Rethink Priorities has been piloting expanding into human-focused neartermist global priorities research. This post is one of three outputs from the pilot program. Open Philanthropy provided funding for this project and we use their general frameworks for evaluating cause areas, but they do not necessarily endorse its conclusions. We don’t intend this report to be Rethink Priorities’ final word on charter cities. We hope the report galvanizes a productive conversation about charter cities within the EA community. We are open to revising our views as more information is uncovered.

If you are interested in doing similar research, please apply to Rethink Priorities’ Global Health and Development Staff Researcher position (deadline 13th June 2021).

Key Takeaways

• The value of charter cities can be divided into three main buckets: (1) direct benefits from providing an engine of growth that increase the incomes and wellbeing of people living in and around the city, (2) domestic indirect benefits from scaling up successful charter city policies across the host country, and (3) global indirect benefits from providing a laboratory to experiment with new policies, regulations, and governance structures.
• We think it is unlikely that charter cities will be more cost-effective than GiveWell top charities in terms of directly improving wellbeing.
  • The nearest currently existing empirical analogue to charter cities are special economic zones (SEZs). These do not seem to grow faster on average than their host countries and if there are spillovers they are limited to immediately adjacent areas.
  • Although some SEZs do outperform their host countries, there do not seem to be policy and regulatory features of these zones that correlate with high levels of growth, suggesting that our ability to design high performance SEZs, and by extension charter cities, is limited.
  • The cities that charter city advocates typically point to in support of charter cities (e.g. Shenzhen, Singapore, Hong Kong and Dubai) also had other factors going for them such as proximity to trading partners and high skilled immigration that may be difficult to replicate in future charter cities.
  • Charter cities do not seem particularly tractable given the large number of actors who have to coordinate over a long time period to develop and manage a city. The initial failures in Madagascar and Honduras illustrate the difficulty of establishing genuine charter cities, though these attempts may not be representative of future efforts.
  • Our cost-effectiveness model suggests that over a 50-year time horizon charter cities would have a 127x return on investment (ROI) in expectation, a 5% chance of having greater than 1000x ROI and a 40-50% chance of having negative ROI. For reference, top GiveWell charities typically have an estimated ROI of around 1000x.
• Scaling up successful policies could substantially increase the ROI of charter cities, with the main datapoint in support of this being China liberalizing after Shenzhen demonstrated high growth rates. However, we are uncertain about how much this can be replicated. We would be very keen to see advocates make a more robust case for the returns here being sufficiently high.
  • We are very unsure what proportion of China’s economic liberalization can be attributed to Shenzhen versus other factors such as external pressures from countries that wanted to trade with China.
  • China was exceptionally far from market-oriented institutions before liberalization and has an exceptionally large population so we should expect future scale-ups to provide benefits to fewer people and provide fewer benefits per person since nearly all countries where charter cities could be founded are much closer to typical market institutions.
• The value of information from experimenting with new policies and governance structures in charter cities may be high, but we are very uncertain about this.
  • The advocates we have spoken to and read mostly emphasised the direct benefits and only briefly mentioned these indirect benefits. We could not find a comprehensive argument for the value of these indirect benefits being large.
  • Charter cities are expensive to build and take a long time to come to fruition, so if your focus was on the value of information, it seems likely that you could more cheaply and quickly generate this by focusing on alternatives which do not require you to build a new city, such as economic development and reform zones.
  • We think it would be valuable for advocates to make a robust case for the value of information benefits of charter cities being high, and in particular, higher than alternatives in the subnational governance space.

Executive Summary

We don’t believe that charter cities are a cost-effective method to lift individuals out of poverty directly. However, charter cities may provide indirect benefits (spillovers, replication, and experimentation) that are larger than their direct benefits, and the addition of indirect benefits may substantially improve the cost effectiveness of charter city advocacy. Unfortunately, these indirect benefits are hard to model, and hence it is difficult to estimate their magnitude. However, we believe that most of the value of the indirect benefits (whatever their size) can be captured by special reform zones that are cheaper and easier to implement than charter cities. Establishing a genuine charter city looks relatively intractable, and the gains are extremely uncertain. Therefore our overall view is that charter cities require more evidence in their favor before we could recommend them as a promising intervention to pursue.

Charter cities require a host nation that is willing to cede substantial authority to a new jurisdiction. Ideally, this country would be low- or middle-income with a growing urbanization rate and a relatively stable political climate. Because charter cities are meant to be entirely new developments, the host nation must also have a suitable greenfield site. This site ought to be sparsely populated, mostly undeveloped, with room to grow, and trade access (ideally access to the sea). Charter cities also require private investors to partner with the host nation to help finance the project, with total costs on the order of $500 million to $10 billion. We believe bringing these components together will be difficult.

The closest current analogue to charter cities are special economic zones (SEZs). Special economic zones have a surprisingly mixed track record: most zones grow at roughly the same rate as the national average, and a significant fraction of zones grow at a slower rate than the national average. Furthermore, zone growth appears difficult to sustain over time, and there is no clear correlation among zone policy and regulation features and zone growth. This suggests that we don’t really know how to design a special economic zone to optimize for growth. Because charter cities are, by definition, significantly more complicated than special economic zones, we are uncertain whether charter cities will be likely to adopt the sort of regulations that could, by themselves, promote robust, long-lasting growth.

More research on the indirect benefits of charter cities, including the extent to which these benefits can be captured by special reform zones, would be valuable. Although we don't think the current evidence supports the case for charter cities being as cost effective as GiveWell charities in expectation, there is a strong case that advocacy for charter cities has positive expected value and more empirical research on the potential indirect benefits could further increase the expected value so we are glad that there are smart, passionate people working on this topic.

Introduction to Charter Cities

Paul Romer developed the modern concept of charter cities^[1] with a small group of like-minded individuals in 2008. Romer launched the idea publicly in a 2009 TED Talk. Since then the concept has evolved and been refined, such that as the term is currently employed, it differs significantly from Romer’s original proposal. In this report we focus on the idea as it is promoted by today’s charter city advocates, which we take to be an improvement on Romer’s model.^[2]

What Is a Charter City?

Charter cities involve two components: a new city and new rules for that city.^[3] Charter cities are semi-autonomous units with significant legal jurisdiction over economic, social, and judicial institutions. While not wholly sovereign, charter cities are intended to be given a blank slate upon which to craft domestic policy as they see fit. As such, charter cities are expected to differ in many ways from the nations that host them. A charter city would likely feature different courts, different schools, different police, and different healthcare. A key aim is for charter cities to offer an improved business environment over their host nations, with “a legal and regulatory regime conducive to sustained economic growth” (Mason 2019: 11). Charter cities are meant to be places where starting a business, registering property, and obtaining construction permits is straightforward and affordable. Charter cities are also meant to feature reliable contract enforcement and impartial dispute resolution. In addition to better institutions, charter city advocates hope that charter cities would possess better physical infrastructure, with dependable electricity, water, and transportation.

Charter cities intend to bring together public and private sector interests in a mutually beneficial way. The process would work as follows. First, the host nation designates a sparsely populated and mostly undeveloped tract of land as a charter city, with the legal autonomy to set its own rules.^[4] Next, a private developer acquires the land and builds the physical infrastructure. The host government and private developer then jointly appoint a city council, with some seats reserved for the local community and/or third-party auditors. The council establishes the rules and regulations for the city. The developer then sublets the land to businesses and individuals. The private developer stands to gain when the value of the land rises and so its representatives on the council are incentivized to adopt rules conducive to economic growth. What follows is a virtuous cycle of new jobs and new investment. If all goes according to plan, the city prospers and the host nation begins to emulate the policies and regulations that led to the city’s growth. By this mechanism, charter city advocates hope to lift millions of people out of poverty.

Although there are as yet no true charter cities (although one might call Próspera in Honduras a charter town), some existing cities resemble the charter city model in important ways. Hong Kong, Singapore, Dubai, and Shenzhen are the most commonly cited examples. Shenzhen, for instance, was a relatively sleepy hamlet of 30,000 when the Chinese government designated a 300 square kilometer special economic zone around the town. The region’s new status enabled provincial officials to undertake significant economic liberalization. At the time, the average annual income per resident was a paltry $122. Over the next four decades, the city’s population grew to over 12 million and annual income per capita increased more than a hundredfold (World Bank, 2010). According to the standard charter cities story, Shenzhen’s success inspired Chinese officials to endorse economically liberal policies in other parts of the country, eventually lifting as many as 800 million individuals out of poverty. The charter cities movement hopes to replicate this success in other low-income countries.

Related Concepts

Charter cities are related to but distinct from a number of other special jurisdictions and urban governance innovations.

