The online Belgian Arithmetic Microsimulation Model (BEAMM) platform is a free service offered by the Center for Applied Public Economics (CAPE) of the UCLouvain Saint-Louis Bruxelles.

The online BEAMM platform is made available in an open access model for instructional purposes, to promote a better understanding of the tax-benefit system in Belgium and to enhance the democratic discourse.

Introduction to the platform and its limitations

BEAMM is a complex model of a far more complex tax-benefit system. The team that develops and maintains the BEAMM platform has been working since 2020 to gradually improve the model and correct errors, and will continue improving and correcting the platform over the coming years. Step by step, we will continue to improve BEAMM over time. Despite these efforts, the BEAMM platform may contain factual errors and ambiguities.

Important limitations of the online BEAMM platform include:

• Every model is a simplification, and the same is true of BEAMM. Some tax and benefit calculations may be simplifications, e.g., because of data limitations. Moreover, the legislation underlying our tax-benefit system changes over time. We try to update the simulation code regularly, but we do not have the resources to keep the model continuously up to date; not all recent policy reforms are necessarily included in the model.
• The online BEAMM platform runs on entirely fictitious data, from a series of micro-datasets that are themselves subject to limitations.
  • The data used in the online version of BEAMM have been simulated by machine learning algorithms to be an accurate reflection of reality, and the similarity between the fictitious and real data has been extensively tested and validated. Nevertheless, this data simulation results in an increased number of errors, and may be imprecise with respect to some specific correlations.
  • Moreover, the underlying dataset results from a process of statistical matching procedures, in which information from different anonymous individuals is combined based on statistical regularities.
  • As with any dataset, the original micro-data can be subject to measuring errors, missing information, underreporting, etc. All of these factors can cause errors in results.
• The world changes constantly, and datasets, by definition, represent a point in the past, i.e., the point at which the data were collected. Given that we are interested in evaluating policies today, or in the near future, we use projection algorithms to reweight these data to better represent the present or near future, depending on demographic and economic projections. However, predicting the future is difficult. Therefore, these projections add additional potential for error to our calculations.
• Changes in the tax-benefit system are also likely to change the behavior of citizens and firms. People typically change their behavior when public policy changes. If this is the case, then our micro-data may no longer accurately reflect reality. To deal with these behavioral changes, we incorporate behavioral models into BEAMM, in order to predict behavioral changes. However, predicting behavior is difficult: it increases the predictions’ potential for error, and it is always a simplification. These behavioral models typically cannot account for the full variety and detail of the public’s behavioral reactions. Instead, they rely on probable behavioral reactions from various types of individuals and households.
• For significant changes in the tax-benefit system, BEAMM, like any other model, will have very significant error margins in its predictions. Major changes in the tax-benefit system are beyond the capacity of this model, and of any model for that matter. These models are estimated from data collected in the current environment, and major tax-benefit reforms tend to change the entire economic context.
• Finally, the online BEAMM platform is a generic, general-purpose model made available for instructional and demonstration purposes. The model is not necessarily fine-tuned and validated to answer all specific questions with the greatest possible accuracy.

Liability

Information provided on the BEAMM platform

The UCLouvain Saint-Louis Bruxelles (USL-B) is bound only by a duty of diligence with respect to the information that it provides on the online BEAMM platform. Under no circumstances can it be held liable for any errors or omissions, nor for any results obtained following use of the information provided on the platform.

Although the university’s CAPE center team uses all useful and reasonable means to ensure that the information provided is reliable and up to date, the USL-B cannot guarantee that the information provided on the BEAMM platform is correct, specifically because of the limitations mentioned above.

The CAPE team reserves the right to amend or update the information on the BEAMM platform, at any time, and to amend or update this notification without prior notice.

USL-B exclusion of liability

All users of the BEAMM platform are acting under their sole responsibility. The UCLouvain Saint-Louis-Bruxelles can in no case be held liable for any direct, indirect, tangible, or intangible damages resulting from modifications to, consultation of, and/or use of all or part of the BEAMM platform.

The USL-B is not liable for decisions and actions that a user may make based on the information found on the BEAMM platform. BEAMM users are responsible for verifying the accuracy of data before making any decisions.

Access to the platform

The USL-B may not be held liable for any service interruptions on the BEAMM platform, or for any outages (maintenance, technical problems) of the services provided on the platform.

Links to external sites

In the event that links to external sites are created by the USL-B or by any research tools, the USL-B may in no case be held liable for the content of these other sites.