第六季第30集：加纳的移动支付

移动货币的市场结构极其有趣。它是一个能进行优质经济学研究的平台，探讨在面向贫困人群的金融服务市场中出现的各种问题。

Mobile money, the market structure is super interesting. It's a platform to do really good economics, studying questions within the marketplace when it comes to financial services for the poor.

欢迎收听VoxDev两期专题访谈的第一部分。我是蒂姆·菲利普斯。我们如何设计数字普惠方案，才能在乡村、低信任度或现金密集型经济中最大限度减少欺诈并最大化社区收益？这是来自加纳的三项研究提出的问题，这些研究的作者是伯克利的弗朗西斯·阿南德。我最近就这项研究与他进行了两次对话。

Welcome to part one of a two part VoxDev talk special. My name is Tim Phillips. How can we design digital inclusion in a way that minimises fraud and maximises the benefits of the community in rural, low trust or cash heavy economies? That's the question posed by three studies from Ghana and the author of those three studies, Francis Anand of Berkeley. I spoke to Francis twice about this research recently.

自研究生阶段起，这就是他职业生涯的重要部分。今天我们将探讨他为减少移动货币市场欺诈与不当行为所采取的创新干预措施。在本周稍后发布的第二部分，我们将聚焦他如何管理与服务提供商及监管机构的关系，从而持续扩大这项工作的影响力。但首先，我请弗朗西斯阐述移动货币为何对发展如此关键。弗朗西斯，欢迎来到Vox Dev访谈。

It's been a big part of his working life since he was a graduate student. Today we talk about his innovative interventions to minimize fraud and misconduct in the mobile money market. In the second part, coming later this week, we will focus on how he managed the relationship with providers and regulators that has allowed him to build on the impact of this work. But first, I asked Francis why mobile money is so fundamental to development. Francis, welcome to Vox Dev Talks.

你好，蒂姆。

Hello, Tim.

这些研究都聚焦加纳日常的移动货币市场。请向我解释为何移动货币在加纳——尤其是乡村环境中如此重要且实用。

These studies, all of them are about the day to day mobile money market in Ghana. Explain to me why mobile money is so important, so useful in Ghana, especially in a rural village setting.

移动货币的核心显然是金融可及性。它真正改善着民众——特别是贫困人群的金融生活。同时我们拥有大量经济学证据表明：它能帮助人们摆脱贫困，提升福利水平，增强储蓄能力，显著赋予人们自主权，实现低成本汇款，并缓解这些地区普遍存在的严重经济冲击。其益处不胜枚举。

Mobile money, obviously, it's about access to finance. It's really about improving the financial lives of people, especially the poor. At the same time, we have a lot of evidence, in particular in economics, that have shown it has the ability to lift people out of poverty. It has the ability to improve welfare, the ability to save, the ability to empower people distinctively, the ability to remit money in a low cost way, the ability to mitigate shocks that are very profound in a lot of these settings. And the list goes on and on and on.

它能增强该市场民众的金融与经济韧性。加纳并非特例。尽管移动货币已成为加纳人日常生活的一部分，但它确实具有显著提升社会福利的深远能力。

It has the ability to build the financial and economic resilience of people in this market. Ghana is not an exception. In as much as mobile money is part of the day to day life of people in Ghana, it has this profound ability to actually improve welfare.

你将当地提供这些服务的人描述为‘人肉ATM机’，能否具体说明什么是人肉ATM机？

You describe the people who are providing these services locally as human ATMs. Can you describe who a human ATM is?

由于缺乏更贴切的定义，我更倾向于将这些直接与消费者对接的下游服务者视为人肉ATM机。在欧美等地的传统金融服务机构中，我们要么去银行大厅办理业务，要么通过ATM机存取款。当移动货币作为金融服务的媒介时，这不过是其现实映射。市场体系本质上由上游参与者（如大型商业供应商、电信公司及部分后端商业银行）提供支持，但如何让银行未覆盖地区的人们也能享受这些服务？

For lack of a better definition, I like to think of these downstream players that actually interface with consumers as human ATMs. In the classic form of financial services institutions in The US and Europe and other places, either we interface at the banking hall, we go to an ATM to make deposits or withdrawals. It's just an analog of that when it comes to mobile money as a medium to provide access to finance. And the markets are in principle organized such that you have upstream players, in this case, large commercial providers, telecommunication companies and some back end commercial banks that was delivering this service. But how do you make these services accessible to people in places where these banks are not located?

这就需要人肉ATM机代表上游服务商，为消费者提供开设移动货币账户、存款（即现金存入）、取款（即现金提取）等服务，并在零售网点配套各类数字工具。这些人肉ATM机实质上是移动货币供应商的下游代理，通过与消费者直接互动来传递庞大的数字金融服务生态。

Well, you're gonna need this human ATM on behalf of this upstream player who offer the ability for consumers to show up, open new mobile money accounts, make deposits, what people call cash ins, make withdrawals, what people call cash outs, and onboard a bunch of other digital tools at these retail outlets. So human ATMs are gonna be these downstream players that act on behalf of these large commercial providers of mobile money that interface with consumers to deliver these massive space of digital tools.

要使这个系统良好运转，我们需要哪些基础设施？此外，市场运作需要建立怎样的人际信任关系？哪些要素必须到位？

For this system to work well, what sort of infrastructure do we need? And also, what kind of human relationships, trust relationships that we need for markets to function? What do we need to be in place?

技术层面首要是网络覆盖——这是基础架构的关键。移动货币在这些市场具有吸引力，很大程度上得益于电信基础设施从城市到乡村的深度普及。当然，网络连接是刚需。但与此同时，人性化因素同样至关重要。

This is about technology, so we need to have access for connectivity. That's one thing which is very structural. And I think one of the key things that makes mobile money appealing, especially in this market is how telecommunication infrastructure has scaled quite profoundly, right, from the city to the village. Clearly you're gonna need the connectivity. At the same time, this is gonna be about humans.

现在我们暂离技术话题，聚焦于消费者与人肉ATM机交互时的人际关系。信任与问责机制变得极其重要——要维持并扩大这种技术驱动型金融服务的影响力，必须建立可信且责任明确的金融市场。遗憾的是，在贫困乡村等历来缺乏正规金融服务的地区，现状远非如此。这些本就脆弱的群体，从非正规金融向正规金融过渡时，将面临更多风险敞口。

So we're gonna move away from technology and now begin to think of human interactions, where consumers engage with this human ATM at some point. So then the notion of trust and accountability becomes very important For you to sustain and maximize the impact of this technologically driven financial service, you're gonna need a trustworthy and more accountable financial marketplace, which unfortunately has not been the case, especially in markets like the village of the poor where historically people have lacked access to formal financial services. And at the same time, these people are also very vulnerable, right, to begin with. Yeah. Having to now transition from informal finance to more formal finance creates a lot of vulnerability.

这个市场可能滋生诸多风险，其存续也依赖于人们对市场经济的信念。若民众对市场风险存在认知偏差，必将侵蚀信任根基——毕竟商业活动仰赖信任而生。当然，我们也必须依托有效的问责机制来维系运作。

And this is a marketplace that could deliver a lot of risks. Course, it also depends on the belief that people have in this market economy. And in fact, if people have the wrong beliefs about the risks that this market poses, then clearly this is gonna undermine trust. Commerce only tries on trust. The same time, of course, we're gonna have to rely on accountability mechanisms, right?

我们究竟该如何监管？如何为这一领域的人们提供救济机制？在发达经济体中，金融领域已有强大的监管力量和非常完善的消费者救济机制——如果你遭受侵害，可以举报并寻求救济。但在这些制度薄弱、功能不健全的环境中，情况则大不相同。嗯。

Really how do we monitor? How do we provide redress mechanisms for people in this space? In advanced economies, when it comes to finance, we have seen a lot of regulatory power and very interesting redress mechanisms that consumers, as a whole, you're victimized, right, you could report and seek for redress. It is less so in this context, these environments where these institutions are shallow and don't function as well. Mhmm.

与此同时，你谈论的是某个村庄，比如生活在加纳北部或肯尼亚北部偏远角落的村落。对每个监管机构来说，亲自到场核查显然不现实。因此必须针对当地问题，制定真正本土化的政策或法规。

At the same time, these are, you're talking of a village, a single village who live somewhere in the corner of Northern Ghana or Northern Kenya, gonna be hard for every regulator, right, to show up in the village and say, look, I'm gonna go look at this. So you're gonna have to think of really local policies or local regulation for local problems.

我们将讨论这三篇论文，它们阐述了这些风险如何破坏信任，并导致市场失灵。首先关注的是服务商超额收费问题。那些人力ATM机提供服务时收取费用——但谁来确定他们可收取的费率标准？

We're gonna discuss these three papers, which are about how those risks can undermine trust and how they can make these markets not work well in some cases. First of all, we're looking at overcharging by vendors. The human ATMs, they charge for their services. Who sets the tariffs that they are allowed to charge?

移动货币市场的设计是垂直化的：上游商业主体拥有服务平台（我们称其为供应商），下游则是代理或商户等中介机构。这些中介代表供应商行事，而交易服务费率由上游供应商预先设定，人力ATM只能接受既定价格——本质上这些价格是由商业供应商设计并确定的。

Markets for mobile money are designed to be vertical in the sense that there's this upstream commercial player who owns a platform and provide a service, call them providers. And then you also have these downstream intermediaries, call them agents or vendors. And this as opposed to act on behalf of the provider. The tariffs or the fees on transactions and services that are delivered are these retail points, ex ante, set by the upstream player, in this case the commercial provider, and the human ATMs as opposed to take these prices as given. Prices by definition or by construct are designed and set by the commercial provider.

因此这本质上是个量能市场——交易量越大，佣金越多。关于这种政策优劣可以长篇讨论，不过这个话题我们留待其他播客节目展开。

So it's basically a volumes market. The more transactions you do, the more commissions you earn. We can talk quite a length about why this is a good or bad policy. But I'll leave that for another podcast.

你们调查的是超额收费现象。这种现象有多普遍？他们又是如何得逞的？

In this case, you're investigating overcharging. How widespread is this overcharging? And how do they get away with it?

跨市场、跨地区的数据显示这简直像是个魔幻数字——每四到五笔交易中就有一笔存在超额收费，尤其在乡村地区极为普遍。

Across different markets, different districts, different regions, it's almost like a magic number. It's remarkably widespread. One in four, one in five transactions are overcharged. Wow. Especially in the villages.

这一点非常一致。顺便说一句，这也具有很高的外部效度。从尼日利亚到乌干达等国家进行的跨国调查——这些调查由我们的姊妹机构如IPA开展——实际上显示消费者报告的水平相似，甚至更高。因此，这种现象确实是这个市场的一部分，极大地限制了如何最大化影响力的可能性。人类ATM有很多渠道可以逃脱惩罚。在许多情况下，消费者甚至不知道自己第一次受害。

And this is very consistent. And by the way, this also has a lot of external validity. There are all these cross country surveys from Nigeria to Uganda and all of that which are being run by our sister institution like IPA, where in fact consumers report similar levels, if not even higher levels of of So the it's a phenomenon that's really it's part of this market that really limits the ability for how to maximize the the impact. There are a lot of channels through which is human ATMs can get away with it. In a lot of instances, consumers are not even aware that they're being victimized for the first time.