Special Economic Zones

Special economic zones (SEZs) come in many flavors and sizes, but generally their mandate extends across a relatively small geographic area with a relatively narrow regulatory remit, often centered on a single industry.^[5] According to the World Bank, “When compared to the domestic economic environment in which they are placed, SEZs can be considered ‘special’ in several ways. They provide infrastructure (access, quality, reliability, cost, flexibility); customs regimes (efficient customs, duty, or value-added tax (VAT), free or deferred); regulatory regimes (efficient licensing, planning, flexibility), and fiscal regimes (capital freedoms, tax incentives, subsidies)” (World Bank, 2017: 12). Unlike charter cities, SEZs generally do not possess the sort of political autonomy that, for instance, would allow the establishment of an independent judiciary. Depending on how one defines them,^[6] there are hundreds to thousands of SEZs across more than a hundred countries (UNCTAD, 2019: xiii).^[7]

Sustainable Development Zones

Sustainable development zones (SDZs) are “new urban communities with special legal and administrative institutions that further sustainable development in line with the UN Agenda 2030” (SDZ Alliance, n.d.). SDZs are a new model for urban governance, sometimes referred to as ‘SEZ+’. SDZs combine the pro-business, economically liberal policies of SEZs with a limited form of political autonomy. In contrast to charter cities, however, SDZs are meant to be zoned on top of existing developments. As such, they don’t involve extensive new construction. SDZ advocates argue that they could be used in refugee camps and other informal settlements to formalize land ownership in a way that jumpstarts economic growth. There are no SDZs with a bottom up inclusive governance approach as yet, but the SDZ Alliance has received very strong Ethiopian government endorsement and support for this concept which addresses uncontrolled urbanization and informality, leading to bankability, micro, small and medium enterprise development and creates jobs for the urban poor, internally displaced peoples and refugees

Economic Development and Reform Zones

‘Economic development and reform zone’ is an umbrella term for special jurisdictions that don’t naturally fall under the scope of SEZs or SDZs but are intended to promote economic growth through better institutions. These zones are typically more ambitious than SEZs but less ambitious than charter cities. For example, Michael Castle-Miller is working with the government of Vanuatu to develop special jurisdictions that would be managed by private companies with authority to reform select laws. As with the charter city model, these companies would profit by subletting land within their jurisdiction, giving them incentive to improve the value of the land under their control. Unlike charter cities, which are given tabula rasa to draft laws from scratch, normal Vanuatu laws would apply by default in these zones. Managing companies would then be granted the authority to adjust certain laws, and, importantly, the appropriate reforms would be determined by the company rather than the central government.

Seasteading

Seasteadings are similar to charter cities in that they involve creating new settlements with a new set of rules, but they differ in that the settlements are created at sea outside of the jurisdiction of any existing nation state, rather than on land with permission from a host country. The Seasteading Institute is the main organisation associated with seasteading. No settlements-at-sea have been successfully created yet and previous attempts near French Polynesia and Thailand have failed after governments opposed them, but Metaculus forecasters believe there is a 20% chance of a successful seasteading venture of at least 100 people by 2035.

Company towns

A company town is a settled area in which a single company controls much of the land and may provide many of the town’s basic services and infrastructure. The company usually employs much of the town’s populace. In conversation, Gyude Moore described charter cities as a cross between company towns and SEZs. Mark Lutter of the Charter Cities Institute has also explored this analogy. In some regions, company towns are empowered to provide independent security, healthcare, and schools. Although company towns have a checkered history, Gyude pointed to the modern success of Harbel, Liberia run by the Firestone Natural Rubber Company which provides evidence that privately run cities can outperform their public neighbors. Unfortunately, we could not find any studies that systematically examined the performance of company towns relative to the national average of their host nations, but there is a short defense of them by Alex Tabarrok.

Three Ways to Think about the Value of Charter Cities

There are (at least) three non-exclusive ways by which one might conceptualize the value of charter cities. The first is by the economic growth premium that charter cities might enjoy over their host nations and the effect that this growth has on the charter city residents. (Ideally, this framing would also include spillover effects on neighboring areas outside the charter city.) The second is by serving as a model jurisdiction that the host nation can replicate and scale up. The third is by the value of information that charter cities might produce through experimenting with innovative and unprecedented policies and governance models that haven't been implemented anywhere else in the world. In our conversations with them and in their published writing, advocates have mostly emphasized the direct benefits of charter cities (i.e., the first type of value). The first type of value is far easier to model, but it’s plausible the majority of the expected value of charter cities flows from the second or third type of value.^[8]

Charter Cities as Growth Engines

The conventional approach to charter cities values them as drivers of economic growth. Charter city advocates typically believe that differences in institutions can account for the bulk of the difference between high-performing and low-performing regions (Mason, 2019: 5-12). What counts as an ‘institution’ is not always clear, but, for example, the Charter Cities Institute often cites the Ease of Doing Business Index as “a reasonable measure” of the sorts of rules and regulations that govern growth.^[9] The goal of charter cities is to import the best institutions into low-performing regions, thereby jumpstarting economic activity and investment. Charter city advocates claim that new, well-governed cities will attract many migrants who would otherwise face bleak prospects.^[10] Advocates believe that founding charter cities is a cost-effective way to alleviate poverty. This source of value is the focus of The Case for Charter Cities Within the Effective Altruist Framework (Mason, 2019).

Charter Cities as Models to Emulate

When a charter city succeeds, its success may inspire otherwise intractable reforms to be adopted across the charter city’s host nation.^[11] In this way, the direct benefits that a charter city confers on its own residents may be dwarfed by the indirect benefits that the city confers on the rest of the nation by functioning as a proof-of-concept that is emulated elsewhere. The Chinese ‘growth miracle’ appears to fit this model. Shenzhen’s remarkable economic growth served as a demonstration of the power of liberal reform and bolstered support for reform among relevant decision-makers in China.

While we’re sympathetic to this conception of charter city value, it’s unclear whether charter cities can achieve anything like the Chinese growth miracle in other parts of the world. There may not be any countries or regions that share enough of the relevant, unique characteristics that defined China’s economic boom. In particular, it may be difficult to replicate Shenzhen-level persuasion even with Shenzhen-like results. That is, there may be diminishing returns on additional demonstrations of the power of economic liberalization to promote growth. If a country continues to push economically illiberal policies, despite ample available evidence that such policies constrain growth, it’s unclear how much one more example of a well-governed city will do to convince the government to change. (It’s not as if the economic success of South Korea has convinced North Korea to liberalize.) Leaders that want to found charter cities on their territory want to do so precisely because they expect the economically liberal policies adopted in the city will promote the city’s growth. But if a leader already believes in the power of economic liberalization, it’s hard to see what counterfactual impact the city will have in convincing the leader to pursue economic liberalization. Even in countries that develop charter cities, broader reform may be off the agenda, either because it is deeply intractable or deeply unwanted.^[12]

If charter cities are going to serve as models to emulate, they will probably produce the most value by exploring new, untested and unprecedented regulatory and political environments, which brings us to the third way to think about the value of charter cities.

Charter Cities as Laboratories of Governance

Apart from their potential to generate economic growth directly, charter cities may prove valuable as locations where officials are free to experiment with radically new policies and governance models.^[13] Even if most of these experiments fail, if charter cities accelerate the discovery of a handful of important generalizable techniques for improved governance and policy-making that are then replicated around the world, they might look attractive from a hits-based giving perspective.

There are, however, several concerns with this conception of the value of charter cities.

First, by neartermist standards, the feedback loop from charter cities is long. It’s unclear when (or if) advocates will be able to bring together the right mix of stakeholders to launch a charter city, but even if all the interested parties began negotiating today, the planning process would span years. Once the relevant details were hammered out, the construction phase could easily take a decade or more. Then, to fairly evaluate the results of experimentation, we would want the city to run at least a full decade before we assess its performance.^[14] In contrast, the feedback loop from reform zones is much shorter, while still offering many of the same benefits from experimentation.

Second, not only is the feedback loop long, it is also expensive. Charter city advocates want to construct brand new cities, which would naturally cost a lot of money. Although only a fraction of that money would come from altruistic sources, the high cost (and other unique requirements) of charter cities will likely limit their number for the foreseeable future. With a small and heterogeneous sample, it will be difficult to extract robust lessons from experimentation. If we expect reality to be underpowered, the case for charter cities as laboratories of governance appears diminished.

Third, the laboratories of governance model means there are additional risks involved. The risk level of charter cities falls on a spectrum, but many charter city projects are likely already riskier than traditional real estate investments. Further focusing on the value that comes from experimentation would even further raise the risk, so it may be the case that there aren’t a sufficient number of investors with a large enough taste for risk. There may have to be some balance between sticking to best practices versus novel approaches to ensure the coalition of public and private sectors remain on board. This likely limits the potential sample size to learn from.^[15]

Finally, the laboratories of governance model may add to the neocolonialist critique of charter cities. Charter cities are not only risky, they are also controversial. Charter cities are likely to be financed by rich-country investors but built in low-income countries. If rich developers enforce radically different policies in their charter cities, that opens up the charge that the rich world is using poor communities to experiment with policies that citizens of the rich world would never allow in their own communities. Whether or not this criticism is justified, it would probably resonate with many socially-minded individuals, thereby reducing the appeal of charter cities.

History of the Charter City Movement

In 2008 Paul Romer made the first of several trips to Madagascar to promote charter cities as a solution to Malagasy^[16] poverty. By the end of the year, Romer had secured the blessing of Marc Ravalomanana, the President of Madagascar, to found not one but two charter cities in the nation. However, less than two months later, Ravalomanana was deposed in a coup. The new government denounced Romer’s project as treason^[17] and the idea was quickly shelved (Mallaby, 2010).

