State of the art

We start by introducing the state of the art on measuring individual well-being, which is the research topic of this project. Subsequently we motivate the relevance of relaxing the underlying assumption of individuals having well-behaved preferences (see below for an exact definition). This will be the main contribution of this project. We end by stressing the importance of acknowledging that individuals live in households in order to obtain (empirically) meaningful measures of individual well-being.

1. Measuring individual well-being

Economists evaluate social and economic policies based on their impact on the well-being of the members of society. This requires a tool to measure individual well-being. Several theories of well-being have been developed and applied. They range from purely subjective to purely objective measures.

Measuring well-being on a purely subjective basis consists in taking only account of the feelings and mental states of the individual. This is in line with the utilitarian tradition, tracing back to Bentham’s pioneering contributions of the 18^th century (Bentham, 1789). More recently, the work of psychologists and social scientists on the measurement of happiness and life satisfaction has revived the use of subjective well-being measures. This has led to the booming field of “happiness economics” (see, for instance, Frey and Stutzer, 2002, and Layard, 2005).

Measuring well-being on a purely objective basis, on the other hand, consists in taking only account of the outcomes for an individual in different dimensions of life and assigning weights to these outcomes to derive a composite index of well-being. When only market-commodities are considered, prices can be used as weights so that well-being can be approximated by individuals’ incomes. Using income as a measure of well-being is the classical approach to, for instance, inequality and poverty measurement (see Atkinson, 1970 and Sen, 1976, for seminal contributions). Recently, Alkire and Foster (2011) have advocated a widely-used approach to measure multidimensional poverty that is based on counting the number of dimensions in which an individual is deprived. The EU 2020 objectives against poverty and social exclusion, for instance, are stated in terms of objective well-being measures such as the so-called at-risk-of-poverty measure and the material deprivation index. The former measure looks at the share of citizens with a disposable income below 60% of the national median disposable income. The latter measure identifies citizens as materially deprived when they cannot afford the consumption of at least three basic items in a list of nine (see Atkinson, Cantillon, Marlier, Nolan, 2002; Decancq, Goedemé, Van den Bossche and Van Hille, 2014).

The strengths and drawbacks of the above approaches are now relatively well understood (see, for example, Decancq, Fleurbaey and Schokkaert, 2015b, for an overview). The purely subjective approach suffers from the drawback that it is hardly compatible with the idea that resources should be allocated fairly. A rich individual consuming a lot more of all items than a poor individual may be viewed worse-off if he declares a lower happiness or satisfaction. This is not only a theoretical possibility. Several surveys have, not surprisingly, shown that the correlation between satisfaction or happiness on the one hand and income or consumption on the other hand is all but perfect. Defining the fight against poverty or inequality on the basis of subjective well-being concepts may thus lead to redistributing from poor to rich citizens.

The purely objective approach, on the other hand, suffers from two main drawbacks. First, it leads to paternalistic judgments, in the sense that there may be a conflict between a policy recommendation based on a purely objective well-being measure and what the citizen herself wishes. Policies based on paternalistic well-being measures could therefore consist in improving individuals measured well-being against their own will. The second drawback of purely objective measures is their difficulty to take heterogeneity in tastes and needs into account. Different individuals may care about different aspects of their life or may have different needs, and a purely objective measure cannot take this into account.

Purely subjective and purely objective measures have their advantages, too, of course. For instance, they are very easy to implement statistically. The main starting point of this research proposal, though, is that scientific research needs to go beyond these well-being measures, first because of the political interest of evaluating policies in a way that is consistent with a fair allocation of resources (which goes against the use of subjective well-being measures) and second because of the democratic necessity to acknowledge heterogeneity in tastes and needs (which goes against the use of objective measures).

To address the drawbacks of subjective and objective measures, an alternative preference-based approach has been proposed (see Fleurbaey and Maniquet, 2011, and Fleurbaey and Blanchet, 2013, for a review). A preference-based approach is a mixed approach that takes account of the objective outcomes of an individual, as well as the subjective preferences of an individual about what is relatively more important in her own life. One such approach is the so-called equivalent-income approach, which constructs a hypothetical income for each individual by correcting her actual income for the well-being losses that she experiences in non-monetary dimensions of life, evaluated according to her own preferences. This well-being loss can either be caused by physical reasons (e.g., if this person is not in good health) or for social reasons (e.g., if the individual is unemployed due to the imperfection of the labor market). See Decancq, Fleurbaey and Schokkaert (2015a,b) and Fleurbaey (2016) for more details.