没错。因为消费者对真实价格缺乏了解，价格透明度有限。要成为人类ATM，你需要接受商业提供商的培训。但不幸的是，消费者从未接受过如何成为消费者的培训。这就造成了巨大的信息不对称或知识差距，如果你愿意这么称呼的话。

Right. Because consumers have poor visibility into what the true prices are, and there's limited price transparency. For you to become a human ATM, you are trained for this commercial provider. But unfortunately, consumers are never trained to become consumers. So that creates this massive asymmetric information or knowledge gaps, if you want to call it.

因此，作为消费者，你只能通过艰难的方式学习。你总会在某个时候成为受害者。另一个事实是，你可能甚至不知道自己在受害或被欺骗，或者即使知道也没有选择。唯一提供服务的人会说：‘看，这就是现实’，尤其是在追索机制非常薄弱的情况下。你唯一的应对方式可能是减少使用，但心里明白自己可能被骗，这并不是...

So as a consumer, you're gonna have to learn it the hard way. You're gonna have to get victimized at some point. The other part is also the fact that you may not be information, indeed you know that you're being victimized, you're being cheated, but you don't have an option. The only guy who offers the services says, look, this is the fear and especially when the redress mechanisms are very weak. So the only way you can respond is, well, I'm gonna use it very less, maybe not more, but taking into cognizance of the fact that I may be cheated, this is not.

监管和监控非常薄弱。市场上缺乏合规的成本极低。零售商或代理商——这些小企业——在做是否欺骗的心理计算时可能会考虑到这一点。成本低，收益高。另一个原因是，想象一下，即使市场上有更多参与者、更多零售商，一旦有人开始欺骗，就会破坏市场信誉，因为其他人不再信任市场，于是每个人都会效仿。这时你就陷入了囚徒困境问题，市场可能维持在一个非常糟糕的状态，仅仅一个‘柠檬’就足以让整个市场变成‘柠檬市场’。

Regulation and monitoring is very weak. Cost of lack of compliance in the marketplace is so very low. Retailers or agents, the small firms, might take that into consideration as they do the mental calculation, of whether to cheat or not. The costs are low, the benefits are high. The other reasons, imagine that even when there are more players, there are more retailers in the market, and one person starts cheating, it destroys the market capital because no one else believes in the market, and therefore everyone So will follow then you run into this prisoner's dilemma problem, where in fact the market can sustain a very bad outcome where just one lemon in the market could actually make the market a lemon market.

因为人们对服务本身（而非特定提供商）缺乏信任。下游如此，因此即使你不想欺骗，欺骗也可能是更有利的选择。我们有很多证据支持这一点。

Because people have beliefs or lack of trust about the service, not about a particular provider. Right? Downstream, and therefore, you are better off also cheating even if you don't want to cheat. We have a lot of evidence of that.

你们通过实验来观察是否能降低欺骗或不当行为的水平。你们具体做了什么？

You used an experiment to see if we could reduce the level of cheating, the level of misconduct. What did you do?

我们讨论过信息不对称，对吧？下游提供商和消费者之间的信息不对称。因此，我们设想了一些与信息相关的干预措施，思考如何通过消除供应商和消费者之间普遍存在的信息摩擦来减少不当行为。这是思考这个问题的起点。其他所有项目都将逐一解决这些不同的摩擦，尝试缓解它们，看看这对解决不当行为有何影响。我们正在深入思考可以向消费者——实际上不仅是消费者，还包括提供商——提供的信息结构。

We spoke about information asymmetry, right, between the downstream providers and also consumers. So you imagine interventions that are information related, thinking how to reduce misconduct by eliminating those informational frictions that prevails between the vendors and also consumers. Okay, so that was the starting point in thinking about this. All the other projects are gonna take these different frictions one by one and try to relax them to see how that matters in resolving this misconduct. We're thinking hard about the structure of information that we can provide to consumers, but not in fact to only consumers but also to the providers.

我称之为双向信息干预。我们希望消费者能获取这些信息，同时我们也希望代理商、供应商知道消费者已掌握这些信息及其具体内容。关于这些乡村经济，我们确切了解声誉机制的重要性。我们真正想设计的是能激活这种声誉渠道的信息干预，从而利用当地机构组织方式的价值。简而言之，这些双向信息项目既让消费者更准确了解真实价格，也让代理商/供应商意识到消费者现在知晓这些干预措施。

What I call two sided information interventions. We want consumers to have access to this information, at the same time we want the agents, the suppliers to also know the consumer has access to this information and here is that information. And there's a precise thing that we know about these village economies, the notion of reputation, which matters a lot. We really want to design information interventions that turns on this reputational channel in a way that can allow us harness the value of how these local institutions are organized. In short, these were two sided information programs where consumers were better calibrated about what the true prices are and the agents, suppliers also made aware of the fact that now consumers are aware, okay, of these interventions.

非常简单、低成本的信息项目，却能显著减少不当行为。

Very simple, very low cost information programmes that has the ability to distinctively reduce misconduct.

那么在你们的实验中，消费者知道合理价格，商贩也知道消费者知情。这有效吗？是否降低了不当行为水平？同时是否增强了消费者对不被欺骗的信心？

So in your treatment, the consumers knew how much it should cost. The vendors knew that the consumers knew. Did it work? Did this reduce the level of misconduct? And also, did it make the consumers more confident that they were not going to get cheated?

是的，产生了多重效果。第一，若在两个相似村庄中，一个实施双向信息干预而另一个不实施，你会发现受干预村庄的代理商欺诈概率降低约72%。这确实极大改变了不当行为发生率，不仅影响发生频率，还降低了超额收费的严重程度。

Yeah. So a lot of things happened. Number one, if you take two villages and one village you introduce this two sided information intervention, in another village you want, and presumably these villages are identical, what you see is that agents in treated villages are about 72% less likely to cheat. So indeed it did a lot of change to the incidence of misconduct. And not just incidence, but even on the severity of how much is overcharged.

