-choice
32:20candidate. And remember cuz there's all
32:22these strongholds, getting your
32:24first-choice candidate often times mean
32:26you're also getting your first-choice
32:28elected member of parliament. So if you
32:30see in the control races, this is if I'd
32:32looked at that bottom line, who would
32:34have won the primary and who did the
32:35parties pick? The parties picked the
32:37voters first choice only 38% of the
32:39time.
32:41And if you think of this as a measure of
32:42representation, that's a really low
32:45baseline, right? In the majority of
32:47races, parties are picking somebody
32:49other than the person that the voters
32:50want the most. Okay, what happens in
32:52treatment? That shoots up to 61%. That's
32:55a sizable magnitude. It's very
32:58statistically significant, but you can
33:00see that's still very far away from
33:01100%, right? Because this is not
33:02actually a primary, it's just free
33:04disposal
33:05of information. Okay, so representation
33:07went up. Great. What happens to
33:10selection on quality?
33:11So this is now we're thinking about,
33:13okay, when we compare the people that
33:15became the candidate at the end of the
33:17day,
33:18in the control races where parties are
33:20picking, you do see positive selection
33:22on quality, right? Those numbers, these
33:24are in standard deviation units, but
33:25they're they're looking at things like
33:27how many development projects has this
33:28person been involved in. These are like
33:30local public goods. So in control races,
33:32the candidate is better than the average
33:34of people that weren't chose chosen by
33:360.17 standard deviation units, but in
33:38treatment races, that jumps up to 0.46
33:41standard deviation units, right?
33:43So overall, this is saying, actually, in
33:46this context, where information problems
33:48were a very big constraint, if you give
33:50parties information on what voters want,
33:53and remember, these are very poor
33:54voters, they have little access to
33:56education, right? But they they know
33:58what they want,
33:59and if you give parties this
34:01information, they're willing to
34:02accommodate it in part because voters
34:05and party leaders in this context are
34:07both pretty interested in similar
34:09characteristics, like how good are you
34:11at providing local public goods. Okay,
34:14so that's that's kind of wrapping up
34:16selection on valence, selection on
34:17competence, and thinking about, you
34:19know, gatekeeping and voters in the lens
34:21of of one specific policy, which is
34:24primaries. So I'm going to end here, and
34:26I'm going to pass over to Rohini, who's
34:27going to pick up and talk about what
34:29about selection on group identity.
34:34>> Great. Thanks, Kate. And I'll take over
34:37now. Let me make this full screen.
34:41Thanks, everyone. And it's great to be
34:43here and I'll be engaged with a bunch of
34:45you on
34:47the Q&A after
34:49I think you've been asking one question.
34:51Are the people who get into government
34:52any good? Are they competent? Are they
34:55honest? These are all what, as you
34:56mentioned, we think of as issues of
34:58valence.
34:59I want to ask a slightly different but
35:01very related question.
35:03You know, who is the leader? Which group
35:05do they come from? And does identity
35:07matter?
35:08And to do that, I'm going to focus on
35:10just one group, which is women.
35:12Um the main reason for this is that we
35:15have by now a lot of good evidence, and
35:18as Kate pointed out, causal evidence is
35:21hard. We have causal evidence in this
35:23area because
35:25uh we've seen a large expansion of
35:27gender quotas, which have provided us a
35:28lot of both natural experiments and in
35:31some cases randomized variation when
35:33women are elected.
35:35And the other thing is that I think the
35:36gate that keeps people out is very well
35:39documented in the case of gender.
35:41Having said that, these are issues that
35:43you can think about as applying to race
35:45and ethnicity as well.
35:47So, let me start by just kind of giving
35:49you uh the entire argument on one slide.
35:53So, as I said, we're going to be
35:54thinking about um
35:56we're going to be thinking about
35:57selection on identity.
35:59So, the first step is going to be to ask
36:01what
36:03uh it matter who is represented in
36:07And that's a question that I saw come up
36:08quite often on the
36:10um
36:10Q&A. And the answer is that
36:12representation will matter whenever
36:14preferences differ by groups. Uh we have
36:17a lot of evidence that for men and
36:19women, these preferences differ. And
36:21importantly, these are not differences
36:23that disappear with development. So,
36:25there's, you know, one view that oh, in
36:27low income countries where there's a lot
36:28of occupational segregation,
36:31you may see these differences, but
36:33they'll differ over time. But there's
36:34recent work, for instance, by Almond
36:36Faulk showing that gender differences
36:39persist even in rich countries.
36:46The second reason why
36:49identity will matter is if leaders
36:51respond more to their own group. What's
36:53the reason for this? Again, I think Kate
36:56alluded to this is that if we believe
36:58that you can't have full policy
36:59commitment, so before election, perhaps
37:02the incumbent to some extent can commit
37:04to policies by saying and making
37:07announcements and starting, for
37:08instance,
37:10you know, schemes to give women
37:12unconditional cash transfers. The
37:14opposition members can't, and this means
37:17that it's very natural for voters to
37:19infer that some of what they're going to
37:22see is going to be a reflection of what
37:25the politician wants, and gender is a
37:27reason gender and many other
37:29aspects of group identity are
37:30informative of policy preferences.
37:34And then the third is that
37:36representation is not automatic. It's
37:38not something that follows just because
37:41there's a group that's reasonably large.
37:43Women is an interesting group to look in
37:45this respect because in most countries
37:47it's close to the majority, if not the
37:49majority of voters.
37:51But as Kate pointed out, you know,
37:52representation doesn't happen just
37:55anyone actually gets voted. It goes
37:57through things like party selectorates.
38:00And because holding power gives private
38:03returns, there's a lot of value for the
38:05incumbents to gatekeep and ration entry.
38:09And
38:10why is does this matter? This matters, I
38:13think, at a societal level, as I'll talk
38:15about, is because
38:17often when you have inadequate
38:19representation of a group, in this case
38:21gender, what we see that there are
38:23returns that are kind of externalities.
38:26These are societal returns. So, the
38:28private privately a gatekeeper can lose
38:31when they see um
38:34you know, their their seats going to
38:35women, but on the other hand, society
38:38may gain, and I'll make this more clear
38:40as we go through the slides. And this
38:42creates a case for identity-based
38:44policies like gender quotas, which we're
38:45going to spend a lot of this lecture
38:47talking about.
38:50So, let me give you just one example
38:53from a paper so that this doesn't remain
38:54like an abstract set of concepts before
38:57we kind of open up the bundle. So, this
38:59is a very nice study by Pascaline Dupas
39:01and Radhika Jain,
39:03who start by looking in a state of
39:06India, Rajasthan, where they have
39:08amazing data on hospital
39:11um
39:13um claims. This is a setting where
39:15there's free health care. And the first
39:18point they make that free hospital care
39:19doesn't mean that there's this equal
39:21care for men and women in the household.
39:24They show that over time,
39:26as the number of annual visits doubled
39:29because there was expansion in the
39:31access to this, the female share
39:33actually fell from 47% to 44%.
39:36So, what this tells you that not just
39:38the infra-marginal, but the marginal
39:40person benefiting as well was male. And
39:43they calculate as a result of this,
39:45there's a large number of missing female
39:47visits.
39:49What they find that if you just cut
39:51costs for all, this doesn't close the
39:53gap. So, when a new subsidized hospital
39:55comes up, we know distance matters for
39:57most things when it comes to public
39:59services,
40:00they find it increases both men and
40:02women's visits, but men's more.
