Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

The price of land

Trends in land prices

Most commentaries on Latin American cities sooner or later aver that land prices are excessively high. Many blame this problem on the process of land speculation and it is certainly true that far too many plots are being held out of the market in anticipation of a rise in price (Kowarick, 1988; Trivelli, 1987).4 They also emphasize how spectacularly land values are rising. For example, Allen (1989: 8), citing IBASE (1982), claims that "the average price of land in São Paulo quadrupled between 1964-1978, and in Rio rose by three and three quarter times its value." In Bogotá, FEDELONJAS (1988) claims that real land prices increased roughly six times between 1959 and 1988, and Villamizar (1982) that they increased annually by 4 per cent per annum between 1955 and 1978.

Until recently, the idea that land prices always rose rapidly in third world cities was conventional wisdom. Such a trend made sense. As Payne (1989: 45) puts it: "Under conditions of sustained high urban growth, urban land has enjoyed a level of demand well in excess of formally sanctioned supply, guaranteeing a good rate of return on investment." However, recent evidence has begun to detect signs of falling real land values in Santiago and in several Mexican cities (CED, 1990; Coulomb and Sanchez, 1991; Gilbert, 1989; Jones, 1991). Limited data from Guadalajara suggests that the Mexican recession had a strong impact on land prices. Between 1975 and 1980, the average price of peripheral plots rose in real terms by 14 per cent; between 1980 and 1985 it fell by 3 per cent (Gilbert and Varley, 1991: 93).5 In Puebla, real land prices fell spectacularly after the crisis broke in 1982 (Jones, 1991) and a similar pattern also occurred in Querétaro, Toluca, and Mexico City (Jones, Jiménez, and Ward, 1993; Coulomb and Sanchez, 1991). In Santiago, land prices in high-income areas of the city rose spectacularly between 1979 and 1982, only to plummet in the next few years (CED,1990).

If land values have fallen the reasons are fairly clear. As Durand-Lasserve (1990: 49) points out: "The contradiction between rapid land price increases and stagnant or declining urban incomes at least during the first phase of the crisis led to the erosion of the formal land market. This erosion is reflected in ... the outflow of capital from the land sector to either other economic sectors or to the up-market real estate business."

Whether changes in land prices matter depends in part on what is being measured. What precisely do average land values measure? Detailed work on an excellent data set of land prices in Bogotá demonstrate the problem perfectly: land prices between 1955 and 1988 rise during one longish period and fall during another; they shoot up one moment and fall the next. As such, the choice of time period is critical to analysis. Even more important is how to interpret the meaning of the average: what does an average land price rise measure when land prices rise rapidly at the periphery and slowly near the centre? Still more troubling is the fact that trends are very sensitive to changes in the way that the annual average is calculated. Dowall and Treffeisen's (1990: 118) reworking of the Bogotá data set turns an annual 4.4 per cent increase into "an annual decline in real terms of 1.4 per cent." Without going into a long technical discussion of the reasons why, my feeling is that few of the figures on trends in land prices are really to be trusted.

Do rising land prices matter?

Mohan and Villamizar (1982: 248) argue that a rise in land values only matters if it "widens the already high levels of inequality that exist in many poor countries." If the poor buy land and benefit from rising land values, then is there any reason to worry? If the poor sell land to the rich, won't incomes have been redistributed? Even if the poor buy from the rich, might they not benefit from later price increases?

Unfortunately, although Mohan and Villamizar's point is important, it is less than clear that the working of the land market in Latin America's cities is helping to redistribute income. Much agricultural land is owned by affluent families, so the sale of that land worsens the distribution of income. Even where agricultural land is owned by the poor, it is uncertain whether they ever receive the full market value.

