Street Vendors Don’t Disobey: They Obey Hotelling, Salop, and Krugman

For the past three to four weeks, operations to clear street vendors have resumed with renewed intensity in Delmas and Pétion-Ville. Drawing on three economic models established over the past century, this article shows that these operations will fail just as they always have. Not for lack of political will, but because they address none of the real causes of the phenomenon. It then proposes five measures that would, by contrast, act on the right parameters.

What everyone sees, and what no one explains

For three to four weeks now, the images have been circulating. On Delmas 19, on Delmas 31, and in Pétion-Ville near the markets and the old cemetery, municipal agents are clearing away the stalls. Equipment is seized. Official statements follow one after another. The sidewalks empty for a few hours, sometimes a few days.

Then the vendors come back.

This comes as no surprise to anyone who has followed the situation for a few years. The cycle has been repeating itself for decades, with perfect regularity. The authorities step in. The vendors leave. The vendors come back. And round it goes. Official markets are held up as the solution. No one settles in them for long. And round it goes again.

The question this article asks is simple: why? Why have evictions failed systematically, without exception, for so long? The usual answer is a lack of firmness, a lack of resources, or a cultural resistance to order. That answer is wrong. What happens on these sidewalks obeys economic laws that science has been establishing for nearly a century. These laws are precise and verifiable, and they predict exactly what we observe: the inevitable failure of any policy that attacks the symptoms without touching the causes.

Before getting into the models, a point of intellectual honesty is in order. The Haitian economics literature has begun to document this phenomenon. Lamaute-Brisson (2002) produced the first rigorous analysis of it in the metropolitan area. Aspilaire (2014, 2017) measured the informal sector at nearly 57% of Haitian GDP in 2007, a share that has only grown since. Jean-Gilles and Paul (2023) showed how the security crisis has redistributed these activities across urban space without shrinking them. This work establishes the scale of the phenomenon. The analysis that follows seeks to lay out its inner mechanics in accessible terms.

You cannot evict an equilibrium. You reconfigure it, or you endure it.

Three economic laws that policymakers ignore

There are three economic models that, taken together, explain everything we observe on these sidewalks. They were written between 1929 and 1991 by economists who were not thinking about Haiti, to solve problems that look very different. Yet their conclusions apply here with a precision that should give pause to anyone designing policy for the management of public space.

Each of the three models answers a different question. Why do vendors always end up in the same places instead of spreading out? Why are there always too many of them, even when their earnings are pitiful? And why do they invariably come back, however much force is used to dislodge them? Three questions, three answers, one shared conclusion.

First law: vendors go where the customers are (Hotelling, 1929)

Here is a scene everyone knows. You pass the intersection at Delmas 32. There are fifteen vendors. You pass the intersection at Delmas 19. There are twenty. Move away from these main roads and you find far fewer. Why this concentration? Why don’t these vendors spread out across the whole neighborhood so as not to compete with one another? That would make sense, wouldn’t it? Fewer competitors around you, more potential customers.

This is precisely the intuition that an American economist, Harold Hotelling, proved wrong in 1929. His reasoning runs as follows. A customer who wants to buy something does not choose a vendor on price alone. The choice depends on total cost: the price of the goods plus the cost of getting to the vendor. That travel cost is time lost, money spent on transport, sometimes risk. And in Haiti, this cost is particularly high: the roads are difficult, transport is expensive, insecurity weighs heavily. In practical terms, this means that Haitian customers buy close to home not out of convenience, but because doing otherwise genuinely costs them too much.

Hotelling turns this reasoning into a formula. Along an urban corridor (say, the stretch from Delmas 19 to Delmas 32), each customer chooses the vendor that minimizes total spending:

Put simply: the customer located at t chooses the vendor i whose price (p) plus travel cost (τ times the distance) is lowest. The symbol τ (tau) represents the cost per meter traveled. The higher this cost, the more the customer prefers the nearest vendor, even one who is a little more expensive.

Hotelling’s conclusion is counterintuitive but rigorously demonstrated: under these conditions, vendors do not spread out. They move closer together. They converge on the point where there are the most customers. Not because they have coordinated, not because they are following a fad, but because any vendor who moves away from that point immediately loses customers to the one who stays. If vendor A moves to the left, vendor B captures all the customers to the right of A. A therefore has no rational reason to move.

That is why the vendor set up outside the school on Delmas 19, at the Delmas 31 intersection, or at the airport intersection did not choose that spot on a whim or to defy authority. That vendor found the place where customers pass. Evicting them means forcing them to leave that place. But the place itself does not move. Customers keep passing through it. So the vendor comes back, or someone else takes the spot. The eviction has not changed the fact that this is where people buy.

It also explains why official markets, even well-built ones, fail to attract street vendors. The problem is neither their quality nor how busy they are. The problem is their location. They were built where land was available, not where people pass. A vendor who moves into one is voluntarily giving up their customers. No rational being would do that.