40:06So, what does matter? They find that
40:07what matters is uh having female
40:09leaders. They find that in a in a
40:12village that has a female leader,
40:15uh the female visit share to hospitals
40:17goes up for both under 15s and over and
40:2115 to 45 year old. And what they
40:24highlight the reason why it goes up is
40:26not because they reduce generalized
40:29um, constraints, but because they they
40:33reduce female specific constraints. So,
40:35they find that what it they do is they
40:37increase
40:38women health workers contacts, monthly
40:41village meetings um,
40:44go up. At the same time, there's no
40:46change in facility supply, program
40:48awareness
40:49or perhaps and this is kind of a common
40:50finding very often, we don't see changes
40:52in gender attitudes.
40:54So, this I think is a canonical example
40:56of what I think of as the externality.
40:59Um, you know, you get a female leader in
41:01place and what you're going to see
41:02there's an externality that benefits
41:05other women in the village, but this is
41:07not something that's a private gain for
41:09the female leader.
41:12So, just to remind you, uh, we see these
41:15externalities in many settings. We see
41:18that women, uh, if you look at this if
41:20you look at the picture on the
41:22on the right, you can see that women are
41:25roughly 50%, but the share of seats held
41:28by women everywhere is relative is less
41:30than 50%. Only 15% of the heads of
41:33government today are women and 27% of
41:36national parliamentarians.
41:38So, this again is just a stark reminder
41:40of the fact that having the ability to
41:43go to the voting booth is not going to
41:45is probably necessary, but is not going
41:48to be sufficient to get, uh, women the
41:50position of power that they
41:53from a purely representation perspective
41:55would probably benefit from.
41:58So, what am I going to do today? I'm
41:59going to walk you through the reasons
42:01through this and I thought it'd be
42:02useful just lay out to start by telling
42:05you what the different, um,
42:08arguments we're going to make is. So,
42:10we're going to talk about the first part
42:11which I think is something Kate talked
42:13about already, is well, why are women
42:15not elected? And what we're going to see
42:17is the women very often face a higher
42:19bar.
42:21What do quotas do in this setting?
42:23Quotas in this setting
42:25reduce the bar mechanically. They force
42:27entry of women.
42:30What are the benefits of having
42:32women come in? So, one reason why women
42:36don't get elected is, you know, you sort
42:39of
42:40traditionally believe that women should
42:41not play leadership roles. You think
42:43you're kind of
42:44biased about it. And so, when you're
42:46exposed to women, learning cuts the
42:49bias. And so, we'll see evidence that
42:51that's one one reason why women face a
42:53higher bar.
42:55We'll also see that
42:58when women get elected, it often reduces
43:01the private rents to male incumbents,
43:04but also reduces misallocation. So, this
43:07speaks to the other argument for why
43:09women may may face a higher bar is
43:12because there are potentially
43:15male incumbents are more threatened by
43:17female newcomers than male newcomers.
43:21And we'll see evidence from Sweden on
43:22this. And then finally, we'll look at
43:24gatekeepers and how do gatekeepers try
43:27to control this margin?
43:29So, let's start by thinking about why
43:31aren't women elected?
43:33Now, I want you bring here a concept
43:35that, you know, is often discussed in,
43:37I'd say, labor economics and perhaps a
43:40little bit less in political economy
43:41than it should be, which is to really
43:43think about the role of norms. We've
43:47seen in political economy in recent
43:49years a lot of discussion about culture.
43:52I think there's a lot of recognition
43:54that political economy is as much about
43:55informal institutions as formal
43:57institutions. And both these factors
44:01are very very important when we think
44:03about gender.
44:06So, just to remind you, this is data
44:07from the World Values Survey, and this
44:09just shows you across countries
44:12what is the percent of respondents in
44:16blue it's uh
44:18uh women and
44:19uh
44:20So sorry, uh it's uh yeah, blue is men
44:23and orange is women who agree that uh
44:27Yeah, who who agree that sorry, I forgot
44:30yeah, who agree that men make better
44:32political leaders than women. And what
44:34we can see that other than OECD and uh
44:37Latin America, in all other parts of the
44:39world the majority of men believe that
44:42men make better political leaders than
44:44women do. And so, you know, the group
44:46that's not affected is has particularly
44:48strong views on uh women's ability to uh
44:52make better leaders.
44:54Now,
44:56let's remind ourselves about what a
44:57gender norm is. A gender norm is, you
45:00know, a gender-specific rule of
45:02behavior. It could be for instance that
45:04uh men should play the role of the
45:08breadwinner, or men should be leaders.
45:10And what a norm holds is when
45:14people prefer to conform to the stated
45:17belief rather than
45:21perhaps what they themselves believe,
45:23because they think others are going to
45:24approve of such behavior, or they expect
45:27sanctions or rewards if they uh are in
45:30line with this uh gender-specific rule
45:33of behavior. What this means about what
45:36matters in uh sustaining uh norms is
45:39beliefs about beliefs. And this is where
45:43um
45:43gatekeepers uh benefit from it.
45:46Gatekeepers often in many societies are
45:48seen as the holders of uh cultural
45:51norms. They're seen as the group that
45:52where be able to inform these uh
45:56uh enforce what is appropriate behavior.
45:59And these expected penalties can drive
46:01conformity.
46:03So, if gatekeepers are able to
46:06um drive such norms and benefit from it,
46:10we can see why they would why these may
46:12continue to live on even if personal
46:14beliefs deviate from it.
46:17So, when we think about the gates that
46:19matter and [clears throat] when we think
46:20about the gates that matter in politics,
46:22there are really two.
46:24First is within the family. We see this
46:26in a lot of settings um that women's
46:29entry into politics is sort of a family
46:32matter and husbands and in-laws decide
46:34whether or not you can enter into
46:36politics. And then as Kate has already
46:38talked about, party selectorates play an
46:40important role and they choose to affect
46:43your ability to enter and access
46:44leadership.
46:46And so, the question then becomes why
46:48would it be that a gatekeeper would want
46:51to use these norms to in particular keep
46:54women out?
46:55So, this is um I think the only
46:58uh
46:59conceptual framework slide I'm going to
47:01go through, but I think it's a useful
47:03one to think about.
47:04So, let's think about uh what are the
47:07benefits to society from a female
47:09leader?
47:10It's going to be some It's going to be
47:12the difference between the value of
47:14having a female leader and a male
47:15leader. This could be positive, this
47:17could be zero, this could be negative.
47:20Plus importantly what we talked about is
47:22there's a representation externality.
47:24This could for instance the case of uh
47:27the Dube and Sen paper be that other
47:29women benefit from the active actions
47:32that women undertake. There are many
47:34other papers showing that
47:36when when women are elected, they have
47:39important role model effects. So, in any
47:41society where there is gender bias,
47:43there appears to be significant evidence
47:45of a representation externality.
47:48But now if we think about um the
47:51gatekeeper's benefit, the gatekeeper
47:53doesn't care about the representation
47:54externality. That's not a private
47:56benefit.
47:57Rather, what he thinks about is his
48:01the first two terms, but these two terms
48:04are now going to be modified by three
48:06terms, B, V, and kappa, which I'm going
48:09to talk about.
48:11So, what are these three terms?
48:13So, B is the belief bias.
48:15So, I may not know I may just believe
48:18literally in the norm and think that men
48:20make better leaders and believe that,
48:22you know,
48:23women are not going to be so good.
48:25This is a case where we can see that
48:27gender quotas that cause exposure to
48:29women can actually erode such a biased
48:33beliefs.
48:34The second is uncertainty.
48:36Um this is just the fact I've never seen
48:38women. I don't know how they will be. I
48:41am at all risk averse. I'd rather go
48:44with a known quantity than a unknown
48:46quantity.
48:47Now,
48:49in this particular case, you could say
48:50that experimentation is a good thing.
