1 - A peak watt (Wp) measures the 'ideal' output of a PV system, when the panel is at 25° C and receiving 1000 W/m2 of solar insolation.

2 - These figures are averages in Kenya (Hankins, 1993). Among the systems surveyed by Acker and Kammen (1996) battery replacement and repair accounted for almost 5%, and the charge controller accounted for under 1% of the total cost.

3 - In 1994 the Duty and Value Added Tax (VAT) resulted in a cumulative tax of 55% on the PV panels.

4 - Environment et Développement du Tiers Monde -- Programme Energie.

5 - The wide range reflects the fact that this is a loan, where some or all of the capital costs could, potentially, be recovered. The lower end reflects only the administrative costs of the project.

6 - As a further example of the need for an expanded interdisciplinary perspective for cost/benefit comparisons, consider the selection of households for intensive sales and marketing of PV systems. The GEF Project Document (1993) recommends a focus on families, "living more 25 kilometers from Harare, or more than five kilometers from the grid." This recommendation reflects the economics of PV systems relative to grid extension. Grid extension in developing nations generally costs US$ 3,000 - 10,000 per kilometer (Foley, 1995). Electrification projects are normally under the auspices of the national Department of Energy, which may view stand-alone PV as an important complement or even precursor to grid extension, or PV may be viewed as a direct threat, "stealing" the most affluent and influential rural customers.

Thus, while PV policy in Zimbabwe intends to focus on customers distant from the grid, in Kenya we found little evidence of a 'grid effect'. With PV generated power at US$ 0.50/kWh and grid extension costs of about $8200/km, one might expect to see the bulk of PV customers beyond the break-even point, in this case about 8 km from the grid. In our survey of households that purchased PV systems, however, we found 85% were less than 5 km from the grid, 50% are within 2 km, and one quarter are under 1000 meters away (Acker and Kammen, 1996; Figure 24). Instead of fighting this trend, the designers of PV programs might look for ways to work with the local preferences, social and market forces.