The thesis offers a novel perspective on the election and political economy. To my knowledge there has not been an examination of its kind (event study approach using RDD and SCM) of subnational German political economy. It has methodological value for SCM - it presents a novel application of SCM to Germany. Moreover, it uses a valid method of inference to assess the estimates. The thesis is also a novelty in the application of RDD as it is applied to the smallest administrative level in Germany in questions of economic policy.
Economists use events to study causality. This thesis rides this wave with an event study of the surprise victory of Green-Red in the German federal election of Baden-Wurttemberg in 2011, using it as exogenous shock in (expected) government policy (short-as well as long-term), going from "pro-market" to more "social democratic" and "interventionist". I apply the synthetic control method (SCM) by Abadie and Gardeazabal (2003) and the regression-discontinuity (RDD) approach by Calonico et al. (2015) and Imbens and Wager (2019) to statistical data from Germany of municipalities, counties, and states. In the RDD approach I use the border between both states as cut-off.
Table of Contents
1 Causality in economics: the promise of event studies
2 Setting the scene: The literature, the thesis builds on
2.1 Economics and politics
2.1.1 Studies on the specific
2.1.2 Studies on the general
2.1.3 The exogeneity argument
2.2 Choices of firms and people
2.3 Competences of German states
2.4 Policies by Green-Red
2.5 Synthetic control method (SCM)
2.6 Regression Discontinuity Design
3 Quantifying the shock: Econometric analysis of the election of Green-Red in 2011
3.1 Data sources and variables
3.1.1 Data sources
3.1.2 Variables and observations
3.2 Assumptions and plausibility
3.3 Hypotheses and aspects
3.3.1 Material hypotheses SCM
3.3.2 Material hypotheses RDD
3.3.3 Methodological aspects
3.4 Synthetic control method
3.4.1 Approach
3.4.2 Identification
3.4.3 Inference SCM
3.4.3.1 Assumptions
3.4.3.2 Implementation
3.4.4 Results
3.4.4.1 State level
3.4.4.2 County level
3.4.4.3 Summary of the hypotheses
3.4.5 Discussion
3.4.5.1 Relevance of other shocks
3.4.5.2 Another explanation for Bavarian economy after 2011
3.4.5.3 Quality of the control group and general design
3.5 RDD Design
3.5.1 Identification
3.5.2 Approaches
3.5.2.1 RDD: The robust estimator - linear local estimator
3.5.2.2 RDD: The data-driven min-max estimator by Imbens and Wager (2019)
3.5.2.3 RDD - placebo-test
3.5.3 Inference RDD
3.5.4 Results
3.5.4.1 Long-term change of conditional average treatment effect
3.5.4.2 The conditional average treatment effect in the treatment year
3.5.4.3 The commercial building argument
3.5.4.4 The residential building argument
3.5.4.5 Results of the placebo exercise for the residential and commercial buildings in 2012
3.5.4.6 Economic impact on businesses
3.5.4.7 Impact on regional migration
3.5.4.8 The price argument
3.5.4.9 Impact on GDP
3.5.4.10 Summary of the hypotheses
3.5.5 Discussion
3.5.5.1 Business activity
3.5.5.2 Personal activity
3.5.5.3 The comparison of both estimators and manipulation at the cut-off
4 Causality in economics: Did it hold water?
4.1 Shortcomings
4.2 Summary of results
Objectives & Research Questions
The thesis investigates the causal economic effects of the surprise election victory of the Green-Red coalition in Baden-Württemberg in 2011, treating it as an exogenous shock towards more "social democratic" and "interventionist" policy. It employs synthetic control methods (SCM) and regression discontinuity designs (RDD) to analyze impacts on GDP, business activity, and personal economic indicators.
- Evaluation of the causal effect of political regime change on regional GDP using synthetic control methods.
- Application of RDD to analyze short- and long-term impacts on business formation, investment (commercial/residential buildings), and migration at the municipal level.
- Comparative analysis of different RDD estimators (robust vs. min-max) in a geographical context.
- Assessment of the "political shock" hypothesis by evaluating economic outcomes near the border between Baden-Württemberg and Bavaria.