48:52You will learn by trying them. But, the
48:54problem is that the gatekeeper is not
48:55the only person who gets the returns
48:57from experimentation. If you learn that
48:59women are as capable as men, all future
49:03employers, all future party leaders
49:05learn about it. And so, this under
49:07experimentation
49:09is going to be a result of the fact that
49:11the gains are to the society even if the
49:14gatekeeper is the first one to bear the
49:15cost.
49:16And the third is just control. The rise
49:19of a woman may erode a male control
49:22within the household if, for instance, a
49:24woman starts
49:25uh gaining political position and
49:27working outside, she may have more
49:29bargaining power within the household,
49:30which reduces the husband's control.
49:33Within a party, as we'll see in Sweden,
49:36women may challenge male leaders a
49:38little bit more, and this can lead to
49:40resistance.
49:41So, you should think of this as really
49:43uh what we're going to try to look at
49:45through the lens of quotas.
49:49So,
49:50gender quotas are interesting from in
49:53one particular sense that they got
49:54implemented well before there was
49:56evidence on them. They started after the
49:591995 Beijing conference, and as you can
50:02see
50:04other than North America
50:06and
50:07uh upper part of Asia, they're pretty
50:09dominant everywhere.
50:12We can take the extreme case of reserved
50:15seats where you have only women can
50:18stand in particular reserved seats. They
50:21can take candidate quotas where parties
50:24have to
50:25in proportional representation systems
50:27have a certain number of women
50:30take can be candidates. Or they could be
50:32voluntary party quotas. This was started
50:34for instance in Scandinavia, where the
50:36parties explicitly announced that they
50:38would have a certain number
50:40representation of women.
50:43I should just note we're going to be
50:44talking a little bit about quite a lot
50:46about India. Uh India has gender quotas
50:49right now at the local levels of
50:51government, not the national, which is
50:52why it's in the gray color.
50:56Okay, so just to remind you what what do
50:59quotas do?
51:00So as you saw from the one page of the
51:03conceptual framework we had, um the
51:06gatekeepers bar is going to be higher
51:08than the society bar by the difference
51:11of the externality.
51:12And so this is the the area where the
51:15quota is going to act.
51:18It's going to act on both margins. It's
51:20also going to both shift beliefs. And if
51:24it shifts beliefs, these are persistent
51:26gains. So you could believe that you
51:27could have quotas in the short run,
51:30which will see persistent gains that
51:32self-sustaining.
51:34On the other hand, if [clears throat]
51:37if the and we'll see evidence of this as
51:38well, if quotas leave parties discretion
51:42and they threaten the rents, then you
51:45you see that
51:48party gatekeepers are going to react to
51:50these quotas by
51:53trying to undermine how effective they
51:55are.
51:58So, let's see what quotas deliver.
52:02So, I want to start with a lot of some
52:05quite well-known evidence. I think
52:06Esther Duflo uh
52:08uh
52:09who won the Nobel Prize a few years ago
52:12in development is well-known for uh some
52:14of the initial work showing that gender
52:16quotas uh
52:18are associated with women's preferences
52:20getting better represented. I then did
52:22more work with her and colleagues to
52:25look to see what it did to beliefs.
52:28So, as I said, Esther had early work
52:30showing that public good spending
52:32follows women's complaints. So,
52:35the advantage of looking at this in the
52:37Indian context was that there was
52:39randomization in which villages were
52:41exposed to gender quotas. So, you could
52:44ask whether uh the public good spending
52:46in those villages was more in line with
52:50what women were complaining about in
52:51village meetings.
52:54There's been work since then suggesting,
52:56and this is I think again a classic
52:57externality example, that uh women are
53:00more comfortable reporting crimes and
53:03police responsiveness also rises when
53:05you have women leaders.
53:08What we found in our work is that uh
53:12voters actually updated their beliefs
53:13about female leaders once they were
53:15exposed to leaders.
53:17Uh they also improved their aspirations
53:20about their daughters' uh future roles.
53:23They saw We saw a rise in the schooling
53:26among girls and a reduction in the time
53:29use of girls at household chores.
53:33And finally, perhaps most importantly,
53:34and this is consistent with the idea
53:36that if it is just belief bias, once you
53:38move it, you should be done, is that
53:40once the quotas were removed, women were
53:43more likely to get
53:45to remain
53:48viable candidates and get selected. And
53:50this wasn't being driven by incumbents
53:52being reelected. This was about new
53:54women entering and having a high
53:55probability of winning.
53:57There's also nice work by Tanu Goyal
54:00showing that one thing that women do is
54:02women are more likely to use female
54:04party workers. So there's also an
54:06expansion in grassroots political
54:09activity by women.
54:13And just to kind of reiterate the point
54:15because I think this is something Kate
54:17talked quite a lot about is
54:20you know, it's always hard to look at
54:23identity effects just because selection
54:25is endogenous.
54:26The kind of areas that are going to
54:28select women are likely very different
54:30from those that are not.
54:31So the advantage in India, and that's
54:33perhaps the reason why there's been
54:35several studies from it,
54:37was the fact that what we saw is
54:40the assignment of whether you were
54:42reserved was done by a serial number
54:44rule on randomized lists, and this was
54:46redrawn at each election. And the nature
54:49of it meant that there was randomized
54:50variation not only in whether you were
54:52reserved, but also in how long you were
54:54reserved for. You either never reserved
54:56for one term or two terms.
55:00So the other part that I think is worth
55:03highlighting is
55:05in looking at
55:07beliefs, there's a huge concern of
55:10social desirability. This is exactly
55:12goes back to what we talked about the
55:14underlying gender norms.
55:16Uh
55:18Suppose I believe that it is appropriate
55:21for me to say that
55:23women are not good leaders, then maybe I
55:26think women are good leaders, but I
55:27won't say that in response to a survey
55:29because I don't believe that is
55:31appropriate. And so this leads to the
55:34concern that if you just go out and
55:35surveys and ask
55:38uh
55:39villagers, you know, do you like your
55:41female leader or do you like your male
55:43leader? The answers may be very tainted
55:46by what they believe you want to hear.
55:49So, what we did here, and I think
55:50there's been a huge expansion in uh
55:53especially in the political science
55:54literature on this, so I think for those
55:56of you thinking of research in this
55:57area, I would encourage you
55:59to engage with this literature of
56:01thinking of ways in which you can try to
56:04reduce social desirability bias.
56:06So, for instance, what you can do is you
56:08can
56:10give people identical speeches or
56:12vignettes where you just randomize
56:15the voice of the person giving the
56:17speech or the name in the vignette and
56:19see if the change in the gender of who
56:23speaks changes their ranking of it. And
56:27that's going to be, you know, some
56:28inherent
56:30bias, not something as a social
56:32desirability bias.
56:34You can also do increasingly what people
56:36call implicit association tests where
56:38you see whether your willingness to
56:40associate men and women with different
56:42tasks varies with um
56:46with how
56:48uh
56:49varies with
56:51how stereotyped that task is. So, do you
56:53associate women with home chores more
56:56than you associate men with home chores
56:58and vice versa when it comes to
56:59leadership roles?
57:01So, when we did all of this, what we
57:03found is that
57:05uh with exposure
57:08there was a significant reduction in
57:10bias. But, what was interesting is in
57:12some it took time. So, the first cohort
57:15did first cohort of women leaders did as
57:18well as male leaders when it comes to,
57:20you know, public good provision quality
57:22of what they provide, they they took
57:24fewer bribes than
57:26in their than comparable male leaders.
57:29But, they were ranked lower.
57:31However, the second cohort was rated at
57:34par with men.
57:35And so what this tells us is that
57:38you can see the shifting of behavior,
57:41but it takes time.
