Improving local citizen participation and tax compliance in Niger

The impact evaluation will focus on identifying experimental impacts from awareness raising activities at the village level. Given the difficulty in randomly assigning households to exposure to different policy instruments, treatment assignment will be clustered at the village level, meaning that all households within targeted villages will be exposed to the same awareness raising activities. The impact evaluation will consist of a control group of villages that are not directly exposed to the dissemination of awareness raising materials and two treatment groups exposed to different types of materials.
The research team will design awareness and sensitization messages on tax compliance and participation in local governance (harnessing social recognition of stakeholders/individuals engaged in role model behavior). Materials (information flyers, two variants to be tested against a control group) will be designed at the commune level and distributed at the village level by village chiefs. Flyers will contain the following information: (i) A visual, concrete summary of the commune’s achievements with financing from the municipal tax budget (e.g. number of schools built); (ii) a calendar of budget participatory meetings and related events; and (iii) messages, visuals, vignettes on tax compliance and its link to participation, including an optional socially motivated message (for one of the two treatment groups).
The expected impact of the materials will be to improve attitudes and behaviors towards the municipal tax (e.g. tax morale and tax compliance) and a willingness and plans to participate in commune meetings by filling information gaps, increasing a sense of civic duty and reciprocity, and updating expectations about the behavior of others (within and outside of the village), among others. Changes in various indicators related to self-reported tax compliance and citizen participation will be measured through a baseline and endline survey of households, where comparisons can be made between exposed and non-exposed respondents (to the communication campaign). Additional outcomes will be considered at the village level through the use of administrative data, where tax collection levels can be evaluated over time.

2021-09-01

2021-12-31

The primary outcomes are participation in local government meetings (i.e., whether citizens attend participatory meetings organized by the municipality governments) and tax compliance (i.e., whether citizens pay the local municipality tax).

The outcomes will be measured using both self-reported household surveys on tax compliance and participation and through municipal government accounts on meeting participation and tax collection.

Secondary Outcomes (end points): Secondary outcomes include awareness of local government affairs and tax obligations, civic engagement, trust and satisfaction with public services supply.

The study is implemented through a stratified multi-level selection process: among the 4 pre-defined regions, 24 communes are randomly selected and within each, 12 villages (and then, for data collection to identify impacts, households) are randomly selected. Within the selected communes, 288 villages are selected at random (stratified by commune) to participate in data collection activities (household survey, 15 households per village) and the implementation of pilot activities (dissemination of information materials). The selected villages are representative within each commune, such that these are not different from non-selected villages in terms of population, distance from the commune seat, and other observable characteristics at the village and household levels. Households within the 288 targeted villages are selected at random for the data collection process, based on figures provided by the village chief. A total of approximately 4,320 households is selected at random (stratified by village) for data collection of key outcome indicators related to citizen engagement and tax compliance. Random selection/assignment at each level (commune, village, household) is executed in such a way that guarantees both internal and external validity of the findings from the study.
Targeted villages are randomly assigned to the communications campaign. The random selection of villages to each of the treatment arms is stratified within each commune to achieve balance in treatment group allocation within and across communes. The 12 selected villages within each commune are randomly assigned to each group, such that four receive no dissemination, four receive dissemination of basic materials, and the remaining four receive dissemination of materials including an element of social influence (social recognition and/or social influence). In total, of the 288 villages randomly selected to participate in the study (across the 24 selected communes), 96 constitute the control group (no dissemination of materials), 96 receive a flyer with basic information, and 96 receive a flyer with the basic information in addition to elements of social influence.
Two surveys will be conducted: baseline in June 2021 and endline in March 2022.

Randomization done by a computer.

The unit of randomization is the village or urban neighborhood.

Yes

288 villages

2880 households (10 households per village)

96 control villages, 96 villages in treatment arm #1, and 96 villages in treatment arm #2.

): Initial power calculations suggest that the impact evaluation design should consider a village-level (clustered) rather than commune level RCT, where the unit of observation is the household, as there is not enough power to identify impacts from an intervention clustered at the commune level. With a minimum detectable effect of 0.2 standard deviations in the rate of tax compliance (identified from the literature) and 10 surveys conducted per village, each treatment arm would need a minimum of 93 villages (930 households) across the selected communes. Our actual sample size is larger allowing us to be slightly overpowered and capable of addressing some amount of attrition.