所有这些都发生了显著变化。当然，在干预实施前，这个市场的关键事实是消费者对真实价格乃至整个市场经济中的超额收费程度都缺乏准确认知。实际上，若客观测量超额收费或代理商不当行为，估计约为20-25%，即每四五次交易中就有一例。

So all of that changed remarkably. Of course, before the interventions were rolled out, one of the key facts about this market is that consumers were poorly informed or even calibrated, not only about what the true prices are, but in fact, what the true level of overcharging is in the market economy. In fact, if you do objective measures of overcharging or agent misconduct, you estimate around, let's say, 20 to 25%. Okay. As I explained, one in five or one in four.

客观数据如此。但当你询问团队认为普遍不当行为水平时，他们会告诉你60%。显然，通过提供这些信息，不仅因人们重视声誉而减少超额收费行为...

Objective. Mhmm. When you ask the team, what do you think, okay, is the prevailing level of misconduct? He tells you 60% of the time. So clearly, are even, so once you provide this information, you do not only reduce the level of misconduct because people care only about reputation and cutting down on overcharging, right?

还使消费者的认知更准确。实际风险并非如他们想象的那般严重。这些修正后的认知非常有用——虽然他们对不当行为水平曾有错误认知，但所有市场决策都基于这些认知。

But you also actually make consumers better calibrated. So if at a risk is not as we think as it is. And these beliefs are very useful, right? Because they may have imperfect beliefs, right, about the level of misconduct. But in fact, all their market decisions are based on that.

因此，让消费者更准确地了解情况是一回事。而代理商或公司根据这些信息做出反应则是另一回事。所以最终，我们估计的这些信息干预措施的效果，实际上是两方面的综合结果。一方面是消除原有认知偏差的程度，另一方面是服务提供商因重视声誉而对信息项目作出的响应。事实上，我们在实验村和对照村都观察到了移动货币服务需求的大幅提升。

So getting consumers better calibrated is one thing. A conditional giving that information, agents or firms responding to this information is also another thing. So you see that eventually, the treatment effects that we estimate, the effect of these information interventions that we estimate is gonna be a decomposition of two things. One is about eliminating the bias in how much indeed there was, and then also the fact that now providers are responding to this information program because of all the reputation they care. And indeed, we saw massive improvement in demand for these mobile money services across our treatment and control villages.

这一变化很大程度上既源于零售商行为的改变，也源于消费者端偏见的转变。

And a lot of this is driven by both the fact that the behavior of the retailers are changing, but also the bias from the consumer side is also changing.

是否需要告知每个使用移动货币的人？需要对所有人进行教育吗？还需要通知所有代理商吗？还是说只需针对部分群体实施，就能通过代理商变得更诚实和人们口口相传产生溢出效应？

Do you have to tell every person who's using mobile money about this? Do you have to educate everyone? Do you then have to inform all the agents as well? Or can you do it for some of them and there'll be a spillover effect because the agents become more honest and people talk about it?

事实上你超前了我的论文思路，但我很欣赏。这正是我们的做法——寻找所谓的低成本信息干预措施。当然，发展领域最核心的问题之一就是如何以更低成本实现同等影响力。

In fact, you're ahead of my paper, but I like that. And that's precisely what we do. We were in the search for what we call low cost information interventions. Obviously Yeah. One of the biggest questions in development is how do we achieve the same impact at a lower cost.

基于此，我们研究了外部性或溢出效应的概念。这意味着我们无需与市场上所有代理商或消费者沟通就能见效。在我们的研究场景中，典型村庄约有3个代理商，服务500至5000人口。我们首先普查这些市场，然后随机选择部分（而非全部）代理商进行干预，从而即使在实验村也能观察未受干预企业对其他竞争者受干预后的反应。

So in that sense, we toyed ourselves with this idea of externalities or spillovers, which actually simply meant that I didn't have to talk to all agents or all consumers in the market for this to work. The way the markets are organized, the typical village in our setting has roughly about three agents, three providers, the typical village. This is with a population of about 500 to let's say 5,000. What we did was start to do a census of these markets and then we only randomly treated some agents or some providers and not all. So that even in the treatment village, we can study the effect of how the non treated firms responded to the intervention conditional on any of the other competitors or other firms in the market being treated.

这确实产生了显著效果。未受干预企业的响应程度达到受干预企业的一半左右。这为低成本部署此类干预提供了思路模板——因为无需覆盖所有人就能见效。

And it did refined a lot of large effects. The untreated firms responded close to half of how the treated firms responded. They have the ability to provide some template in thinking about how to deploy these interventions at a low cost precisely because you don't need to talk to everyone for them to work.

转到你的第二篇论文，你提到不当行为的部分原因是人们缺乏替代选择。他们只能找代理商，若被多收费也无可奈何。所以后续研究聚焦于引入新供应商后的变化。如果没记错的话，最初是有禁止招募新供应商的禁令？

Onto your second paper because you did mention that one of the reasons why there is misconduct is because people don't really have many alternatives. They go to the agents, and if the agents overcharges them, well, that's tough. They don't have a choice. So your next study, this focused on what happens when we recruit new vendors. At the outset of this, if I recall correctly, there was a ban on recruiting new vendors.

这是正确的吗？为什么会这样？

Is that correct? Why was that?

完全正确。当市场上只有一个服务提供商时，实际上可能会产生激励过度收费的现象。这与我们上一篇论文中描述的信息结构完全正交或独立。同样真实的是，当存在多个提供商且他们共谋采取相同行动时，市场上仍可能持续存在不当收费行为。从先前的研究数据来看，有一个因素对不当行为具有显著预测性。

That is exactly correct. When you have only one provider in the market, you could actually create incentives for overcharging to prevail. And that would be very orthogonal or independent of the information structure we described in the previous paper. It is also very true that when you have multiple providers and they all collude to do the same thing, you could still end up sustaining misconduct overcharging in the marketplace. From the previous study, if you look at the data, there was one thing that was very predictive of misconduct.