57:43Importantly, what is true is that you
57:44can see that your willingness to
57:46kind of vote for a good candidate can
57:49change without your taste changing.
57:52I think this is important because very
57:53often we fixate on saying, you know,
57:56what will it take to change gender
57:57norms?
57:58You know, maybe people don't like women
58:00very much as leaders, but they also want
58:03the best qualified leaders, and so they
58:06may be willing to sort of put up with
58:08women because they realize they do much
58:10better, and that's what we saw in West
58:12Bengal.
58:14And so we see when when you look at
58:1710 years after reservation what it was
58:19in place at seats that were no longer
58:22reserved,
58:23if you go from being in an area that had
58:26never seen reservation for women that
58:27were reserved twice, you see one and a
58:31half times increase in the likelihood of
58:33having female leaders.
58:35It seems that beliefs first take time to
58:38update, and then entry follows after
58:40that.
58:42So last couple of things I want to
58:44quickly talk about is first, what is it
58:46that we learn about a binding quota?
58:50So this is a paper by Besley et al. from
58:52Sweden, where they basically took the
58:55Swedish case, which was a case of
58:57voluntary party quotas. So the Swedish
59:00uh
59:01one of the mainstream Swedish parties
59:02announced a zipper law. So it said every
59:05alternate seat has to be male or female.
59:08And the model they put forward is that
59:11the party leader controls the list. This
59:13is not a system with primaries.
59:15But the party leader recognizes that in
59:18who it he gives tickets to is going to
59:21affect who is going to
59:25support him in the next leadership
59:26contest within the party. So, that's the
59:29private return they're trying to
59:30protect.
59:31They hypothesize that um there is
59:36homophily, that men like men, and so a
59:39party leader would rather go for what
59:40they call mediocre male candidates who
59:43are worse leaders because they know
59:45they'll be loyal to them in part because
59:46they realize the reason they got the
59:48ticket was because the party leader
59:50favored them.
59:52Notice that this is a model with no
59:54bias. This is just a case of a party
59:56leader who wants to maintain that
59:57position. And this gate then binds even
1:00:00when there is zero bias, and citizens
1:00:02bear the cost.
1:00:04One of the questions that came up
1:00:06repeatedly I saw in the Q&A's how do you
1:00:08measure competence? How do you measure
1:00:09quality? Sweden is a good example for
1:00:12this. Um they can basically look at the
1:00:15administrative data and look at
1:00:17everyone's earnings relative to their
1:00:19peers. They define someone as competent
1:00:21if controlling for a set of
1:00:23characteristics uh you earn more than
1:00:26you would be expected to earn. And then
1:00:28for men they can validate it against
1:00:30cognitive scores at the time of entering
1:00:32the military.
1:00:34What do they find? They basically find
1:00:36that
1:00:38uh when you have the zipper quota put in
1:00:41place, you see no change in the
1:00:44competence of women, but a huge increase
1:00:46in the competence of men. So, recall
1:00:48with the zipper quota a set of men are
1:00:50now disqualified, so there fewer men who
1:00:52are left. Those men are much more
1:00:55competent. And so,
1:00:57the suggestion is that what was getting
1:00:59displaced were uh less competent men for
1:01:03women who are at least as competent as
1:01:04those being displaced.
1:01:06Um
1:01:09so,
1:01:11again, I think this brings us back to
1:01:12what Kate was saying is you know, the
1:01:15quality of who's elected reflects the
1:01:16gate, so you're going to need the whole
1:01:18chain. You're going to have to
1:01:19understand who was available, how they
1:01:21got chosen, and how they entered, not
1:01:23just look at probability of winning
1:01:26conditional on actually running.
1:01:29Let me spend I'm slightly over time, but
1:01:31let me spend the last couple of minutes
1:01:33just talking about backlash.
1:01:36So, as I said, candidate quotas exist
1:01:38across the world, but what this graph
1:01:41shows you is that the majority of the
1:01:42world candidate quotas under deliver.
1:01:45So, what this is telling you is zero
1:01:46would be where as many candidates get
1:01:50selected who are women as are
1:01:53seats that are supposed to be assigned
1:01:56to them. Right? So, if a party says 50%
1:01:59of my candidates are going to be women,
1:02:02this number would be zero if we see 50%
1:02:05of those elected are women.
1:02:07But, in the majority 43 out of 57
1:02:09countries fall short of their own
1:02:11candidate quota.
1:02:13And the reasons are kind of have been
1:02:15documented a number of papers, and some
1:02:17of them are striking.
1:02:19So, there's a paper from for Spain that
1:02:23sort of highlights that in Spain we know
1:02:26that very often there are what we call
1:02:27ballot order effects that voters are
1:02:30more likely to choose candidates who are
1:02:32higher in the ballot. They find that
1:02:35when this was put in place, women are
1:02:38more likely to be put on the ballot who
1:02:40have late alphabet surnames, which means
1:02:42they come lower in the court in the
1:02:44list, and are more likely to lose. This
1:02:46you can say is like an extreme form of a
1:02:48strategic play by parties.
1:02:50Um in Nepal, we found in work with Mike
1:02:54Callen and others that
1:02:56uh parties faced a binding quota, but
1:03:00they recognize it was on one of they had
1:03:02to either have a mayor or deputy mayor
1:03:04candidate who was female, but they
1:03:07recognized if they formed a coalition
1:03:08with another party, and each of them
1:03:11fielded only one of the two candidates,
1:03:12the quota was not binding. And so, we
1:03:15saw this took place a lot in 2022 with a
1:03:18decline of eight percentage points in
1:03:20female representation.
1:03:22In France and Germany, there's evidence
1:03:24that women are run. So, you may have a
1:03:2650% candidate quota, but you know where
1:03:28you're less likely to win. And that's
1:03:30where you run women.
1:03:33And in Mexico, uh it seemed to be there
1:03:35was some evidence that this was
1:03:37pressuring women. So, elected women
1:03:40would resign after being elected, and
1:03:42then male alternatives would take over.
1:03:45This was
1:03:46uh recognized so broadly that in 2009, a
1:03:49Juanitas law was passed, which now
1:03:51requires that if a woman resigns, she
1:03:53has to be replaced by another woman.
1:03:56So, let me just end with uh just briefly
1:03:59talking about I think are some of the
1:04:00design implications, and then open it up
1:04:02with some research questions that Kate
1:04:05and I thought were on the frontier right
1:04:07now.
1:04:08So, when you think very specifically
1:04:09about gender, if one really believes in
1:04:12the
1:04:13value of quotas in
1:04:15creating exposure, leading to
1:04:17experimentation, and that there is an
1:04:18externality, then to the extent
1:04:20possible, what you want to do is mandate
1:04:22seats or candidacy because voters are
1:04:25parties are able to manipulate when they
1:04:27can.
1:04:32I think the second part, and I think
1:04:33this resonates with also what I think
1:04:36Kate found, is the selectorate is really
1:04:38important. We want to think about how do
1:04:41we actually get women onto selection
1:04:43committees? How do we get people into
1:04:45the party
1:04:46stronghold positions that really matter
1:04:48and scrutinize where women run.
1:04:51And then I think like something like
1:04:52this Juanitas rule in Mexico tells you
1:04:55that
1:04:56if all that fails, you want to design
1:04:57policies to limit avenues of reversal.
1:05:01So, let me just end by putting forward
1:05:03some set of questions that I think are
1:05:05at the research and policy frontier.
1:05:09So, the first three come from Kate's. Uh
1:05:11you know, what interventions and
1:05:12policies are going to induce better
1:05:13people to run for public office? I think
1:05:15this is an evergreen question.
