La regulación de la subcontratación en Chile

Saelensminde (1999) applied the CE exercise with attributes such as in-vehicle time, fuel cost, level of noise and air pollution, using different exercises based on different combinations of the attributes. The author provided definitions of air and noise pollution along with other environmental factors in addition to the information on damage and nuisance caused by road traffic. Using demand for alternative fuels as the base and the different fuel costs as the monetary measure, the author employed different exercises based on varying combinations of the attributes. Respondents were informed on the definitions of local air and noise pollution that comprised the study and the attributes air and noise pollution were represented in the CE exercise as a percentage change (20, 40 and 60% reductions compared to the current level).

While the noise variable had a significant value with the correct sign, the author found the value to be higher for lower level of noise reduction compared to larger reductions, inducing the conclusion of declining marginal utility. Using the WTP estimated from these exercises, the author computed the WTP for a 50% reduction which was further multiplied by the total number of households in Oslo to obtain the total WTP. In order to compute the ‘WTP per annoyed person’, the total WTP was divided by the number of people annoyed by noise, which was captured during the survey through the question on respondent’s experience of various forms of nuisance. This study was a pioneering application of CE for air and noise pollution in Norway. However, a major shortcoming of the application lies in the representation method of these attributes through percentage change as it is difficult for respondents to understand the level of environmental change based on this representation form. The authors rightly acknowledged this problem however and suggested the use of ‘no choice’ or ‘don’t know’ alternatives in the choice experiment in order to elicit any preference uncertainty caused by the lack of attribute level understanding.

To assess the benefits of traffic calming measures from speed reduction, noise and community severance, Garrod et al. (2002) conducted a SP study on road traffic

noise. Three noise levels were incorporated in the choice experiment – 60, 70 and 80 dB which were presented to the respondents through audio recordings of traffic noise. In order to accomplish this method of representation, noise was measured at the pavement and the nearest residence to the line of traffic. The authors allowed for interaction term between the distance from the traffic and noise levels showing that the benefits derived from increasing the distance and having a lower noise exposure exceeded the benefits derived from avoiding the increase in the effective speed limit. An evident advantage of this study was the possibility for respondents to hear the noise levels which would have aided their understanding. However, the effect of respondent’s socio-physical characteristics such as age or lifestyle as well as the ambient sound levels during the survey could have an effect on the interpretation of the experimental noise level. Moreover, the level of loudness and annoyance could also possibly vary. However, this study offers an interesting alternative to noise representation compared to other exposure based methods and especially the use of percentage reduction.

Wardman and Bristow (2004) applied the SP method to evaluate air and noise pollution under residential choice context in Edinburgh. While the authors employed proportionate change as well as the location method (where respondents were offered different locations based on the level of air quality) to represent air pollution, noise levels were again represented as percentage change in the SP choice scenario. Though the authors focussed more on the size, sign and level effects, to examine whether gains and losses are valued differently and whether there is a significant effect dependent on the size of the change, they also noted the difference in perception caused by different methods of representation. In case of noise with 50% variation, the authors found that respondents valued increase in noise higher than reduction while in the case of 100% change, a large variation in interpretation was found in the case of 100% improvement ‘twice as good’ than in 100% deterioration ‘twice as bad’ prompting the authors to conclude that respondents did not interpret this form of representation in a way that was intended.

The paper thus further emphasised that the percentage reduction method of noise representation is not very conducive for respondent’s understanding of the attribute.

While Saelensminde (1999) conducted the CE application on noise with a relatively new form of representation thus providing insight on its effect, the results showed that the method employed posed a significant challenge on respondents’

understanding. The application of CE to value air and noise quality in residential context by Wardman and Bristow (2004) formed one of the earliest applications in this field. Moreover, the authors used two different methods of representation for air quality along with different forms of representation in terms of gains and losses.

However, in this case too, the representation of noise as a percentage change proved to be a shortcoming as a clear variation in respondents’ understanding in the case of improvement and deterioration was not observed.