As the coup in Madagascar closed one door, a coup on the other side of the globe, in Honduras, opened another. In June 2009 President Manuel Zelaya was forcibly removed from office and eventually replaced by the more conservative Porfirio Lobo Sosa. By early 2011 Romer had convinced the new Honduran government to pass a constitutional amendment allowing for special development regions in which charter cities could be built on Honduran territory. The cities were to be managed by a government-appointed Transparency Commission, which Romer agreed to join (Kroth, 2014).^[18]

The idea became mired in controversy almost immediately. Residents in the proposed sites complained that officials had not consulted them. The government signed an investment deal without the consent of the Transparency Commission.^[19] The Supreme Court ruled the new zones unconstitutional, only to have the justices who voted against the zones summarily dismissed on dubious grounds. Shortly thereafter, Romer distanced himself from the project (Kroth, 2014).^[20] The project lurched forward without Romer, eventually taking the form of Próspera, which was scheduled to break ground in 2020.^[21] It’s not clear if the project in its current guise qualifies as a charter city due to its small size (58 acres as opposed to Charter City Institute’s minimum recommendation size of 10 square kilometers / 2,470 acres). You can read recent media coverage of Próspera in Bloomberg and Astral Codex Ten and see Metaculus forecasts of Próspera’s population in 2021 (currently 18% chance greater than 1000) and 2035 (currently median <100).

After the failures in Madagascar and Honduras, Romer seemingly lost interest in charter cities. Other figures have since taken up the mantle of charter city advocacy (see below). Criticism of the idea also persists. Sarah Moser, professor of cultural and urban geography at McGill University, is perhaps the most prominent critic. In a 2018 essay, she argues that:

aside from the many failed startup city projects, there is a wide range of negative impacts that deserve careful study. There is growing evidence that the collateral damage produced by many projects is severe and unacceptable to many: there are ecological disasters; many have demonstrated little respect for human rights, dignity, or lives; they are not accountable to any public, and they are sites of unprecedented corruption. There is a consistent disregard for legally protected tribal and indigenous land. They are certainly zones of experimentation with new forms of governance and unique levels of autonomy, but at what cost and to whom?” (Moser, 2018).

We haven’t vetted Moser’s assertions but note that some of her criticism is aimed at projects that may not fall under the charter city umbrella. More importantly, Moser also denies that China’s economic success supports the claim that charter cities are a good way to achieve economic growth, labelling it “cherry picking of the highest order.”^[22] We examine this claim in greater detail later in the report, when, in the section Identifying Suitable Rules, we look at a more representative sample of SEZ performance, rather than just focusing on the most successful SEZs.

Advocacy Groups

Charter city advocacy is an extremely small space. We’re 90% certain that total annual spending is less than $2 million. The main organization promoting charter cities is the Charter Cities Institute. Other relevant organizations include the SDZ Alliance, Greenfield Cities, Startup Societies Foundation, and NewCities. However, these organizations are generally smaller than the Charter Cities Institute and/or focus on related but distinct ideas.

Charter Cities Institute

Mark Lutter founded the Charter Cities Institute (CCI) in 2017.^[23] The organization began with the aim to coordinate the diverse stakeholders who might be interested in charter cities. Since its inception, the organization’s mandate has expanded beyond coordination to include general advocacy for charter cities, research on governance and institutions, and technical assistance to new city projects. The organization runs a blog and a podcast, hosts conferences and meetups, holds essay contests, maintains reference guides, writes white papers, engages in direct outreach, and consults on projects. In 2020 CCI grew from a team of three to a team of eight. In conversation, Mark Lutter reported that with more funding, CCI would hire a senior researcher with ties to the international development community.^[24] CCI received 500 ETH from Vitalk Buterin on 12th May 2021 (worth $1.9 million on that day, $1. 2 million on 12th June 2021)

SDZ Alliance

The SDZ Alliance, a US registered non-profit organization, is a joint venture among individuals and groups promoting sustainable development zones. Joachim Rücker, Michael Castle-Miller and Kilian Kleinschmidt are the main advocates involved in the organization. The Alliance offers technical assistance and networking to countries and municipalities interested in tackling informality and poverty through SDZs as an instrument. The organization is currently donor- and project-driven causing considerable delays in moving forward. The Alliance would like to have the discretion to develop their own projects up to investment stage, but they are constrained by a lack of funding.

Greenfield Cities

Greenfield Cities is a small Dutch nonprofit that aims to stem forced migration^[25] by building new, sustainable, and prosperous cities in regions with many international migrants, especially Middle Eastern and North African countries.^[26] These cities would not have the political autonomy of charter cities, and in practice it seems Greenfield Cities isn’t promoting cities at all. Instead, they advocate for “self-supporting ‘pop-up’ campuses to kick-start sustainable urban development.” These campuses only appear aimed at hosting a few thousand residents at most. Greenfield Cities is currently developing its first campus in Mafraq, Jordan, on the border with Syria and only ten kilometers from the massive Za’atari refugee camp. Greenfield Cities is funded in part by the Dutch Ministry of Foreign Affairs.

Startup Societies Foundation

Startup Societies Foundation is a small, U.S.-based organization that aims to cultivate an international community of individuals interested in special economic zones, eco villages, micro states, intentional communities, seasteds, and smart cities. (Their website doesn’t explicitly highlight charter cities, but charter cities appear to be a natural fit.) Startup Societies runs a blog, hosts conferences, and has published a book. The foundation is described as “the mother organization” of the Institute for Competitive Governance, but it’s not clear how the two groups differ in practice.

NewCities

NewCities is a Canada-based nonprofit that researches issues relevant to urban development and advocates for sustainable urban growth. The organization hosts an editorial platform, publishes a blog, runs a podcast, and coordinates events. Charter cities are not a major focus, though Mark Lutter has written for their editorial platform.

Background Assumptions that Motivate the Case for Charter Cities

There are a few core assumptions that charter city advocates tend to share and these seem central to their support of charter cities. To even get the case for charter cities being one of the most cost-effective interventions off the ground, we think you have to agree with the broad thrust of these assumptions. Here we summarise these assumptions as they appear in CCI’s The Case for Charter Cities Within the Effective Altruist Framework (Mason, 2019) and provide a very shallow overview of the evidence supporting and opposing them.

Economic growth is the best way to alleviate poverty

Charter city advocates typically believe that economic growth is substantially better than alternatives in its ability to tackle poverty. Mason (2019: 4) writes that “long-run economic growth is unrivaled in its power to alleviate poverty”. He points to Pritchett (2018a) as the academic article supporting this claim and the rapid growth of India and China and the consequences this had for the proportion of the world’s population living in extreme poverty as examples. Pritchett argues that the income gains of the Graduation approach^[27] are 40 times smaller than income gains from migration to a high productivity country like the USA and further dwarfed by large growth accelerations such as those experienced by China and India, and as such, we should focus on growth or migration rather than typical GiveWell-style interventions.^[28]

Pritchett’s broader body of work is also the foundation for many of the claims in the popular EA Forum post, Growth and the case against randomista development, by Jonathan Halstead and Hauke Hillebrandt. We won’t summarise this post, but suffice to say, it was well received amongst the EA community and the community now seems to think that it is plausible that focusing on economic growth is an effective approach. However, not much practical work on interventions to increase growth seems to have been done. The absence of direct EA work to improve growth is partially explained by the practical difficulties surrounding interventions aiming to boost growth. Stephen Clare and Aidan Goth (of Founders Pledge) summarize these difficulties in their post Can we drive development at scale? An interim update on economic growth work.

Our take is that economic growth is the most important factor in increasing incomes and bringing people out of poverty. However, we sympathize with the view that improving growth has low tractability because it is unpredictable and very hard to influence, even for large government actors. As such, our prior (or outside view) is that in general, interventions aimed at increasing growth have low tractability and a strong inside view case needs to be made for specific interventions such as charter cities being tractable.

CCI’s view is that we are “substantially underestimating the tractability of charter cities. For example, CCI is drafting legislation for a country that elected a president sympathetic to charter cities and has been giving strategic advice to a stealth charter city development that is negotiating substantial legal authority and concessions on hundreds of square kilometers. [They] anticipate similar opportunities arising in the coming years.” (Mark Lutter, personal communication)

Economic growth is mostly a result of good institutions

Perhaps the most critical claim is that institutions are the primary determinant of growth. Mason (2019: 5-13) writes that “Many have strongly argued that the principal cause of differences in economic growth both across and within countries are differences in institutions across and within countries” and then goes on to summarise these arguments. The academics and key articles commonly cited in support of this claim are: North (1991); Acemoglu, Johnson & Robinson (2004); Acemoglu & Robinson, Why Nations Fail (2012); Rodrik (2004); and Rodrik, Subramanian & Trebbi (2004). These articles by economists and political scientists use varying methodologies to make the claim that institutions are important for growth. In some cases they compare the relative importance of institutions against other factors such as geography, climate, culture and trade and find that institutions win the horse race. We haven’t looked into these papers specifically for this project, but are familiar with this broader literature and we note that these articles are by academics well-respected in the economics profession, the papers themselves are influential and highly cited, and they are often a core part of an undergraduate or graduate macro-development economics course. However, they often rely on coarse data, contested identification assumptions and relatively weak instrumental variables to make their causal claims, so interpreting them requires significant care, especially when they are crucial considerations for action. See Vollrath (2014)’s Skeptics Guide to Institutions (part 2, part 3, part 4) for an accessible critique of this literature.