1:05:18I think a question that I think
1:05:19researchers have seen increasingly as
1:05:21important is how do we actually help
1:05:23them get through gatekeepers? You can
1:05:25motivate
1:05:26citizens to want to run, but what if
1:05:28they're completely stemmed through the
1:05:30gatekeepers?
1:05:32And then for voters, how do we
1:05:35understand the trade-off that voters
1:05:36have between balance, group identity,
1:05:38and ideology? And today more than ever
1:05:40this is a question that seems to be
1:05:42extremely important in understanding
1:05:44whether democracy will work.
1:05:46Turning to quotas,
1:05:48uh
1:05:49how do we ensure that quota design stays
1:05:51ahead of circumvention?
1:05:53I think this is
1:05:55you know, I think economists love
1:05:56thinking about unintended consequences,
1:05:59so I think we can write
1:06:00more and more sophisticated papers
1:06:02trying to see how it can be gamed, but
1:06:05from a policy perspective, we really
1:06:06want to try to think about what can be
1:06:08done to give the world the fairest
1:06:10chance to reduce bias and ensure that
1:06:15kind of societal externalities are
1:06:16incorporated.
1:06:18Uh how long how long should
1:06:21uh quotas last for? If we think that
1:06:25we belief is the main channel,
1:06:27how much exposure is needed? And that
1:06:29may also help with the political
1:06:31viability of introducing quotas.
1:06:34And then finally, trying to understand a
1:06:36little bit from the perspective of
1:06:37voters,
1:06:38you know, what signals update their
1:06:40beliefs? What perpetuates the status
1:06:42quo?
1:06:43So, these are some of the questions we
1:06:44wanted to leave you with and uh
1:06:47thank you very much and we'll open it up
1:06:48now for questions. Let me stop sharing.
1:06:54>> Thanks so much, Rohini and Kate. Um
1:06:58I'll first ask the panel if anyone has a
1:07:00question, please raise your hand. And if
1:07:03there's anyone in the audience that
1:07:04would like to ask one of your questions
1:07:06live, um you can raise your hand as well
1:07:08and I can select a few more, um, to be
1:07:12to be panelists and then we'll just kind
1:07:13of go back and forth like we normally
1:07:15do.
1:07:16Um, let's start first with Andrea.
1:07:20>> Uh, hi everyone. Thank you very much for
1:07:22the presentation. They were really
1:07:23interesting. I think I learned a lot.
1:07:25Uh, I had a question about
1:07:26representation. Uh, I'm from Peru and
1:07:29something that typically happens is that
1:07:31now illegal economies are prominent. So,
1:07:33technically when they choose politicians
1:07:36that have uh, illegal connections, they
1:07:38are being good representation but at the
1:07:40same time they are producing negative,
1:07:43uh,
1:07:45externalities for the
1:07:47country as a whole. So, uh, I was
1:07:50thinking how do these like illegal ties
1:07:53or uh,
1:07:54negative externalities produced by by
1:07:59uh, local representation
1:08:01can be asserted as something positive or
1:08:04negative because technically you have
1:08:05representation but bad outcomes. Thank
1:08:08you.
1:08:12>> I I guess I I could jump in quickly on
1:08:15that one. I mean, I think that's sort of
1:08:17a
1:08:17what we talked about is this question
1:08:19like when, if, where is there a
1:08:22trade-off between representation and
1:08:24quality? And and you're suggesting one
1:08:26where it seems pretty obvious that there
1:08:27is, right? So, if you have like large
1:08:30criminal element, giving them
1:08:32representation is coming at a real cost
1:08:34for selection on quality if they're
1:08:35promoting, you know, bad policies and
1:08:37have negative externalities on the
1:08:38economy. So, I think that's a great
1:08:39example where you would be really
1:08:40worried about that trade-off.
1:08:47>> Great. Thank you. Um, there's a question
1:08:50in the Q&A from Sierra Wang.
1:08:54Uh, she says in countries with autocracy
1:08:59or in her case in countries like China
1:09:01where political selection is driven by
1:09:03institutional appointment rather than
1:09:04popular election,
1:09:06gender targets often lead some in the
1:09:08public to view female officials as quota
1:09:11fillers. I think to be fair, even in the
1:09:14US, in places where there are democracy,
1:09:16people can can view that or have a
1:09:18similar view.
1:09:20Um,
1:09:21does the positive learning effect of
1:09:24quotas require a competitive election
1:09:26setting to prove um, valiance? And what
1:09:29are your more general thoughts around
1:09:32um,
1:09:33your research in auto- autocratic
1:09:35settings?
1:09:37>> Yeah, it's a it's a great question.
1:09:39Uh, and I'll just start by giving an
1:09:41example of something that
1:09:43I think I saw, uh, when I was once I was
1:09:46covering, uh,
1:09:48election campaigning by women, you know,
1:09:50running in quota seats. Uh,
1:09:53my first reaction was to be very
1:09:54disparate when I saw one of the women
1:09:56campaigning with a picture of her
1:09:58husband next to her, who was the person
1:09:59who had the seat before it had been kind
1:10:01of
1:10:02put up close for quotas. And then I went
1:10:05to talk to her afterwards, and she was
1:10:06like, you know, in a very conservative
1:10:08society, the only way I can be
1:10:10competitive is to show that I'm not
1:10:12threatening the norm. And so, once I'm
1:10:14elected, I can I know I can do what I
1:10:16want, but for the process, you know, I
1:10:18have to kind of be part of the system.
1:10:21So, I think it's something that requires
1:10:23more research. There is certainly, I
1:10:26think it's certainly the case that
1:10:28uh, trying to directly tackle, uh, norms
1:10:31by
1:10:33putting, uh,
1:10:35counter-stereotypic examples of say
1:10:36women running for seats through quotas
1:10:38can very much immediately lead to an
1:10:41easy portrayal of them as
1:10:43quota-protected women, uh, who don't do
1:10:46much other than what say their husbands
1:10:48want or aren't as able. Uh,
1:10:51I think the two counters are a lot of
1:10:53the evidence, for for instance, the
1:10:54evidence I talked about from Sweden, the
1:10:56evidence I talked about from India,
1:10:57shows that that doesn't seem to be the
1:10:58case. So, I think there is anecdotal
1:11:01evidence versus this. However, on the
1:11:03other hand, I would say what is true,
1:11:05especially in the case of India, is that
1:11:06over time, the amount of resources going
1:11:09through local governments has increased
1:11:10a lot. And so, I think, you know, why do
1:11:13I mean, it's first it's an interesting
1:11:15question, why do male politicians allow
1:11:17gender quotas to get passed? You know,
1:11:20in India, we have not been able to pass
1:11:22or have implementation of gender quotas
1:11:23at the national level.
1:11:26Partly because perhaps they're seen as
1:11:27very I mean, they're not passed, but
1:11:28they haven't been implemented. It's seen
1:11:30as this very valuable at the when they
1:11:32were introduced at the local level, that
1:11:34was a point in time when they weren't
1:11:35seen as very valuable. Now they are.
1:11:37There's a lot of money going through
1:11:39village councils. And so, there's
1:11:40perhaps more potential to usurp them.
1:11:44I think
1:11:45some work like political scientists like
1:11:46Soledad Peralman have been doing is to
1:11:49argue that
1:11:50the these kind of programs are
1:11:52successful when they are backed by
1:11:55community measures to create collectives
1:11:58of women who can come together. So,
1:12:00things like self-help groups or other
1:12:02ways of bringing women together means
1:12:03that's not just a single woman kind of
1:12:05trying to run in a very male society,
1:12:07but feels that there are also other
1:12:09systems. And I think that's right. You
1:12:11can't you have to
1:12:13at the same time as putting these
1:12:14political quotas in place, you want to
1:12:16think about what you can do to
1:12:18strengthen the role
1:12:20of the collective available to the
1:12:22women.
