A city’s population, the ratio of its debt to the assessed value of its taxable property, its per capita taxes and tax collection rate, the average salary of its inhabitants and the percentage of its population employed in one major industry are all relevant to the investment quality of the city’s municipal bonds. But bond rating is not an exact science, and there is some leeway for factors that are irrelevant to investment quality to creep into municipal bond ratings.

Using a sample of 381 cities in 50 states, the authors compare bond ratings predicted by a model employing the six relevant variables mentioned above with actual bond ratings. The differences between the actual and predicted ratings are examined to see if one irrelevant factor—the city’s location—plays any role in the rating of its municipal bonds. The results indicate that state and regional biases do exist: Actual bond ratings for the south are lower than the model estimates they should be, whereas actual ratings for states in the central plains are higher.