|Title||THE EFFECTS OF EFFICIENCY-INCREASING REFORMS IN RURAL GOVERNANCE: EVIDENCE FROM A QUASI-EXPERIMENT IN TURKEY|
|C1 Background and Explanation of Rationale||Since the early 1990s, many countries went through a rural government reform to increase efficiency and effectiveness of service and infrastructure provision as well as land-use planning, support to agriculture and education. Yet, in the existing cases, reforms changing rural government models to increase the efficiency in the provision and management of infrastructural services either amalgamate smaller rural governments into larger regional agencies or transfer decision-making power from central agencies to small rural administrations. Yet, Local Administration Reform (Law No. 6360) that the Turkish government introduced in 2012 followed neither of the above-stated paths. The reform changed the rural governance system in 30 out of 81 governorates of the country by shifting the responsibility and resources for management of rural infrastructure from regional agencies specialized in rural services to metropolitan municipalities specialized in provision of services to urban areas. Villages lost their autonomous administrations and became, de jure, urban neighborhood units, even though they de facto remained rural. The governorates were assigned to the reform in a discretionary manner (based on the total population of the governorate, not population density) which means rural regions were assigned to reform or no-reform governorates in an “as-if random” manner. Taking the opportunity of this “quasi-experiment,” we will have a chance to compare regions that are affected by the reform and regions that are not affected by the reform and explore how the reform changed agricultural production and population trends. Specifically, we aim to find answers to the following questions: What happens when the villages are run by governance institutions that have been governing urban regions? Can urban institutions adjust their delivery structure and priorities fast enough to answer the needs, for example, agriculture-related demands, of rural communities? Or do they harm rural sectors by approaching governance of rural areas with an urban logic?|
|C2 What are the hypotheses to be tested?||
Hypothesis 1: Governance of rural infrastructure provision by institutions with urban bias leads to decrease in agricultural production. Therefore, we will observe a disproportionate decrease in agricultural production in treatment group villages.
Hypothesis 2: Population in villages that are governed under the new system will decrease faster as opposed to the villages that are governed under the old regional governance system. This effect will occur because agricultural production in villages under the new governance will decrease disproportionately which will then lead to faster migration of the younger generations.
Hypothesis 3: The electoral support for the party in government will increase more in the villages within the scope of the reform as opposed to the villages that stayed under the previous regional governance system in the first elections after the reform passed but before implemented, due to the increased expectations from the new governance system.
Hypothesis 4: The electoral support for the party in government will decrease more in the villages within the scope of the reform as opposed to the villages that stayed under the previous regional governance system in the second, third and fourth elections after the reform passed, due to the negative consequences of the reform on agricultural production and village economy.
|C3 How will these hypotheses be tested? *||
Our analyses will draw on a difference-in-difference design. We will check the robustness of our findings by replicating our analysis with samples selected by alternative population bandwidths and selection criteria.
To examine the effect of the Local Government Reform on villages’ economy, out-migration, and political preferences, we employ a difference-in-differences estimator using the following specification:
〖Y_it=α+〖βT〗_t+γD_it+θT_t D_it+δX_it+ϑZ_dt+〖ε_it〗_^ 〗_
In the above formula, Yit is the outcome variable in village i at time t. Tt is a dummy time variable that equals 1 for post-treatment periods and 0 for pre-treatment periods. Dit is the treatment variable that equals 1 for treatment villages and 0 for control villages. We interact Tt and Dit to produce the coefficient θ, which is the average treatment effect (ATE) of the treatment on outcomes over time. Xit is a vector of control variables that are employed to minimize variance and capture minor over-time imbalances. Zdt is also a vector of control variables at the district level. In some of the designs where village-level data is not available, Y_itand D_itwill be replaced by district-level variables, Y_dtand D_dt.
Villages cannot be treated as independent as they are clustered within districts, the first unit above villages in the administrative hierarchy, even after the government reform. We will therefore cluster standard errors at the district level. We will also control for the within-year correlation in the standard errors across units which will limit biases caused by the correlation of errors within a year.
|C5 Scale (# of Units)||Approximately 12,100 villages in 335 districts.|
|C6 Was a power analysis conducted prior to data collection?||No|
|C7 Has this research received Insitutional Review Board (IRB) or ethics committee approval?||not provided by authors|
|C8 IRB Number||not provided by authors|
|C9 Date of IRB Approval||not provided by authors|
|C10 Will the intervention be implemented by the researcher or a third party?||This is a quasi-experimental study. The intervention has been implemented by the Turkish government since 2014 without any intention for research.|
|C11 Did any of the research team receive remuneration from the implementing agency for taking part in this research?||No|
|C12 If relevant, is there an advance agreement with the implementation group that all results can be published?||not provided by authors|
|C13 JEL Classification(s)||not provided by authors|