The preference-based approach also has its drawbacks. A first practical drawback is that one needs a lot of detailed information on both the objective outcomes as well as on the preferences of the concerned individual. The second drawback is that the preference-based approach is grounded on the assumption of “well-behaved’’ preferences. Well-behaved preferences are a complete and transitive ranking of the situations that an individual can experience. Completeness means that any two situations can be compared (to conclude that one is superior or that they are both equally good). Transitivity means that if situation a is ranked above situation b, and situation b is itself ranked above situation c, then situation a is also ranked above situation c.

The interest of economists in well-behaved preferences is largely due to the scientific power of the assumption that individuals behave according to some specific choice rules. The interest of welfare economists into preferences is different. It comes from the fact that the evaluation of policies (being able to claim that policy a is socially preferable to policy b) seems to require to have a well-being measure that is complete (the effect of all available policies over all relevant individuals can be evaluated) and transitive (if an individual would be better-off under policy a than under policy b and better-off under policy b than under policy c, then consistency requires to also claim that this individual would be better-off under policy a than policy c).[1]

Policy evaluation based on preference-based well-being measures requires to have information on preferences. It can be obtained from laboratory or field experiments, which is then typically referred to as revealed preferences, or drawn from interviews or stated choice experiments, which is labeled as stated preferences. As we will explain below, we can in this respect make use of a unique and detailed data set, which is representative for the Belgian population. The underlying assumption is that what an individual prefers is also what increases her well-being, possibly after her preferences are somehow laundered to remove idiosyncratic mistakes. All the existing preference-based methods to measure well-being are built on the assumption that (revealed and stated) individual preferences are transitive (or more generally well-behaved).

2. Non “well-behaved” preferences

The aforementioned preference-based approach to defining and measuring well-being has been challenged by recent contributions in psychology and behavioral economics. These contributions have shown that individuals do not always behave as if they are maximizing well-behaved preferences (see, among many others, Kahneman, 2003, and Fehr and Schmidt, 2006). There is ample evidence that individual behavior often fails to be consistent with transitivity, for instance. Let us briefly discuss two well-established examples of such departures from transitivity.

First, it has been well documented that individuals typically choose after having first restricted the set of alternatives they consider to a limited set of alternatives, the so-called consideration set, and this set of considered alternatives may change from one choice to another (see Eliaz and Spiegler, 2011, Caplin, Dean and Martin, 2011, Manzini and Mariotti, 2012 and 2014a, Masatlioglu, Nakajima and Ozbay, 2012, and Demuynck and Seel, 2017, to give a non-exhaustive list of examples). Consequently, an evaluator trying to measure this individual’s well-being may conclude that her well-being fails to be transitive. Second, individuals may adjust the preferences with which they are consistent to the context in which they must choose. This context-dependency may come from a possible status quo bias, in which case the individual feels a strong preference towards keeping the same original situation. This may also come from some adaptation in aspirations, anchoring, or reference-dependence. See Kalai, Rubinstein and Spiegler (2002) and Manzini and Mariotti (2007, 2015, 2016) for examples of context-dependent preferences and variations.

In the first example, it could remain meaningful to speak of ‘true’ underlying preferences (i.e. variables not related to the preferences explain the observed intransitivities), and the effort of the evaluator should be to retrieve these true preferences. This is not always the case in the second example. Further, it is not clear how and which preferences should intervene in the definition of well-being.

Two main reactions to this transitivity failure have been observed among welfare economists. The first reaction consists in identifying this failure as a further argument to favor either the purely subjective or the purely objective approaches to well-being measurement (see, Dolan and Kahneman 2008, for instance). We consider these reactions as non-starters, given the fundamental drawbacks of these approaches that we briefly discussed above.

