An overview of the mechanisms and evaluation tools. Among the main results, note that:

• The main statistical data concern means-tested benefits and are drawn from surveys that are not necessarily on-going. At the same time, the inventory drawn up indicates that whole sections of welfare systems are excluded, e.g. medical insurance and unemployment benefits, although they are clearly affected by the NTU phenomenon.
• Regular surveys on NTU/EF are rare (and performed primarily in the UK, the Netherlands and Germany).
• The range of information sources mobilised in the different surveys is wide, especially when they are local or territorialized. The large amount of data produced by NGOs (surveys on groups in precarious situations received by these organisations) and their relevance to the subject are noteworthy.
• Statistical data on NTU/EF are drawn from several types of ad hoc or not ad hoc survey.
• Two types of indicator are generally used. First “activity indicators – policy indicators” that indicate the effectiveness of the application of programmes or policies (e.g. number of beneficiaries, share of funds used). Second, “results indicators – performance indicators” produced from surveys and not on existing administrative data. The statistical data produced on NTU/EF often combine the two.
• Professional practices sometimes develop NTU/EF indicators. This is the case for Education (with the measurement of early exit from school before obtaining a qualification) and Health (with the measurement of NTU of care).

All these initial observations provide suggestions in conclusion.

Indeed, our exploratory work on the measurement of NTU shows something important, it seems, and that is that we probably know very little on available sources of information and data, but do know enough to be sure that no immediate and reliable comparison of results is possible from a methodological or scientific point of view. The EXNOTA Consortium's work feeds doubt on the limitations of available sources, and strengthens the conviction that at this stage it is impossible to propose an indicator of NTU of welfare benefits that could correspond to the “Laeken indicators”.

Two solutions seem promising to promote the measurement of NTU:

• The first consists in facilitating the creation of an inventory of available sources of information and data bases in each country, which could help to develop an approach to NTU. We are by no means sure that our inventories in the six countries studied are exhaustive. In France the ODENORE experimental device shows, in its own way, that the observation of NTU can be based on a large number of information sources and data bases which are not initially known. It would be interesting to examine in more detail the resources of the different actors, especially the NGOs. An approach targeted at a type of population seems useful as a start, so that from there one moves on to other actors and groups and a diversity of welfare benefits. Specific work could be undertaken from Caritas in Germany, for instance, or from the Federatie Opvang in the Netherlands which receives large numbers of homeless people.

  Drawing up an inventory on the local authority scale (infra-national territories) at first, could have the objective of producing operating reports on NTU by collecting (or preparing) data at the level of networks of actors working on the same types of population or unclaimed benefits.

• The second suggestion concerns the “interessment” of administrative actors and NGOs in measurement of NTU. Becoming aware of the existence and importance of the phenomenon is still very much a necessity. This question is highly complex and tricky. The only example that we have, the CBSS in Belgium, clearly shows that the political will can be strong enough to set up large-scale national ad hoc devices to provide social welfare institutions with the maximum amount of information on the population. Since the extension of this convincing experiment to other European countries is still not on any government's agenda, neither at national nor at European level, it could prove useful to start showing institutions that they do stand to gain by knowing about NTU. One of the obstacles to the measurement of NTU is often the fact that social welfare actors and institutions prefer to push aside an unfamiliar problem, especially when they assume that to deal with it they would need more funds, something that is hardly conceivable today. Thus, to avoid this “deliberate myopia” that simply worsens the situation, it would be judicious to demonstrate the value of probabilistic approaches to “groups at risk of NTU”. >From there a policy of prevention of exclusion from social benefits could be organized, albeit at the cost of probable redefinitions of priorities and goals.