Low-Cost Computers for Education in Developing Countries

In traditional demand theory preferences are defined over products. If a product is chosen the consumer is said to have revealed a preference for it. Products cannot be compared directly and it makes little sense to talk of distributional gains. In the newer approach by contrast products are viewed as bundles of characteristics over which individual and collective preferences are defined. The distributional impact of products then depends on the matching between preferences and characteristics. At the one extreme will be cases where a small group of high-income individuals will gain at the expense of the low-income majority whereas at the opposite extreme a wider spread among low-income beneficiaries is achieved. The basic model is set out in Fig. 1.

Each of the two goods, X and Y, is represented by a ray from the origin. How far along each ray one can go depends on the price of the good and the budget. As demonstrated in this figure, for illustrative purposes, the consumer can reach A on good X and B on good Y. Which of the two he will choose depends on his preferences for the two characteristics.

The person that favours characteristic 1 (as represented by the shape of the indifference curve that goes through A) will choose product X and the person with opposite preferences will choose product Y. Assume for example that characteristic 1 represents functional features of a product and characteristic 2 more luxury-type features, then a low-income person may tend to favour product X while the opposite is true for a relatively high-income person. Imagine that X is a simple bar of soap without any packaging and Y is a detergent with expensive packaging and other ‘luxury’ features such as optical whitening.

An important part of my framework is the recognition that there is a close relationship between income and the nature of product characteristics. More specifically as shown in Fig. 2 there is a tendency for products to become more ‘high-income’ over time.

As (Stewart 1997, p. 17) puts it:

Technical change in products in developed countries occurs in line with the rising incomes (and is partly responsible for those rising incomes), so that the balance of characteristics offered by new products corresponds to the changing demands of consumers as their incomes rise. The rising incomes have the effect of shifting demand towards different products with more sophisticated, labour-saving, higher–quality, etc. characteristics: to summarize this complex of changes we may say that the characteristics of the new products have more high-income characteristics.

One common feature of new products for example is that they save time compared to older products as incomes rise. Take the case of cooking. In developing countries this is generally undertaken on a time-intensive basis but with rising incomes the balance of characteristics changes. Time becomes more scarce as wages rise and goods become more time intensive, through the use of capital-intensive modes of cooking (e.g. with coffee machines and micro-wave ovens). Note that this tendency is true not only of the physical properties of goods but also the mode of access to them, as will shortly become evident.

‘In the case of public policy’, writes Stewart (1997)

particularly public goods, there is a direct connection between the nature of the product produced and the distribution of income. The provision of particular types of public good itself affects the distribution of income. Given a limit in total public expenditure the allocation of funds to high-income inappropriate products automatically removes resources from low income consumers. For example, the provision of tarmacked highroads is at the expense of improving roads for the rural areas.

Consider from this point of view the situation depicted in Fig. 3. On the vertical axis is the characteristic safety, the standards of which in the developed countries reflect in a rough way the increased premium on this characteristic as incomes have grown (e.g. the increased safety of cars, airplanes and medical drugs). Assume in panel (a) of Fig. 3 that OP is the amount of safety associated with the developed-country product X (assumed to equal the minimum standard required),³ The local product Y, by contrast is intensive in low-income characteristics and yields only OQ units of safety which is below the minimum required (OP). The result is that the average consumer in poor countries would be on a lower indifference curve because he or she is denied the chance to buy the cheaper, albeit less safe product that better accords with these average preferences. As shown in panel (b) however the category of inhabitants that favours safety in relation to low-income characteristics would benefit from the newer good X imported by the developing country. The outcome is that a minority of individuals with high-income preferences gains at the expense of the majority. From a distributional point of view this can only described as unequal.

As another example of the situation in Fig. 3 consider healthcare and in particular manually operated X-Ray equipment which is no longer approved for use in developed countries because it is considered unsafe. This type of equipment has been replaced by more complex and safer machines. In developing countries, however, it is tempting to assume that ‘the benefits of having X-Ray equipment available for use outweighs the potential risk of unnecessary exposure from manual operation’ (Kennedy 1979). Or again expenditure of the government health budget may take the form of relatively few high-cost, low-risk products from developed countries or cheaper, older and more risky alternatives. The latter option would permit a much higher percentage of the population in developing countries to get at least some drug therapy. And it would accordingly represent a much more equal distribution of well-being.

Finally, I turn to the field of housing where developing country governments are notorious for clinging to irrelevant developed-country standards, which the majority of the population cannot afford, ‘Purchase of the housing with ‘too-high’ standards involves a disproportionate expenditure on this item compared to other items in the household budget. i.e. it leads to imbalances in consumption (or to concentration of public expenditure on housing to the exclusion of other social needs)’.⁴ For example, in a slum upgrading project in Kibera, Kenya, ‘One cannot blame Kibera residents for assuming that the project is deliberately attempting to create housing for the middleclass and deprive current—residents of their right to a convenient location in Nairobi’ (Huchzermeyer 2006, p. 18).

In the next and final element of the analytical framework I focus specifically on the two characteristics that seem best to discriminate between the alternative models of computing described above.

Generally, the relationship between per capita income and individual access (as a percentage of population) looks something like the line 0Z portrayed in Fig. 4.

In developed countries (at say Y2) most people are easily able to afford the cost of owning a computer. At developing country levels of income (Y1) this is also true of the upper-income deciles of the population. But for the relatively poor majority in these countries the main alternative mode of access to computers is sharing. Indeed numerous authors, across different developing regions, speak of a culture of sharing. For example, ‘the Philippines have a culture in which household members generally share resources’ (Rangaswamy and Singh 2009). In Asia more generally, ‘material resources are often shared at the level of household and neighborhood…..Asian cultures privilege the family over the individual’ (Rangaswamy and Singh 2009). Or again in a different region and in relation to the mobile phone ‘in some African countries it is regarded as the property of the community, because there is a culture of sharing the tools of communication’ (Lopez 2000).

James (2010a) argues that the amount of computer sharing in developing countries should indeed be greater than in developed countries, to an extent that depends on the income differences between them. The frame of reference is a developed country, the UK, which at a per capita income level of $42,740 has 5 students per computer (this is the actual situation there and also the level recommended by education authorities in other developed countries such as the USA). To arrive at the 1:1 (Rangaswamy and Singh 2009) relationship proposed by the OLPC, the British income has to be multiplied by 5, yielding a figure of over $200,000 (see Table 1). In effect the OLPC is basing its solution on an income level that no country on earth comes close to attaining. Take the case of Kenya shown in Table 1. The recommended amount of students per laptop of 312.5 is reached by calibrating the per capita income of Kenya ($680) against $213,700, ‘In particular, the ratio of the former to the latter level of gross national income (GNI) per capita is equal to 0.32% which, in the ratio of 100:32, gives the balanced amount of computer spending per child (in this case, 1 computer should be shared by 313 Kenyan students)’. (James 2010a). For more developed countries such as Peru, however, the amount of sharing drops to 27 students per computer. With the ‘N Computing’ model this amount is easily accommodated whereas the OLPC solution would require a per capita income that is 29 times the figure of Peru (i.e. $213,700). At lower levels of income the purchase of ‘XO’ computers for all students, would result in serious imbalances in the education budget and even in other sectors. This is the inevitable consequence of applying (higher than) developed country standards to poor countries (or relying on a decidedly high-income characteristic to solve a poor country problem).