The second reaction consists in correcting actual preferences to make them transitive. This can be done in many ways. At one extreme, one way of correcting preferences is completely agnostic to the reason why the observed preferences fail to be transitive (see, for example, Salant and Rubinstein, 2008; Bernheim and Rangel, 2009; Green and Hojman, 2009; and Fleurbaey and Schokkaert, 2013). The key requirement in this approach consists in claiming that an alternative is better than another one for an individual if he never chooses the latter when the former is available, independently of the context. The main strength of this approach is that it is fully general and can be widely applied. The drawback, on the other hand, is that being completely agnostic about why decision makers depart from transitivity may seem unreasonable in many circumstances. Moreover it also implies that completeness of the observed preferences is lost. At the other extreme, one may claim that the precise reason why decision makers depart from transitivity is perfectly known and captured by a model of behavior. Thus, the researcher treats the information he has about choices to recover the true preferences (beyond the intransitivities) by making use of the behavioral model (see Rubinstein and Salant, 2012, for a concrete example and Manzini and Marioti, 2014b, for more discussion).

Between these two extreme approaches, there is a possibility for an intermediary position and analyze how (imperfect) information about the reasons why decision makers depart from transitivity allows to impose restrictions on the set of true (well-behaved) preferences that are consistent with the revealed or stated preferences. Indeed, the situations in which individuals find themselves may tell us that some reasons why individuals depart from transitivity are more plausible than others. In matters involving health, for instance, it could be that low expectations about the likelihood of a full recovery induces individuals to underevaluate the cost for them of being sick (see, for example, Dolan and Kahneman, 2008). Such an intermediary approach has not yet been fully developed in the context of measuring individual well-being.

As a final remark, the current literature on measuring well-being has restricted itself to the question of the possibility to retrieve transitive preferences from observed intransitive behavior. In other words, it focuses on the question of whether well-being can after all be defined at the individual level. The second step of the construction of a well-being index consists in being able to compare well-being across individuals. When the input to this second step comprises of well-behaved preferences, the construction of interpersonal comparability consistent with the ideal of fair resource allocation is reasonably well explored (see, for instance, Fleurbaey and Maniquet, 2011 and 2017, Fleurbaey and Blanchet, 2013, Decancq, Fleurbaey and Schokkaert, 2015a and Decancq, Fleurbaey and Maniquet, 2015). When the input to this second step comprises preferences that fail to be complete, which is the case under the agnostic attitude toward the source of the boundedness of rationality, the theory is much less developed (see, however, the notable exception by Fleurbaey and Schokkaert, 2013).

3. Individuals and households

Another challenge for the preference-based approach to well-being is the fact that most data sets with measures of material well-being (like consumption and income) contain information at the household level rather than the individual level. In order to obtain individual measures, it is therefore often assumed that resources are evenly distributed among household members (a policy-relevant example is the apparatus to measure poverty in Europe). A long literature in family economics has made it clear that this assumption does not hold in reality. This literature has developed many ways to reconstruct individual well-being from household data of which the collective model is nowadays the most widely used model (see, for instance, Chiappori, 1988 and 1992, for seminal contributions and Browning, Chiappori and Weiss, 2014, for a state of the art).

This collective model explicitly accounts for the fact that multi-person households consist of different household members, who have their own, possibly different, preferences. The model further assumes that the household members choose Pareto-efficient allocations of time and money, which implies that it is impossible to make a household member better off without making another one worse off. See Chiappori and Ekeland (2009), Cherchye, De Rock and Vermeulen (2012), Dunbar, Lewbel and Pendakur (2013) and Cherchye, De Rock, Lewbel and Vermeulen (2015) for a sample of recent theoretical and empirical work on the collective model.

Recently, Browning, Chiappori and Lewbel (2013) have introduced the concept of indifference scales. Indifference scales allow to retrieve both the individual levels of consumption and individual preferences so that the well-being of individuals living in households of different sizes can be compared, even with a typical household data set. These well-being comparisons are based on the level of consumption that makes an individual indifferent (or equally well-off) between living in households of different sizes. In essence, indifference scales are a specific example of the aforementioned notion of equivalent income, and are therefore perfectly integrated with the existing methods for measuring individual well-being. Once more, however, the use of indifference scales is based on the assumption of well-behaved preferences.

