Title: Structural labour supply and preferences for job attributes.
Other Titles: Structural labour supply and preferences for job attributes.
Authors: Jara Tamayo, Holguer
Issue Date: 26-Nov-2013
Abstract: For a long time, economists have considered earnings as the main variable of interest in labour market research. However, jobs are characterised by a variety of attributes which individuals consider important. Once this multidimensional nature of jobs is taken into account, the issue of individual’s preferences for different job attributes comes soon into scene and with it the question of how to derive information about such preferences. Traditionally, economists have opted for a revealed preference approach in order to elicit people’s preferences. The idea behind the concept of revealed preferences is that information about individuals’ preferences can be recovered by analysing the choices and decisions those individuals make in the market. Among the most popular models in economics, based on the idea of revealed preferences, are hedonic price models (see Rosen, 1974). Such models have been widely used in the context of housing markets not only to derive preferences for housing attributes but also for environmental amenities such as air or water quality. In the context of labour economics, hedonic wage models have been mainly used to test the theory of compensating wage differentials but to a lesser extent to recover preferences for job attributes. Another type of models, also based on the idea of revealed preferences, are discrete choice models. As in the case of hedonic models, discrete choice models have been mostly used for the valuation of environmental resources but much less, and only recently, in order to analyse preference information for job attributes. Finally, a different approach used to recover preferences is based on information from subjective well-being data. Analyses based on subjective well-being have become increasingly popular in economics, mainly due to the fact that subjective well-being measures have been used as proxies for utility. In this dissertation, we focus on the potential of discrete choice models of labour supply to derive preference information for different job attributes. Most discrete choice labour supply models use only wages and hours of work as variables of interest. Here, we propose to extend these models in order to incorporate additional job attributes in the analysis. In particular, we incorporate job insecurity as an additional job attributes affecting individuals’ labour supply decisions. Moreover, most studies interested in deriving preferences information have used one specific approach, without necessarily making the link between the results obtained using different methods. Such comparison is not straightforward. In fact, differences in the underlying assumptions of the approaches and the econometric methods used render the results not fully comparable. Moreover, there might even be more fundamental distinctions among the underlying concepts related to the different approaches. Kahneman (2000) makes a distinction between two concepts of utility related to the approaches mentioned earlier. On the one hand, the revealed preference approach is related to the concept of decision utility, defined in terms of individuals’ choices. On the other hand, the subjective well-being approach is related to the concept of experienced utility, defined in terms of the levels of pleasure and pain an individual experiences from different events. A comparison of approaches related to these different concepts of utility can be useful to understand to which extent these concepts are related to each other. Based on behavioural microsimulation models, widely used in the context of policy evaluation, in Chapter 2 we propose extensions, which allow taking into account the effect of different nonpecuniary job attributes on labour supply. Due to the fact that job insecurity has been considered one of the most important aspects at work in the literature and thanks to the availability of such information in many datasets, we use this variable in our analysis. In this sense, in addition to extending discrete choice models of labour supply, we contribute to the literature on the consequences of job insecurity on labour market outcomes. The estimation of a conditional and a nested logit model of labour supply show that nonpecuniary job attributes, in our case job insecurity, play a significant role in labour supply decisions. In particular, a decrease of job insecurity increases the probability of participation by around 2%. This is a finding that deserves further analysis as employment policies could turn to improvements of (nonpecuniary) working conditions in order to create incentives for labour market participation. Our analysis further shows, that an assumption often used in discrete choice labour supply models needs particular attention when incorporating additional job attributes in the analysis. Independence between wages and working conditions needs to be tested in order to construct the discrete choice alternatives for the extended structural labour supply models. The relationship between wages and working conditions has been widely analysed in the literature, following the idea of compensating wage differentials. In Chapter 3, we showed that compensating wage differentials can be found for aspects such as job insecurity and hard-demanding work. Evidence of the presence of a small compensation for job insecurity was found with three different datasets used in this dissertation. Moreover, we illustrated how hedonic wage models can be used to recover worker’s preferences for job attributes, based on Rosen’s two-step procedure (Rosen, 1982) and recent work by Ekeland, Heckman and Nesheim (2004). Individuals’ preferences for job attributes become, an issue when we consider a model incorporating various job characteristics. For this reason, in Chapter 4, we compare information about preferences for job attributes obtained using our structural labour supply models, on the one hand, and subjective well-being regressions, on the other hand. We consider this an important contribution given that these two streams of the literature have been used to recover preference information, and that therefore, it seems necessary to know to which extent the information obtained is comparable. Interestingly, some similar patterns across different population groups were found in terms of preferences for hours of work and job insecurity. For instance, under both methods, single females with young children presented lower preferences for hours of work compared to other groups. However, important differences between both models were also observed. Such differences can be attributed to different factors. First, the two models are related to two different concepts of utility: decision utility versus experienced utility. Moreover, the results are based on estimation of different econometric models. Finally, both approaches are subject to some limitations. On the one hand, structural labour supply models are partial equilibrium models, where demand side factors are not taken into account. Demand side constraints certainly play a role in people’s choices in the labour market and might affect estimations of individuals’ preferences. On the other hand, subjective well-being regressions are influenced by aspirations and expectations and the information about preferences obtained depends on how well such factors are controlled for. The extent to which these different factors play a role in explaining the differences between the two approaches compared constitutes, in our opinion, the basis for future research. Labour market research has progressively moved towards the analysis of a variety of job attributes and this trend will certainly continue. The incorporation of multiple attributes in structural labour supply models represents a step forward in this direction. At this point, many possibilities are opened for future research. The introduction of demand side factors into the analysis of structural labour supply is of great interest. In terms of policy evaluation, this will allow to represent more clearly individuals’ preferences and therefore to provide a better picture of the impact of different policy reforms. As previously said, the effect of nonpecuniary job attributes on labour supply, could serve to design policies to create incentives for labour market participation. Finally, the comparison of preferences information obtained with different models seems important to better understand the relationship between different concepts of utility and will provide an original validation exercise for different approaches.
Publication status: published
KU Leuven publication type: TH
Appears in Collections:Research Center of Public Economics, Leuven

Files in This Item:
File Status SizeFormat
Jara_Dissertation2013.pdf Published 897KbAdobe PDFView/Open Request a copy

These files are only available to some KU Leuven Association staff members


All items in Lirias are protected by copyright, with all rights reserved.