We note that the concept of “institutions” is not typically well defined and some claims about charter cities and growth more broadly may hinge on the exact definition being used. Most definitions typically include a reference to “the rules of the game” and “humanly devised constraints that shape human interactions” North (1991). These definitions allow for many subclassification schemes, with formal (e.g. constitutions and laws) and informal (e.g. customs and traditions) being the most common. “Culture” is usually posited as another potential fundamental determinant of development and there is often overlap between culture and informal institutions. Furthermore, when the concept of institutions is operationalised for quantitative study, it often ends up being much narrower. For example, Acemoglu, Johnson & Robinson (2001), in their study of the links between settler mortality, institutions, and GDP, use a 0-10 measure of property rights (or protection against expropriation risk in their terminology) from Political Risk Services as a proxy for institutions more broadly. The operationalisation Mason (2019: 11-12) relies most heavily on is the World Bank’s Doing Business Index, which they recognise is not perfect but is a reasonable measure of a country’s business environment.

We think charter city advocates overstate the consensus of economists on the primary importance of institutions for growth. For example, Mason (2019: 6) references the work of Rafael La Porta, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert Vishny (LLSV) on legal origin, saying that “Legal systems are a key piece of the institutional story”. However, LLSV and co-authors are typically framed as arguing against the importance of institutions, not in support of them. In Glaeser et al. (2004, 271) they write that “human capital is a more basic source of growth than are the institutions”. In Gennaioli et al. (2012: 108) they argue that human capital is the key driver of differences in development within regions of a country, but for measures of institutional quality, “their ability to explain cross-regional differences is minimal.”^[29] Furthermore, our impression is that there is not a consensus amongst economists on the claim that institutions are the most important factor for growth and we have not been able to find any polling of economists on this topic. That said, we think there probably is consensus that institutions are important for growth, if not the most important factor.

As such, our broad take is that institutions do play a substantial role in economic growth. However, we take charter city advocates to be claiming something like formal and informal institutions are 80% of what causes growth. We favour a more pluralistic view and would assign institutions a weakly held 30-50% of the credit.

Direct reforms to institutions at a country level are not tractable

If you accept the previous two premises, then you might think that we should be aiming at reforming national institutions to effect growth, since these would increase your impact by affecting more people than subnational institutions. One main reason advocates choose to focus on cities instead is due to tractability concerns. This is typically framed in terms of special interest groups who profit from existing institutional arrangements and thus lobby against and resist changes that are beneficial from a social welfare perspective but harmful to those groups. Special interests groups being smaller and more coordinated than society at large means they can more effectively resist institutional or policy change that harms them. Charter cities avoid these concerns by being set-up on greenfield sites and starting from a blank slate or charter. Both of these features mean that in theory there are no existing interest groups to oppose new institutions.

An alternate response we have heard is that charter cities and similar concepts actually are a tractable way to indirectly reform national institutions. The thought here is that if the charter city can experiment with a new set of institutions and show that they work well, then national governments will be more willing to implement those institutions on a broader scale. The example most commonly raised in support of this point is the special economic zone in Shenzhen, China, described in more detail in our Charter Cities as Models to Emulate section above.

However, there are a number of ways to reform subnational institutions as an alternative to charter cities, discussed above in Related Concepts. Ideally, charter cities should be at least as tractable as these alternatives such as reform zones. Given that charter cities require the expense, time and logistics of building a new city on top of a greenfield site, which alternatives typically do not require, we are skeptical that charter cities would be more tractable than the alternatives. To think that charter cities were more tractable than the subnational alternatives, you would have to think that the blank slate feature of charter cities makes it much more likely for subnational reforms to happen, e.g. as charter cities are X times more expensive than reform zones, they have to be X times more likely to achieve the reform to be more tractable.

We agree that reforming institutions at subnational levels is likely to be more tractable than reforming institutions at national levels. However, we’re very unsure on how much more tractable it is and whether it is worth the tradeoff in terms of impact. Furthermore, it’s not clear to us that reforming subnational institutions by setting up an entirely new city is more tractable or cost-effective than reforming them by some other means such as a special economic zone (or one of the many other options) placed on top of an existing city or region.

Charter Cities: Path to Impact

For charter city advocacy to be successful, it must increase the probability that a new city will be built with significantly different (hopefully better^[30]) governance than the host nation. Below we outline this path to impact.

Advocacy Makes a Counterfactual Impact

Charter city advocacy can succeed in two (non-exclusive) ways: increasing interest and improving implementation. Because charter city advocacy can take so many different forms,^[31] it is difficult to put a price tag on advocacy in general. It is also difficult to estimate the probability that different forms of advocacy will succeed.

Increasing Interest

Charter cities require the cooperation of many stakeholders in order to be established. Most obviously, a charter city requires a host nation willing to cede substantial authority to the city. Charter city advocacy can succeed by identifying governments willing to cede such authority or by helping convince governments to cede such authority.

Charter cities also require investors and developers to finance the construction of a new city. Charter city advocacy can succeed by identifying such investors and developers and connecting them with relevant governments. Where such investors and developers do not already exist, charter city advocacy can succeed by helping to convince investors and developers to partner with relevant governments.

Finally, charter city advocates often point to the benefit of partnering with a credible international non-governmental organization or intergovernmental organization, such as UN-Habitat. The NGO or IGO can serve as a third-party auditor and mediator among the host nation, charter city, and private developers. Charter city advocacy can succeed by identifying relevant NGOs or IGOs and connecting them to relevant projects, or, again, by generating interest among NGOs and IGOs.

Improving Implementation

In addition to promoting the idea of charter cities, advocates can achieve counterfactual impact by facilitating cooperation and providing technical assistance that improves the odds that a charter city, once established, will succeed.

Charter city advocates might influence the selection of a charter city site, advising on both the location and initial size of the site. Charter city advocates might also influence the rules that govern the charter cities. This influence might affect the original legislation that grants the city its autonomy and/or the laws are first adopted in the city. CCI has developed a template upon which countries could model the legislation authorizing charter cities, as well as an example city charter itself. We did not have time to examine these documents in detail.

A New City Is Established

For a charter city to succeed, land must be acquired on a suitable site in a willing host country. This process requires multiple components to come together in the right way.

Host Nation

First, a suitable host country must be identified. The charter city advocates with whom we spoke focus on establishing charter cities in low- or lower-middle-income countries with growing urban populations.^[32] Such countries are likely to benefit more from a new, well-governed, fast-growing city than rich countries with stable urban populations. However, if one believed the primary value of charter cities is the information they would generate from experimenting with new rules, there appears to be less reason to focus on developing countries.

The host nation must be willing to cede substantial authority to the charter city. Crucially, the desire to cede authority must be relatively stable across administrations. Charter city negotiation and development would be a long process that could span the tenure of multiple officials. One way for a charter city to fail is for the initial enthusiasts in government to leave office before the city is complete and for their replacements to be much less keen on the idea. Rahul Sagar, a political scientist at NYU Abu Dhabi, observes, “Political elites, especially those who succeed the original visionaries, could easily lose interest in fostering a competitor whose success starts to undermine support for the rules that benefit them at home” (Sagar 2016: 527). Paul Romer experienced this failure mode in Madagascar after the coup in 2009.

The host nation must also be able to cede substantial authority. The constitutional framework of a nation may not be able to accommodate a semi-independent city within its territory. As such, even if the desire to cede authority is stable across administrations, a nation’s courts could constrain the administration’s ability to cede authority. Paul Romer experienced this failure mode in Honduras in 2012, when the Supreme Court ruled that charter cities were an unconstitutional abrogation of Honduran sovereignty (though this decision was reversed two months later when the National Congress removed the four justices who had ruled against charter cities).

Finally, the host nation must refrain from confiscating charter city assets after the city has been established, and, again, this abstention must be stable across administrations. In this regard, charter cities face the same threat as any sort of substantial and immobile foreign investment. In conversation, charter city advocates maintained that the threat of confiscation could be kept acceptably low through current international law, which would, for example, give private developers the right to seize the nation’s overseas assets if charter city assets were confiscated. One could use megaprojects funded by foreign investment (e.g. oil extraction) as a way to construct a base rate for how often these sorts of major projects succeed, although we note that charter cities would typically require giving up more sovereignty than these projects.

In conversation, Gyude Moore (former Liberian Minister of Public Works and current Senior Policy Fellow at the Center for Global Development) emphasized how difficult it would be to get current leaders to cede substantial authority, at least in Africa. He thinks that absent some concrete model to emulate which had spectacular success elsewhere, the only other thing that would get leaders to cede substantial authority would be a major crisis such as a complete loss of natural resource rents for countries dependent on commodity exports. For example, if Nigeria's oil revenue dried up or if climate change started having really dramatic effects. In conversation, several advocates noted that, since building charter cities requires governmental stability, charter cities might have more luck in autocratic regimes because autocrats face fewer constitutional constraints and tend to enjoy longer tenures in power.^[33] Hence, founding charter cities could require close cooperation with unsavory dictators. If correct, then that fact probably raises the risk that support for charter cities becomes a public relations liability.

City Site

Once a host nation has agreed to cede enough authority to found a charter city, a suitable location must be found for the new city.^[34] One of the defining features of a city charter is that it is a genuinely new city. As such, charter cities must be built on greenfield sites: relatively undeveloped and sparsely populated.^[35] And because charter cities are meant to be genuine cities, the land area required may be considerable. As a lower bound, CCI suggests a minimum of 10 square kilometers to support a population of 50,000 residents (Mark Lutter, personal communication).^[36] However, one would probably want to acquire the purchasing rights to an area much larger—potentially over 100 square kilometers—for two reasons: 1) the 1000 acre estimate assumes a population density that may be unrealistically high for a new city^[37] and 2) the city will require room to grow.