While Wardman and Bristow (2004) applied the location method only for air pollution, this method in residential CE survey to evaluate noise from road traffic in Portugal can be found in Arsenio (2002) who represented noise as a perceived stimuli for different apartment locations. Thus in this case, respondents’ noise perception for the specific apartment locations was considered in the modelling process. The CE was held in the context of residential choice with different characteristics of the apartment based on view, noise, sunlight and housing service charge levels. With the location method applied in these studies, the author asked respondents to provide a numeric rating (from 0-100) for their perception on view, noise and sunlight attributes. Along with the location method for road traffic noise, the author also measured the physical noise level which was incorporated in the modelling process. Using binary and mixed logit models, Arsenio (2002) found that models based on respondent’s perceptions statistically outperformed those based on physical noise measure. Moreover, Arsenio et al. (2002) reveal that models based on indoor noise measurement outperformed model on outdoor noise measurement, implying that outdoor noise measure is not a true proxy of respondents’ noise perception. Apart from the relative significance of the location method compared to the physical noise measurement, it was also stated that households at the quieter façade had a higher marginal value for quiet. The application of this representation method for noise valuation was one of the first examples of using spatial references to convey noise levels. This method thus provided a unique approach to noise representation.

Carlsson et al. (2004) evaluated aircraft noise in Sweden using a residential postal CE survey based on the aircraft frequency at different times of the day and different days of the week. Two separate CE surveys were conducted based on increase and decrease in take-off as well as for the weekday and weekend. In case of the increased aircraft frequency version, the authors used compensation as the monetary measure while the payment was used in case of noise decrease. Using a random parameters logit model with normal distribution and accounting for preference heterogeneity and household characteristics, the authors found that respondents were likely to choose the opt-out choice if no one from the household flew from the airport, in the increased take-off version while less likely to do so in case of decreased take-off. Moreover, households with a detached house chose

‘current situation’ in fear of changes. While the main objective of the paper was to evaluate the value of aircraft noise based on aircraft movement at different times of the day and different days of the week, the authors also laid emphasis on the ‘opt-out’ alternative indicating that most residents were satisfied with the current state of the airport. However, a shortcoming associated with the study lay in the different factors considered to capture preference heterogeneity. While the authors considered the socio-economic characteristics as indicative of preferences and the usage of the airport in the household, factors such as the number of hours spent at home, the presence of children or the elderly and the perceived level of annoyance and attitudes towards noise, which would have been important factors in choice, were not considered in the study. Nonetheless, this study employed a different approach to noise representation through the use of, what will be termed subsequently, as the ‘proxy method’.

In a study conducted to evaluate aircraft noise from Manchester and Lyon airports, Bristow and Wardman (2006a) applied the stated choice (SC) as well as the priority ranking (PR) methods in a residential survey. The PR method comprised of various quality of life attributes along with aircraft noise (which was specified as aircraft movement during day and evening time) which the respondents were asked to rank in order of preference in terms of improvement from the current state. The quality of life variables included ‘traffic noise at home’ which was represented as

‘extremely noisy’, ‘very noisy’ ... ‘not at all noisy’.

Using the SC method which comprised of eight binary choice scenarios, the authors conducted the survey by characterising the noise level using the ‘proxy method’

which was based on the type of aircraft and the frequency of movement. In the first SP exercise (PR), noise was represented as total aircraft movement across the day and evenings while in the other case it was represented as aircraft movements (based on different aircraft types) across different time periods. Developing logit models with these two representations as well as with physical noise measure, the authors found that higher perceived noisiness increased the WTP in case of Manchester airport while in case of Lyon, people living in quiet areas had a higher WTP to reduce noise. In case of the PR method, models were formed based on aircraft movement as well as the Leq measure for day and evening time. The authors found that most of the PR models improved by substituting aircraft movements by Leq measure.

This study applied the ‘proxy’ approach in the SC experiment, already mentioned in Carlsson et al. (2004) albeit with slight variations. While the variations to the method were an interesting extension to the previous method, several shortcomings can be observed in the noise representation technique for the SC method. Using the SC method, the authors note that though the current level of noise is experienced by individuals, the other levels of aircraft movement and noise are only experienced to the extent that they may occur in other time periods. While this has been cited as a shortcoming by the authors in the paper, this in itself is not a major problem as long as the levels are realistic. A more significant problem lies in using different aircraft types as it might not be possible for respondents to know the variation in noise level based on the aircraft type without that information provided by the interviewer.

In another paper reporting the comparison of SC and PR methods to evaluate aircraft noise, Wardman and Bristow (2008) represented noise as the number of different aircraft movement in an hour for specific time periods for the SC exercise.

The PR method consisted of noise represented as different aircraft movement every hour for evening and day time, along with other quality of life variables or as categorical semantic levels such as: extremely noisy, very noisy, moderate, slightly noisy and not at all noisy. In order to examine the different methods of aircraft noise representations applied in the PR method, the authors compared the monetary

values obtained under the aircraft movement scenario as well as for the different categorical values, concluding that the two different means of presenting aircraft noise exhibit an encouraging degree of similarity. While the authors examined the applicability of the semantic method of noise representation by comparing the values obtained from the aircraft movements model, they failed to examine which method is more suitable for noise representation based on the relative ease in respondent’s understanding. However, the results from this study do indicate that the linguistic representation of noise can provide logical valuation estimates.