1:12:26>> Hello. Can I ask a question, ma'am,
1:12:27directly to you?
1:12:31>> Hello.
1:12:31>> Sure, Kanchan. Go ahead.
1:12:34>> Yeah, ma'am, I am from an India,
1:12:36particularly from West Bengal. I'm a
1:12:39civil servant, has been working in this
1:12:41field, particularly the women
1:12:42empowerment, for the last 10 years. My
1:12:45point is that in one kind of an negative
1:12:48aspect that we have seen
1:12:51I have seen in the many cases that
1:12:53during the panchayat visiting as a civil
1:12:55servant that that you have raised a
1:12:58question that an woman have taken their
1:13:00picture of the male and male are taking
1:13:03and dictating the woman what to do what
1:13:05not to do. Being a researcher what could
1:13:07be your suggestions in this field for
1:13:10the particular woman empowerment through
1:13:12the pushing quota.
1:13:14>> I think what I was saying was that I
1:13:16think we have to this is where research
1:13:18on evidence becomes valuable. You can
1:13:20look at that and say you know it looks
1:13:21terrible but sometimes when you look at
1:13:23the data you realize that's not the only
1:13:25thing that's going on. That somehow
1:13:27when you look at the actual policies
1:13:29being passed they do seem to have been
1:13:32moved a bit towards women maybe not as
1:13:34much as they could have been
1:13:35but I think the one of the most
1:13:36important things you can do is to have
1:13:38supportive organizations like self-help
1:13:41groups like others that mean that it's
1:13:43not a woman leader is not functioning in
1:13:45a vacuum.
1:13:48>> Thank you ma'am. Could you suggest
1:13:50certain books and articles that can be
1:13:52published and that is for further
1:13:54development in this area?
1:13:56Could you suggest some books and
1:13:57articles? Okay.
1:13:59>> I think there's some readings that would
1:14:00be helpful that I think have been
1:14:01provided already on the
1:14:04>> Yes. Thank you so for your eye-opening
1:14:06presentation. Thank you ma'am.
1:14:09>> Thank you. I shared the link in the chat
1:14:13and I I'll share that again by the end
1:14:15but they're all the past reading lists
1:14:18including Remi and Kate's
1:14:21are are available there so you can check
1:14:23those out.
1:14:25Sanjana, would you like to ask your
1:14:26question?
1:14:29>> Yeah.
1:14:30Thank you Professor Alswel this
1:14:32wonderful presentation. I think my I'm
1:14:34forcing a PhD candidate at Duke studying
1:14:37political economy of implementation
1:14:40of environment and development policies
1:14:42but I was very interested in two points
1:14:44that were made
1:14:46in the end slides.
1:14:47One was about, you know, thinking about
1:14:49targeting the selectorate.
1:14:51Um,
1:14:52so when we look at these effects of
1:14:54good, positive effects of women leaders
1:14:57on provision of public services,
1:15:00how can we think about the interaction
1:15:02of female leaders with
1:15:04with junior political leaders and senior
1:15:06political leaders? In some sense, you
1:15:09know, the point of targeting the
1:15:10selectorate. And also their interaction
1:15:12with
1:15:13uh, bureaucrats. Because the provision
1:15:16of public services is happening through
1:15:18them. So,
1:15:20is there any, uh,
1:15:21like, is there any work on thinking
1:15:23about that interaction that once you
1:15:25select female leaders, they interact in
1:15:27a certain way to push those boundaries
1:15:29or are they still,
1:15:31um, you know, uh, uh,
1:15:32these local institutions are still
1:15:34stifling their, you know, potential to
1:15:37actually attain the kind of public goods
1:15:38expansion that they want for
1:15:41females. And the second point about, um,
1:15:43uh, future questions was the, you know,
1:15:47dynamic effects of quotas. Like quotas
1:15:49for how long?
1:15:50And
1:15:52um,
1:15:53um, how I was thinking about this was
1:15:55that can we can the length of quotas in
1:15:57the system affect both party dynamics,
1:16:01um, and household dynamics of voting?
1:16:05Um, and like, how do we how do we think
1:16:07through that if if, uh, to sort of like
1:16:10build a question around this for future
1:16:12research?
1:16:13Thank you.
1:16:17>> I can answer quickly on the selectorate
1:16:19one and then pass Rohini for the
1:16:20duration of quota. So, so there are
1:16:22papers that show like if you put women
1:16:24on the selection committee, more women
1:16:27candidates go through, right? So, so
1:16:28that's like a very clear
1:16:31uh, policy that that leads to changes in
1:16:33the identity of who gets selected. I
1:16:35think your point about, you know,
1:16:36working upwards and downwards, like
1:16:38junior, more junior, more female, you
1:16:40know, that's kind of mentors and groups
1:16:42of women. Rohini and co-authors are
1:16:43doing work on, you know, bringing women
1:16:45politicians together so they can help
1:16:47each other.
1:16:48I think on the how politicians interact
1:16:52with bureaucrats. So, not specific on
1:16:55the agenda angle, but Pierre Raffler has
1:16:56an interesting paper on this in Uganda
1:16:58if you want to just check out how some
1:17:00of the things to think about when local
1:17:02politicians are overseeing local
1:17:04bureaucrats. And Rohini, I'll let you
1:17:05talk to the quota one.
1:17:07>> I think of the dynamics of quotas
1:17:10this
1:17:11a decreasing sense that, you know, as
1:17:14once you have quotas, there's obviously
1:17:15both there's a bias that gets changed,
1:17:18but there also, hopefully over time,
1:17:21underlying uh
1:17:24investments such things like educational
1:17:25investments change. So, a good example
1:17:27is if you think about education,
1:17:29the gender gap in education in both
1:17:30settings are now close to reversed. And
1:17:33so, that the the flip question arises,
1:17:35does this mean that we need to have
1:17:37quotas offered in the opposite
1:17:39direction? Uh and you know, there are
1:17:41some settings, for instance, there's a
1:17:43very nice paper on Finland that shows
1:17:45that uh
1:17:47gender quotas for uh primary school
1:17:50teachers that favor men had very
1:17:52positive effects because both, you know,
1:17:54male and uh female kids require role
1:17:57models.
1:17:58At the same time, uh gender quotas that
1:18:02uh perpetuate
1:18:04potentially that are not merit-based.
1:18:06So, for instance, there's another paper
1:18:07saying that in China,
1:18:09uh when they introduced meritocratic
1:18:11uh exams for tax collectors, uh women
1:18:14started doing much better. So, then they
1:18:16started introducing some amount of
1:18:18quotas for men, but that just kind of
1:18:21led back to old boy cronyism. So, I
1:18:23think kind of there is going to be no
1:18:25single answer for how long you should
1:18:27have a gender quota. It's going to
1:18:28depend on how the underlying dynamics
1:18:30change as well.
1:18:32Uh
1:18:33and it's going to I think it you you may
1:18:35need to change the group you have it for
1:18:36whether it's for men or for women, but I
1:18:38think what you want to always kind of
1:18:39think through is is there argument to
1:18:41believe that there is some kind of
1:18:42either belief bias or role model role or
1:18:45externality
1:18:46that is going to be served by the quota.
1:18:52>> Right. Thank you so much.
1:18:55>> Brilliant. Thank you.
1:18:56>> Yes.
1:18:57>> Um
1:18:58there's
1:18:59uh
1:19:00a question in in the chat I think is is
1:19:02linked to this
1:19:04asking if
1:19:05you know obviously
1:19:07political selection is is
1:19:09multi-dimensional.