Plausibly one of the most important determinants of a city’s economic potential is its access to trade. Charter cities are meant to feature world class business environments that attract multinational firms. But firms that rely on access to the global market for purchasing raw materials and selling finished goods will be less likely to set up shop in a charter city that doesn’t have the infrastructure and geography conducive to international trade. It’s no coincidence that many of the successful cities that advocates hope to emulate—Hong Kong, Singapore, Shenzhen—are all situated on the coast. And we know that poor locations sometimes constrain special economic zones. A 2019 United Nations Conference on Trade and Development report on SEZs identified poor locations as one of the main reasons SEZs underperform: “The success of individual SEZs depends on getting the basics right. Most failures can be traced back to problems such as poor site locations that require heavy capital expenditures or that are far away from infrastructure hubs or cities with sufficient pools of labour” (UNCTAD, 2019: xv).

Given the somewhat unique requirements for a charter city, it’s natural to wonder how many suitable sites exist. In a 2019 conversation with Tyler Cowen, urban planner Alain Bertaud said:

Cities need a good location. This is a debate I had with Paul Romer when he was interested in charter cities. He had decided that he could create 50 charter cities around the world. And my reaction — maybe I’m wrong — but my reaction is that there are not 50 very good locations for cities around the world. There are not many left. Maybe with Belt and Road, maybe the opening of Central Asia. Maybe the opening of the ocean route on the northern, following the pole, will create the potential for new cities. But cities like Singapore, Malacca, Mumbai are there for a good reason. And I don’t think there’s that many very good locations.^[38]

The standard response from charter city advocates is to note that only 3-4% of earth’s land surface area is currently urbanized,^[39] and that brand new cities are being built all the time. To know which side of this debate to trust, it would help to get a better sense of the percentage of earth’s surface area that is suitable for charter city development (but we haven't had time to do this).

Private Investors and Developers

In conversation, several experts emphasized the importance of finding the right private partners for charter city development. Once again, a unique combination of traits is needed. Although the private investors and developers will expect to turn a profit on the project, these partners should also be aligned with the broad altruistic goals of charter city development. Value alignment is important because the development of a charter city will be a long and complicated process involving negotiation among many stakeholders. This negotiation must occur in good faith, without the threat that private interests will exploit inevitable loopholes to maximize their returns. Environmental, social, and governance (ESG) investors appear to be natural targets for charter city development, though it remains to be seen if charter city critics pushing the neocolonialist critique will dissuade ESG investors from becoming involved.

In addition to value alignment, the group of private investors and developers ought to have deep pockets. Building an entirely new city is, of course, an expensive enterprise, and many low-income countries will rely on the private sector to foot a large portion of the bill. Deep-pocketed investors are also preferred because they can operate on a longer time horizon. Most large real estate projects are debt-financed, but, as charter city advocates admit, the return on charter city investment will typically pay off over a longer period of time than a typical real estate investment (Lutter, 2020).

This fact points to a third desirable characteristic for investors: appetite for risk. As the early failures in Madagascar and Honduras suggest, charter cities may be a risky venture. Even if the initial stages of charter city development go well, private investors will be expected to finance large-scale infrastructure projects with a breakeven point potentially measured in decades rather than years.^[40] Because charter city development in even its most basic form is novel and untested, investors may be less likely to compound their risk by investing in projects that plan to experiment with new rules. Private sector risk aversion thus probably limits the “laboratories of governance” model for charter city value, at least in the early decades of charter city development.

When we spoke to Scott Foushee, a partner at Catalyst Cities LLC, he emphasized the criticality of establishing a charter city’s “enabling environment” - the set of laws and regulations which allow a charter city to attract and protect the rights of investors and residents over the long term. His position was that this foundation is more important than a detailed development or master plan, which will necessarily evolve organically over time

New Rules Are Implemented

According to advocates, charter cities will outperform their host nations because the laws that govern them will be different and better. There are two components needed to achieve this goal: identifying a suitable set of rules and implementing those rules.

Identifying Suitable Rules

Charter city advocates point to the success of cities like Shenzhen as evidence that better institutions can deliver fast-paced economic growth. In 1980 the Chinese government designated the Shenzhen area a special economic zone, thereby devolving significant administrative and policymaking authority to provincial officials, with a mandate for economic liberalization.^[41] At the time, annual per capita income in the area averaged around $122; by 2008, that figure had jumped a hundredfold to $13,196 (World Bank, 2010: 58). Certainly, if the growth that Shenzhen has achieved over the last four decades could be reliably replicated in charter cities around the world, that would make charter city advocacy an attractive philanthropic intervention. On the other hand, if Shenzhen is an extreme outlier, then the attention it receives in the charter city movement may be misplaced (except as an example of one side of a heavy-tailed distribution).^[42] Hence, it’s helpful to look at a more representative sample of special economic zones.

Fortunately, a 2017 World Bank report, Special Economic Zones: An Operational Review of Their Impacts, does just this.^[43] The study constructs a dataset of special economic zones, their features, and nightlights data as a measure of economic activity.^[44] The study analyzes factors that drive SEZ performance and how SEZ performance drives growth in nearby areas. The headline result is that “Rather than catalyzing economic development, in the aggregate, most zones’ performance has resembled their national average” (World Bank 2017: 5). (See graph below.) We focus on the relative performance measure of ratio of SEZ performance to national growth partly because it seems to be the authors’ preferred measure, but mostly because this gives us some idea of what the counterfactual might have been, whereas a focus just on the absolute measure of SEZ growth doesn’t inform us about the counterfactual. In addition, the authors note that “Zone growth is difficult to sustain over time. Generally, the economic dynamism of the most successful zones happens in their early years and slows over time, leading to the zones’ economic performance becoming similar to that of their surrounding areas” (World Bank 2017: 5).^[45]

Furthermore, the authors conclude that “The majority of SEZ program features have had little bearing on zone performance” (World Bank 2017: 5).^[46] They also find that, “Experimentation with alternative measures of the quality of institutions at a national level, such as the Ease of Doing Business Rank, also deliver insignificant results (appendix E). In other words, the business environment seems to have limited sway over the performance of SEZs.” (World Bank 2017: 75). This suggests that it’s not obvious how to design special economic zone policies to optimize for growth, which in turn suggests that it will be difficult to design a charter city to optimize for growth. However, the report does note that “The performance of SEZs in emerging economies has been affected first and foremost by the zones’ country- and region-specific contexts” which suggests that location may be a good predictor of success, whereas the standard case for charter cities relies on policies and institutions being predictors of success, especially given that the best sites for cities are already occupied.

Notes: SEZ absolute performance is just the growth rate of the SEZ. Stable means total growth from 2007 to 2012 was between -5% and 5%, while shrinking is below -5% and growing is above 5%. SEZ performance relative to national growth is the ratio of SEZ growth to national growth. Equal means the ratio is between 0.9 and 1.1, while slower is below 0.9 and faster is above 1.1.

A couple limitations are worth noting. First, the study limits itself to studying zones between 0.5-10 square kilometers. (By way of comparison, the Shenzhen zone was about 300 square kilometers.) Larger zones probably resemble charter cities more than smaller zones, and CCI argues that a charter city requires a minimum of 10 square kilometers, and preferably 100 square kilometers. Within the 0.5-10 square kilometer range, the authors find that larger zones tend to perform better than smaller zones. This fact suggests that the report is more favorable to charter cities than it initially seems.

The second limitation is that the authors don’t actually show the full distribution of SEZ relative performance so we don’t know how fat the right tail is. They show that 65/346 SEZs outperform the national average and they provide a couple of summary statistics: the median SEZ/national ratio is 0.95 (so the median SEZ grows slower than the national average) and the standard deviation of performance is 0.22. We can make some limited inferences about the absolute performance distribution (i.e. GDP growth) since the authors have scatter plots of the absolute performance measures (see Figure 4.4 below). The distribution is not symmetrical around the median with the SEZs with the highest absolute growth rates (150% over 6 years, an annual growth rate of 17.2%) being 3x higher than the absolute^[47] value of those with the worst growth rates (-30% over 6 years, an annual growth rate of -5.8%). But we don’t know what this implies for the relative performance measure since we don’t know whether those at the top of the distribution are there because they are in countries with high national growth rates, which would mean that their relative performance isn’t as impressive.

Notes: SEZ growth is the total growth from 2007 to 2012.

The third limitation is that the econometric analysis is not great. Nightlights measures are a noisy proxy for growth so their precision is hampered. They don’t exploit any random variation in SEZ location, instead using a selection on observables approach to identification. Our first thought was that this would likely lead to an overestimate of the effect of SEZs on growth due to a positive selection story whereby SEZs are placed in locations which we would expect to do well anyway, perhaps as a way for politicians or bureaucrats to claim credit for the growth in that area. But a negative selection story whereby SEZs are placed in an attempt to improve performance in a struggling region thereby leading to underestimates also seems plausible.

