Thinking aloud about how to estimate impact of libraries in U.S.

In my class on the economics of gender in developing countries we were discussing a recently published paper by Nancy Qian that estimates the effects on the sex ratio (how many boys there are compared with girls) of increases in the relative incomes of women.  Not surprisingly, in 1980s China the effect is negative and pretty large.  The interesting thing about the paper is the method.  Normally a correlation between higher female income and lower sex ratio would be viewed skeptically: maybe some other factors are responsible for both lower sex ratios and higher incomes.  Qian is able to use the market liberalization in China's agriculture, in 1979, to do what is known as a difference-in-difference analysis.  That is, she can estimate the effects of the higher incomes after liberalization in tea growing areas by comparing the changes there to the changes that happened in non-tea growing areas.  Tea growing areas favor female labor, natch, so higher price of tea after liberalization meant higher demand for female labor and that then meant more girls survived (sex ratio fell). 

After class, I was trying to think if there was any analogy to apply the method to library impact studies.  Hard to do anything like this in the places where FAVL operates... there are simply not enough libraries to properly study their impact, nor are there any non-library measures of reading habits and availability.  But in developed countries with lots of libraries, it occurred to me that the move of library systems to enable bar-coded self checkout and online renewals meant a huge reduction in the cost of using libraries.  I know that in my own personal behavior, I have not waited in a line to check out books since, well, since 1995 I suppose... just before San Jose library system switched to self-checkout.  Suddenly there were no more lines, ever.  Amazing!  So since different library systems probably switched at different times due to budgetary reasons, the changes in their trends of borrowers from before and after their switch to self-checkout will be largely exogenous to user reading habits.  If so, then those changes can be used to estimate the effect of reading more through libraries on school test scores.  Someday, maybe someone will pursue this line of reasoning.  Maybe someone already has!