This appendix serves two purposes. First, I provide additional details regarding data issues, definitions, and case selection. Second, I present a variety of additional analyses supporting the main findings in the text and discuss more technical econometric issues. The reader who is not concerned with the data and econometric details can skip this appendix without missing any important substantive content.

DATA ISSUES

COG Data on Special Districts

The primary source of data for the government finance and organization variables used in my analyses is the COG. Like any data source, this one has strengths and weaknesses. The strengths of the COG are its comprehensive coverage of U.S. local governments, its historical comparability from the 1970s on, and the breadth of its coverage of government finance variables. The weaknesses of the COG are a matter of some dispute, as evidenced in the exchange between Leigland (1990) and Sacks (1990). Foster (1997, pp. 81–84) also reviews data quality issues related to the COG.

Leigland (1990) is highly critical of the COG and raises two primary concerns about the data on special districts. First, he argues that the Census Bureau overcounts special districts that are functionally inactive. For example, he shows that in the 1982 COG, 24 percent of special districts reported no expenditures, but the Bureau classified only 1.5 percent of districts as inactive.