1:19:11Um
1:19:12and uh
1:19:14as we shift kind of the needle on
1:19:18getting more women into government uh
1:19:21there are some
1:19:22people that are are wondering are there
1:19:25any
1:19:26um adverse effects
1:19:29uh
1:19:30if it's like for class or
1:19:34um
1:19:36diversity or you know certain
1:19:38backgrounds that are are being
1:19:40compensated for when there are more
1:19:43women coming in are there um
1:19:45other variables that are shifting
1:19:47downward and if so how can we try to
1:19:51control for those?
1:19:57>> Uh
1:19:58let me say one thing and then hand it to
1:19:59Kate and I think this question about uh
1:20:02is is central to representative
1:20:03democracy right? Very often anything
1:20:06you're going to do to represent a group
1:20:07is going to take away from another.
1:20:09There are some cases that I think I
1:20:10tried to highlight those where we think
1:20:12there are positive externalities where
1:20:13societal gains
1:20:15may be large enough.
1:20:17But in general I think we have to just
1:20:18recognize at the heart of it
1:20:20representation will have some amount of
1:20:24uh
1:20:24you know gains and losses. I mean Kate
1:20:26and California they're about to think
1:20:28about a billionaire tax soon which I
1:20:30think is going to be one of the very
1:20:31heated examples of this. let her speak
1:20:33to it a bit more but I think we have to
1:20:35recognize that
1:20:37kind of class conflict and gender
1:20:39conflict may not be that different in
1:20:41the end.
1:20:42And Kate, what do you
1:20:45>> Yeah, I mean I think you can think about
1:20:47so the one is if you privilege
1:20:49representation of any kind, what are you
1:20:51trading off, right? So we were talking a
1:20:53little bit, you know, with the Andrea's
1:20:54example about, you know, quality and and
1:20:57illegal activity. I mean I think so like
1:21:00the the California is we have, you know,
1:21:02we all have an election coming up but we
1:21:04have these this particular procedure
1:21:06where you can put a proposal on the
1:21:09ballot that goes directly for voters to
1:21:11decide, right? So the billionaires taxes
1:21:13is this one that's that's getting a lot
1:21:14of attention and then, you know, the
1:21:16question is
1:21:18is that a good policy? Do voters have
1:21:19good information to adjudicate like
1:21:22whether this is a good idea or not? And
1:21:23there's many intricacies to this policy
1:21:26but it comes down to the idea of like
1:21:29representation, I mean you are asking
1:21:31people to make decisions on your behalf.
1:21:32So the the thing that maximizes
1:21:35the idea of representation, one person
1:21:36one vote, is direct democracy, which is
1:21:38what California is experimenting with
1:21:41right now and for some types of issues,
1:21:44you feel like actually the average
1:21:45person A is very well informed,
1:21:47understands the tradeoffs and their
1:21:49preferences and the intensity of their
1:21:50preferences are a great way to make the
1:21:52decision. There are others that are so
1:21:55convoluted and arcane and involve these
1:21:57tradeoffs that, you know,
1:21:59even like the most sophisticated voter
1:22:01really struggles with and those really
1:22:03aren't well adjudicated by one person
1:22:06one vote. So I I think you can think
1:22:07about representation as this, you know,
1:22:09scale from do you want everyone to have
1:22:11to make every decision to thinking
1:22:13about, okay, we're choosing some people
1:22:14to make decisions on our behalf. What is
1:22:17important to us in terms of who's at the
1:22:19table and what are their competencies?
1:22:25>> Perfect, thank you.
1:22:27Um
1:22:28Chikezie, I'll go over to you for your
1:22:30question.
1:22:34>> Okay, thank you so much for your
1:22:35lectures.
1:22:36I'm more particularly interested in
1:22:39uh political representation in West
1:22:41Africa. I'm currently in Nigeria, so
1:22:42I'll ask
1:22:44uh how much does political
1:22:45representation matter for development
1:22:47when government institutions are weak?
1:22:49For instance, can we having a better
1:22:51representation lead to economic uh
1:22:54change if the institutions responsible
1:22:56to enforcing these uh
1:22:58policies uh cannot deliver?
1:23:02Don't have Does not have the capacity to
1:23:03deliver.
1:23:06>> Maybe I I can start a little bit on
1:23:08that. So, I think
1:23:10what what you're saying is like does a
1:23:12representation of groups or or
1:23:14individual people, can that have a
1:23:17positive influence on government
1:23:18performance when the overall structures
1:23:20of government are weak, right? When you
1:23:21have weak institutions, weak capacity?
1:23:24So, I think a kind of interesting point
1:23:26that that um Jones and Olken paper made
1:23:30is that it it depends on both, right?
1:23:34So, their their results are showing
1:23:36you know, a single leader can have a
1:23:38really large impact on on economic
1:23:41policy, particularly when they face no
1:23:43checks and balances from government,
1:23:45right? So, an autocracy. And this is for
1:23:47better and for worse, right? So, if you
1:23:49have an autocrat that makes really bad
1:23:50decisions, you're going to have very bad
1:23:52macroeconomic impacts, but if you get a
1:23:55great autocrat that makes really great
1:23:57decisions, you can have you can have
1:23:59positive impacts, right? So, I think the
1:24:02idea of institutions and capacity, these
1:24:05things are working together and are
1:24:06tackling different dimensions of this.
1:24:08So, I tend to think of like the
1:24:10political institutions, like you know,
1:24:12what is the selection and incentives
1:24:13that is going to bring good people into
1:24:15leadership? And then the capacity
1:24:17question is
1:24:18what is the state able to deliver the
1:24:21policies that the leadership chooses,
1:24:23right? And so I think your your question
1:24:25is a good one because even if you have a
1:24:27really great president who has a really
1:24:29great policy platform, but the
1:24:30ministries are so lacking in capacity
1:24:32they can't deliver on it or there's no
1:24:35government revenue,
1:24:37the ability of the leader to implement
1:24:39great policies is is going to be
1:24:41constrained and I think that's that's
1:24:43perhaps what you're you're worried
1:24:44about.
1:24:50>> Yes, that's what I worry about.
1:24:52Thank you.
1:24:54>> Yes.
1:24:55>> Thanks, Chikezie.
1:24:57Um yes, but
1:24:58Sorry, one second. There's a couple
1:25:01other questions in the chat that um
1:25:05that I just wanted to touch on and part
1:25:09part of I think what resonated for me
1:25:11from your presentation was uh
1:25:13essentially this idea that I
1:25:15experience is the best teacher. And so
1:25:19when we have an experience, we can see
1:25:23uh
1:25:24are my biases reinforced or are they
1:25:26broken down?
1:25:28Um and kind of relating to that,
1:25:31Lutufullah Khan asked um whether there
1:25:36is any longitudinal evidence of
1:25:38interventions, whether educational or
1:25:42participatory um that have helped
1:25:45communities be more receptive to women's
1:25:47leadership or significantly
1:25:50shifted prevailing norms around
1:25:52leadership. Um and maybe I'll throw in
1:25:54kind of tribalism in there as well
1:25:57around the social norms that Renee
1:25:58talked on earlier.
1:26:00Um whether there these are in fragile
1:26:03settings or not. Um
1:26:06In addition to experience, what other
1:26:09um
1:26:10things have have you guys seen that can
1:26:12help shift that needle?
1:26:19>> Wait, this is
1:26:21and it's good to highlight it even
1:26:22though
1:26:23they don't come to mind the names don't
1:26:25come to mind immediately. There's a
1:26:25large cultural literature that
1:26:28highlights different forms of kind of
1:26:30pre-colonial
1:26:32institutions that may have been positive
1:26:35or not in favor of women.