The challenge that transitivity failures raise to well-being measurement is even stronger when one considers that individuals may choose to live in couples and may choose to have children. The first obvious challenge consists in rebuilding individual transitive well-being measures from household data that may aggregate possible intransitive individual choices, including those with respect to the household composition. The few empirical applications of this recent literature of well-being measurement with intransitive individual behavior use data that are obtained at the household level and do not take into account the insights of collective models (see, for instance, Bernheim, Fradkin and Popov, 2015).

There is an additional challenge, though. It is clear, indeed, that individual well-being is not only unequally distributed inside the household, but may also depend on the marriage itself. Since Becker (1973) argued that the institution of marriage can be analyzed by means of modern microeconomic theory, a rich literature has been developed, under the assumption that each individual looks for the best mate subject to the restrictions imposed by the marriage environment. A recent example along these lines is the marriage model proposed by Cherchye, Demuynck, De Rock and Vermeulen (2016), which combines the collective model with the assumption that marriage choices lead to a stable allocation of individuals into households (Gale and Shapley, 1962, Shapley and Shubik, 1972, and Becker, 1973) to analyze the impact of the marriage environment on the intra-household allocation of resources. The stability of a marriage is assumed to depend on each partner’s outside options, which depend in this paper on the age difference between potential mates as well as on wages.

This approach is in line with the widely empirically observed positive dependence of the individual shares in the household’s resources on wages (see, for example, Chiappori, Fortin and Lacroix, 2002, Blundell, Chiappori, Magnac and Meghir, 2007, and Cherchye, De Rock, Lewbel and Vermeulen, 2015). It also optimally allows to investigate the dependence between marriages and individual well-being. However, to model the outside option of the bargaining process, it is assumed that individuals know and consider all the potential partners in their marriage environment. This is clearly a rather strong assumption and it would be more plausible to assume that individuals consider only a limited and context-dependent set of the potential mates, which can perfectly be addressed by the concept of consideration sets discussed above.

Objectives and research hypothesis

Until now, the preference-based approach to individual well-being measurement assumed that observational data were generated by idealized individuals with well-behaved (i.e. complete and transitive) preferences. As discussed above, a large, and rapidly expanding, body of evidence from behavioral scientists challenges the core assumption of transitivity, however. This observation raises serious concerns about the real-world policy-relevance and applicability of the preference-based approach to well-being measurement. In this project, we want to take the concerns of behavioral scientists seriously and address the following ambitious research question.

Research question: How to define and measure individual well-being when revealed or stated preferences are not well-behaved?

In short, the overarching objective of this project is to develop a second-generation preference-based approach to the measurement of well-being. This second-generation framework extends the existing framework that only fits perfectly rational individuals. To do that, we have interrelated theoretical and empirical objectives.

The theoretical objective of the current project is to address the question of how to construct an interpersonally comparable measure of well-being for individuals whose preferences are not necessarily well-behaved. We plan to favor an intermediary approach between the completely agnostic approach, which assumes that nothing is known about why individuals’ preferences fail to be transitive, and the other extreme approach, which assumes that the reasons why individuals depart from classically rational decision making are known.

The empirical objective of the current project is to use specific models of intransitivities and bring them to data to retrieve true individual preferences. Subsequently we will use these preferences to measure individual well-being. We will restrict our attention to four specific and important dimensions of well-being: health, material well-being, employment status, and the household situation. Importantly, we will adopt the individualistic point of view, which means that we do not restrict our attention to a notion of household well-being, but, on the contrary, we aim at evaluating well-being of each adult member of the household.

For the empirical part, we will extensively use a new and largely unexplored data set that was recently developed in the frame of a Belspo BRAIN program (i.e. the MEqIn project in which all the current PI’s were involved). This unique and attractive data set contains information on a representative sample of the Belgian population. It is structured in such a way that all major subjective, objective, and preference-based approaches to well-being measurement can be tested and this for every adult member of the household. Given the focus of this project on intransitivities, we will complement these data with a panel of additional modules that will allow us to shed light on the behavioral mechanisms underlying the observed intransitivities.

The ultimate objective is that the methods that we will define can be used by practitioners. It means that we aim at identifying the minimal information that is necessary for the policy evaluator to evaluate the impact of a social policy on the well-being of all relevant individuals, to be able to conclude whether the policy is socially justified or not. Our empirical applications of well-being measurement involving health and employment will already allow us to evaluate health care, family and labor market policies. Summarizing, all this leads to the following specific objectives.

Objectives:

1) Define a method to measure well-being at the individual (as opposed to household) level by taking account of the reasons why individuals’ revealed or stated preferences fail to be well-behaved.

2) Identify the information that is necessary to apply the new methods for the evaluation of social policies in general.

3) Apply the method to evaluate individual well-being when the relevant dimensions include health, material well-being, employment status, and family situation.

Methodology

This project involves developing theoretical and empirical work and gathering data. We describe these different parts separately, but it should be clear that they are highly interrelated. It should also be clear that the general topic of well-being measurement when individual behavior departs from the classical completeness and transitivity assumptions may lead to more research questions than the ones we list below. More specifically, we largely leave aside the issues related to decision making under uncertainty. We focus here on the ones we currently consider most promising, but we are of course well aware that they may need to be adjusted along the way.

1. Theory

A first set of theoretical questions deal with cases in which choice data are available at the individual level. First, we will assume that individuals are characterized by incomplete preferences (as in Salant and Rubinstein, 2008; Bernheim and Rangel, 2009). The observed incompleteness can have several sources: imprecise preferences of the respondents (Butler and Loomes, 1988; 2007), but also anchoring and framing in which choices depend on features of the choice situation rather than the alternatives considered. Moreover, the incompleteness can come from the use of a random component in random utility models (McFadden, 1999; Herriges and Kling, 1999; and Dagsvik and Karlström, 2005). We will study the conditions under which an interpersonally comparable well-being measure can be constructed based on these incomplete preferences, extending the earlier work by Fleurbaey and Schokkaert (2013).

Second, we will explore the assumption that individuals are characterized by their type. A type corresponds to a function that determines the preferences of an individual in each context. These preferences are related to each other, that is, there are regularities in the way preferences vary across contexts. Such regularities capture the reasons why the observed behavior fail to obey transitivity. For instance, if an individual has a relatively low substitution rate among two goods in one context, compared to other individuals in the same context, he will keep having a relatively low substitution rate between these two goods in another context, even if this rate has changed.

In a type-based approach, measuring well-being does no longer amount to assigning numbers to individual situations as a function of preferences, but to assign numbers to individual situations as a function of types. We will axiomatically study this problem and derive a general theory of well-being measurement when individuals can be characterized by types, that is, a family of preferences. This encompasses the possibility that preferences are a function of the marital status of an individual. When this individual forms a household with a partner, assuming preferences are modified in a regular way (e.g., preferences towards goods that are consumed in common increase), how do we define well-being and how do we measure it? Alternatively, we could assume that preferences remain stable but that the household technology is different, but in a measurable way, across types.

The second set of theoretical questions deal with cases when choice data are observed at the level of the household and when they depart from rationality in well-known ways. For instance, one question consists in analyzing whether it is possible to replicate the model based approach of Rubinstein and Salant (2012) when choices are made within the household. We will also address the question of the possibility of retrieving the true preferences and measuring well-being when the potential partners of individuals may be context-dependent. Therefore we will assume that the partner choice is consistent with the model of consideration sets discussed above.

2. Empirics

The results obtained in the theoretical part of this research proposal will be used to measure the well-being of Belgian individuals. For this we will keep concrete policies in mind and focus on the life dimensions health, material well-being, employment status, and family situation. The MEqIn data set will be the prime data source to address the empirical questions of this project. This data set has been collected in 2016 by several of the researchers involved in the current consortium with the specific purpose of measuring well-being and eliciting individual preferences in Belgium. It contains 3,404 respondents in 2098 households. Some questions can be addressed with the data that are currently available, for others new modules are needed. In the latter cases, we will invite the existing MEqIn respondents to participate to some innovative new follow-up modules involving new questions, checks of test-retest validity of the elicitation methods, and a novel field experiment. We will provide incentives to reassure sufficiently high response rates. It is reassuring that more than 80% of the original MEqIn sample mentioned to be willing to participate to follow-up surveys. By bringing the original data set together with these novel follow-up modules, we will obtain a unique and tailored panel data set.

A first set of empirical questions deals with determining the extent to which the elicited preferences over health, material well-being and employment status are inconsistent according to the different methods. The MEqIn data set allows us to measure the trade-off between labor time and consumption, and health and consumption in different ways: based on observed outcomes (i.e. stated labor time and health expenditures), satisfaction data (i.e. how satisfied are you with your job or health status?) and contingent valuation or willingness-to-pay questions (what is the money equivalent for you of modifying your labor time or of having a different health situation). When we apply the different (subjective, objective and preference-based) approaches to preference measurement, we obtain conflicting views on the individual’s underlying preferences using the different methods. Using the structure of the data set, we can study the regularities in the differences and then apply our general model of well-being as a function of types. This will provide us with a measure of well-being that will offer a synthesis among the different approaches to the definition of well-being.

A second set of empirical questions deals with the possibility that individuals’ stated preferences exhibit context dependency. First empirical results using the MEqIn data set suggest indeed that context dependence may hold with respect to health and employment status. As an example, unemployed individuals may understate how important it is for them to have a job, while employed individuals may overstate it. To understand the regularities underlying the context dependency, we will recontact the respondents of the MEqIn data set with (currently missing) questions about what having a job, or having perfect health means to them. For this we will elicit the preferences using a discrete choice set-up in line with pioneering contributions by MacCrimmon and Toda (1969) to check whether changing contexts (in particular for health and employment) can explain some of the observed preference reversals.

When considering employment status, we are particularly interested in eliciting the reservation wage to understand the reported importance of having a job for employed and unemployed individuals. Consequently, we will invite our MEqIn respondents to participate in a large field experiment. In this experiment, respondents are offered a job that they can do from home. We use the Becker-DeGroot-Marschak mechanism to elicit reservation wages, a standard tool in experimental economics to elicit reservation values (see, for example, Bohm, Lindén and Sonnegård, 1997). Thus, we can obtain an incentivized measure of reservation wages for a standardized task for both employed and unemployed individuals. This is an advantage compared to standard hypothetical questions for which each respondent has a different job in mind. Typically, job characteristics play an important role in determining reservation wages (Hofler and Murphey, 1994), but are not observed by the econometrician. Our measure instead elicits reservation wages for the same job, thus preventing any bias associated with unobserved heterogeneity in job characteristics.

A final set of empirical questions will focus on the effect of introducing bounded rationality in the choice of one’s partner. As indicated above, one potentially fruitful avenue is to integrate the concept of consideration sets in the marriage market model of Cherchye, Demuynck, De Rock and Vermeulen (2016). Many studies in this literature make use of a discrete choice framework. This implies that they are an ideal starting point for their integration with the model of Cherchye, Demuynck, De Rock and Vermeulen (2016). The MEqIn data set already contains some relevant information regarding the reconstruction of the individual specific marriage market. However, to fruitfully study the impact of consideration sets, we will again revisit part of the MEqIN respondents to ask specific questions related to the potential partners they consider. More specifically, we will reconstruct the marriage market history of the respondents to obtain insights into their consideration sets and how these sets vary with their specific contexts.

Bibliographical references

Atkinson, A. (1970), “On the measurement of inequality”, Journal of Economic Theory, 2, 244-263.

Atkinson, A., B. Cantillon, E. Marlier and B. Nolan (2002), “Social Indicators: The EU and Social Exclusion” Oxford University Press.

Alkire, S. and J. Foster (2011), “Counting and multidimensional poverty measurement”, Journal of Public Economics, 95, 476-487.

Becker, G. (1973), “A theory of marriage: Part I”, Journal of Political Economy, 81, 813-846.

Butler, D. and G. Loomes (1988), “Decision difficulty and imprecise preferences”, Acta Psychologica, 68, 183-196.

Butler, D. and G. Loomes (2007), “Imprecision as an account of the preference reversal phenomenon”, American Economic Review, 97, 277-297.

