A revealed preference analysis of collective consumption models – Bram de Rock

Should our policy recommendations depend on ad-hoc choices to capture the behavior of households?
Can a reallocation of the household budget, improve the well-being of some household members without decreasing the well-being of the other members?


Prof. Bram de Rock is awarded with the Francqui Prize 2019 (also known as the ‘Belgian Nobel Prize’)


bramBorn in 1977. PhD in Economics (KUL), PhD in Mathematics (KUL). Professor of mathematical economics at ECARES (ULB) and Professor in statistics at KUL. Honorary Senior Research Associate at University College London. He is specialized in modeling multi-member household consumption and revealed preference theory.

He published on this subject articles in the leading economic journals as American Economic Review, Econometrica, Journal of Polictical Economy, Review of Economic Studies, Operations Research,… He is promotor of numerous research projects on national level and has benefited from prestigious funding sources such as the European Research Council Starting Grant (2010), FNRS, FWO, BNB Grant. He received also the Research Council Award (KUL), and the Prize of the Vereniging voor Economie.



Screen Shot 2013-12-05 at 17.49.34

The word ‘economics’ stems from the Greek ‘oikos nomein’, which is literally translated as ‘running the household’.

This reflects the fact that the household is the cornerstone of our society and that, as a consequence, decisions made by households are driving a huge part of the economy.
Remarkably, however, neo-classical economics has tended to disregard the household by treating its decision process as a black box.
The household is mostly treated as if it were a single decision maker that maximizes a household utility function subject to the budget constraint.

There is a growing consensus that this starting point can seriously flaw the analysis and its corresponding policy recommendations.


There is a growing consensus that the consumption behavior of multi-person households should no longer be treated by the unitary decision making model.
In contrast, the analysis should explicitly take into account the individual preferences of each household member and the resulting household decision process.

In this project we want to create a variety of household consumption models that allows to the empirical analyst to select the most appropriate model for the data and setting at hand.
This in turn will lead to policy recommendations that much better anticipate the behavior of the household’s members in response to tax changes, cash transfers, new environmental laws,…

For the design and empirical applications of these models, the standard econometric approach will be used as well as the revealed preference approach.

While the standard approach allows much better for dealing with empirical issues, such as measurement error and unobserved variables, the revealed preference approach leads to much more robust results by avoiding maximally ad-hoc choices made by the researcher.

Interestingly, there is a recent literature trying to combine both methods, which is also the final objective of this project.

3. The PROJECT: Grasping the behavior of multi-person households and development of new consumption models.

Screen Shot 2013-12-05 at 17.50.04

There are two complementary components in this project, which are both multidisciplinary (inside the field of economics) in nature.

The methodological component

In order to grasp optimally the behavior of multi-person households, we have to model the decisions taken by individuals as well as the interaction between the household members.

For modeling the individual behavior, we will first focus on neo-classical models that assume that individuals are perfectly rational.

In a second step, we will relax this assumption by using insights of behavioral economics, which explain that individuals are not time consistent (i.e. preferences may change), that framing matters (i.e. the context, and thus not only the prices, influence the decisions),…
For modeling the interaction between the individuals inside the household, several options will be considered, going from Pareto optimal decisions (i.e. making one household member better off will hurt the other member) up to strategic behavior (i.e. all members try to free ride on the efforts/choices of the other members, members hide income,…).

Finally, we want to use the insights developed in the consumption setting in a production context.

More precisely, many institutions (firms, countries, public entities,…) produce multiple outputs while using multiple inputs. Analyzing if these institutions are performing optimal if we compare them to their peers, is surprisingly very related to the models developed  for households.
Screen Shot 2013-12-05 at 17.50.18

We also plan to further investigate this cross-fertilization to obtain an interesting side-product and to give an extra policy dimension to this project.

The empirical component

The ultimate goal of this project is to bring the developed models to the data to answer empirical and policy relevant questions related to household decisions.
In this respect it will be important to develop welfare indicators that extend the current indicators by taking the implications of our household consumption models into account.

Moreover, to obtain the robust answers provided by the revealed preference theory, we will have to develop the revealed preference theory for our new models.

We will also try to integrate this theory with the standard econometric approach. We have some preliminary promising results in this respect that could influence significantly future empirical research on households.[/su_spoiler]

Screen Shot 2013-12-05 at 17.50.30 Screen Shot 2013-12-05 at 17.50.38

4. European Center for Advanced Researchin Economics and Statistics , ECARES, ULB

Bram de Rock is a part of ECARES, the European Center for Advanced Research in Economics and Statistics (ULB).

Bram de Rock is the head of two research groups, which cross-fertilize each other.
The first research group is located at ECARES (ULB) and is mainly funded by the ERC Grant obtained by Bram de Rock in 2011. This research group is multi-disciplinary.
This research group focuses on applying state of the art consumption models in a developing context to answer questions related to education investments, the impact of conflicts, the consumption behavior of households,…

The second research group is located at KULeuven and focuses on the theoretical topics namely: developing new household consumption models and revealed preference theory.

Screen Shot 2013-12-05 at 17.50.51

Budget over 4 years

With his ERC grant, Bram De Rock established at ECARES a research group of 7 members (the PI, 2 Post-docs and 4 PhD students). To attract students we use seeding grants of one year and encouraged them to find their own funding afterwards.

The budget for personnel covers the following:

The system of seeding grants will be used to keep attracting promising PhD students.

A senior post-doc will assist Bram De Rock in scientifically managing the extensive and multi-disciplinary research group.

Two junior post-docs will be used to support the PhD students in disciplines not covered by myself.

A part-time administrative staff should deal with the daily management. A second part-time position will be used to help the research group with the coding.

Buy-out the teaching duties of Bram De Rock to allow him to optimally disseminate the research results of his team.

The remaining part of the requested budget will be used to cover the equipment and working budget used by the research group.

4 Phd students for 1 year 140.000 €
Senior Post-doc for 4 years 400.000 €
2 junior/international post-doc for 2years 180.000 €
Teaching buy-out for the Principal Investigator 140.000 €
Computers (8×2000) Specialized softwares (8×3000) 24.000 €
Administrative/technical staff 10.000 €
PhD students (4×2500) 10.000 €
Principal Investigator (4×10.000€) 40.000 €
TOTAL 1.170.000 €


The project in PDF accessible here.

Partager cet article :