Comparing the valuations obtained from the SC and the PR methods, the authors found valuation variations ranging from ratios of 1.5 to 3.0 for Lyon and Manchester which when examined across a range of reasons was thought to arise from strategic bias caused in the SC exercise.

In Bristow and Wardman (2006b), application of the ‘proxy method’ was again cited with Bucharest airport along with Manchester and Lyon. In this case, the authors applied the method to conduct ‘within period’ (WP) and ‘between period’

(BP) SP experiments. In case of the ‘within period’ experiment, trade-off was required for different aircraft movements within a given time period while in the case of the ‘between period’ experiment, more emphasis was laid on the flight frequency for the whole range of time periods across the two options. Reporting results from the same study as outlined in Bristow and Wardman (2006a), several similarities can be found in the two; however, an important point of shortcoming in Bristow and Wardman (2006b), lies in the application of the BP method where the CE employs eight different time periods with flight frequency in each scenario, with a potential to cause increased task complexity. This caused the authors to adopt a ‘priority evaluator’ type question where current flight, deterioration and improvements were offered to the respondents with different weekly tax to elicit 25 rankings on what they most prefer. Moreover, in addition to some lexicographic choices with the BP experiment, respondents were unable to rank the alternatives in logical order, implying that this is a difficult method for respondents to understand.

Using the ordered logit model for the BP method and the standard logit method for the WP exercise, high degrees of similarity were found between the relative valuations by time period across the WP and BP results in Manchester while lower similarity was found in case of Lyon. However, though the authors mention the

difference in the modelling method employed across the two representation methods, the different model structures are not explicitly specified.

To evaluate aircraft noise in the context of airport relocation, Thanos et al. (2006) conducted several residential SC studies around two different airports in Greece.

As the study context involved opening of the Eleftherios Venizelos airport and closure of the Hellenikon airport, the authors employed ‘noise as of now’ and ‘not subjected to aircraft noise’ as the means of noise representation in the SC experiment. Pooling the data across three different SC experiments and developing different logit models with income and socio-economic characteristics, the authors found that sensitivity to aircraft noise varies for the different airports. Moreover, differences in sensitivity was also observed in case of socio-economic characteristics and the level of reported noise annoyance, with respondents who considered noise annoyance in their neighbourhood extreme, had a greater sensitivity to the ‘no aircraft noise’ variable. This study adopted an extreme variation scenario with ‘noise’ or ‘no noise’ as the alternatives. While this approach is suitable for the specific case-study due to its context, the application of this method could not be relevant in other studies where such representation of noise is either unrealistic or infeasible. Incorporating different socio-economic and attitudinal characteristics in the SC model, the authors found that these factors affect noise valuation and the willingness to pay amount. However, this induced the authors to conclude that these factors influence aircraft noise annoyance (rather than aircraft noise valuation), thus causing confusion between aircraft noise annoyance and aircraft noise valuation.

In a unique application of SC methods to evaluate rail noise, Nunes and Travisi (2007) conducted a residential CE survey to assess rail noise annoyance in Italy.

The key attributes employed in the SC exercise included noise dB reduction, height of the trackside barrier, investments in trains and tracks technology and the cost of the noise mitigation programme. While reduction in the dB (A) levels were explained to the respondents using visual aids, the main method of representation along with dB (A) in the SC experiment consisted of the specification of the distance of the house from the railway. Thus, for example, in case of a 9-11 dB noise reduction, the level was described as ‘10 times increase in distance of the

residence from the railway’. The authors found that rail noise was ranked more important by the respondents than traffic noise but less important than air pollution.

Preferences of respondents for different noise abatement program was found to be sensitive to the type of instrument, noise abatement target and the price associated with each policy. Moreover, the authors found that the intermediate level of noise reduction (12-14 dB (A)) was preferred more than the other levels. While this is one of the first studies applying SC to evaluate rail noise abatement, the representation of the noise variable can be difficult for respondents to understand as the noise level is characterised by dB (A) and different levels of distance of the residence from the railway, both of which could be difficult to imagine. However, the authors have attempted to alleviate some of these effects by providing visual aids in relation to the noise levels, prior to the SC experiment.