When looking at the drivers of SEZ performance, they consider a set of features related to characteristics of the zone (operator, infrastructure & location) and regulations in the zone (fiscal and nonfiscal incentives & investment requirements). They choose which features to include in the model on the basis of whether they were significant in univariate analyses, which is generally a bad idea (Sun & Kay, 1996; Heinze & Dunkler, 2016). They have no exogenous variation in the zone features so causal identification is limited. They cluster their standard errors at the regional level (i.e. subnational), but since SEZ features are correlated within a country, they ought to cluster at the country level; this would likely further reduce the significance of their results.

Despite all these concerns, the paper’s descriptive work is specifically answering our question of what the distribution of SEZ performance looks like. There are many papers on case studies of particular, high performing SEZs, but we haven’t been able to find anything else which looks quantitatively at performance more generally, which is why we lean more heavily on this paper than we would necessarily like to.

Importantly, charter city advocates often appeal to the purported success of SEZs in making their case. For instance, in The Case for Charter Cities Within the Effective Altruist Framework, CCI writes, “The SEZ model has proven that new jurisdictions operating under a well-designed legal framework can successfully promote growth. Charter cities take this model further by focusing on a much broader set of deep reforms, while also building a new city that people are going to want to inhabit” (Mason, 2019: 13). Taken at face value, the World Bank report appears to undermine this claim.

Implementing Suitable Rules

The identification of growth premium-inducing rules means little if those rules aren’t implemented properly. A core part of the case for charter cities is that by building a new city on a greenfield site with a blank slate to adopt new policies, officials can escape the vested interests that constrain reform at the national level. This feat is easier said than done. The initial implementing board of a charter city will consist of a mix of representatives appointed by the central government, the private developer(s), and potentially members of the local community and/or members of a multinational NGO or IGO. These individuals will likely bring their own interests to bear on the rule selection process. Central government representatives may be loath to deviate much from current policies, lest the nation’s non-charter city residents come to suspect that their poverty is the product of poor governance. The private developer representatives will be motivated to maximize profits, and, while every effort will be made to align their profit motive with the city’s public benefit, such alignment is difficult to achieve consistently. NGO/IGO representatives may want to promote environmental or human rights protections that other representatives oppose. The local community, supposing they are given a seat at all, may have entirely different interests. All told, a delicate balance must be struck, and it’s unclear in advance how likely it is that such a balance can be struck. Where compromise is politically necessary, it’s unclear how far such a compromise will take the city away from the ideal set of rules.

Special economic zones again offer a cautionary tale. Because charter cities are complicated projects involving many stakeholders and myriad moving parts, it will be important to monitor the city’s performance and adjust policies accordingly. But such self-assessment is rare even in comparatively simpler special economic zones. A 2019 United Nations Conference on Trade and Development report notes:

In fact, the performance of many zones remains below expectations. SEZs are neither a precondition nor a guarantee for higher FDI [foreign direct investment] inflows or GVC [global value chain] participation. Where they lift economic growth, the stimulus tends to be temporary: after the build-up period, most zones grow at the same rate as the national economy. And too many zones operate as enclaves with limited impact beyond their confines. Only a few countries regularly assess the performance and economic impact of zones. Doing so is critical, because the turnaround of unsuccessful SEZs requires timely diagnosis, especially when there has been a significant level of public investment in zone development” (UNCTAD, 2019: xiv).

A final failure mode is expropriation. If a charter city is successful, the central government may feel compelled to renege on its commitments and reabsorb the city under national rule.^[48] This scenario seems especially likely if the politically liberal laws of the charter city allow its residents to agitate for greater freedoms for the rest of the country. The prosperity of the charter city offers some protection, but, as the world has witnessed in Hong Kong, a central government may be willing to sacrifice a bit of prosperity in the pursuit of absolute stability. The Charter Cities Institute has developed a Risk Mitigation Guide offering advice they claim would reduce the danger of expropriation, but we have not had time to vet the guide.

Cost Effectiveness

In this section we describe: CCI’s cost-effectiveness model, our alternate model, and further modelling extensions that could be made in modeling the direct benefits of charter cities. None of these models include any of the potential indirect benefits discussed above and so potentially miss out on the largest source of value from charter cities. It’s important to note that all of these models are simplistic in that they assume charter cities have some growth premium and project this forward to look at different growth paths over time. As such, these models are very fragile, with the key parameters of the charter city growth premium and population being based on limited information. We strongly encourage you not to take these models literally and just treat them as weak signals of what the value or ROI on charter cities might be.

CCI’s model

CCI’s model is set-up as follows (parameter values are in the table below):

• There’s a fixed population size in the charter city and the same size population in the host country. This is just to provide an appropriately sized counterfactual.
• There’s a growth rate of 4% in the host country
• Charter cities have a growth premium relative to the host country
• The growth premium is delayed for some time (further discussion of how this is modeled below)
• CCI is responsible for some percentage of the success of the charter city
• CCI incurs some costs

CCI key parameter assumptions

Notes: The key parameters as they appear in Mason, 2019 and CCI’s model

Under the various scenarios they calculate GDP per capita in the charter city and host country each year. They calculate the ratio of charter city GDP per capita to host country GDP per capita and then take the log of this. They then multiply this by 1.44, as this is “GiveWell’s value assigned to increasing log consumption by one unit for one person for one year” to create value units. Value units are then discounted at an annual 4.2% discount rate. They then divide the cumulative sum of discounted value units by the project cost and calculate the return per $100,000. They look at cumulative value units over various time horizons, in 5-year increments from 5 to 50 years. We focus on the 50 year numbers.

We were initially confused by how they modeled the growth delay. In our model, if the charter city receives funding, it starts achieving the growth premium 10 years earlier. In the CCI model, the value produced in year t is discounted by 1.042 to the power of t+ the number of years of growth delay, instead of just to the power of t. This gives approximately similar values.

CCI cost-effectiveness results

Their neutral assumptions give a cost-effectiveness similar to Deworm the World, while their pessimistic assumptions give a cost-effectiveness similar to GiveDirectly. Their optimistic assumptions on the other hand, give a cost-effectiveness estimate roughly 40x higher than Deworm the World.

Our model

We have two models. The first one is a simple adaption of the CCI model where we (1) model the intervention as causing the growth premium to kick in 10 years earlier (rather than doing additional discounting), (2) we calculate returns on investment using by valuing the doubling of income at any level at $60,000^[49], and (3) we otherwise use CCI’s pessimistic parameter estimates. If we don’t discount future income gains, then this suggests that the ROI on charter cities is 262x, while if we do discount them at the 4.2% rate used in the original model, the ROI is 97x. We include this just to help understand how the CCI calculations convert to our preferred ROI units. For reference, GiveWell charities typically have a 1000x return in this framework (where a DALY averted is valued at $50,000) so this is the bar we would want charter cities to clear.

Our second model is our preferred model. We implement this in causal.app to allow us to use distributions rather than point estimates. You can see our core assumptions graphed at the top of the page and their implications for incomes and ROI over different time horizons at the bottom of the page. If you wish to input your own assumptions, you can either edit the inputs as they appear on the screen, or press “Use this template” at the top which will allow you to make structural modifications to the model as well.

First, we describe the extensions we make to the model which make it more complex but we think better captures the key concerns when thinking about the cost-effectiveness of charter cities as a tool for increasing incomes. Note that we still do not include the potential value from indirect benefits outside the city, coming from experimenting, finding good policies and scaling these up to the country as a whole. We think this sort of value is particularly hard to formally model, so should have a separate model entirely devoted to it. Secondly, we use our own beliefs about the parameters of the model instead of CCI’s beliefs. Where possible we specify distributions of beliefs rather than point estimates. We are generally more pessimistic than even CCI’s set of pessimistic assumptions.

The general set-up of the model is the same. We estimate a growth path with a charter city, with a charter city delayed by 10 years and with no charter city. We focus on the comparison of incomes between a charter city versus a charter city delayed by 10 years (fast:slow) but we also include the comparison of a charter city and no charter city (fast:no) for reference. We calculate how many income doubling equivalents there are each year and convert this into an annual ROI valuing an income doubling at $60,000. We then discount future values (but also show the non-discounted values) and take the cumulative sum of these ROIs over time. This gives us an ROI over various timeframes, with the maximum time-frame being 50 years after the charter city is set-up, as in the CCI model. However, we change some features of the model and some of the parameter values as described below.^[50]

Firstly, instead of having a fixed population we allow the population to grow over time. We specify a distribution over the initial population and a distribution over the population growth rate and project these forward to get the population over 50 years. We assume a lognormal distribution for the initial population and a normal distribution for the population growth rate with a mean of 2%. Importantly, we implicitly assume that population growth is independent of the charter city growth premium. In reality these things are almost certainly positively correlated since people are more likely to migrate to fast growing cities. Adding a positive correlation to the model would reduce population in bad scenarios and increase it in good scenarios, thus increasing the ROI of charter cities.

We keep the same initial GDP per capita ($5,000) and counterfactual growth rate (4%) as in the CCI model. These parameters don’t matter for calculating the ROI since we look at income doublings relative to this counterfactual growth path.