如果提供商不仅提供金融服务，还涉及其他非金融服务，他们过度收费的可能性就会降低。一个既卖大米又提供金融服务的人，实施不当行为的概率更低。这让我开始思考：或许在服务提供专业化与多元化捆绑之间，存在某种影响不当行为结构的机制。我们可以在一定程度上规范市场，这实际上取决于人们选择专业化还是多元化经营。本文最大的动机之一在于，纵观全球金融市场发展，其核心理念正是追求这种低成本的服务提供方式。

If the provider is involved not only in providing these financial services, but also being other non financial services, they are less likely to overcharge. The fact that a guy sells rice and also provide financial services is less likely to commit misconduct. That for me was a trigger in thinking a little bit about maybe there's something to be done about whether people specialize in service provision or people bundle in service provision and how, to what extent that matters in understanding the structure of misconduct in this We can discipline the market to some extent. It really depends on whether people specialize or do not specialize or bundle. One of the biggest motivations for this paper is that if you look across the world globally, how a lot of financial markets have evolved in principle is to pursue this low cost way of providing services.

我们看到富国银行关闭了在美国的分行，转而寻求零售店或超市作为新的服务点。因此从某种意义上说，从发达国家到发展中国家，低成本扩展金融服务的传统方式正是将金融与非金融服务相互关联。试想ATM机通常设在何处？餐厅里、商店中——人们不仅来购物，还能同时取款。

We've seen Wells Fargo closing out their branches in The US and actually looking for retail outlets or maybe some supermarkets to actually now begin to offer services within the supermarket. So in some sense, the traditional approach in scaling financial services at a low cost from developed to developing countries is really this idea of really interlinking financial to non financial service. You even imagine, you ask yourself, where are typically the ATMs located? Typically, they are in some restaurant, they are at some stores, right? People show up not only to buy, I don't know, some grocery, but also take money at the same time.

同样的变革也发生在移动货币市场的组织方式上。现代扩展移动货币业务的核心策略，就是说服街角商店将金融业务与其主营业务（非金融服务）相结合。这有望改变移动货币市场的不当行为结构。禁令的存在是福也是祸：虽然限制了更多提供商进入，但若更多提供商本可降低不当行为，这对消费者实则不利。

And it turns out the same revolution is happening in the way mobile money markets are organized. The modern approach in scaling mobile money is really about finding a corner store and convincing this corner store, encouraging the corner store to then interlink financial with the non financial stuff, which is the core business. So this has the ability to change the structure of misconduct in the mobile money market. There was a ban. It's both a blessing and a curse because ban probably meant that we couldn't have more providers.

但禁令也有好处，它让我们能退后一步思考：或许可以通过研究市场的随机准入机制获得启示。这对研究有利，但对产品推出前的消费者可能不利。市场已发展到这样的阶段：首先，提供商无法确定市场中代理人的最优数量——一个村庄应该设多少分支机构？

And so if more providers meant that misconduct is gonna be lower, then actually it's bad, right, for consumers. But it's also good because we can carefully take a step back and ask ourselves, maybe there's something to be done here to learn more about randomized entry into marketplace. So it's good for research, but maybe bad for consumers before the start of the product. The market had evolved to some point where, first of all, providers were unsure about what the optimal number of agents to have in the marketplace is. How many branches should I have in the village?

他们不得不持续尝试，不断招募，直到发现市场因ABCD等因素过度饱和。问题的关键在于，他们无法确定村庄中代理人的最优数量，这取决于诸多因素。

I don't know. They're gonna have to try. Okay? They're gonna have to keep going, keep recruiting up until they get to a point where they think, oh, well, the market is too saturated because of ABCD. Got to the point where they were not sure about what's the optimal level of agents that you have in a village, it all depends on a bunch of things.

但市场已发展到出现黑市的地步，人们获取代理执照后以更高价格转卖给他人。这对他们而言是个明确信号——这里确实存在问题。加之不断涌现的违规行为报告，甚至部分代理因这类黑市交易而处于停业状态。

But the market has evolved to the point where there was a black market that was evolving, where people get the license to become an agent and they sell the license at a higher price to other people. So for them, was a clear signal. Oh, something is really happening here. Of course, combined with the fact that there were these issues that were coming up regarding misconduct. And even in some instances, some of the agents being inactive, precisely because of this black market activity.

这让他们意识到必须退后一步，重新评估整个代理网络：我们是否处于最优状态？如果不是，该如何在持续运营的同时解决已发现的问题？因此服务商暂停了执照发放和招募，正是出于这些担忧。于是我提议：或许我们可以采取更随机化的方式重构现有体系。

So for them, was a clear signal that they need to take a step back and reevaluate their entire agent network to ask the question, are we at the optimal point or we are not? And if we are not, what do we have to do to ensure that we can keep going but also to help fix the issues that we have identified? So there was a pause in recruiting or providing licenses precisely because of some of these signals that the provider was worried about. So I went in and said, maybe we can do something here. Let's think a little bit more about what it means to redo what you have been doing in a more randomized way.

这促成了我与CEO的对话，他表示公司层面其实正讨论向其他市场扩张。我说那正好，我们一起行动——提供市场/村庄清单，我们共同推进这个规模化项目。这就是本项目的起源。

And that led to my conversation with the CEO, and he was like, well, in fact, we actually have been talking at a corporate level that we wanted to actually scale our presence in other markets. So said, okay, then wait. Let's do this together. Give me the list of markets or villages, and then you and I can work together in actually scaling this. And that's gave birth to this particular project.

你说服他们招募新供应商进入市场以观察后续影响，是这样吗？

You managed to convince them to recruit some new vendors into the market to see what then happened to those markets. Is that correct?