And, if they buy plots which subsequently rise in value, it is uncertain whether they will be able to capitalize on the increase. Evidence from several cities is beginning to suggest that it is very difficult to sell plots of land once they have consolidated houses on them (Gilbert, 1993). Land is normally sold only when it contains little more than a shack (Gilbert and Ward, 1985); it is difficult to sell a proper house because banks are reluctant to offer mortgages on property in low-income settlements and sometimes there is no legal title. Better-off families prefer to buy empty plots on which to build houses to their own design. As such, it is difficult for a poor owner to reap the benefits of any rise in land values.6

But, even if the poor benefit from rising land values, this hardly helps those trying to buy a plot for the first time. It is the relationship between land prices and wages that is the key to housing affordability. The problem is demonstrated by evidence from São Paulo, where Kowarick and Campanário (1988: 39) report that the cost of land rose 3.6 times as quickly as the minimum salary between 1959 and 1986. New households without property were increasingly frozen out of the formal land market.7

New entrants to the housing market also suffer in so far as plot sizes tend to decline when land costs rise faster than incomes. In Bogotá, plot sizes fell quite dramatically during the 1970s and 1980s both for formal and informal construction sites. Between 1973 and 1975 formal construction sites averaged 155 square metres; ten years later the average had fallen to 73 (Molina et al., 1993: 59).8 A similar trend was apparent for Bogotá for informal-sector housing plots, which fell from 157 square metres during the 1960s to 131 square metres in the later 1970s (Gilbert and Ward, 1985: 118). By the mid 1980s, plot sizes had fallen to 75 square metres (Molina, 1990: 304). Under such circumstances, population densities are almost bound to rise as families economize on their use of land.

The increasing price of land, therefore, is clearly a major problem in a city such as Bogotá and is even more of a problem in cities like Santiago, where a poor family does not have the same option of buying land without a full range of services. In certain Mexican cities, however, poor-quality peripheral land is not particularly expensive; in the middle and late 1980s a plot of land cost between two months' and five months' earnings at the minimum salary (Colomb and Sanchez, 1991; Gilbert and Varley, 1991). The land may be of poor quality, it may be distant from the main centres of employment, it may be lacking services, but it is cheap. In Caracas or Lima, where the poor generally obtain land through invasion, plots are virtually free.

The key point to recognize is that there is a great deal of variation between Latin American cities in the way that the poor acquire land. These different methods have a pronounced impact on the affordability of land in each city (Gilbert and Varley, 1991). What we need to know is not what is happening to some mythical average land price but what is happening to the cost of plots on the periphery relative to incomes. Unfortunately, it is just this kind of data that is rarely available. Until we have such a data set, we will be unable to judge whether the cost of low-income plots of land is increasing or decreasing over time relative to what the ordinary family earns.

Residential segregation

In the nineteenth-century city, most people lived close together. In the absence of motorized transport, the city had spread little and population densities were high. By the turn of the century, the rich had begun to move out of the city centre into detached suburban homes. As transport facilities improved, the pace of suburbanization accelerated and a wider range of social groups began to participate. Even the poor began to move out of central rental accommodation into the self-help suburbs. Of course, the process of suburbanization was very different for the poor: the rich moved one way, the poor another. By the 1950s, a definite pattern had been established in most cities, in which the rich lived in the most environmentally attractive areas of the city and the poor somewhere else. In São Paulo in 1970, "over three quarters of families in the top income bracket . . . lived in seven districts within a radius of approximately 7 kilometres to the south and west of the city" (Batley, 1983:102).

Latin American cities remain highly segregated and few locals will have any difficulty in reciting the names of the richest and the poorest settlements in the city. In Lima, everyone knows that San Isidro and Miraflores are rich while Comas and Villa El Salvador are poor; in Santiago, the extremes are found in affluent Providencia and Vitacura in the north-east and poor La Pintana and La Granja in the south. In Rio de Janeiro, the rich live in Leblon and Ipanema and the poor in the Baixada Fluminense. However, the pattern of segregation today is much more complicated than in the past. Figure 4.2 shows that it is no longer possible to state quite so categorically that the rich live in the north of Bogotá and the poor in the south. In Caracas, rich and poor live in most zones of the city, albeit in clearly demarcated areas.