We modify the charter city growth premium in two ways. Firstly, we are more pessimistic about the initial growth premium. We assume it is log normally distributed with a median at 0 and a long right tail (see CC growth premium figure below). This therefore allows for the possibility of charter cities being bad for growth, but recognises that there is potential for them to cause much higher growth rates. Our median value is lower than CCI’s pessimistic assumption for the growth premium but this reflects our reading of the World Bank (2017) report described above, which found the median effect of SEZs on growth was 0. We take this to be the best empirical evidence on growth rates of a close empirical analogue to charter cities, although we still think it is weak evidence. We allow for charter cities to have worse growth performance than their host country since this is shown to be a possibility for SEZs and also seems possible for charter cities too. We recognise that there are arguments that charter cities may have higher growth rates than SEZs in expectation both due their different institutional arrangement and the fact that they are planned to be larger and therefore potentially benefit more from agglomeration.

Secondly, we then also assume that this growth premium trends toward 0 over time. We assume an average 2% annual decline in the growth premium, so if the growth premium was 1% in year one, it would be 0.98% in year two, 0.961% in year 3 and so on. We are very uncertain about what value this parameter should take so we model it as a uniform distribution between 1% and 3%, but nonetheless think it is important enough to include in the model. This is also based on the World Bank (2017) report which found that older SEZs had lower relative growth rates.^[51] We attempted to estimate this parameter from the results of the report, but there wasn’t sufficient information to do so. This decline has a limited effect when the initial growth premium is close to 0, at the centre of our distribution, but its effect is more significant in the tails where initial growth rates are 3 to 5%.

We continue modelling the impact by considering a counterfactual where the charter city (and its associated growth premium) comes into existence 10 years later if they don’t get philanthropic funding. We include CCI’s ‘marginal contribution to success’ parameter as ‘funding credit for success’ but use a uniform distribution between 15-35%.

We don’t discount future value by a generic 4.2% as in the CCI model and in practice we use a lower discount rate. Instead we split up discounting future benefits into two parts. Firstly, we apply an annual 0.25% discount rate to account for existential risk and epistemic uncertainty. Secondly, we also account for the risk of city failure by discounting future benefits by an average of 1.5% annually.^[52] We specify a uniform distribution between 1 and 2% to account for uncertainty over the annual risk. This is meant to capture scenarios like the host country expropriating the city after a change in government or private developers pulling out with no one to replace them. This results in an average annual discount rate of 1.75%, as opposed to CCI’s 4.2%. We are also very uncertain over this parameter.

We then divide the value by the project costs to calculate the ROI. For costs, we also use a lognormal distribution, shown below. We assume higher costs than CCI do because we think it is plausible that CCI could be in operation for 20+ years without causing the development of a charter city. This only includes the costs for a non-profit advocating for a charter city and providing technical assistance in the set-up. It does not include the potentially hundreds of millions of dollars that would be spent by governments and private developers to build and manage the city.

The main output of this modeling process is the graph below. This gives us an estimate for the expected ROI over different time horizons in the scenario where a charter city is sped up by 10 years and future benefits are discounted at a 1.75% rate due to existential risk, epistemic uncertainty and risk of city failure.

We think there are two main take-aways from this modeling exercise. Firstly, under our assumptions, the direct income benefits of charter cities don’t reach the 1000x ROI bar in expectation, even with a time horizon of 50 years, where they reach an expected ROI of 127x. If we extend the model, it takes approximately 170 years before the expected ROI goes above 1000x. Secondly, even this basic model contains a lot of uncertainty about the potential ROI on charter cities. There is a 5% chance that the ROI is greater than 1000x. On the other hand, there is also roughly a 50% chance that the returns to charter cities are negative, rather than positive. We think these estimates should be taken very tentatively given the additional uncertainty outside the model, but they reflect our general pessimism of charter cities as a direct intervention to boost incomes.

Potential extensions

Extensions 1 & 2, which would increase the ROI of charter cities, come from CCI’s write-up. Incorporating the extensions from 5 onwards would likely reduce the ROI. Extension 5 itself is from Stephen Clare’s shallow investigation of charter cities for Founders Pledge. The remaining suggestions are our own.

1. Including spillover effects in the form of growth boosts to nearby areas that trade with the city
2. Including spillover effects from the country adopting similar reforms to those made in the city
3. Including the benefits from novel experimentation (e.g. legalizing reproductive cloning, or a good regulatory scheme for prediction markets)
4. Including non-pecuniary benefits as part of the value (e.g. increased freedom)
5. Model population growth and the charter city premium as positively correlated rather than independent to capture the notion that people are more likely to move to cities with higher economic growth
6. Including a risk of failing to successfully implement the charter city
7. Including opportunity costs of governments and private developers in setting up and managing the charter city (appropriately discounted to account for the lower effectiveness of their marginal use of the money relative to a philanthropist)
8. Better modelling of the decline in the growth premium over time, especially as cities approach the growth frontier and can no longer rely on catch-up growth
9. Account for a positive correlation between the growth premium and population growth, as people are more likely to move to a fast growing city

2 and 3 are arguably the largest additional source of value for charter cities but unfortunately they are also the ones that we have the least amount of information on and seem hardest to model.

Ultimately, we’re not convinced that extending this particular modelling exercise further is valuable. We focused on building a more complex model of the direct income growth benefits of charter cities, since this was the focus of Mason's (2019) write-up and model and by the lights of direct benefits, charter cities don’t seem exceptionally cost-effective to us. However, it’s entirely plausible that the majority of the benefits of charter cities come from exploring alternative policy and governance possibilities and scaling up successful ones. This was not the main focus of Mason (2019) and has not been the focus of charter city advocates in conversations we’ve had with them. We think calculating an ROI from these sorts of benefits is substantially harder and would require a totally separate model. We would be keen to see charter city advocates explicitly make the case for charter cities by these lights and with formal modelling rather than anecdotes.

A sketch for how one could model the benefits from 2 follows, proposed by Peter Favaloro of Open Philanthropy. We haven’t done this exercise but we think it would be worthwhile for charter city advocates to do something similar to this, perhaps in a different context, and quantify just how much value they think charter cities might produce along these lines.

1. Find a high quality estimate of China’s national income gains from the Deng reforms
2. Estimate what proportion of the reform can be attributed to Shenzhen. Include some concept of expected value e.g., “2 out of 3 times, China wouldn’t have captured all these benefits”
3. Pick a target country that would gain from a charter city or special economic zone today.
4. Estimate what could be gained in the country by analogy to the example in step 1 and 2. Adjust for the fact that China was exceptionally far from a market economy before the Deng reforms, has a large population, the likelihood of success and other factors.
5. Produce a statement like, for example, “there’s a 1-in-A chance this city’s growth is roughly as impressive as Shenzhen’s; a 1-in-B conditional probability that if the city is impressive, it will inspire nationwide policy changes; and a 1-in-C conditional probability that, given all the above, the broader nation will experience a roughly China-analogous acceleration in growth and given the current national income and projected growth in this country this will produce X amount of expected value.” It would likely be worth further breaking down the chain of conditionals and using Monte Carlo simulations to correctly propagate the uncertainty across steps.
6. Compare those estimated gains to the expected philanthropic (and potentially other) costs of founding a charter city.

Uncertainties and Open Questions

1. How much do charter city advocates need to spend to increase the probability by one percentage point that a charter city will be founded? Alternatively, how much do charter city advocates need to spend to accelerate charter city development by one year?
2. How many governments are willing to cede enough authority to create a charter city and how might this change over time? What sort of cascading effect should we expect after the first potential wins?
3. What percentage of earth’s land surface area is suitable for city development? What percentage of this area is both sparsely populated and geographically suited for easy access to international trade? What percentage of this area is in low-income countries?
4. How many investors are sufficiently value-aligned, deep-pocketed, and risk-tolerant enough to potentially finance charter cities? How much money do these investors collectively control?
5. How easily can growth premium-inducing rules be identified for a charter city? What are these rules and how generalizable are they?
6. How likely is it that implementing boards will establish the rules that have been identified as growth premium-inducing?
7. How easy is it to align the profit incentives of private actors running the city with the social welfare of the citizens?
8. How likely is it that the policies that prove successful in charter cities will be replicated elsewhere?
9. What is the expected value of encouraging charter cities to experiment with novel governance structures rather than adhere to current best practices? How much of this value can be captured in reform zones? How tractable are reform zones relative to charter cities? What risks, if any, does this sort of experimentation increase?
10. How likely is expropriation?
11. How politically controversial will the idea of charter cities become over the next decade?
12. How much of the proposition value of charter cities is in non-economic benefits? How should we value/model these benefits?
13. What types of current charter city advocacy (blogs, podcasts, newsletters conferences, white papers) make the most impact? How much do these methods differ in cost? What other types of advocacy could be pursued?

Research Ideas

Because there is so much uncertainty surrounding charter cities, further research could unlock valuable information at a modest cost. This could either be in the form of in-depth research or small-scale projects that test the core assumptions behind the charter city movement.

Commissioned Report on Charter Cities in Africa

In conversation, Gyude Moore suggested commissioning the African Union Development Agency New Partnership for African Development (AUDA-NEPAD) or a similar group to write a report about the potential for charter cities in Africa. AUDA-NEPAD has a reputation for investigating speculative interventions to promote African development, and Gyude thinks the group would be a good fit for this sort of research. We believe the report could provide stronger, and more credible evidence about the current and potential viability and benefits of establishing charter cities in Africa than we are able to offer as non-experts without the necessary local context. Unfortunately, we do not know how much it would cost to commission such a report.