Excerpt from the Book
3.1.2 Variables and observations
For SCM I use as dependent variable GDP per capita. The "growth predictors" are more or less the same as in Abadie and Gardeazabal (2003) and will be elaborated in the analysis part. For the state-level exercise the number of observations is n=16, which is not uncommon in the literature. For the county-level exercise I use at most n=110 counties (of the 140 of BY and BW), for which data across the relevant variables is available. The relevant "growth predictors" for the "counter-factual" county are: price per sqm of land and share of the industrial sector as percentage of the workforce. The pool from which the synthetic controls are constructed is of varying size depending on the method.
In the RDD part the running variable will be the shortest distance to the border between BW and BY (in the univariate case) and the x- and y-coordinates of the centroids of the respective entities in the application of the min-max estimator.7
Table 2 presents the variables used in the RDD part. I standardized both building variables and net-migration by stock of residential buildings and population, to make it consistent.8 Note also that commercial buildings are formally "Nichtwohngebäude" and defined as buildings not primarily used for residential purposes. Still, this would only bias the results if for some reason municipalities to the left and right of the border would build different "kinds" of "Nichtwohngebäude".
Jobs, firms, establishments, corporate tax revenues, commuters and prices are standardized by taking the growth rate. This makes all variables flow variables, as building data and net-migration are itself the absolute growth of stock variables (buildings and population).9
Summary of Chapters
1 Causality in economics: the promise of event studies: Introduces the methodological ambition to derive causal inference in economics from observational studies and defines the "event study" framework used in this thesis.
2 Setting the scene: The literature, the thesis builds on: Provides a comprehensive overview of existing literature on political economy, specifically examining how politics and electoral outcomes affect economic performance at sub-national levels.
3 Quantifying the shock: Econometric analysis of the election of Green-Red in 2011: Details the empirical strategy, explaining both the Synthetic Control Method (SCM) and the Regression Discontinuity Design (RDD) applied to the Baden-Württemberg data, and presents the findings regarding various economic policy indicators.
4 Causality in economics: Did it hold water?: Reflects on methodological shortcomings, summarizes the empirical findings, and concludes that while the thesis provides interesting insights into sub-national economic impacts, the "political shock" hypothesis yielded mixed and largely insignificant causal results.
Keywords
Economics, Causality, Event Study, Synthetic Control Method, SCM, Regression Discontinuity Design, RDD, Baden-Württemberg, Bavaria, Regional Policy, Business Activity, GDP, Political Shock, Election 2011, Economic Growth, Econometrics
Frequently Asked Questions
What is the core focus of this thesis?
The thesis examines the causal impact of the 2011 Green-Red election victory in the German state of Baden-Württemberg on various economic outcomes, utilizing it as an exogenous shock to government policy.
Which primary research themes are explored?
The central themes include the performance of regional GDP, changes in business activity (firm/establishment formation), real estate and residential development, and regional migration patterns.
What is the key research question?
The primary inquiry is whether a "surprise" shift toward more social-democratic and interventionist economic policy is causally observable in regional economic data (GDP, business activity, personal behavior) across different administrative levels.
Which scientific methods are employed?
The author uses the Synthetic Control Method (SCM) by Abadie and Gardeazabal (2003) for macro-level analysis (state and county GDP) and the Regression Discontinuity Design (RDD) for micro-level analysis (municipal data) using the state border between Baden-Württemberg and Bavaria as a cut-off.
What does the main body consist of?
The main body focuses on data description, the establishment of research hypotheses, the detailed econometric estimation procedures (including placebo tests), and an extensive discussion on the robustness and interpretation of the observed effects.
What are the primary keywords describing this work?
Key terms include Causality, Event Study, Synthetic Control Method, Regression Discontinuity Design, Baden-Württemberg, Political Shock, and Regional Economics.
How does the author treat the border between Baden-Württemberg and Bavaria?
The border acts as a geographical "cut-off" for the RDD approach, allowing the author to compare municipalities with similar socio-economic characteristics that are separated only by different state-level political environments.
Are the results statistically significant?
Overall, the findings are largely insignificant for most variables examined, with the important caveat of some significant short-term effects on commercial buildings and evidence of relative regional GDP stagnation, though a definitive "political shock" consensus remains elusive.
- Quote paper
- Martin Schaller (Author), 2019, Geographic RDD design and synthetic control method. Application to the election of Green-Red 2011 in Baden-Wurttemberg, Munich, GRIN Verlag, https://www.grin.com/document/1270711