Figure 4.2 Bogotá: Residential segregation, 1973 and 1981

The pattern of segregation has become more complicated for a number of reasons. First, as urban areas have expanded in population and area, affluent areas have come into contact with lower-income areas. Sometimes, the expansion of these elite suburbs collides with working-class areas that were once some distance away. In Bogotá, for example, the affluent north has reached the low-income settlements of the periphery, clustered around the quarries of Usaquén (figure 4.2). Since the authorities could not be persuaded to move such well-consolidated areas, the rich had little choice but to live close to the poor.

Second, the growth of the middle class has complicated the pattern of residential segregation. The middle classes have increasingly taken over land close to low-income areas because they cannot afford land elsewhere. During the 1960s, several large construction companies bought land in the west and south-west of Bogotá which they developed into middle-class suburbs during the 1970s (Gilbert and Ward, 1985: 116). Today, such suburbs occupy land that would have been inconceivable some years ago.

Third, variegated topography has also helped to blur the neat pattern of segregation; indeed hilly cities are arguably less clearly polarized than flat cities. This is obvious in Caracas and in Rio de Janeiro, where many steep slopes have long been considered unsuitable for formal-sector construction. Since the land had little commercial value it was allowed to remain in the hands of the municipal authorities and during periods of rapid urban growth came under pressure from low-income groups; barrios and favelas often developed in close proximity to high-income and middle-income areas. Indeed, in Caracas, this process has continued in the highly accentuated terrain that is now being developed in the south and south-east of the city. Here, every exclusive residential development appears to have its low-income neighbour next door. A functional symbiosis has developed; the urbanización provides work for maids, shoe menders, laundresses, and the like, and the barrio provides the cheap labour.

At the same time, there is little sign of residential segregation disappearing. Indeed, in some respects there is greater polarization. As crime rates rise, the fear of burglary, kidnapping, and other crimes has encouraged many rich families to move into protected areas. Some move into blocks of luxury apartments located in the elite suburbs, offering, in São Paulo, "security in a city where violence is an acute problem" (Taschner and Bruna, 1994: 100). Other rich families move into what are effectively armed camps: residential developments surrounded by high walls with entry controlled by security guards. It is increasingly likely that rich Latin American children will grow up without having any contact with the urban poor. They attend exclusive private schools and universities, they play in luxury clubs, they shop in exclusive commercial centres, and they are never allowed out on their own. Even their traditional form of contact with the poor, through the live-in maid, has changed; most domestic servants now prefer to live with their own families and commute to work.

Urban planning has contributed little to the process of segregation, even if it is sometimes argued that zoning regulations were introduced in Latin American cities primarily to help keep poor people out of the elite suburbs (Amato, 1969). The most likely explanation of residential segregation has always been market forces; elite suburbs have long been protected by high land values and the costs of installing full services before housing is constructed. If governments have accentuated residential segregation, it is through their response to market pressures. Governments have given priority to the servicing of areas that could afford to pay for them. The water, telephone, and electricity companies have not ignored low-income areas because they were full of social lepers but because installing services in poorer areas is more complicated. It is more difficult to install infrastructure in settlements where the houses have already been built, especially when so many of their inhabitants have difficulty paying the bill.

Of course, some governments have gone out of their way to increase the level of residential segregation. Over the years, several military regimes have attempted to remove poor people from "slums" in high-income residential areas to "proper homes" in peripheral areas of the city. Orgies of squatter relocation occurred in Caracas in the 1950s, Rio de Janeiro between 1967 and 1973, and Buenos Aires between 1976 and 1981 (Valladares, 1978; Portes, 1979; Dwyer, 1975; Pajoni, 1982).

The most recent example of such "social cleansing" was practiced by the Pinochet regime in Santiago between 1979 and 1985. In those six years, some 28,700 families were moved from campamentos located on private land (CED, 1990). Most of those families were moved from the central areas and from the affluent eastern suburbs into the communes of the far south and the north-west of the city. An administrative reorganization, in 1982, aggravated the effects of the removal programme. New municipal authorities were established with much greater powers. Unfortunately, the revenues available to each municipality differed considerably; the authorities facing the greatest social problems had only a fraction of the resources available to affluent districts such as Providencia and Vitacura. Portes (1990: 24) claims that "it is now difficult to talk about Santiago as a single city because groups thus segregated lead widely divergent lives and remain confined - by choice or force - to their distinct spatial locations."

