22 Transparency: GA, if properly done and used, is useful to both voters and those elected. It raises the awareness of administrators about gender issues and points out "weaknesses" in women's well being in the community they govern. It informs voters about how much is spent on specific issues as percentage of total expenditure. It increases accountability. GA provides the means for achieving the goal of better control over the administration. How well this goal is achieved depends on the support that GA has from the civil society. What cannot be done: Too many factors determine gender inequality. The assessment of the real impact of a policy on gender inequality is rather difficult and requires sophisticated econometric methods which are beyond a "low cost" GA exercise. Beware of local administrations which claim to have affected macroeconomic variables (such a decrease of the employment gender gap). Most of the time it is not down to them alone. Lesson 3. GA at a local level, however useful, must be supported by an initiative at national level. This is nee- ded for the following reasons: ? Cost: calculations of efficiency cannot be carried out by people without the necessary expertise and experts cost. ? Data collection: The data which are necessary for GA can be collected, at a small addi- tional cost, by the national institute for statistics when it performs its usual surveys. ? Coordination: methodology and benchmarks must be coordinated. So far each initiative has followed its own rules. This may have been fruitful at an initial stage, but it is ineffi- cient in the long run as it makes comparisons difficult and hinders the spread of "best practices". A standard is required. There are many reasons for demanding a GA exercise at national level, and indeed it is a peculiar- ity of the Italian case that it has never been attempted. The experience of GA in the province of Sie- na has provided a further reason: GA at national level is also necessary for better understanding the gender bias of local policies.