完全正确。不仅是招募新供应商，还要随机化各市场的供应商密度，从而追踪新进入者带来的均衡效应。

That is exactly correct. And not just recruit new vendors, but to randomise the saturation of how many vendors we can have across the market so that we can trace out the equilibrium effects of this new entry.

那么在均衡状态下，这些新供应商是否影响了违规行为？是否改变了业务量？

And so in equilibrium, when you got these new vendors on board, did it have an effect on misconduct? Did it have an effect on the amount of business they were doing?

发生了许多变化，我重点说两个主要影响：首先在移动货币金融市场，当随机引入新便利店作为竞争者后，原有供应商实施欺诈等违规行为的概率降低——新进入者对现有商户形成了促进消费者保护措施的市场压力。

Well, a lot of things happened, but let me tell you two of the major ones. One is something happened in the actual financial market, the market for mobile money. Incumbent mobile money vendors were less likely to commit misconduct, to cheats if the market got a new entrant, okay, who is now this new corner store that has randomly been onboarded. So actually, was some pressure on incumbents to promote consumer protection practices in the marketplace. So that's on the mobile money side.

在非金融方面，这正是我们现在将那些非金融店铺与移动货币代理商相互关联的地方。我们看到非金融员工的收入大幅增长，这非常有趣。我认为这确实说明了移动货币市场不仅有能力建立金融韧性——这更倾向于改善消费者和家庭的生活——还展示了传统金融市场中我们未曾想到的对非金融市场的溢出效应。因此，我们在移动货币市场看到了巨大影响，尤其是在与不当行为和消费者保护相关的事项上。

On the non financial side, which is again, this is where we interlink those non financial stores now to become mobile money agents. We saw large increases in the revenues for the non financial staff, which is very interesting. And I think it's really the story that mobile money markets do not only have the ability to build financial resilience, which is more along the lines of improving the lives of consumers and households. But I think this is a result that shows the spillover effects in the non financial markets that traditionally we don't think of financial markets have. So we saw large impact on the mobile money market, more so on things that relates to misconduct and consumer protection.

我们还看到了业务成果的显著改善，以及非金融服务对当地经济的一些整体影响，我认为这在思考家庭货币市场时显得尤为独特。

We also saw much improvement in business outcomes and some aggregate local economy effect on the non financial services that which I think is super distinct in the way that we like to think of markets from home money.

你提到已经发现那些已有非金融业务的代理商较少涉及不当行为。能否解释一下原因？是因为声誉对他们在那方面也更重要，还是有其他因素？

You were saying that you'd identified that agents that had a nonfinancial business already were less likely to be involved in misconduct. Could you explain why that is? Is it because reputation is more important to them in that area as well, or is it something else?

他们的故事稍微复杂一些，但有几种假设。一种显然如你所描述的，关乎他们的声誉。因为一旦在市场上受到打击，你不仅在移动货币服务上受损，如果你卖的是大米，那大米业务也会遭殃。是的。因此，从这个意义上说，声誉影响在那里可能会大得多。

Their story's a bit more nuanced, but there are a couple of hypotheses. One obviously is, as you described, they came about their reputation. Because once you get hit, right, in the market, you don't only suffer in your services for mobile money, but you also suffer in your services for rice, if rice is what you sell. Yeah. So, that sense, the reputational impacts are likely to be much larger there.

所以这是一个假设。另一个假设与这些公司非常聪明的事实相关。作为一家杂货店，你的主营业务是卖大米，这是你关心的。在大米行业，价格没有监管。

So that's one hypothesis. The other hypothesis is related to the fact that, and of course, these firms are very smart. The main line of business for you as as a bundle store is you sell rice. That's what you care about. In rice, there's no regulation in prices.

你实际上可以为大米设定更高的价格。嗯。那么为什么要欺诈呢？你可以想象通过提高大米价格来降低移动货币的价格，从而总体上保持更高的利润。

You could actually set higher prices for rice. Mhmm. So why cheat? You can imagine reducing your prices for more money by increasing your prices for rice. Overall, you keep higher profits.

我们在论文中看到了很多这样的情况。大米的市场进入价格略有上涨。他们不太可能通过欺诈来实现从金融市场到非金融市场的这种传导。

We see a lot of that in this paper. Prices for entrance for the rice went up a little bit. They were less likely to cheat to see this pass through from the financial to the non financial market.

那么让我们看看你这组引人入胜的研究中的第三项研究，你正在探讨性别对不当行为的影响。我对此的第一个想法是，你如何从性别角度识别这种影响——既要识别哪些交易发生在女性之间、男女之间或男性之间，又要确定如果存在不当交易，是否是性别因素导致的。

So let's look at your third study in this fascinating collection of research, where you're looking at the effect of gender on misconduct. My first thought about this is how can you identify the effect of gender either from the point of view of how do you identify which transactions are being done between women, between men and women, and between men, and also identify that if there is any misconduct transaction, it is gender that is causing it.

当然，我们一直在讨论市场不当行为的基本前提。嗯。我们对这个市场中不当行为的本质知之甚少。而性别似乎是其中可能起关键作用的因素之一。因为在经济学理论中，确实有许多理论表明性别在风险态度上存在差异。

Of course, the general premise we have been talking about the notion of misconduct in the marketplace. We know very little about the nature of misconduct Mhmm. In this market. And gender seems to be one of the key things that could be at play here. Cause obviously, in economic theory, we have a lot of theories suggesting gender differences in attitudes towards risk.

风险厌恶可能具有性别特征，复杂程度可能因性别而异，竞争反应或许也存在性别差异，还有其他诸多方面。因此在理论上，这些市场基础要素确实存在许多差异，这能帮助我们更合理地思考为何不当行为也可能具有性别特征。再想象一下，在类似这些村庄的市场中，若不当行为不成比例地针对某一特定性别群体，那就可能构成歧视。

Aversion may be gendered. Sophistication could be gendered. Competition responses might be gendered and a bunch of other things. So there are a lot of really, in theory, differences in these market based primitives that actually could help us think more conceivably about why misconduct could be gendered as well. Also imagine that in a marketplace like these villages, if misconduct is committed disproportionately against one particular gender group, then that could be discriminatory.