If the Pinochet regime went out of its way to cleanse the city of its slums, are the poor and rich are any more segregated in Santiago than in any other major Latin American city? Is Santiago really any different from São Paulo, where "the poor are removed to the periphery, ever further and less wellserviced, while the rich live in central neighbourhoods well provided with infrastructure in housing of good quality, even of great luxury" (Sachs, 1990: 36)? Are any of the other major cities any different?

Population density

Population densities in most major Latin American cities today average between 100 and 130 persons per hectare. There are, of course, substantial variations between cities. As table 4.12 shows, the poorer cities, Bogotá, Lima, and Mexico City, have particularly high population densities, although Caracas, confined to its narrow valleys, is also densely populated. More affluent Santiago and São Paulo have lower densities.

Ingram and Carroll (1981: 265) claim that Latin American cities "do not appear to be significantly more dense than older central cities in the US. The peripheral densities of Latin American cities are similar to those found in North American cities."9 This rather surprising finding is explained by the fact that North American cities have much more high-rise building. Indeed, what is worrying about population densities in Latin America is not the overall average but where the figures reach their peak. Most of the highest figures are found not in areas containing high-rise apartments but in consolidated low-rise, self-help areas. Thus, although areas of high-rise middle-class housing in Rio de Janeiro, such as Copacabana, have densities of over 300 persons per hectare, these densities pale in comparison with the high values of some of the favelas (Allen, 1989). Similarly, in Santiago, the highest values of between 200 and 250 persons per hectare are found in the low-rise suburbs (Bähr and Mertins, 1985; INK, 1986b). In Bogotá, the peak densities are also found in the poorer suburbs (Pineda and Jiménez, 1990), where poor families have attempted to economize on the high cost of land.

But, if this is the current pattern, what are the dynamics of change? Are Latin America's cities becoming less densely packed through time? Mills and Tan (1980: 317) suggest that this is almost inevitable, given that international evidence shows that "large urban areas are more decentralised than small urban areas,... high income urban areas are more decentralised than low income urban areas (and) ... transportation improvements almost certainly lead to decentralisation. "

Table 4.12 Population density and growth of urban area

City Year Population
Bogotá 1900 100 909 110
1928 235 1,958 120
1938 330 2,514 131
1964 1,730 14,615 118
1985 4,177 32,866 127
1991 4,960 30,300 164
Caracas 1936 235 542 430
1950 694 4,586 151
1971 2,184 15,000 145
1981 2,583 19,750 131
1990 2,989 23,300 128
Lima 1940 618 2,100 294
1952 1,105 5,540 199
1959 1,603 8,500 189
1972 3,303 13,054 253
1990 5,826 54,000 108
Mexico City 1930 1,049 8,609 122
1940 1,560 11,750 133
1950 2,872 24,059 119
1960 4,910 47,070 104
1970 8,455 68,260 122
1980 12,140 91,211 133
São Paulo 1930 878 15,000 59
1980 12,184 125,800 97
1980 12,184 137,000 89
Santiago 1940 952 11,017 87
1952 1,350 15,351 88
1960 1,893 21,165 90
1970 2,861 31,841 90
1982 4,318 42,800 101

Sources: Coulomb and Sanchez, 1991; Dietz, 1978; Dowall and Treffeisen, 1990: 17; Gilbert, 1978; Gilbert, 1993; IBGE, 1984; INK, 1986a; Kross, 1992; UNCRD, 1994.

