According to the theoretical literature, it is reasonable to assume that air pollution enters into the utility function of potential house buyers. It is therefore no surprise that hedonic house price models that incorporate environmental variables among the set of explanatory variables are becoming increasingly popular.
In the context of the hedonic price theory, the traditional approach to this problem has been to use the housing market to infer the implicit prices of these nonmarket goods (see Freeman 1993 for a comprehensive review of property value models for measuring the value of environmental amenities). Under standard assumptions of perfect competition, information and mobility and the maximization of well-behaved preferences, hedonic theory unambiguously predicts that the implicit price function relating housing prices to an environmental amenity will be positively sloped, all else equal.
But the question is that empirical research does not confirm the hedonic theory. Smith and Huang’s (1995) meta-analysis suggested that a one-unit reduction of PM10 (mg/m3) results in a 0.05-0.10 percent increase in property values. Boyle and Kiel (2001) found that air studies produced mixed results and posited that measurement factors are not generally known to homebuyers. Jackson (2001) offered no final observations on the consistency of findings and he called for a more systematic study. Simons and Saginor (2006), who included air pollution together with concentrated animal feeding operations, obtained a different sign in the coefficient of the variable depending on whether the model included positive amenities or not.
In summary, the literature concerning the effects of contamination on property values reveals that the effect of air pollution on property value is far from being conclusive. What is more, there are serious doubts that air pollution significantly affects the price of properties. Additionally, the study type may also generate different results.
Recently a successful line of research has emerged that includes the spatial argument in the hedonic specification. As Straszheim (1988) stated many years ago, it may not be appropriate to assume that the implicit prices of housing attributes are stationary ?we use the same term as Carruthers and Clark (2010)? across geographical space, even within a big city. The rationale behind this is that, on the supply side, homes
near each other tend to be similar and, on the demand side, homebuyers regularly emulate one another’s behavior. The result is a process of spatial interaction among market participants, which at least suggests that the first stage hedonic price function should be modified to include a spatial lag of its dependent variable (Anselin, 1988; Anselin and Bera, 1998). This spatial lag can be interpreted as a flexible fixed effect that absorbs the existing and unobserved spatial correlation structure of supply and/or demand. Recommended literature that considers the spatial argument in the specification of the hedonic model includes the pioneer works of Can (1990), Can (1992), Kim et al., (2003), Theebe (2004), Brasington and Hite (2005), Anselin and Le Gallo (2006), Anselin and Lozano-Gracia (2008) and Osland (2010), among many others.
But again, results are not conclusive. In the best of cases, clean air has a negligible influence on housing prices, which does not fit the hedonic theory.
What is wrong in the preliminary studies? Maybe nothing is wrong but, as stated in Chay and Greenstone (1998), exogenous differences in air quality are extremely difficult to isolate, because the “true” relationship between air pollution and property prices may be obscured in cross-sectional analysis by unobserved determinants of housing prices that co-vary with air pollution. For example, areas with high levels of pollution tend to have well educated populations with higher per capita income and population densities. Economic activity is also a driving force in the determination of property values, but differences across space in the level of activity may also be behind changes in the level of pollution. Of course, the above circumstances lead to a spurious positive relationship between pollution and property values. When Chay and Greenstone use the conventional cross-sectional estimates of the relationship between property values and PM10 they conclude that the relationship is weak, unstable and indeterminate.
The other possibility is that the concept of air pollution that enters into the utility function of potential house buyers is perceived pollution rather than objective (measured) pollution. It is important to bear in mind that indirect methods like the hedonic strategy are based on actual transactions and empirical measurements and assume that decision makers possess all the necessary information and always act rationally, attempting to maximize their personal utility. However, when in the process
of deciding their location, house buyers weigh up one property or location against another, their choices are not necessarily rational. As Berezansky et al. (2010) states, their rationality is essentially bounded by available information, limited processing capacity, errors of judgment, and an inability to foretell the future. Additionally, perceived pollution is not an instantaneous concept like, at least during a short period of time, the price of a dwelling, but one that is created over a long period and needs another long period of time (due to the above mentioned factors) to be modified.
In our opinion, one more question remains to be answered: When measuring air quality in large cities, what do house buyers (and citizens in general) understand as clean air? Our previous research (see Mínguez et al., 2010) suggests that, at least in European cities, people decide the location of their house according to a range of factors, including: personal or family income, commercial properties, communications, schools, medical centers, etc. but not according to the level of pollution, as living in the city alone implies a polluted environment. What do they mean by a “clean air” location in a large city? Probably a neighborhood located near parks and open areas. This suspicion has been confirmed by Berezansky et al. (2010) in the case of the City of Haifa. This is another important reason in favor of using subjective air pollution variables in spatial housing price hedonic models.
In brief, house buyers do not decide how much they are willing to pay for clean air on the basis of the complete information provided by monitoring stations (they probably do not even know how to interpret it), but rather according to their perceptions, perceptions that are not instantaneous but generated over a long period of time. Additionally, in the case of using environmental interpolated variables as explanatory variables, “objective" and "subjective" air pollution maps could be quite different.
As a consequence, if there is not a strict positive correlation between objective and subjective measurements of pollution, the literature, regardless of whether or not it includes the spatial argument, is not using the right pollution variable or index.
The above statements suggest classifying neighborhoods according to subjective environmental measures (resident perceptions), which could be enormously more
informative than objective environmental measures when it comes to accounting for the willingness to pay for clean air or other environmental factors.
We have focused our empirical analysis on Madrid (Spain). There are several important reasons for choosing Madrid as a study case: (i) Most of the empirical research in the literature refers to American cities. This is the main reason; (ii) population is highly concerned with the environment in general and air quality in particular (iii) construction, particularly residential building, is of great importance to the overall economy; and (iii) it can be said that in Madrid there is almost perfect information about air quality all over the city (both excellent ratios monitoring sites/population and monitoring sites/surface), which makes it possible to ignore the problems related to how much is known about air quality variables, because, as is well known, the impact of air quality variables on the hedonic price function depends on how much is known about them (Clark and Allison, 1999).
We constructed a massive data base in 2009 including 11,796 dwellings (after depuration). Apart from price, we have registered more than thirteen core variables for each along with subjective and objective air pollution indicators representative of the level of pollution.
After this introduction, section 2 is devoted to both briefly delineating the way we propose to construct APIs based on interpolated objective pollution measures and the construction and comparison of both objective and subjective environmental air quality maps. In section 3 we briefly describe the Spatial Durbin Model, which is the specification that we use in this article to measure the impact of clean air on housing prices. Section 4 is devoted to the case study: Madrid City. First we give some details about air pollution (both objective and subjective measures) and the housing sector in the study area. Second, we comment on both the air quality and housing data sets used in this research. Third, we report the main results obtained from the...