Second law: when entry is free, there is always overcrowding (Salop, 1979)

Hotelling explains where vendors set up. The second question is one of numbers. Why are there so many of them on these sidewalks? Why does a spot that is already saturated attract newcomers? And why, despite this saturation, do incomes stay as low as if no one were really making a living? The economist Steven Salop answered these questions in 1979.

His explanation comes down to a few simple steps. Picture a busy intersection where a first vendor arrives. Business is good; the vendor makes money. A second sees this and sets up nearby. The newcomer makes money too, a little less than the first since the two share the customers, but makes money all the same. A third arrives. Then a fourth. Each newcomer mechanically reduces the share of customers of everyone who was there before. The process stops only when there is nothing left to gain, that is, when revenue just covers costs. That is the subsistence level. In this setting, each vendor’s profit can be written as:

In words: each vendor’s profit (π) equals the margin (price p minus cost c) divided by the number of vendors n, minus the initial outlay F (goods, equipment). As n rises, each vendor’s share shrinks. New vendors stop entering when profit falls to zero: there is nothing left to gain.

This formula says something rarely heard in Haitian public debate. Street vendors are not poor because they are lazy, or because they do their work badly. They are poor because the very structure of the market, where anyone can set up without any formality, condemns them to share a clientele that cannot grow indefinitely. Aspilaire (2014) confirmed this result empirically for Haiti: however long people spend in the informal sector, incomes converge toward the subsistence minimum.

In Haiti, the situation is extreme because the barrier to entry is nil. There is no commercial register that is actually enforced, no real business license, no minimum capital requirement. Anyone can set up at any time. In the language of the model, the initial outlay F is close to zero. The equilibrium number of vendors then becomes:

In words: as F tends toward zero (entry without barriers), the equilibrium number becomes very large. When τ is high (customers do not travel), it is larger still. Both conditions hold in Haiti. This expression is indicative and derives from the zero-profit condition under free entry.

Salop’s most important finding is not in the formula itself. It lies in what he proves next: the number of vendors reached at equilibrium is exactly twice the number that would be best for everyone, vendors and buyers alike. A planner seeking to maximize collective welfare would choose half as many. Free entry spontaneously produces twice as many vendors as are needed, each earning half of what they could. This is not peculiar to Haiti. It is a general law.

The direct consequence for evictions: they remove vendors but do not alter the conditions of entry into the market. F stays at zero. Demand stays concentrated at the same intersection. The equilibrium has not changed. As soon as the pressure lets up, new entrants fill exactly the gap left behind. The sidewalk returns to the same density as before, sometimes with the same people, sometimes with others.

Third law: the busier a place is, the more activity it attracts (Krugman, 1991)

The first two laws tell us where vendors set up and why there are too many of them. The third explains something that even an attentive observer finds puzzling: why are these configurations so robust that they rebuild themselves in full after intense, repeated operations? Why has the security crisis not shrunk the informal sector, but merely shifted its points of concentration to other corridors? Paul Krugman, who won the Nobel Prize in economics in 2008, supplied the answer in 1991.

His starting observation is this: in an economy, activity tends to concentrate at certain points in space, and that concentration reinforces itself over time. Why? Because a busy place offers advantages that an empty place cannot. An intersection where twenty vendors are at work is a place where prices are known, where wholesalers deliver directly, where customers are in the habit of coming, where vendors extend credit to one another through sols (rotating savings clubs). An empty intersection has none of this. Each additional vendor who sets up in a busy place makes it even more attractive to those who follow. This is what economists call a virtuous circle, or more precisely a self-reinforcing equilibrium.

Krugman measures the relative attractiveness of two zones (the busy sidewalk versus the official market, for example) with the following index:

In words: ω compares two zones, the central zone (c) and the alternative zone (p). Y measures the level of disposable income in each zone, that is, the customers who buy. φ measures commercial accessibility: suppliers, habits, networks. σ measures how easily a customer can switch from one vendor to another. When ω exceeds 1, the central zone wins. And the more it wins, the more it attracts, and the more its advantage grows.

What this formula captures is crucial to understanding why evictions fail. A clearance operation temporarily disrupts an intersection’s equilibrium. But it changes nothing in the parameters that make that intersection attractive: customers keep passing through, wholesalers keep delivering, buying habits do not shift in a day. As soon as the pressure eases, all these forces regain the upper hand and the configuration reassembles itself.

Jean-Gilles and Paul (2023) observed this mechanism directly on the ground in Haiti. The security crisis of recent years has not shrunk the informal sector. It has redrawn its geography: wherever a corridor became dangerous, vendors moved to another one where centripetal forces could come into play again. The equilibrium moved. It did not dissolve.

The same mechanism explains why official markets struggle to attract street vendors, even when they draw a fair number of shoppers. Their ω index is lower than that of the busy sidewalks, because they lie off the corridors of heaviest traffic and do not benefit from the credit and information networks that have built up over time at the major intersections. For an official market to become truly attractive to vendors, its ω would have to exceed that of the competing sidewalks. That does not happen by decree.

Repression displaces an equilibrium. It does not dissolve it. Only a change in the underlying conditions can do that.

What each eviction really costs

The three economic laws we have just seen explain why evictions do not work. We need to go further and assess what they destroy, in concrete terms, in the households affected.

For the vast majority of Haitian street vendors, this trade is not a supplementary income. It is the income. The International Labour Office (ILO, 2020) estimates that more than 80% of non-agricultural employment in Haiti is informal. For most of these people, there is no formal employer waiting for them if their sidewalk is taken away. In economic terms, the constraint their household faces can be written as:

In words: what the household can spend, the price of goods multiplied by the quantities consumed, is financed from two sources: an alternative wage w, if there is one, and the profit from street trade. In Haiti, for most street vendors, w is close to zero. If the street profit disappears, everything disappears.

An eviction does not gradually squeeze this budget. It cuts it off outright, without notice, without compensation, without an alternative. Studies conducted by WIEGO in ten cities in sub-Saharan Africa and South Asia (Roever, 2014) document immediate and steep drops in household income in the period following a clearance operation. For Haiti, such longitudinal data do not yet exist, a gap that Haitian economic research should fill as a matter of priority.

Nor do the figures show the destruction of the informal networks that allow these households to survive. Vendors are not isolated individuals. They lend each other money through sols. They keep one another informed of the prices charged by the wholesalers in Croix-des-Bouquets. They guarantee one another’s small loans to get through the lean weeks. To break up a street market is to break up this network, and rebuilding it takes time and resources that the most vulnerable households generally do not have.

At the national level, the picture is more complex. The sector plays a deeply ambiguous role, as the following table shows:

What the informal sector provides and what it costs, across five dimensions. Sources: Aspilaire (2014), IMF (2023), Loayza and Rigolini (2011).

Loayza and Rigolini (2011) formalized this paradox: the informal sector is at once the best safety net available to poor households and a lasting brake on the growth of national productivity. The IMF (2023) puts the long-term cost of this unregulated informality at between 1.5 and 2.5 points of GDP a year. The answer, then, is not to eliminate the sector, which would amount to removing the only shock absorber that hundreds of thousands of families have, but to create the conditions for a gradual transition toward more productive forms of activity.

What other countries did instead

Several countries have faced the same situation and found responses that did act on the right parameters. These examples are not recipes to be copied, but illustrations of what a policy grounded in the right diagnosis actually produces.

Singapore in 1968 resembled Port-au-Prince today: tens of thousands of street hawkers, free entry into the market, subsistence incomes, constant conflict with the authorities. Rather than step up repression, the government built 113 Hawker Centres, covered spaces placed where demand was highest, in keeping with Hotelling’s logic, and rented out at an affordable price. That rent introduced a barrier to entry (F turning positive, in Salop’s model), which mechanically brought the number of vendors down toward the social optimum, without a single eviction. A renewable three-year concession replaced precariousness with an asset that could be formalized. These spaces are now inscribed on UNESCO’s list of intangible cultural heritage.

Between 2008 and 2015, Accra tried an even simpler approach: an identification number issued within a day, with no immediate tax obligation, granting access to a legal spot marked out on the ground. The result, documented by Lindell and Appelblad (2009): a 40% reduction in conflicts between vendors and the authorities, a 25% increase in the incomes of registered vendors thanks to easier access to microcredit, and the emergence of local taxation where none had existed. Minimal registration makes visible an asset that was invisible, which opens the way to formal credit.

Mexico City set up formal negotiations between vendors’ associations and city hall, covering zones, hours, and sanitary conditions. This is the idea Elinor Ostrom (Nobel Prize, 2009) theorized: rules drawn up by those who use a space are observed; rules imposed from outside are circumvented. Bogotá under Antanas Mockus (1995 to 2003) added a further dimension: transforming public space requires legitimacy, not just force.

Haitian vendors do not resist rules out of anarchism. They resist because no promise of a credible alternative has ever been kept.

Five measures that would address the real causes

Here are five measures that follow directly from the three models. Each one changes a specific parameter that evictions leave untouched. None requires substantial resources. All require understanding the problem before intervening.

1. Map demand before building anything

Hotelling’s logic is clear: vendors will use an official market only if it is placed where customers pass, not where land happens to be available. No Haitian city hall today has a detailed map of pedestrian flows. A 90-day survey of the target areas, conducted by the Haitian Institute of Statistics and Informatics (IHSI) in partnership with the city halls concerned and estimated to cost between $150,000 and $200,000, is the prerequisite for any coherent intervention.