Since Latin American cities have become both more affluent and more decentralized over recent decades, there should have been a clear reduction in overall densities. They have also improved their transportation systems, which, Echeñique (1982: 256) argues, should lead to an increase in the supply of land and therefore to declining densities. In fact, the figures in table 4.13 show anything but a clear trend. If there are signs that densities have fallen in Caracas and Lima, the opposite appears to have occurred in Bogotá, Santiago, and São Paulo. In Mexico City, densities first fell and then began to rise. 10

I would argue that the reason why densities have not fallen universally is closely linked to the dynamics of land prices and local patterns of housing development. First, because land prices have remained high in most cities, and possibly even risen in some, it has been difficult for population densities to fall. Indeed, as the last section showed, the average size of building plots has been declining in several cities over time. Second, although extensive self-help development has brought lower densities in cities such as Lima, such accommodation does not guarantee low densities. Indeed, as I have already shown, some of the most densely packed suburbs are not high-rise residental areas but low-rise consolidated self-help areas. Third, many Latin American governments have tried, admittedly with varying success, to restrict the expansion of the urban area. This has been particularly common in cities where urban functions compete with agriculture for land. In so far as official attempts to restrict suburban growth have met with success, they have pushed up population densities.

Latin America's major cities, therefore, have been subject to different processes, some raising densities and some reducing them. The combination of processes has clearly affected different cities in differing ways. To understand what has happened, it is essential to recognize that although Latin America's major cities share many similarities, there are vital differences between them. For example, while all have large areas of self-help housing, the contribution that self-help accommodation makes to the housing stock differs considerably. The form of self-help also differs. In Lima, invasions have occupied vast areas of land in ways that have not been permitted in Bogotá or Mexico City. Arguably, the proliferation of self-help settlement on invaded land in Lima in recent years (see table 4.1) has reduced population densities in that city, whereas the land alienation processes characteristic of Bogotá and Santiago have encouraged denser forms of residential occupation. In short, Latin America's cities show diversity within similarity.

Table 4.13 Population decline in the central city

Bogotá 1964-73 1973-85  
Central area1 -2.2 - 1.3  
Inner ring2 0.0 - 1.1  
Lima 1961-72 1972-81 1981-92
Central area3 0.6 0.4 - 0.9
Inner ring4   0.2 - 1.6
Mexico City 1960-70 1970-80 1980-90
Cuauhtémoc -1.4 - 2.0 - 2.2
Inner ping5 0.2 - 1.4 - 2.0
Second ring6 2.5 - 0.1 - 1.6
Santiago 1960-70 1970-82 1982-92
Commune of Santiago7 - 2.2 - 1.8 - 1.4
Inner ring8 n.a. 0.0 - 0.7
São Paulo 1960-70 1970-80 1980-87
Historic centre9 - 1.4 0.0 n.a.
Inner city10 0.7 2.2 0.9
Inner ring 0.1 1.3 3.6
Intermediate ring 2.8 1.3 3.6

Sources: INK, 1986b; Kowarick and Jacobi. 1986: 200; Mohan and Villamizar, 1982; Pineda and Jiménez, 1990; UNCRD, 1994: 96; national censuses.


1. Comuna 31: an area of 411 hectares.

2. Comunas 32, 41, 61, 71 and 81: an area of 1,406 hectares.

3. Lima and Breña.

4. Pueblo Libre (Magdalene Vieja), Jesus Maria, Lince, and La Victoria.

5. Cuauhtémoc, Miguel Hidalgo, Benito Juárez and Venustiano Carranza.

6. Acapolzaclo, Gustavo Madero and Iztacalpo.

7. The area of the commune of Santiago as recorded for 1960-70 was much larger than that covered in the later periods. In 1982 it was 22.3 square kilometres.

8. Communes of Lstación Central, Independencia, Ñuñua, Pedro Aguirre Cerda, Providencia, Quinta Normal, Recoleta, San Joaquin and San Miguel; a total area of 108.2 square kilometres. The administrative areas of Santiago were changed in 1982 and calculation of comparable areas for the period before 1970 is not possible.

9. Seven innermost sub-districts of central ring.

10. Thirteen surrounding sub-districts.

Contents - Previous - Next