ECONOMIC DEVELOPMENT QUARTERLY / November 2001
about the cities’ economic development policies and programs. The initial draft of the question- naire was reviewed by a local economic development practitioner. The purpose of this review was to ensure proper professional language and the reasonableness of the questions. After revisions, the questionnaire was mailed to 20 cities outside the study’s sample. The results of this pretest indi- cated a solid survey instrument with a potentially high response rate (65%). The questionnaire needed only minor revisions for clarity reasons prior to implementation of the survey.
Optimization of the response rate was a major objective of the survey design. Therefore, we fol- lowed Dillman’s (1978) method for a mail survey as it is known to produce good response rates. This systematic approach involves identification of a reliable respondent mailing list, a persuasive cover letter, and multiple mailings.
The survey achieved a 58.3% response rate, with 413 cities responding to the questionnaire. Although this response rate is very good for a mail survey, two issues need consideration. First, the purpose of the sampling design is to produce results that represent the population of cities. If, how- ever, respondents differ from nonrespondents in a systematic manner, then the results will not be generalizable to the population. To test for nonresponse bias, a logistic regression model using response as the dependent variable (1 = responded, 0 = no response) was run in SAS using citywide characteristics as independent variables. These variables included the unemployment rate, reve- nue-to-expenditures ratio, per capita income, percentage below poverty, population growth rate from 1980 to 1990, region dummies, and population category dummies. Only the coefficient for the unemployment rate was statistically significant (p = .01) in the response bias model. Therefore, cities with higher unemployment rates were less likely to respond to the survey. In addition, the largest category of northeastern cities contained only one city. It was necessary to collapse this information into an adjacent category. For this reason, generalization to the population must include a warning that analytical results may not apply to the largest cities in the northeast region or to cities with high unemployment rates.
Second, the variable total economic development expenditures was missing for many of the cit- ies. In fact, only 332 cities provided adequate expenditure data.1 The response bias test relies on citywide characteristics that were available from U.S. Census sources. It may be that the cases used in the analysis are biased toward cities more heavily involved in economic development. In other words, cities omitting the expenditure data simply may not be expending funds on economic devel- opment, and therefore, the respondent left the expenditure section of the questionnaire blank instead of entering a zero. Given the information available, we cannot confirm this supposition.
Data and Variables
The survey provides the key variables for this research including the dependent variables. These survey data were merged with selected variables downloaded from the 1990 U.S. Census of Popu- lation and Housing summary tape files (U.S. Department of Commerce, 1992, 1993) and the 1994 County and City Data Book (1995). Specifically, policy variables, for example expenditures, and opinion variables, such as the perception of business influence on economic development deci- sions, were collected in the sample survey, and city contextual factors came from the U.S. Bureau of Census sources.
This research investigates the factors thought to influence the support for economic develop- ment by cities. Support for economic development has been operationalized in several different ways. The existence of a particular tool, such as tax abatements, as a dichotomous dependent vari- able, is used in some studies; that is, the locality either employs the tool or not (Cable et al., 1993; Sharp & Elkins, 1991). Wolman (1996) criticized this approach as failing to capture the extent of economic development. The count of tools also has been used to measure local economic develop- ment effort. Wolman emphasized that the measure fails to account for the amount of resources put into economic development. Furthermore, he noted that there is no evidence that there is a correla- tion between the number of tools and amount of resources devoted to economic development.
The survey of economic development professionals conducted for this research asked for expenditures data as well as use of inducements. Table 1 displays a list of economic development inducements used by cities. Regulatory relief was used by more than 75% of the cities in the