Notes

1. Self-help housing is not easy to define. Most planners, however, would probably agree that the distinctive characteristic of self-help housing is that it always begins as a flimsy form of shelter lacking all kinds of service and is developed on land which either lacks planning permission or has been invaded. The adjective "self-help" stems from the fact that the occupier has built some or all of the accommodation, even if some form of professional help has almost always been involved. The typical architect is the local jobbing builder or bricklayer; the building manual is the advice received from family and friends.

2. Of course, two of the eight cities under discussion here, Bogotá and Santiago, largely escaped recession during the 1980s.

3. The ejido is a form of community landholding established in Mexico as a result of the Revolution. In law, the land cannot be sold, although in practice much of the land occupied by housing in many Mexican cities belongs to ejidos.

4. Trivelli (1987: 105) claims that "vacant lots represent one-third of the building space of Brazilian cities."

5. Based on a sample of 446 plots in the three years, drawn primarily from the western, higher-income areas of the city. Not surprisingly, land prices within a given settlement increased more rapidly; lower prices were found in more peripheral locations.

6. In any case, it is uncertain whether prices continue to rise in low-income settlements once they have begun to develop. Indeed, data from Bogotá and Mexico City suggest that plot prices in low-income settlements often fall over time (Gilbert and Ward, 1985). They do not rise partly because the best plots are sold first and partly because it is so difficult to sell such plots.

7. Of course, Brazilian experience during much of that period as hardly typical. The economy was growing rapidly but the poor suffered the consequences of a deliberate attack by the government on their living standards. Under these circumstances, the purchase of land was almost bound to become more difficult over time.

8. For low-income groups the average fell from 132 square metres to 68 over the same period.

9. The latter observation is more true of the eastern US cities. The problem with Ingram and Carroll's argument is that they use figures which record population over a rather wide central area. They also include some figures of dubious value.

10. Table 4.13 presents data on gross population densities over time. While such figures are never totally reliable, they do make sense. The figures have the advantage that they record the urbanized area rather than the administrative area of the city. As such, they record real changes in density over time. The figures contradict the findings of both Portes (1990: 17), who claims that "with the exception of Buenos Aires, densities have continued to increase in urban centers, regardless of their legal boundaries," and Ingram and Carroll (1981), who show that densities increased between 1950 and 1970 in Mexico City, São Paulo, Rio de Janeiro, Bogotá, Recife, Belo Honzonte, Guadalajara, Monterrey, and Cali. Lee (1985) shows the same tendency occurring in Brazil's eight largest cities. Ingram and Carroll's figures also show dramatic increases in density in some cities. The problem with all these sets of calculations is that they have used a constant administrative area of the city over time. Since much of this area was empty at the beginning of the period, any expansion of the city outwards into this empty area is almost bound to increase the population density. All the authors admit the problems with calculating figures in this way but none have modified their figures to take account of them.