Experimental Design
We implement an online survey experiment with German mayors using a conjoint design to study decision-making regarding municipal public investments and the take-up of intergovernmental grants.
In the main task, respondents evaluate a series of hypothetical scenarios describing a public investment project and an associated grant program. Respondents are instructed to consider a typical investment in their municipality and to assume that (i) for each investment project a respective grant program would be available and that the best program for the respective municipality was identified, (ii) the project is not decided upon yet and competes with other municipal tasks and investments, such that financing would rely on additional revenue or limitations on other spending, (iii) funding will be approved conditional on compliance with the program requirements and once an application was filed, and (iv) the investment is generally sensible and technically feasible. Further, subjects need to assume that, such investment projects can generally happen in their own municipality and they need to consider, how well their municipality would be able to implement the project under actual limitations of adminstrative and fiscal capacities (v).
Each scenario is characterized by seven attributes that vary along two dimensions: (i) characteristics of the investment (investment volume, urgency of investment, and expected support in the population), and (ii) characteristics of the grant program (pages of the grant application, discretion how the funds might be used, documentation intensity after project completion as well as co-financing rate).
The possible values for each respective attribute are as follows:
Investment volume: €0.2 million; €1 million; €3 million
Urgency in the affected policy area: Very high urgency; High urgency; Moderate urgency
Expected public support: High support; Moderate support; Low support
Application form length (number of pages to be completed): 2 pages; 10 pages; 25 pages
Expected documentation requirements after project completion: Detailed documentation with receipts; Simplified documentation without receipts
Flexibility on use of funds: High (funds can be used largely at own discretion); Low (use of funds is largely pre-defined)
Co-financing rate: 50%; 60%; 70%; 80%; 90%
Each respondent evaluates six scenarios presented sequentially in a tabular format.
While the attributes are shown in fixed order for a given subject, their order is randomized between respondents. Similarly, we have attributes that are related to the investment project (investment volume, urgency in the affected policy area and expected public support) and the grant (application form length, flexibility in use of funds, co-financing). Each of these blocks also is shown in random order across subjects but is in fixed order across scenarios for a given subject. Attribute values are randomly assigned across scenarios and respondents. Randomization enables identification of the causal effects of grant and investment characteristics on stated policy support of surveyed politicians.
The design uses a single profile evaluation format. For each decision for an individual scenario, respondents indicate whether, in their role as mayor, they would support carrying out the investment. By identifying the respective treatment effects of investment- or grant-related attributes, we can differentiate the relevance for each attribute for public investment support of the surveyed political decision makers.
To study mechanisms, a subset of two scenarios is revisited after the main task. For these scenarios, respondents assess perceived administrative burden and risks associated with the grant process, including expected effort, perceived stress, and perceived risk of implementation or compliance problems.
These tasks are embedded in a questionnaire, structured in several blocks. On the first screen, we ask for their confirmation to participate. The survey itself begins with an introduction that emphasizes the relevance and policy impact of the study. Next follows the main conjoint experiment. We also ask about the applicability of scenarios to policy areas in a follow-up question. Subsequently, we use two scenarios shown before in the mechanism block asking about perceptions about grant features. The survey then includes follow-up questions on municipal characteristics, administrative capacity, and respondents’ background. It concludes with open-ended questions on motivation and feedback.
The survey is administered online. The target population consists of all mayors in Germany, who are invited via personalized contact information by mail and email.