2. Build covered markets in the right places

On the Singapore model: light, covered structures with water and toilets, positioned at the points where the survey finds demand to be concentrated. With a rent that creates the barrier to entry needed to bring the number of vendors toward the social optimum. The rule of thumb from Salop’s model gives the order of magnitude of this rent:

In words: the optimal rent rises with travel cost and falls with the square of the desired number of vendors. Precise calibration requires the data from the pedestrian survey.

3. A vendor card in a single day

A formal occupancy permit with a photo and a unique number, issued for a token fee and conferring three rights: protection against arbitrary eviction, access to microcredit, and access to hygiene training. The permit turns an invisible asset into a recognized one. It introduces a minimal barrier to entry, with F turning positive, which organically reduces overcrowding on the sidewalks without the need for any police operation.

4. Legally recognized street committees

Committees bringing together vendors, local residents, and a municipal representative, tasked with jointly negotiating hours, the width of the pedestrian walkway, waste management, and the rotation of spots. Ostrom’s lesson has been borne out in dozens of countries: rules that people draw up themselves, they respect. Rules imposed on them from outside, they get around. An hour of negotiation costs infinitely less than a clearance operation and its aftermath.

5. Bring the cost of formalization below the threshold that matters

Djankov et al. (2002) established that once the cost exceeds the equivalent of about two weeks’ income, no one formalizes: rational calculation points toward informality. In Haiti, this cost amounts to several months’ income for a small vendor. The objective is simple: bring it below the two-week threshold, through a municipal one-stop shop, a radical reduction in paperwork, and a 24-month tax grace period for new registrants.

What the coming weeks will confirm

The operations under way in Delmas and Pétion-Ville for the past few weeks will follow the script the three models predict. The sidewalks will empty. Then they will fill up again. This is not an opinion: it is what Hotelling has predicted since 1929, what Salop has demonstrated since 1979, what Krugman has explained since 1991. And it is what Jean-Gilles and Paul (2023) confirmed empirically on the ground in Haiti.

The question is not whether the vendors will come back. They will. The question is whether, this time, the Haitian authorities will be ready to acknowledge that this outcome is not a failure of law enforcement, but the predictable result of a policy that ignores the laws of economics. For several decades now, the same intervention has been repeated, with the same result. Persisting in this approach does not reflect the strength of the state: it reflects the absence of an analytical framework suited to the phenomenon.

Haiti is not condemned to this cycle. It is condemned to repeat it until it decides to understand rather than to repress.

References

Aspilaire, R. (2014). L’économie informelle en Haïti: un impact contracyclique sur le PIB? Mondes en développement, 166(2), 101–112.

Aspilaire, R. (2017). Économie informelle en Haïti, marché du travail et pauvreté: analyses quantitatives. Doctoral dissertation, Université Paris-Est Créteil.

ILO (2020). Perspectives mondiales de l’emploi et du secteur informel: focus Caraïbes.

De Soto, H. (2000). The Mystery of Capital. Basic Books.

Djankov, S., La Porta, R., Lopez-de-Silanes, F., and Shleifer, A. (2002). The Regulation of Entry. Quarterly Journal of Economics, 117(1), 1–37.

IMF (2023). Haiti: Selected Issues, Informal Sector and Fiscal Capacity. Country Report No. 23/174.

Fujita, M., Krugman, P., and Venables, A. J. (1999). The Spatial Economy. MIT Press.

Hotelling, H. (1929). Stability in Competition. Economic Journal, 39(153), 41–57.

IHSI (2010). Enquête sur l’emploi et l’économie informelle 2007: premiers résultats.

Jean-Gilles, S., and Paul, B. (2023). Crise sécuritaire et transformation des places marchandes dans l’économie informelle en Haïti. Études caribéennes.

Krugman, P. (1991). Increasing Returns and Economic Geography. Journal of Political Economy, 99(3), 483–499.

Lamaute-Brisson, N. (2002). L’économie informelle en Haïti: de la reproduction urbaine à Port-au-Prince. L’Harmattan.

Lindell, M. K., and Appelblad, J. (2009). Preventing Livelihoods: Urban Planning and Street Trade in Accra. Urban Studies, 46(11).

Loayza, N., and Rigolini, J. (2011). Informal Employment: Safety Net or Growth Engine? World Development, 39(9).

Ostrom, E. (1990). Governing the Commons. Cambridge University Press.

Roever, S. (2014). Informal Economy Monitoring Study Sector Report: Street Vendors. WIEGO, Cambridge.

Salop, S. C. (1979). Monopolistic Competition with Outside Goods. Bell Journal of Economics, 10(1), 141–156.

Singh, I., Squire, L., and Strauss, J. (1986). Agricultural Household Models. World Bank / Johns Hopkins University Press.