Truth or Dare? Detecting Systematic Manipulation of COVID-19 Statistics

(with Asli Cansunar and Gozde Corekcioglu), Journal of Political Institutions and Political Economy, Vol. 1: No. 4, pp 543-557.

Which countries manipulate COVID-19 statistics? Does the party ideology of local governors affect the probability of data manipulation at subnational levels? How does democratic quality affect statistical transparency during the pandemic? In this article, we apply election fraud detection methods — various digit-based tests that exploit human biases in generating random numbers — to the daily announced official numbers of new and cumulative coronavirus infections. First, we use digit-based tests to identify countries that likely manipulated their pandemic statistics. We then move on to examine the empirical relationship between democratic quality and data transparency. We find suggestive evidence that data manipulation occurred in China, the United States, Russia, and Turkey. Second, we show that non-democracies, as well as countries without free and fair elections, are more likely to release data that display signs of statistical malpractice.

Featured on Independent Türkçe

Download here

Out of Sight, Out of Mind? Electoral Responses to Proximity of Healthcare

(with Asli Cansunar and Gozde Corekcioglu), R&R at the Journal of Politics

Existing work on electoral returns to public goods provision has investigated changes in government expenditure aggregated at levels that do not have any bearing on geographical access. In this paper, we focus on the political economy of the catchment areas of public services. Rather than investigating the binary relationship between public goods provision and electoral returns within formally drawn borders, we ask whether decreases in walking time to a public service attract votes for the incumbent. Leveraging the Family Medicine Reform in Turkey, which gave rise to an exogenous variation in voter proximity to the free health clinics in Istanbul, we find that communities whose walking distance to the closest clinic decreased voted significantly more for the AKP, the incumbent, between 2011 and 2015. We also show that poorer and healthcare dependent communities were more responsive to improvements in spatial accessibility to the local clinics.

Download here

Politics and Tree Cover Loss: Evidence from High-Resolution Satellite Data

Under Review

Environmental issues have gained saliency in Turkish politics over the last decade, especially after the Gezi Park demonstrations. However, no systemic empirical evidence exists to inform us about the relationship between politics and tree cover in Turkey. Although Turkey witnessed significant tree loss over the last decades, we do not know how much of this damage is attributed to politics. Using high-resolution satellite data, this paper provides the first empirical relationship between local politics and tree loss. The results show that districts with Justice and Development Party (AKP) mayors have higher tree loss by around a combined area of 62 football pitches on average. These results imply that local governments can have a substantial impact on the environment despite their limited effect in the design of environmental policies.

Download here

Indirect favor exchanges and pro-government slant

Working Paper

Previous studies on media capture strategies focused on either coercive measures (e.g., censorship) or direct favor exchanges in which the media outlet adopts a pro-government stance in return for state-advertising. However, most government favors to media owners occur in non-media sectors since advertising markets are tiny, while states frequently grant high-value contracts. This paper analyzes around 315,000 Turkish newspaper articles published between 2007 and 2009 to present evidence on indirect favor exchanges in which the government reciprocated the favorable coverage by granting state contracts to the media owner’s non-media businesses. I constructed two slant measures to show that contractor-owned newspapers are significantly more pro-government due to the opportunities to access state contracts. One such measure uses partisan phrases identified by analyzing a vast corpus of deputy speeches from the Turkish parliament. The other context-aware slant measure was developed from a state-of-the-art neural network model. Moreover, I leverage a legal change in the procurement law granting the government more discretion in the construction sector to show that contractor-owned newspapers became even more pro-government, and started using more pro-government partisan phrases after the change.

Feel free to contact me ( for the most recent draft

Institutional Gridlock and Democratic Backsliding: explaining popular support for aspiring autocrats

Working Paper

In recent years, the world has seen a wave of democratically elected leaders move their countries in undemocratic directions. Why do people support leaders who remove checks and balances? I argue that aspiring autocrats are more likely to gain popular support when they present these institutions as obstacles to getting things done. In doing so, aspiring autocrats exploit a critical tension between the possibility of gridlock and the abuse of power, which is inherent in democratic institutions. An original survey experiment conducted in Turkey supports these arguments. More interestingly, respondents perceive the aspiring autocrats’ gridlock justification as a pro-democratic attempt to remove the obstacles to a policy-responsive regime. These results show that democratic backsliding is strategic, and its leaders exploit a tension in democracy that makes it harder for citizens to perceive the threat they face.

Feel free to contact me ( for the most recent draft

New Interest, New Measures, Old Problems: An Analysis of the Latent Dimension(s) of Democracy

(with Scott deMarchi, Jeremy Springman, Mateo Villamizar-Chaparro, Erik Wibbels), Working Paper

The third wave of autocratization has brought renewed attention to the study of regime type. This attention has been accompanied by a proliferation of new and more nuanced measures of democratic characteristics. We combine recent approaches using machine learning for dimensionality reduction with new democracy measures to investigate the latent dimensions of democracy. Using data from eighteen distinct democracy indices, we explore variation in theoretically-motivated dimensions of democracy across historical periods.

The Islamic Waqf: Instrument of Personal Security, Worldly and Otherworldly

(with Timur Kuran), Working Paper

Until the modernizing reforms of the 19th century, the Islamic waqf played a massively important role in the economy of the Middle East, the Balkans, and North Africa. Formally, it was a trust that an individual established in perpetuity to provide some service through the income of endowed assets. Most case studies of this institution focus on the largest waqfs, and especially on those founded by ruling families. For this reason, they give the impression that the waqf was primarily an instrument of social service, charity, generosity, and piety. Using a huge original data set covering the period 1500-1900, we challenge this common generalization. The data set draws on the records of Islamic courts in Istanbul—over the analyzed period, the Ottoman capital and the leading economic hub of the Eastern Mediterranean. The city’s largest waqfs were established by high elites to provide social services now supplied by municipalities or charitable corporations. Examples include those supporting hospitals, colleges, and major soup kitchens. But the typical waqf, whose endowment and architectural footprint were more modest, was established primarily to provide material security to its founder. As a rule, piety played a much smaller role than the literature tends to convey. In a social setting characterized by weak property rights and a biased legal system, economically vulnerable demographic groups tended to use waqfs as instruments of material security. Such groups included commoners and females.