Abstract: The consumer cost of carbon pricing is globally regressive, more so across countries than within—it falls harder on average consumers in poor countries than on poor consumers in average countries. I show this using a novel, global approach to estimating the consumer incidence of carbon pricing. On the demand side, I allow consumption to differ both between countries and across income levels within them. On the supply side, I model substitution of inputs along global value chains. I identify all model parameters from data on bilateral trade flows. Similar to a global carbon price, the introduction of the EU Emissions Trading System (ETS) in 2005 was likely regressive. The results are different for a carbon price on traded goods. The cost of a hypothetical Border Adjustment to complement the EU ETS follows an inverted U-shape—the richest and the poorest consumers in the EU incur the largest cost.
Abstract: Does air quality influence road safety? We estimate the effect of increased air pollution on the number of road traffic accidents in the United Kingdom between 2009 and 2014. To address concerns of spurious correlation we exploit atmospheric temperature inversions as a source of plausibly exogenous variation in daily air pollution levels. We find an increase of 0.6 - 1.0% in the number of vehicles involved in accidents per day for each additional 1mg/m3 of PM2.5. The finding suggests that less safe roads may present a large and previously overlooked cost of air pollution.The results are robust to a number of specifications and across various sub-samples.Working Paper: GRI Working Paper Series, WP 251, October 2016. Newer draft, February 2018Media: Financial Times, The Times, The Independent, The Telegraph, The Economic Times, The Hindu, BBC Radio 4, and others.
Abstract: In this paper, I investigate the relationship between income inequality and the carbon dioxide (CO2) content of household consumption. I quantify household carbon by linking household expenditure to carbon content input-output analysis. I then estimate Environmental Engel curves (EECs) that describe the relationship between household income and CO2 in the United States between 1996 and 2009. A second-degree polynomial specification in income approximates well the fit of more flexible nonparametric models. Using these parametric EECs, I decompose the within-year household carbon inequality as well as the evolution of household carbon over time. In both cases, income is a main driver of carbon consumption. Finally, I describe a potential “equity-pollution dilemma”—progressive income redistribution may raise aggregate emissions. I estimate that progressive income transfers may raise household carbon by 5.1% at the margin and by about 2.3% under complete redistribution. Working Paper: GRI Working Paper Series, WP 285, November 2017; CCCEP Working Paper Series, WP 319, November 2017. Abstract: Many empirical studies have examined various determinants of crime. However, the link between crime and air pollution has been surprisingly overlooked despite several potential pathways. In this paper, we study whether exposure to ambient air pollution affects crime by using daily administrative data for the years 2004-05 in London. For identification, we mainly rely on the panel structure of the data to estimate models with ward fixed effects. We complement our fixed effects analysis with an instrumental variable approach where we use wind direction as an exogenous shock to local air pollution concentrations. We find that elevated levels of air pollution have a positive and statistically significant impact on overall crime and that the effect is stronger for types of crime which tend to be less severe. We formally explore the underlying mechanism for our finding and conclude that the effect of air pollution on crime is likely mediated by higher discounting of future punishment. Importantly, we also find that these effects are present at levels which are well below current regulatory standards and that the effect of air pollution on crime appears to be unevenly distributed across the income distribution. Overall, our results suggest that reducing air pollution in urban areas may be a cost effective measure to reduce crime and that air pollution forecasts can be used to improve predictive policing in cities.Working Paper: IZA Discussion Paper Series, DP 11492, April 2018; GRI Working Paper Series, WP 295, April 2018. Media: The Independent, iNews, The Economic Times, WSJ Real Time Economics blog.
  • "Strategic Thinking in All-Pay Auctions: A Level-k / Cognitive Hierarchy Analysis."
Abstract: This paper analyses strategic thinking in initial responses to all-pay auctions. It is shown that a level-k / cognitive hierarchy (CH) model of non-equilibrium strategic thinking can account for the systematic deviations from (Bayesian) Nash equilibrium observed by three experimental studies. These step-level models of reasoning reproduce the bimodal distribution of bids observed in all-pay auction experiments with common values and complete information. An extension to the standard level-k / CH model is proposed for incomplete information settings, limiting the maximum bids of non-strategic types to their private value signals. Such a truncated level-k / CH model is shown to account for the discontinuous bidding behaviour observed in all-pay auctions with independent private values and costs. Maximum Likelihood estimation consistently favours large population shares of low-level non-equilibrium types over equilibrium ones.September 2016. Draft available on request.


Abstract: This paper assesses the residential segregation of German immigrants from Turkey, Italy, the Balkans and eastern Europe with a special focus on the link between social and ethnic segregation. Microdata from the German Socioeconomic Panel Study (SOEP) are used. A new dataset provided by the microm Micromarketing-Systeme und Consult GmbH makes accessible information on participants immediate residential environments at the micro-neighbourhood level where an average neighbourhood contains only eight households. Substantial levels of residential isolation in the form of own-group overexposure are found for all four migrant groups. Based on previous research, an enhanced methodology is proposed to measure the effect of socioeconomic neighbourhood sorting on ethnic residential segregation. It is shown that differences in income, education, language skills and village/city size have the potential to account for 2984 per cent of the residential isolation of the four migrant groups.