Work in progress

Bill Complexity and Voting [latest draft]

Current research on the quality of politicians identifies education as a significant factor leading to good performance. However, the role of education still remains to a large degree unexplored. This paper argues that one of them is the ability to understand complex agenda. Using a theoretical framework, I model the voting behavior of parliamentary politicians and study the response to changes in the complexity of the underlying bill. I identify three groups of legislators. First, those with low information costs are not affected. Second, legislators with high information costs deviate from the party-line more frequently with higher complexity. Last, some legislators selectively acquire additional information when in doubt about signal. Therefore, for these, complexity increases information acquisition and results in a decline in party-line breaks. I verify these results empirically using records from the Danish parliament. Using two complexity measures, I show that, for low-educated legislators, an increase in complexity of the bill report is associated with a higher probability of party-line breaks. The opposite holds for legislators with secondary education. For university graduates, there is no variation associated with changes in complexity. Thus, these findings correspond with the theoretical predictions of the model and suggest education is a relevant factor for complex information processing in a parliamentary setting.

School Performance and Retrospective Voting

with Nicola Maaser

Brief summary: We use Danish register data to link polling-station-level results from local elections with school grades of children of local electorate in order to estimate the effect of bad school performance on electoral punishment (by average income and other polling-area characteristics).

Abstract coming soon

Populism and Inequality: Does Reality Match the Populist Rhetoric?

with Andrea Sáenz de Viteri, Martin Rode, Christian Bjørnskov

Abstract coming very soon -> a brief introduction to the paper by Christian (in Danish or Google Translate)

Loan Provision Methods for Microcredit Group Lending

with Karel Janda

(based on my MA thesis)

Over the years, the lending procedures of microcredit have evolved. The original joint liability group lending with simultaneous financing (loans provided at once) has been replaced by sequential financing (loans provided one by one). Moreover, recent studies suggest individual liability lending in groups to be the optimal framework. While there are numerous theoretical studies, none presents a comparative analysis of relevant trade-offs across these models. In this study, we model three schemes within the framework by Van Tassel (1999). Our findings prove that, in the presence of information asymmetry, group lending with joint liability dominates individual liability lending. Furthermore, the interest rate of the sequential model is more sensitive to changes of monitoring costs or opportunity costs of capital than in the sequential setting. On the contrary, sequential approach allows for higher degree of within-group heterogeneity of loan sizes. None of the models strictly dominates others in terms of profit and lower interest rate. Therefore, the optimal approach is primarily determined by the characteristics of borrowers.


Practical Introduction to Mathematics, Logic, and Statistics for Sociology and Social Sciences, Karolinum, Charles University Press, 2019

(In Czech: “Základy matematiky, logiky a statistiky pro sociologii a ostatní společenské vědy v příkladech”), with J. Hendl, M. Moldan, T. Ranošová, J. Siegl, and J. Žáček

Introductory textbook for courses at the Institute of Sociological Studies of the Faculty of Social Sciences of Charles University

Smoking Czechs: Modelling Tobacco Consumption and Taxation, Prague Economic Papers, 2019, 28(1), 3-29.

with Karel Janda

(pre PhD - based on my BA thesis)

We model the future tobacco consumption, size of smoking population and governmental tax revenues in the Czech Republic. The main model assumption states that smokers determine their future tobacco consumption behaviour as adolescents. Further assumptions make the model applicable to the data from the Czech National Monitoring Centre for Drugs and Drug Addiction. Future teenage smoking rates and average consumption are the inputs to the model; consumption growth coefficients for each age category are estimated using zero-inflated negative binomial regression. Several scenarios are built to model possible developments, including extreme cases. All our scenarios show that all model outcomes are going to grow until 2028 in a very similar pattern. In particular, the projected number of smokers in 2028 is by 4-8% higher than in 2013, the total daily tobacco consumption and tax revenue by 7-26%. This increase is induced by ageing of large birth cohorts.