Bentham, J. (1789, reprinted 1907), An Introduction to the Principles of Morals and Legislation, Clarendon Press, Oxford.

Bernheim, D. and A. Rangel (2009) “Beyond revealed preference: choice theoretic foundations for behavioral welfare economics”, Quarterly Journal of Economics, 124, 51-104.

Bernheim, B., A. Fradkin and I. Popov (2015), “The welfare economics of default options in 401(k) plans”, American Economic Review, 105, 2798-2837.

Blundell, R., P.-A. Chiappori, T. Magnac and C. Meghir (2007), “Collective labour supply: heterogeneity and non-participation”, Review of Economic Studies, 74, 417-445.

Bohm, P., J. Lindén and J. Sonnegård (1997), “Eliciting reservation prices: Becker-DeGroot-Marschak mechanisms vs. markets”, Economic Journal, 107, 1079-1089.

Browning, M., P.-A. Chiappori and Y. Weiss (2014), Economics of the Family, Cambridge University Press, Cambridge.

Browning, M., P.-A. Chiappori and A. Lewbel (2013), “Estimating consumption economies of scale, adult equivalence scales, and household bargaining power, Review of Economic Studies, 80, 1267-1303.

Caplin, A., M. Dean and D. Martin (2011), “Search and satisficing”, American Economic Review, 101, 2899-2922.

Cherchye, L., B. De Rock, and F. Vermeulen (2012), “Married with children. A collective labor supply model with detailed time use and intrahousehold expenditure information”, American Economic Review, 102, 3377-3405.

Cherchye, L., T. Demuynck, B. De Rock and F. Vermeulen (2016), “Household consumption when the marriage is stable”, American Economic Review, forthcoming.

Cherchye, L., B. De Rock, A. Lewbel and F. Vermeulen (2015), “Sharing rule identification for general collective consumption models”, Econometrica, 83, 2001-2041.

Chiappori, P.A. (1988), “Rational household labor supply”, Econometrica, 56, 63-89.

Chiappori, P.-A. (1992), “Collective labor supply and welfare”, Journal of Political Economy, 100, 437-467.

Chiappori, P.A., and I. Ekeland (2009), “The micro-economics of efficient group behavior: identification”, Econometrica, 77, 763-799.

Chiappori, P.-A., B. Fortin and G. Lacroix (2002), “Marriage market, divorce legislation and household labor supply”, Journal of Political Economy, 110, 37-72.

Dagsvik, J.K. and A. Karlström (2005), “Compensating variation and Hicksian choice probabilities in random utility models that are nonlinear in income”, Review of Economic Studies, 72, 57‑76.

Decancq, K., M. Fleurbaey and F. Maniquet (2015), “Multidimensional poverty measurement with individual preferences”, CORE discussion paper 2015/08.

Decancq, K., M. Fleurbaey and E. Schokkaert. (2015a), “Happiness, equivalent incomes, and respect for individual preferences”, Economica, 82, 1082-1106.

Decancq, K., M. Fleurbaey and E. Schokkaert. (2015b), “Inequality, income, and well-being” in: Anthony B. Atkinson and François Bourguignon (eds.) Handbook on Income Distribution. Volume 2A Elsevier. pp. 67–140.

Decancq, K., T. Goedemé, K. Van den Bosch and J. Vanhille (2014), “The evolution of poverty in the Europe Union: concepts, measurement and data” in: Bea Cantillon and Frank Vandenbroucke (eds.) Reconciling work and poverty reduction: how successful are European welfare states? Oxford University Press.

Demuynck, T. and C. Seel (2017), “Revealed preference with limited consideration”, American Economic Journal: Microeconomics, forthcoming.

Dolan, P. and D. Kahneman (2008), “Interpretations of utility and their implications for the valuation of health”, Economic Journal, 118, 215-234.

Dunbar, G., A. Lewbel and K. Pendakur (2013), “Children's resources in collective households: identification, estimation and an application to child poverty in Malawi”, American Economic Review, 103, 438-471.

Eliaz, K. and R. Spiegler (2011), “Consideration sets and competitive marketing”, Review of Economic Studies, 78, 235–262.

