Identifying the Effects of Low Emission Zones, ch. 1

1. Introduction

Air pollution is one of the earliest environmental problems recognized in Germany and its regulation has a long history. Already in 1961, then-chancellor Willy Brandt demanded in a speech that “the sky over the [river] Ruhr has to turn blue again”. This speech is today regarded as the starting point for environmental policy in Germany, and a series of laws to secure clean air, water and a healthy environment followed. As air pollution is of transboundary nature, also the European Union (EUhas taken considerable regulative action to counteract it. Particulate Matter (commonly known as “fine dust”) is one of the pollutants having received major attention during the last few years, being suspect of causing a range of adverse health effects, such as respiratory problems or lung cancer.  In a reaction to this, the EU has defined legally binding limit values for the ambient concentrations of small particulate matter (PM10). Ever since these values came into force in 2005, many German cities are struggling with compliance and try to develop effective measures to decrease particle concentrations. One measure that became especially popular is the so-called “Low Emission Zone” (LEZ). LEZs are supposed to reduce PM pollution from road traffic by excluding vehicles with high exhaust emissions from entering certain areas within cities. The first LEZs were introduced in January 2008 and their number steadily increased up to 55 as of today. Even though the effectiveness of the policy is contested and each establishment of a new LEZ is accompanied by heated discussions in media and public, an overall evaluation of the policy with respect to decreasing pollution levels has not been undertaken so far.

This paper  aims to provide such an evaluation by using the method of difference-in-differences, which compares developments in PM concentrations in cities having introduced LEZs with cities that did not. Thus, this study will first of all contribute to the ongoing political debate on LEZs by trying to establish a causal effect from LEZs on changes in PM concentrations. The results can be used to evaluate the question if cities struggling with PM limit values should continue to rely on LEZs as a counteracting measure, and whether their use should be expanded or not. Moreover , the findings can also contribute to the discussion on policy design directed towards other air pollutants, such as for example from even finer particles (PM2.5). Those particles were regulated by the EU in 2010 and cause significant problems to European cities (European Environmental Agency, 2011).

Furthermore , the results of this study can serve as a basis for comparing LEZs with other policy tools to reduce air pollution from road traffic. As summarized by Schmutzler (2011), a variety of approaches to counteract air pollution from traffic exist. Apart from driving bans (for which LEZs are one example), public transportation subsidies or road pricing schemes are common measures to improve air quality. While some empirical evaluations of local public transport subsidies and road pricing approaches exist 2, the evidence on driving bans consists largely of case studies in single cities. This study can thus contribute to closing this empirical gap and lead to a better understandingof different local transportation policies and their implications for air quality.

The paper also contributes to the broader literature on environmental policy evaluation. As noted for example by Ferraro (2009), few environmental programs have been subject to an evaluation based on techniques making use of counterfactual thinking, especially in comparison to other policy fields such as education or labor market policy. In contrast , evaluations of environmental programs rely largely on a monitoring of indicators which do not allow for an establishment of a causal effect. In the same line , Frondel and Schmidt (2005) claim that especially in Germany, environmental policy evaluation is mainly based on plausibility considerations or modeling, which easily miss out on problems caused by selection bias or behavioral changes. While for example the (arguably older) US Clean Air Act Amendments, whichwhich target similar pollutants as the European Air Quality Framework Directive (Council Directive 96/62/EC), have been subject to a number of causal evaluations3, low emission zones have so far only been evaluated by modeling techniques or case studies.[/annotax]

In contrast , this study uses data on PM10 concentrations at the level of individual pollution monitoring stations for the years 2004 – 2010 to test for a causal effect from LEZs on changes in PM10 levels. With a total number of 122 measuring stations, of which 42 are located in cities that implemented an LEZ and 80 in cities that serve as control units, a large panel data set was obtained where the use of different specifications becomes possible. Apart from carefully selecting the control group from the potentially large number of untreated cities, the length of the observation period and different LEZ start dates within the treatment group allow for the use of cities both as treatment and control observations, thus further strengthening the validity of the estimation strategy. Moreover , the large number of units permits testing for heterogeneous treatment effects with respect to different city characteristics. Testing for heterogeneity can help to understand under which conditions LEZs work – or do not work – as a policy tool to reduce air pollution, hence allowing for a more detailed evaluation of the measure. Moreover, the availability of measuring stations located close to, but outside, low emission zones makes it possible to test for spillover effects. The presence of such spillover effects could in turn help to understand the impact channels through which the policy works, as well as whether they are important to take into account if a more general welfare analysis of low emission zones should be done. Such an analysis is out of the scope of this paper, but could be interesting for future research.

The empirical analysis addressing the effects of low emission zones on PM10 concentrations is relevant to the two different limit values that are set by the EU. The first one regulates yearly average concentrations, while the second limits concentrations within a period of 24 hours. The results indicate no effect on yearly average pollution concentrations, but the introduction of an LEZ reduces the number of days in exceedance of the 24 hour limit by approximately 5 days on average. Heterogeneity in this treatment effect is found both with respect to the city size and the initial pollution levels. The results show that the treatment effect is larger for cities with over 500.000 inhabitants and cities struggling with a high number of days in non-compliance with the 24 hour limit, while it is insignificant in smaller and less polluted cities. It is also found that the policy does not cause spillover effects to close-by locations. Thus , it can be concluded that LEZs have a positive, but rather small and local effect on PM levels on days when concentrations are high, but do not reduce average concentrations significantly.

The remainder of the paper will be organized as follows: Section 2 will give a background description of air pollution by particulate matter, the legal requirements, and the policy design. In section 3, a short overview on the theory of local air pollution policies is given while the data and data selection are discussed in section 4. Section 5 provides some graphical analysis of the data and serves as a basis for the empirical strategy outlined in section 6. Results of the evaluation will be discussed in section 7, followed by a discussion and conclusion.


2  See for example the study on public transport subsidies by Chen and Whalley (2010) and studies on road pricing by Atkinson et al. (2009) for London and Eliasson et al. (2006) for Stockholm; both studies admit that the establishment of causality is difficult.

3  See for example Auffhammer et al. (2009) on its effects on PM10 concentrations, Greenstone (2004) on SO2 concentrations, Becker and Henderson (2000) on industrial pollution from Ozone, or Greenstone (2002) on industrial pollution from particles and number of gaseous pollutants.