Commissioned Model of Indirect Charter City Benefits from CCI

In their own cost-effectiveness modeling, the Charter Cities Institute only accounts for the direct economic benefits that charter cities might provide to their residents. Our view is that these benefits alone are not high enough (in expectation) to justify effective altruists supporting charter city advocacy. However, the vast majority of charter city benefits may lie in the indirect benefits these cities could generate. To examine whether this claim is true, it would be nice to have a formal model that attempts to roughly capture the value that charter cities might generate through spillovers, replication, and experimentation. The Charter Cities Institute seems like a natural organization from which to request such a model.

Improved Analysis of Special Economic Zones

The World Bank’s 2017 report on special economic zones has played a large role in shaping our views on charter cities. However, as we noted above, the report was subject to several limitations. When asked about the report, advocates have leaned on these limitations as reasons to discount the report’s findings (while still maintaining that the success of certain SEZs support the case for charter cities). An improved study of special economic zones would either vindicate charter city advocates (by finding that SEZs of a predictable stripe reliably improve growth) or undercut the general case for charter cities (by finding similar results to the original). Getting clearer on the features of SEZs that are most analogous to charter cities and constructing a data set of just those SEZs would therefore be valuable.^[53] (For instance, the World Bank report limits their analysis to SEZs no larger than ten square kilometers, whereas some of the most successful SEZs have been much larger. If size matters a lot, then, since charter cities would probably be a bit larger than ten square kilometers, the World Bank results may not generalize to charter cities.) Unfortunately, we do not know from whom to commission such a report.

Sustainable Development Zones in Ethiopia

The SDZ Alliance, based on recent negotiations by Kilian Kleinschmidt has agreed with the Ethiopian government to implement sustainable development zones in Ethiopia, labelled as SME Cities. The goal of the project is for a special purpose social enterprise to rezone the land around several informal settlements housing individuals that have migrated from rural villages to the peri-urban areas or been displaced by recent violent conflict in Ethiopia. The special purpose vehicle held by the developer, the residents and the municipality will help ensure adequate infrastructure and social services financed by a blend of finance sources ranging from investors to donors and the state. A feasibility proposal for the project can be found here.

The next phase of the project, planned as of July 2021, will require approximately $600,000 to establish a physical presence in Ethiopia and conduct detailed feasibility studies at proposed locations to be agreed with the government. The $600,000 would pay for a small Ethiopian team supported by SDZ Alliance experts to refine current feasibility studies and coordinate with government officials and investors and set up an initial Project Implementation Unit for 6-8 months. The Ethiopian government is supporting the fundraising effort with donors and investors by the SDZ alliance, which is currently operating on a shoestring budget.

The project itself appears promising, but the main value of funding it would be to see if we could use a reform zone to spark the sort of virtuous cycle of economic growth that charter cities might furnish. If the project is successful, that success would provide modest evidence that we do not need brand new cities to generate the benefits charter city advocates advertise.

Acknowledgements

David Rhys Bernard and Jason Schukraft jointly researched and composed this report. Thanks to Michael Aird, Holly Elmore, Sofia Fogel, Peter Wildeford, Mark Lutter, Jeffrey Mason, David Moss, David Reinstein, and Linchuan Zhang for helpful comments on earlier drafts. Further thanks to Mark Lutter, Kilian Kleinschmidt, Scott Foushee, Michael Castle-Miller, Stephen Clare, & Gyude Moore for taking the time to speak with us. Open Philanthropy provided funding for this project and we use their general frameworks for evaluating cause areas, but they do not necessarily endorse its conclusions.

References

Acemoglu, D., Johnson, S., & Robinson, J. A. (2001). The Colonial Origins of Comparative Development: An Empirical Investigation. American Economic Review, 91(5), 1369–1401. https://doi.org/10.1257/aer.91.5.1369

Acemoglu, D., Johnson, S., & Robinson, J. A. (2005). Chapter 6 Institutions as a Fundamental Cause of Long-Run Growth. In P. Aghion & S. N. Durlauf (Eds.), Handbook of Economic Growth (Vol. 1, pp. 385–472). Elsevier. https://doi.org/10.1016/S1574-0684(05)01006-3

Alexander, S. (2014, June 7). Archipelago and Atomic Communitarianism. Slate Star Codex. https://slatestarcodex.com/2014/06/07/archipelago-and-atomic-communitarianism/

Banerjee, A., Duflo, E., Goldberg, N., Karlan, D., Osei, R., Parienté, W., Shapiro, J., Thuysbaert, B., & Udry, C. (2015). A multifaceted program causes lasting progress for the very poor: Evidence from six countries. Science, 348(6236). https://doi.org/10.1126/science.1260799

Center, M. (2019, September 25). Alain Bertaud on Cities, Markets, and People (Ep. 76—Live in NYC). Medium. https://medium.com/conversations-with-tyler/tyler-cowen-alain-bertaud-cities-urbanism-3321136c71d8

Gennaioli, N., La Porta, R., Lopez-de-Silanes, F., & Shleifer, A. (2013). Human Capital and Regional Development *. The Quarterly Journal of Economics, 128(1), 105–164. https://doi.org/10.1093/qje/qjs050

Gibson, J., Olivia, S., Boe-Gibson, G., & Li, C. (2021). Which night lights data should we use in economics, and where? Journal of Development Economics, 149, 102602. https://doi.org/10.1016/j.jdeveco.2020.102602

Glaeser, E. L., La Porta, R., Lopez-de-Silanes, F., & Shleifer, A. (2004). Do Institutions Cause Growth? Journal of Economic Growth, 9(3), 271–303. https://doi.org/10.1023/B:JOEG.0000038933.16398.ed

Heinze, G., & Dunkler, D. (2017). Five myths about variable selection. Transplant International, 30(1), 6–10. https://doi.org/10.1111/tri.12895

Kroth, M. (2014, September 1). Under New Management. Foreign Policy. https://foreignpolicy.com/2014/09/01/under-new-management/

Mallaby, S. (2010, June 8). The Politically Incorrect Guide to Ending Poverty. The Atlantic. https://www.theatlantic.com/magazine/archive/2010/07/the-politically-incorrect-guide-to-ending-poverty/308134/

Mason, J. (2019). The Case for Charter Cities Within the Effective Altruist Framework. Charter Cities Institute. https://assets.website-files.com/5d253237e31f051057dc0a2b/5d88effe42420361c5e0c3c2_The%20Case%20for%20Charter%20Cities%20Within%20the%20Effective%20Altruist%20Framework.pdf

Moser, S. (2018, July 13). The Future Is Already Here. Cato Unbound. https://www.cato-unbound.org/2018/07/13/sarah-moser/future-already-here

North, D. C. (1991). Institutions. Journal of Economic Perspectives, 5(1), 97–112. https://doi.org/10.1257/jep.5.1.97

Pritchett, L. (2018a). Alleviating Global Poverty: Labor Mobility, Direct Assistance, and Economic Growth - Working Paper 479. Center For Global Development. https://www.cgdev.org/publication/alleviating-global-poverty-labor-mobility-direct-assistance-and-economic-growth

Pritchett, L. (2018b). Paths to High Productivity. Cato Unbound. https://www.cato-unbound.org/2018/07/18/lant-pritchett/paths-high-productivity

Rodrik, D. (2008). 6. Getting Institutions Right. In One Economics, Many Recipes (pp. 184–192). Princeton University Press. https://www.degruyter.com/document/doi/10.1515/9781400829354-008/html

Rodrik, D., Subramanian, A., & Trebbi, F. (2004). Institutions Rule: The Primacy of Institutions Over Geography and Integration in Economic Development. Journal of Economic Growth, 9(2), 131–165. https://doi.org/10.1023/B:JOEG.0000031425.72248.85

Sagar, R. (2016). Are Charter Cities Legitimate? Journal of Political Philosophy, 24(4), 509–529. https://doi.org/10.1111/jopp.12089

Shepard, W. (2017, December 12.). Why Hundreds Of Completely New Cities Are Being Built Around The World. Forbes. https://www.forbes.com/sites/wadeshepard/2017/12/12/why-hundreds-of-completely-new-cities-are-being-built-around-the-world/

Sun, G.-W., Shook, T. L., & Kay, G. L. (1996). Inappropriate use of bivariable analysis to screen risk factors for use in multivariable analysis. Journal of Clinical Epidemiology, 49(8), 907–916. https://doi.org/10.1016/0895-4356(96)00025-X

United Nations Conference on Trade and Development. (2019). World Investment Report 2019: Special Economic Zones. UN. https://doi.org/10.18356/8a8d05f9-en

Voarintsoa, N. R. G., Raveloson, A., Barimalala, R., & Razafindratsima, O. H. (2019). Malagasy’ orMadagascan’? Which English term best reflects the people, the culture, and other things from Madagascar? Scientific African, 4, e00091. https://doi.org/10.1016/j.sciaf.2019.e00091

World Bank. (2010). Building Engines for Growth and Competitiveness in China: Experience with Special Economic Zones and Industrial Clusters. https://openknowledge.worldbank.org/bitstream/handle/10986/2501/564470PUB0buil10Box349496B01PUBLIC1.pdf?sequence=1&isAllowed=y

World Bank. (2017). Special Economic Zones: An Operational Review of Their Impacts. https://openknowledge.worldbank.org/bitstream/handle/10986/29054/P154708-12-07-2017-1512640006382.pdf?sequence=1&isAllowed=y

Footnotes