Fehr, E. and K. Schmidt (2006), "The Economics of Fairness, Reciprocity and Altruism - Experimental Evidence and New Theories," Handbook on the Economics of Giving, Reciprocity and Altruism, Elsevier.

Fleurbaey, M. (2016), “Equivalent income”, in M. Adler and M. Fleurbaey (Eds) The Oxford Handbook of Well-Being and Public Policy, Chapter 16, 453-475.

Fleurbaey, M. and D. Blanchet (2013), Beyond GDP: Measuring Welfare and Assessing Sustainability, Oxford University Press, Oxford.

Fleurbaey, M. and F. Maniquet, F. (2011), A Theory of Fairness and Social Welfare, Cambridge University Press.

Fleurbaey, M. and F. Maniquet (2017), “Fairness and well-being Measurement”, Mathematical Social Sciences, forthcoming.

Fleurbaey, M. and E. Schokkaert, (2013) “Behavioral welfare economics and redistribution”, American Economic Journal: Microeconomics, 5, 180-205

Frey, B. and A. Stutzer (2002), Happiness and Economics: How the Economy and Institutions Affect Human Well-Being, Princeton University Press, Princeton.

Gale, L.S. and D. Shapley (1962), “College admissions and the stability of marriage”, American Mathematical Monthly, 69, 9-15.

Green, J. and D. Hojman (2009), “Choice, rationality and welfare measurement”, working paper.

Herriges, J.A. and C.L. Kling (1999), “Nonlinear income effects in random utility models,” Review of Economics and Statistics, 81, 62‑72.

Hofler, R. and K. Murphy (1994), “Estimating reservation wages of employed workers using a stochastic frontier”, Southern Economic Journal, 60, 961-976.

Kahneman, D. (2003). Maps of bounded rationality: A perspective on intuitive judgment and choice. In T. Frangsmyr [Nobel Foundation], (Ed.), Les Prix Nobel: The Nobel Prizes 2002 (pp. 449-489). Stockholm, SE: The Nobel Foundation.

Kalai, G., A. Rubinstein and R. Spiegler (2002), “Rationalizing choice functions by multiple rationales”, Econometrica, 70, 2481-2488.

Layard, R. (2005), Happiness: Lessons for a new Science, Penguin, London.

MacCrimmon, K. and M. Toda (1969), “The experimental determination of indifference curves”, Review of Economic Studies, 36, 433-451.

McFadden, D. (1999), “Computing willingness‑to‑pay in random utility models,” in: J.K Mervin, R. Riezman, and J.C. Moore (eds.), Trade, Theory and Econometrics. Essays in honor of John S. Chipman, Routledge: London, 253‑274.

Manzini, P. and M. Mariotti (2007), “Sequentially rationalizable choice”, American Economic Review, 97, 1824-1839.

Manzini, P. and M. Mariotti (2012), “Categorize then choose: boundedly rational choice and welfare”, Journal of the European Economic Association, 10, 1141-1165.

Manzini, P. and M. Mariotti (2014a), “Stochastic choice and consideration sets”, Econometrica, 82, 1153-1176.

Manzini, P. and M. Mariotti (2014b), “Welfare economics and bounded rationality: the case for model-based approaches”, Journal of Economic Methodology, 21,343-360.

Manzini, P. and M. Mariotti (2015), “State dependent choice”, Social Choice and Welfare, 45, 239-268.

Manzini, P. and M. Mariotti (2016), "Dual random utility maximization", working paper.

Masatlioglu, Y., D. Nakajima and E. Ozbay (2012), “Revealed attention”, American Economic Review, 102, 2183-2205.

Rubinstein, A. and Y. Salant, (2012), “Eliciting Welfare Preferences from Behavioral Datasets”, Review of Economic Studies, 79, 375-387.

Salant, Y. and A. Rubinstein (2008), “(A, f): Choice with Frames”, Review of Economic Studies, 75, 1287-1296.

Sen, A. (1976), “Poverty: An ordinal approach to measurement”, Econometrica, 44, 219-231.

Shapley, L. and M. Shubik, (1972), “The assignment game I: the core”, International Journal of Game Theory, 1, 111-130.