Feel it or measure it. Perceived vs measured noise in hedonic models : Cahier de recherche

Baranzini, Andrea ; Schaerer, Caroline ; Ramirez, José V. ; Thalmann, Philippe

The aim of this paper is to compare the use of scientific and perceived noise measures in a hedonic model framework. Although in theory the use of subjective variables is recommended, most of the empirical applications use scientific noise variables. Using three different databases, we are able to obtain a relative large sample of about 2 800 apartments located in Geneva, Switzerland. The sample... More

Add to personal list
    Summary
    The aim of this paper is to compare the use of scientific and perceived noise measures in a hedonic model framework. Although in theory the use of subjective variables is recommended, most of the empirical applications use scientific noise variables. Using three different databases, we are able to obtain a relative large sample of about 2 800 apartments located in Geneva, Switzerland. The sample contains both scientific measures and perceived noise data, as well as structural and accessibility variables. However, in order to compare scientific and perceived noise measures we need them expressed in comparable units, e.g. dB(A). Since no such data is available, we refer to the acoustic literature to transform a categorical variable on perceived external noise into a measure directly comparable to a scientific measure of noise in dB(A). We show that a clear relationship appears between the scientific measure of road traffic noise and the transformation of the perceived external noise in dB(A). From this relationship, it is interesting to note that: 1) There is a great variability in the perception of noise when the scientific noise level is relatively low; and 2) The systematic bias of subjective overestimation of low probability events that has been highlighted in the risk literature also parallels with our findings here concerning the perceived exposure to external noise. Having both noise measures in comparable units, we are able to analyse their performance in the context of the hedonic model. We conclude that for moderate to high noise levels, the scientific measures of noise safely approximate the individual perception of it. This last result is very promising from a policy-making point of view, for which the use of scientific measures are more convenient and transferable than subjective perceptions.