嗯。不仅是歧视，还可能造成效率低下。尤其当不当行为将资源从效率较高的性别群体重新分配给效率较低的群体时。对吧。唯一能说的是，它甚至可能是高效的，因为这仅仅增加了交易的边际成本，而这可能引发我们在早先论文中讨论过的一些问题，即人们实际上认为风险比实际情况更高。

Mhmm. And not just discriminatory, it could also be inefficient. Especially if misconduct reallocates resources from the more efficient gender to the less efficient gender. Right. Only thing you can that, it could be even efficient because it just increases the marginal cost of transactions, and this could lead into some of the issues that we discussed in the earlier paper, where people actually think that the risks are higher than they are in practice.

因此，性别如何与不当行为相互作用的观念，在我们思考的背景下，无论是理论上还是实证上都是合理的。我们所做的是提出这项非标准审计研究：我们招募审计员，称他们为消费者。这些是生活在该市场的消费者，数量众多。鉴于我们按性别对所有代理商进行了普查，因此每个代理商都有普查数据。

So, again, the notion of how gender interacts with misconduct is something that is both theoretically and empirically rationalized in the context that we are thinking about. What we do is to propose this non standard audit study where we recruit auditors, call them their consumers. These are consumers that lives in this market. A bunch of them. And given that we have a census of all these agents by gender, so for each agent we have a census.

我们知道操作员的性别、店铺所有者等所有信息。当然，因为我们随机派遣招募的消费者前往不同地点进行交易。他们经过培训和校准，能在每笔交易后立即记录交易体验。这些数据通过他们随身携带的平板电脑在离开代理商场所后上传，这使我们能够克服某些巨大的数据限制，特别是在考虑不当行为与性别时，同时也让我们能够对分选现象进行论述。

We know the gender of the operator, the owner of the store and everything. Now, of course, because we recruit our consumers to send them at random to go to places, right, and conduct transactions. And they are trained and calibrated in a way that right after every transaction, actually they record the experience of the transaction. In those tablets, that is usually in their backpack once they leave the premises of the agent. That allows us to overcome some of the massive data constraints that we have, especially if you wanna think of misconduct and gender, but also in a way that actually allows us to say something about sorting.

在传统市场中，人们可能会自行选择交易对象。但在我们设计的实验中，实际上是随机分配人员。消费者不能选择去向，而是由我们指派。因此我们能够真实测量不同消费者面对不同性别产生的不同不当行为反应时的实际影响。

In the traditional marketplace, people may sort. You decide where to go to. But in the way our experiment is designed, actually we randomly assign people. You don't decide where you go, but we assign to places where you go. So we can really measure the true effect of different consumers being faced with different things, different misconduct responses by gender.

我想我们称之为神秘顾客的女性，她们是否更频繁地遭遇欺诈？

Were the female, I suppose we could call them secret shoppers, were they cheated more often?

是的。这可能并不令人意外。我们在其他市场也见过类似情况。不，很遗憾。

Yeah. And this is probably not an unsurprising result. We've seen that in other markets. No. Sadly.

是的。无论是租房、买车，还是其他消费领域，女性审计员遭受了严重的歧视。与男性消费者相比，她们被欺诈的概率高出好几个数量级。这种现象非常显著。

Yeah. Shopping for housing, shopping for automobile, like a car, or you've seen a lot of that. So, female auditors were massively discriminated against. They were much more likely to be cheated in several magnitude relative to their male consumer counterpart. And this is very massive.

而且当女性顾客光顾女性商家时，欺诈行为更为严重。什么？相比之下，女性顾客面对男性商家时反而好些。

And the cheating is more so when female customers visit a female vendor. What? Compared to female customer visiting a male gender.

不可能吧。

No way.

这显示出女性之间存在性别歧视，而男性之间则存在性别偏袒。换句话说，如果让男性顾客去进行移动货币交易，虽然所有人都会受骗，但女性消费者被骗得更厉害。明白吗？这是第一个事实。所以这纯粹是性别歧视。

So see this evidence of within gender discrimination for females, but within gender favoritism for males. In other words, if you were to send a male customer to go out there and engage in this mobile money transaction, of course, everyone is being cheated, but female consumers are being cheated more. Okay? That's the first fact. So there's really purely discrimination by gender.

除此之外，相比光顾男性商家，受害更深的女性顾客往往是去了女性商家那里。你看这种性别动态非常复杂，不仅涉及消费者端的性别歧视，更体现在女性内部的歧视，而男性之间则存在某种形式的偏袒。结果就是我们看到的所谓'性别不当行为差距'——事实上女性商家更可能实施这种不当行为。

Beside that, female customers who were victimized more if they had visited a female vendor compared to if they had visited a male vendor. You see this very complicated gender dynamics, not only about discrimination by gender on the consumer side, but this within gender discrimination more so on females, but also within gender favoritism, some version of that for male to male matches. The result is that we see a lot of what I call gender misconduct gap. The fact that female vendors were more likely to commit this misconduct.

这完全颠覆了我之前的认知。

This goes so much against my prior belief.

没错。市场上大多数人都会告诉你，我们认为男性商贩确实会比女性商贩更容易出现不当行为。但我们发现了截然不同的结果——存在巨大的性别不当行为差异。存在这种性别歧视现象，女性顾客更容易被欺骗。你可以看到针对女性的同性别歧视证据，同时也有针对男性的同性别偏袒现象。

Yeah. Most of the people in the market are gonna tell you we think male vendors are literally gonna commit more misconduct than female vendors. We find very different result that there was this massive gender misconduct gap. There was this gender discrimination where female customers were more likely to be cheated. And you see this evidence of within gender discrimination for females, but also within gender favoritism for males.

