20 fants who can attend a cr?che - show figures which range 6 per cent to 27 per cent, with a differ- ence of 21 points! A "benchmark" which comes from places which share culture, laws and regula- tions, is far more convincing and acceptable than a benchmark drawn from national and interna- tional standards. Comparison also helps us to put together "checklists" about the quality of a service. A service which is provided in one Comune may be missing in another. This calls for an explanation and the result is an increase in the transparency of the choices of the public administration. For example, when we discussed the problem of immigrant women, we explained that recent research had shown that these fall into two different categories. One category consists of women who join their husbands, usually from countries were women do not enjoy the same rights as Italian women. Another cate- gory consists of women who come to Italy alone, leaving their families behind, usually to work for Italian families as babysitters or badanti (carers for the elderly). The two groups have completely different needs: for the former the main problem is integration, while for the latter it is keeping in touch with their families at home and saving as much as possible. We handed out a form where we suggested possible solutions to the two different kinds of problems. Other solutions were added by the participants, and the list can now be used by all administrators as a reference for dealing with the issue. The "network" approach has other potentialities besides that of providing a tool to assess perform- ance in a specific field and spreading "best practices", but we were unable to exploit them. We hoped to bring to light duplications of services and other sources of inefficient use of resources. An allocation of resources that was efficient 40 years ago, when people did not have cars, mobile pho- nes and internet, may not be so any longer. We know that there is a lot to say in favour of more co- ordination among Comuni in the same area, but we also know that the issue has strong political im- plications and can be very unpopular. We did not even begin to discuss it. Lessons to be learned Three lessons can be learned from this initiative. The first is what is required for a "low cost" GA. The second is about what can be achieved in terms of the four objectives of a GA exercise: ef- fectiveness, efficiency, transparency, equity. The third is the importance of an initiative at a na- tional level, without which GA at a local level cannot thrive. Lesson 1: The most important ingredient of a successful GA at a local level is the involvement of the adminis- trators. So far they have not been under any obligation by law and the only incentive they had was the possible use of GA as a tool of political propaganda. It is not by chance that the few Comuni which underwent a GA belong to the best administered parts of Italy. Even then we did not find any real interest in the results of the GA. We had the impression that for some of them it was more im- portant to show that GA was attempted than to use its results for a change of policies. If GA does not become a constant and mandatory practice I do not have much hope that it can go beyond a few other pilot experiments. The problem of incentives for the administrators is far more important and serious than the problem of incentives for the employees, which can be solved more easily. It is clear that we can rely on vol- untary work only exceptionally, and that small economic incentives (or enhanced career opportuni- ties) must be offered. This should be the only additional cost of GA for the municipal budget. The first GAs have been very expensive, since they required massive intervention of "experts", given that nobody really knew what had to be done. However, the intervention of the experts can be mi- nimized if standard procedures are introduced and local administrators are given guidelines from the central level. On this, see lesson 3 below.