1:26:38I think you want to stay away from the
1:26:41the trope that you know before
1:26:42patriarchy there was matriarchy. I don't
1:26:44think we have any evidence of that
1:26:46overall but we certainly have variations
1:26:48both because of
1:26:51the kind of
1:26:52occupational choices that men and women
1:26:54made in different settings
1:26:56and things like population density about
1:26:59the relative roles that men and women
1:27:01played outside the
1:27:02household and that has often you know
1:27:05translated over time matriarchy
1:27:07matriarchal versus patriarchal
1:27:08inheritance systems being a common one.
1:27:10So
1:27:11I think there's a there's a growing and
1:27:13I encourage people to kind of go and
1:27:15look for for those people in the culture
1:27:18in the culture literature thinking about
1:27:20it. There's a good handbook of cultural
1:27:21economics that has some of these
1:27:23references.
1:27:27>> Yes, you mentioned inheritance. There's
1:27:29a lot of good work on like what happens
1:27:31when you give women property rights in
1:27:32very unequal societies. So that's that's
1:27:34a totally different set of policies but
1:27:37we can also make some progress.
1:27:41>> Thanks.
1:27:42Benjamin over to you. Can you ask your
1:27:45question?
1:27:46>> Yes, thank you. So um thank you very
1:27:49much for presentation. I'm Benjamin from
1:27:51Freetown Sierra Leone. So I
1:27:54was just
1:27:56curious to ask how does a voting
1:27:59a voting system affect the election
1:28:01results and then eventually better
1:28:03social economic and social political
1:28:05outcomes. I'll give an example of
1:28:07Sierra Leone has now a
1:28:09wants to adopt Saudi Supreme Court
1:28:11because there was
1:28:12in the parliament of Sierra Leone as to
1:28:14what system of voting should be
1:28:16implemented. Whether it's the
1:28:18proportional representation or the first
1:28:19past the post. And then of course,
1:28:22during the proportional representation
1:28:23the voters have direct access to vote
1:28:25who they want to vote in parliament and
1:28:27in at state house. And then of course,
1:28:29during the
1:28:30at
1:28:31in in in
1:28:33in the PR system you only have to vote
1:28:36for your political party and then of
1:28:37course the political party would have to
1:28:39choose
1:28:40parliamentarians
1:28:42that they will slot in based on the
1:28:44percentage of votes they get. So, how
1:28:46does
1:28:47this pattern, these the the the election
1:28:49process
1:28:50the election rule affects the outcome
1:28:53the social economic outcome and the
1:28:54social political outcome?
1:28:57And of course, what is the effect of
1:28:59funding money, somebody that has money,
1:29:01how does it affect the result?
1:29:03Especially for sub-Saharan African
1:29:05countries like Sierra Leone.
1:29:07Thank you.
1:29:09>> I can I can take that one. Benjamin
1:29:11Kouchner, my buddy. Glad to have you
1:29:13here.
1:29:14Um
1:29:14>> Thank you.
1:29:16>> [laughter]
1:29:17>> So, this is a huge change, right? So, we
1:29:19actually some of these same co-authors
1:29:21that um I've been working with for a
1:29:22long time, we we did another another set
1:29:25of experiments in the most recent
1:29:27election where they made the switch to
1:29:29PR.
1:29:30And I think
1:29:32So, this is coming back to like how does
1:29:35political selection fit into the broader
1:29:37system of political institutions and and
1:29:39the rules and procedures through the way
1:29:41elections are are determined, right? So,
1:29:45there's some very high-level differences
1:29:47between first past the post majoritarian
1:29:50and proportional representation, right?
1:29:52So,
1:29:53typically people think about well, if
1:29:55you switch to PR you have space for
1:29:57smaller parties like more specific
1:30:00interest to come through like green
1:30:01parties, right? So, you get a lot more
1:30:02green parties in in Europe than you do
1:30:05in the United States. You you have this
1:30:06plethora of sort of parties with a lot
1:30:08on the ideological spectrum.
1:30:10I think um
1:30:13So,
1:30:14the trade-off is that well, the other
1:30:17thing there's variations in how PR is
1:30:19done, right? So, so the PR system that
1:30:22was used in Sierra Leone in the last
1:30:23election is a closed list.
1:30:26And the closed list means the party
1:30:28picks the candidates and ranks them.
1:30:30And then and then like you said, you're
1:30:31right, you just choose between party A
1:30:33and party B.
1:30:34Other countries do PR with an open list.
1:30:37So, voters can both vote for the party
1:30:39and then within the party they want,
1:30:40they can vote for individual candidates.
1:30:43Right? So,
1:30:45one big trade-off when you have the
1:30:47closed list is your your severing the
1:30:49connection between individual voter
1:30:51choices over individual candidates.
1:30:54Right? So, if it's a closed list, voters
1:30:57have no direct say over the performance
1:30:59of individuals, right? So, that's going
1:31:01to cut voters out from the
1:31:03accountability channel and from the
1:31:05selection and quality channel at the
1:31:07individual candidate level. And it puts
1:31:09all of that back into the hands of
1:31:12parties.
1:31:14So, I think what one of the things we're
1:31:16seeing is that
1:31:18in Sierra Leone in particular, when they
1:31:19did this big shift to PR, one of the
1:31:22biggest changes that we see is that
1:31:26incumbents reran at much higher levels
1:31:29than they have historically over the
1:31:31last 20 years in Sierra Leone. And they
1:31:33won at much, much higher rates than they
1:31:36ever have before, right? So, then you're
1:31:38thinking about okay, the party is making
1:31:40decisions that reflect things perhaps
1:31:43beyond what voters want, right? So,
1:31:45maybe the the party is making decisions
1:31:47over incumbents in a way that at least
1:31:49in the previous system, voters were not
1:31:51rewarding incumbents as much as they are
1:31:53now. So, I think
1:31:55um I think your question is that these
1:31:57issues of of selection and also of
1:31:59accountability incentives um they're
1:32:02very strongly determined by the overall
1:32:05electoral system and there's some very
1:32:07sharp trade-offs, positives and
1:32:09negatives from moving between one system
1:32:11and the next.
1:32:13But great great question and a and a
1:32:14live issue in Sierra Leone about whether
1:32:17they'll stick with this or whether
1:32:18they'll change the system
1:32:20for the next election.
1:32:24>> Great. Well, thank you so much and I
1:32:26just wanted to to fight everybody
1:32:30will post these slides along with the
1:32:32recording of this on the website. I
1:32:34would encourage everyone to go back look
1:32:36over the slides especially those those
1:32:38last couple.
1:32:39Um
1:32:41just talking about what the most
1:32:43pressing questions are in the field. So
1:32:44those that are in a PhD or thinking
1:32:47about dissertations, it's it's great to
1:32:49understand kind of what the the key
1:32:51questions are right now and and where
1:32:54kind of the gaps are that that need to
1:32:55be focused on. So this is a really
1:32:57helpful resource for that.
1:33:00Um thank you Kate, thank you Rohini and
1:33:03everyone.
1:33:03>> Hello sir.
1:33:05>> Sorry, I'm sorry.
1:33:06>> ask a question?
1:33:08>> No, sorry. We're out of time now.
1:33:11But thank you everyone and we'll see you
1:33:12next week for our eighth lecture on
1:33:16conflict and fragility.
1:33:18>> Thanks for your watch.
1:33:19Thanks so much everybody.
1:33:22Have a good day. Bye.

About this transcript

This page contains the full transcript of BREAD-IGC virtual PhD-level course 2026 | Lecture 7 Political Selection and Representation by International Growth Centre, generated from the public captions YouTube serves with the video. The transcript has 15,720 words across 2,608 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.