弗朗西斯，你能解释下为什么会这样吗？

Can you explain why this is happening, Francis?

有三种假设或潜在解释。第一种更多涉及市场认知偏差而非误解。实际上，正如我之前解释的，当你询问市场时，所有人都认为男性商贩会比女性商贩更不守规矩。这是市场的传统认知。想象你是那个女性商贩，你知道市场对你的真实行为存在错误判断，你会怎么做？

Three hypothesis or three potential explanations. One is more about beliefs than misbeliefs in the marketplace. In fact, when you ask the market, as I explained before, everyone thinks that male vendors are gonna be non compliant compared to female vendors. This is the conventional wisdom in the market. Imagine that you are that female vendor, and you know the market is poorly calibrated about your true behavior, what will we do?

你会变本加厉。对吧？是的。我们有证据非常吻合这个解释：市场低估了女性商贩实际不当行为水平。例如，你可能会看到女性不当行为上升。

You do it more. Right? Yes. We have some evidence that is very consistent with the fact that beliefs about, with the fact that the market under perceive the true level of misconduct by female vendors is one explanation for Wow. For example, you may see the female misconduct up.

确实如此。

Of course.

对。通过更广泛的性别歧视针对女性。这些结果尤其出现在女性赋权程度低的村庄。我们采用标准的女权衡量指标，结果明确显示：如果你是女性顾客，生活在女性没有话语权的村庄，男女之间存在严重社会隔阂，那么这些女性消费者就更容易成为受害者。

Yeah. Through wider discrimination by gender against females. The results are mostly particularly driven by villages where there's low female empowerment. Are standard measures of measuring female empowerment, and we see exactly that. The fact that if you're a female customer, you live in a village where females are not empowered, there's a lot of social distance between males and females, then those female consumers are likely to be victimized.

第三个故事实际上与收入中的性别差异有关。我们发现这更是性别不当行为差距的关键驱动因素之一，即如果你是女性商贩，或者你生活在一个收入较低的家庭，由于女性收入本就偏低，那么你对这种通过不当行为产生的额外收入的依赖程度会更高。因此，女性商贩更可能依赖不当行为带来的额外收入。在乡村地区这种情况尤为明显，那里的性别收入差距巨大。我本应提到，市场上几乎所有人都认为男性顾客比女性顾客更精明。

The third story actually relates to gender differences in income. And we see this more so as one of the key drivers for why, again, we have that gender misconduct gap, which is the idea that if you're a female vendor, and because to begin with incomes are low for females, or you live in household where incomes are low, then your dependency on this extra income that gets generated with misconduct is much more so. So female vendors are more likely to be more dependent on the extra income from misconduct. And this is more so in villages, where there's massive differences by income between gender. So I should have mentioned that almost everyone in the market believes that male customers are more sophisticated than female customers.

没错。这也意味着，你自然更想对女性顾客实施侵害，因为她们不够精明。因此，这种观念结构影响重大，我们在后续与市场人员深入交流的调研中清楚地看到了这一点。好，这是第一点。

Right. Well, that also implies that of course you wanna do more victimization on female customers because they're less sophisticated. So the belief structure matter a lot, and we see this when we do follow-up surveys really engaging with the people in the market. Okay. That's number one.

第二部分是关于性别赋权差异，第三点则是收入差异的存在实际上也解释了为何女性商贩——因为她们更依赖这笔额外收入。这里有很多结论，对吧？你可以想象，社会距离的概念，无论是通过观念结构、赋权结构，甚至收入差异，都可能成为市场中不当行为的重要源头，这种歧视性方式我们认为会破坏金融市场的正常运作。

Second part is about differences in empowerment, k, by gender, and then the third one being the fact that there are these income differences that actually also rationalizes why female vendors, because they are more dependent on this extra income. There are lot of conclusions here, right? You could imagine that the notion of social distance, either through beliefs, structure, through empowerment structure, or through even income, could act as an important source for misconduct in the market in this discriminatory way that we think can undermine the functioning of financial markets.

弗朗西斯，我想论文中可能还有很多内容我们今天没时间讨论，但非常感谢你向我描述这些发现。

Francis, I think there's probably quite a lot more in the papers that we haven't had time to talk about today, but thank you very much for describing them to me.

当然。也谢谢你的提问。移动货币的市场结构非常有趣。它是一个研究贫困群体金融服务市场问题的绝佳经济学平台。

Absolutely. And thank you too for asking. Mobile money, the market structure is super interesting. It's a platform to do really good economics, studying questions within the marketplace when it comes to financial services for the poor.

我们讨论的三篇论文——《不当行为与性别》《移动货币实地实验》，本期Vox Dev谈话就到这里。确保不错过每期节目的最佳方式是关注或订阅我们。你可以在任何播客平台找到我们。往期节目一如既往地可在voxdev.org上获取，那里有大量与此主题相关的文章。

Those three papers we were discussing, misconduct and Gender and A Field Experiment on Mobile Money. This has been a Vox Dev Talk. The best way to make sure you don't miss an episode is to follow us or subscribe. You find us wherever you get your podcasts. Our past episodes, as always, are at voxdev.org, where you will find plenty of articles about topics like this one.

如果你喜欢我们的内容，请向他人推荐并为我们留下评价。VOXDEF谈话由Talk Normal制作，助理制片人是梅根·比伯，编辑安德烈·扎尔加里安。感谢VOXDEF的奥利弗·哈尼和伊曼·西迪克。

If you like what you're hearing, then please tell someone else about us, and also leave us a review. VOXDEF Talks is a Talk Normal production. The assistant producer is Megan Bieber, and our editor is Andre Zargarian. Thanks to Oliver Hany and Iman Siddik at VOXDEF.