Here’s how they came to that conclusion. To figure out how and whether corruption affected state spending, they first needed a solid, consistent measure of corruption itself. They found that in the Department of Justice’s Public Integrity Section, which maintains annual numbers of public official convictions by state. Mikesell and Liu focused on 50-state statistics from 1997 through 2008, which covers more than 25,000 convictions.
There were two concerns they sought to address. First, they focused only on convictions that resulted from violations of federal corruption laws. State and local laws can vary in strictness and scope and might thereby account for differences in conviction counts. Federal law, on the other hand, is uniform across states. Second, they had to account for differences in resources. Prosecutors and courts in different regions have different caseloads and work schedules, which might affect how many convictions they can secure. After controlling for factors such as work hours, judges per citizen, and case pending rates, the researchers found that resources didn’t have an impact on the corruption conviction rate. In other words, the Justice Department’s statistics provide a valid and consistent yardstick for corruption.
Oregon was the least corrupt state in the union, per public employee, they found. The 10 least corrupt states are spread across the nation, in fact: Washington, Minnesota, Nebraska, Iowa, Vermont, Utah, New Hampshire, Colorado and Kansas all joined Oregon atop the list. The 10 most corrupt states, in order, where: Mississippi, Louisiana, Tennessee, Illinois, Pennsylvania, Alabama, Alaska, South Dakota, Kentucky, and Florida. Here’s a map showing each state’s relative level of corruption over that roughly decade-long timespan.
Reverse engineering a model
Armed with what they felt was a valid, reliable measure of corruption, Mikesell and Lui then put together a model to track how closely a basket of variables tracked to corruption, based on models used by other corruption researchers.
A number of variables—some obvious, others not—were linked to state spending in statistically significant ways. Those were: past spending, interparty rivalry, fiscal centralization, governor’s party, political ideology of citizens, the size of the population aged 18 to 64, urbanization, population, unemployment rates, personal income, and—you guessed it—corruption.
Past research has found a lot of effects of corruption. It has been linked to a reduction in capital investment, inefficient use of public resources, reduced economic productivity, lower output per worker and exacerbation of income inequality. According to Miksell and Liu, it appears also to be tied to overall spending and how spending is allocated.
States with higher levels of corruption are likely to spend more per capita on construction, highways, wages and salaries, borrowing, corrections and police, they found. They also tended to spend less per capita on education, public welfare, health and hospitals. Here’s their list of corruption by state: