Introduction

1A good thirty years ago, on 1 March 1993, begging and homelessness were decriminalised after many centuries. Although the phenomenon of begging is no longer a criminal offence, it has continued to stir up emotions on a regular basis. Especially in cities like Brussels, where people begging are visibly present in the streets, the confrontation with this phenomenon frequently leads to heated political and social debate. The commotion surrounding the introduction of a ban on begging in the presence of children under the age of 16 in the city of Brussels (2022) is a recent example of this [Schrooten et al., 2023]. Even more drastically, there are calls from some quarters to reopen the debate on the criminalisation of homelessness and begging [Dive, 2024].

2Debates on begging often focus on the – frequently assumed – link between begging and criminality or a lack of safety. More often, however, begging is experienced as a nuisance, inappropriate or undesirable, aside from any impact on public order [Colombo et al., 2016; de Coulon et al., 2015; Hermer, 2019]. Begging is also strongly associated with extreme poverty and the debate surrounding that, including discussion of the extent to which people who beg are genuinely experiencing precarious poverty. Some see begging as primarily an easy way of acquiring an income, whereas others consider it to be a survival strategy for people who have few other options, if any [Flåto and Johannessen, 2010; McIntosh and Erskine, 2000]. The term “survival activities” is often used in the literature as a concept [Deckard and Auyero, 2022; Nunez Basante, 2023; Ojebuyi and Ajao, 2023], but rarely given an explicit definition. In the context of this research, it refers to activities that people and communities at the bottom of the income scale engage in to provide themselves with a livelihood. These activities are often situated in the informal economy [Portes, 2010; Venkatesh, 2006], and frequently criminalised [Wacquant, 2009].

3The aspect of criminalisation appears to be a recurrent theme in the debate on begging. All too often, the debate seems to be split down the middle, between begging as a threat to safety and begging as a canary in the coalmine called poverty. It comes down to the contrasting images of the beggar as an uninvited guest or the victim of exclusion. The differing claims made in discussions of begging and people who beg, for example about the background, motives and income of beggars, are highly diverse and often irreconcilable [Donovan, 2008; Fitzpatrick and Jones, 2005]. Incidentally, the word “income” should be understood in the wider sense here: begging does not only yield cash, but also gifts in kind such as cigarettes, food or clothing [Adriaenssens and Hendrickx, 2011; Reinhard, 2023].

4It is striking that the debate on begging is often conducted with great confidence, although there is actually little well-founded knowledge about people who beg [Adriaenssens and Soare, 2023; Leeson and Hardy, 2022]. The group of people who beg belong to what is referred to in the literature as a “hidden” or “hard to reach” population [Adhikari and Bryant, 2018; Tourangeau, 2019; Wallengren, 2021]. Hidden populations are groups about whom little systematic and reliable evidence exists, certainly when it comes to evidence based on existing population surveys. This has to do with the fact that many of these hidden populations are found at the bottom of society [Cobham, 2020]. A few examples of people who are underrepresented or absent from mainstream population surveys are those living in absolute poverty, homeless people or people without a valid residence permit. Besides the hidden nature of the group and the activity, another reason for the lack of knowledge may lie in the illegal or stigmatised nature of their activities. As such, people who beg can also be identified as a hidden population conducting an activity that is not highly valued.

5What knowledge do we possess of begging in Brussels? The previous systematic research into begging in Brussels, now almost twenty years old [Adriaenssens and Hendrickx, 2011; Clé and Adriaenssens, 2007], is based on questionnaires. This study was largely repeated in 2021-2022 [Adriaenssens and Soare, 2023; Soare et al., 2025]. The research documented the fact that, in the study period, the living conditions and background of people who beg were fundamentally different for two groups. Firstly, there was a predominantly male, homeless, local population of beggars; secondly, there were Roma beggars, most of whom were women, who had recently migrated from Romania. The latter were sometimes accompanied by children, and often continued to travel back and forth between the two countries [see also Geurts 2015]. In the past twenty years, however, much has changed. Consider the further expansion of the EU and an increase and diversification in migration flows [Vandekerckhove et al., 2022]. It is only to be expected that these developments will also have impacted Brussels and the population who beg in the city. This has been confirmed by stakeholders in social work. Despite a recent survey [Adriaenssens and Soare, 2023; Soare et al., 2025], a gap remains in many areas between reliable information and scientific research on the one hand, and the social debate on people who beg in Brussels on the other.

6In the light of this gap between the availability of data and the need for it and given the difficulty of effectively reaching groups such as people who beg, the best approach would seem to be one in which several strategies are combined. To begin with, there is a need for a plurality of methods, using both qualitative techniques (such as in-depth interviews and participant observation) and quantifying research (such as questionnaires). Existing administrative and other sources that may contain information about this group are severely underexploited, however.

7In other research on poverty, administrative data has been used with considerable success, for example to document the effectiveness of policy [Meyer and Mittag, 2019; Pattaro et al., 2020]. However, with hidden populations this presents us with a paradox: groups who are not reached by policy are often not included in administrative records either. Consequently, administrative data may be distorted, for example due to a low uptake of rights. In social policy, for example, it often occurs that a significant proportion of those entitled to benefits do not get or apply for them [Noël, 2021; Van den Broeck et al., 2024]. This limits both the effectiveness of policy and the representativeness of the data [Ko and Moffitt, 2020]. A similar problem occurs in criminological research, where victims of crime do not always report the crime to the police [Brame et al., 2010]. Here, for example, it is possible that people in the margins of society are less willing to report crime. If this attrition occurs at sufficient scale and non-randomly, this data on reporting crime or receiving benefits will only give a limited picture of the envisaged population. For a hidden population such as people who beg, therefore, the use of administrative data risks simply reproducing the underrepresentation of this hidden group.

8Besides direct forms of surveying, recording data kept by social workers who focus specifically on vulnerable groups may be a valuable complementary source. For example, the recent survey-based study [Adriaenssens and Soare, 2023; Soare et al., 2025] did not make it possible to compare people who beg with other street people who do not beg. As such, the underlying mechanisms and associations cannot be fully explored. This contribution is intended to help fill that gap in the knowledge of the living circumstances of people and families who beg in the Brussels-Capital Region. We will do so based on an analysis of the database kept by Diogenes vzw, a specialised organisation for the homeless in the Brussels-Capital Region. For this article, we will analyse the data from the past six years (2018-2023). This unique source of information complements what we already know, mainly with respect to the basis of the cross-sectional survey study referred to above [Adriaenssens and Soare, 2023; Soare et al., 2025]. The Diogenes data makes it possible to monitor changes in the composition of the group of beggars over time, and also to compare street people who beg with others who do not.

9A broad definition of begging is asking for handouts in public space [Dean and Melrose, 1999; Fitzpatrick and Kennedy, 2000]. According to the literature, this mainly occurs in the form of a unilateral, non-reciprocal exchange between those who beg and the people who give something. The techniques vary: from addressing passers-by, with or without a story, to the use of a sign or cup to make it clear that a donation is being requested from passers-by. The profiles of people who beg are also diverse [Bose and Hwang, 2002; Ferguson et al., 2016]. In a broader conceptualisation, the activities of street musicians or hawkers are sometimes also considered as begging [Hawkins, 2012; Östman, 2022], because their interaction with the public implies a form of (symbolic) market exchange. Diogenes vzw applies a broad definition of begging: an informal activity, whether or not it is productive, that takes place in public, semi-public or private spaces, where one or more beneficiaries solicit financial or material donations, explicitly or otherwise, from passers-by. This definition is used to compile the data, and as such it constitutes the starting point of this analysis.

10The intention of this article is to provide better documentation of the role that begging may play within a group of people who generally experience a high degree of exclusion. Even within the population covered by Diogenes, there are meaningful differences in the nature and the seriousness of people’s problems. Inspired by the existing literature [for example: Adriaenssens and Hendrickx, 2011; Reinhard, 2023], we assume that the decision to beg is mainly the result of a lack of alternatives, for example due to a precarious residency status or addiction issues. By way of a working hypothesis, we will therefore document whether certain forms of exclusion and problems increase the likelihood that people will beg.

1. Data collection by Diogenes

12Diogenes aims to build bridges between the streets and the rest of society. The target group consists of people who spend much of their time in public spaces. To support this mission, the organisation systematically collects data from three channels:

1. • 2 In comparison: the most recent estimate of the number of homeless people in the Brussels-Capital Re (...)

   Social workers’ diaries: All social workers keep a daily diary, in which they note whom they have seen, where and when; what they did with the person, which organisations they are liaising with and which network meetings they attended. In 2023, Diogenes recorded 1,451 distinct individuals in this way2;

2. Assistance files: When sufficient information is available to identify a person, a file is opened. It contains personal details such as the person’s name, date of birth, nationality, place of birth, gender and place where the person was first encountered, etc.;

3. Annual profiles: Every year, all the outreach workers create a profile of each person with an assistance file. This profile contains about fifty questions on matters including the person’s living circumstances, health and addictions, income and work, family circumstances, accommodation and administrative situation. This information reflects the situation on the last day of the year, or the last known situation in the past year.

13The analyses in this article are based on this latter form of data collection. Diogenes has been issuing this annual profile in its current form since 2017. The data is collected as reliably as possible, partly thanks to training sessions for the social workers.

14The population that Diogenes covers is precisely the population that our data represents. It constitutes a generally vulnerable group of people who spend a lot of time on the street or elsewhere in public space. Not everyone lives permanently on the street. As the data will show, approximately half of the documented group sleep rough in public spaces, and as such they are completely roofless (see also Table 1).

15The data from Diogenes vzw gives a unique insight into the street population of Brussels. Figure 1 shows the number of files filled in for people, whether or not they beg, from 2018 to 2023 inclusive. We see a gradual increase in the number of people recorded. This may be an indication of a growing group of people who have to survive on the streets, and thus a sign of rising, visible precarity in Brussels. However, other interpretations are possible. An improved or extended presence on the streets on the part of Diogenes, or a larger number of contacts per working hour for the social workers, may have the same effect. Over the years of observation, the number of social workers working for the organisation has indeed grown. It is striking that the increase in the number of completed files mainly concerns people who do not beg, which slightly reduces the proportion of people who do.

16To keep the many characteristics somewhat organised and retain sufficient observations, the choice is usually made to simplify the rich data into variables with fewer categories, often even dichotomies. For a person’s housing situation, for example, it has been chosen to distinguish between those who live on the streets and in public spaces with no fixed abode, and others. This is in line with the ETHOS classification [Edgar and Meert, 2006]. The first, most excluded group corresponds to ETHOS category 1, and will be referred to henceforth as “rough sleepers”. It is also important to note that the different forms of homelessness and precarious living situations are usually highly dynamic [Elkins et al., 2024; Scutella and Wood, 2024], whereas the data set only represents a snapshot of the living situation.

17Furthermore, of the many nationalities, the most prevalent have been retained (Belgian, Romanian, Polish) or grouped into a meaningful geographical group (Maghreb). In the data concerning addictions and mental health problems, the many, often subtly different problem have been recoded into the question of whether the social worker has observed the problem.

Table 1. Descriptive statistics

Variable/category	%/median	n
TOTAL (2018-2023)	100 %	5,634
Begs?	37 %	2,106
Year
2018	12 %	683
2019	15 %	872
2020	16 %	895
2021	18 %	1,021
2022	20 %	1,101
2023	19 %	1,062
Gender
Male	68 %	3,848
Female	27 %	1,526
Other	5 %	260
Age (in years)	46	3,452
Roma ?	18 %	994
Nationality (grouped)
Belgian	33 %	1,882
Romanian	20 %	1,116
Polish	12 %	669
Maghreb countries	9 %	485
Other	14 %	776
Unknown	13 %	887
Rough sleepers?	56 %	3,164
Drug or alcohol addiction?	53 %	2,967
Mental health problems observed?	46 %	2,591
Income (grouped)
Full benefits	23 %	1,284
Illness/Invalidity benefits	10 %	586
Unemployment benefits/pension	6 %	364
Income from work	4 %	245
No income	37 %	2,083
Unknown income	19 %	1,072

18Before we look more closely at the people who beg, we would like to present a general picture of the people that Diogenes vzw works with. Table 1 offers a summary of some of their characteristics. In the distribution of personal characteristics, we see that the street population that Diogenes reaches is predominantly male, which is in line with an observation based on the survey of homeless people in Brussels [Pierre, 2025] and in most other high-income countries [OECD, 2024]. Although two thirds of the population are not Belgian, the modal group are Belgian nationals. Poles and Romanians are the other common nationalities. A large minority have Roma ethnicity. The only national group with a background in non-EU countries that is somewhat on the larger side is people from the Maghreb region, mainly of Moroccan origin. Among the non-Belgians are many people with a precarious or undocumented residency status. For EU citizens, the lack of an official address is often an extra obstacle, certainly for those who work informally or beg. This can lead to a vicious circle of exclusion [Aradau et al., 2013; Consoli, 2023; Debruyne, 2022].

19The social workers take stock of several background characteristics of the population that will be relevant to the rest of our analysis of begging. More than half of the group are rough sleepers. Approximately half struggle with an alcohol or drug addiction. The social workers also try to assess whether the people being assisted have mental health problems. Such problems are reported in almost half of them. The type of problem varies greatly in nature and severity, with frequently occurrences of depressive feelings and psychoses. Because begging is aimed at gaining an income, in cash or in kind, it seems logical that possible other sources of income would be relevant to predict this activity. Hence the expectation is that those who can count on an income from other sources are less likely to beg as well. A clear majority of the documented group had either no income (37 %) or their income was unknown (19 %). Only a small minority of 4 % have an income from work, whether formal or informal. All the others have one or another form of benefits. In the regression analyses later on, all incomes from benefits are collated into a single category.

2. Focus on people who beg

20In this section of the analysis, we will focus specifically on the group of street people who beg. We will compare their profiles and living circumstances with those of people who do not beg, based on the data from Diogenes. To begin with, we will explore the data from the perspective of begging, by homing in on people who do and do not beg from a bivariate perspective. That means that we will investigate in each case whether a given characteristic does or does not correspond to begging or not begging as a survival activity. We will test the significance of these differences using standard bivariate correlation measurements: chi-squared tests if the independent variable is also categorical, and a Mann-Whitney test for the relationship between age and begging.

21In the central analyses, we will now bring this collection of covariates together in a single analysis. The goal is the same: to investigate which characteristics of the street population predict whether or not they beg. In this regression, age has been recoded into a categoric variable, because of the large number of missing observations (which are moreover not random). Here, the known ages are divided into four equal quartiles, and a group has also been added of people whose age is unknown.

22We will use a “Linear Probability Model” (LPM) to estimate these relationships. This approach is robust as long as the binary division is not too strongly concentrated in a single group. With a distribution of 37-63 % for those who do beg or do not beg respectively, this does not seem to be the case. To avoid bias, we will estimate the LPM with robust standard errors. Table 2 shows the descriptive statistics for the central variables. In the basic model, the central variables are estimated together. We observed that access to an income is unequally distributed across the characteristics of rough sleeping, addiction issues, mental health issues and ethnicity (see Table 4 in the appendix). For this reason, a second model estimates whether these vulnerability indicators interact with access to an income and the likelihood of begging. This is done by adding interaction terms between income and the four other indicators.

3. Exploration: a profile sketch

23A significant minority of approximately one in three observations concerns people who beg at least now and then. The increase in the number of recorded street people over the years is mainly attributable to people who do not beg (see also Figure 1). This means that the relative share of beggars drops slightly in 2022 and 2023. This relative decrease must be understood, however, against the backdrop of a nominal increase in the number of observations.

24There are clear differences between men and women: a little under half the women beg (45 %), whereas for men that figure is only one in three (35 %). So although men dominate the group as a whole, the likelihood is higher that women within that group will beg. That distinction also remains in the regression. There is no great age difference between beggars and non-beggars. The former are slightly younger, but although the difference is significant, it is small: people who beg are an average of one year and eight months younger than those who do not.

25The most salient characteristic of the begging population is surely that there is a high concentration in terms of ethnicity and nationality among the Roma and people of Romanian nationality. This is largely to be explained by the fact that 88 % of Roma people are identified as Romanians.

26Of all the observations, the social workers report that approximately half have an alcohol or drug addiction. Within this group, the number of beggars is slightly higher than among people without an addiction. It turns out that there is only a limited correlation between rough sleeping and begging; this relationship is marginally significant at most. However, the income situation is a far stronger predictor: people with no income at all beg far more often than people with a different source of income.

Table 2. Bivariate relationship between begging and background characteristics

Variable/category	Share / median begging per group	Relationship with begging (significance)
Entire population	37 %	NA
Year		31.22 (p<0,001)
2018	40 %
2019	38 %
2020	36 %
2021	43 %
2022	35 %
2023	33 %
Age (years)	X̅beg = 44	3.36 (p<0,001)
	X̅others = 46
Gender		110.77 (p<0,001)
Male	36 %
Female	45 %
Other	13 %
Roma?	60 %	267.59 (p<0,001)
Nationality (grouped)		289.90 (p<0,001)
Belgian	33 %
Romanian	57 %
Polish	38 %
Maghreb countries	24 %
Other	37 %
Unknown	25 %
Drug or alcohol addiction?	40 %	20.42 (p<0,001)
Rough sleepers?	38 %	3.21 (p=0,073)
Mental health problems observed ?	38 %	0.48 (p=0,50)
Income (grouped)		534.26 (p<0,001)
Full benefits	33 %
Illness/Invalidity benefits	28 %
Unemployment benefits/pension	15 %
Income from work	14 %
No income	55 %
Unknown income	26 %

For binary variables, the percentage of this group that begs is shown; where there are several groups, the proportion is displayed per group; for the continuous variable of age, the median is given.
Tests: Chi-squared test for categoric variables; Mann-Whitney test for the relationship between begging and age.

4. Results: What links exist between begging and background characteristics?

27The regression analysis (see Table 3) offers a more complete picture of the factors linked to the likelihood that someone will beg. As suggested above by the descriptive and bivariate results, there has been a slight decrease in the proportion of people who beg over the years. It is important to note that there is not so much a drop in the number of people begging on the street (for absolute figures, see Figure 1) as a relative decrease within a growing population of street people who are in contact with social workers.

28As suggested above, the greater likelihood of beggars being female still stands. In terms of age, the differences are small. The likelihood of begging hardly varies between age groups, although it is striking that social workers often have no age data for people who beg. This indicates that social workers have less information on average about who begs; thus those who resort to begging as a survival strategy have a less well-documented status. This may indicate that people who beg are also a more difficult group on average for outreach workers to maintain intense contact with.

29Of the attributed characteristics, ethnicity is the strongest predictor. The group that is most strongly present in begging is without doubt the Roma. If we consider this ethnic group separately, aside from their nationality, this characteristic is one of the best personal predictors of begging or not begging, at least in comparison to the reference category of Belgian nationals who beg.

30The connections between begging and all kinds of situational problems that people who live on the streets may experience are also substantial. Specifically, addictions or mental health problems and sources of income are relevant. Among those for whom the social workers have observed addiction issues, the chance that they will also beg rises considerably. This may be based on all kinds of underlying mechanisms, one of which is probably that people with addiction issues often find it more difficult to obtain an income based on other activities. A comparable dynamic may be at play in people with mental health problems. Likewise for them, there is a significantly higher chance that they will beg. This confirms the image that people often resort to begging when they have complex problems and difficulty accessing other forms of income. It is striking, however, that rough sleeping does not significantly increase the likelihood of begging according to the regression analysis, which suggests that other factors weigh more heavily.

31The strongest predictor of begging is the lack of another source of income. Any source of income whatsoever significantly reduces the likelihood of begging. This supports the thesis that begging is an income-generating activity engaged in as a last resort. The major underlying narrative seems to be that people who have access to other opportunities or income will resort to begging far less readily. However, those who have no form of income whatsoever and live on the streets will often beg because this remains an accessible option when the alternatives are more difficult to attain.

32In a second model, a number of interaction terms were added to investigate whether certain combinations of vulnerabilities further increase the likelihood of begging: between not having access to an income from any source and the characteristics of Roma ethnicity, rough sleeping, addiction issues and mental health issues. In particular, street people with a Roma background and rough sleepers often have no income at all (see Table 4 in the appendix). The interaction terms certainly indicate that a combination of no income with a Roma background and/or addiction issues do indeed increase the likelihood that someone will start begging. Surprisingly enough, it turns out that people with mental health issues and an income beg more often than those without an income. This may indicate a more complex relationship between psychological vulnerability, assistance and begging behaviour. In general, those with no income will beg more often, irrespective of whether they are sleeping rough or not. Of course we must always remember here that the entire sample consists of people who live on the streets, and practically all of them live with no or in precarious accommodation.

Table 3. Regressions of background variables to begging as an activity

	Model 1	Model 2
Year (ref.: 2018)
2019	-.028 (.023)	-.031 (.023)
2020	-.072 (.026)*	-.072 (.026)**
2021	-.045 (.025)	-.049 (.025)
2022	-.122 (.026)***	-.120 (.026)***
2023	-.127 (.026)***	-.118 (.026)***
Age in quartiles (ref.: 1st quartile)
Quartile 2 [36-45]	.046 (.023)*	.043 (.023)
Quartile 3 [46-54]	.050 (.023)*	.044 (.022)*
Quartile 4 (≥ 55)	.012 (.022)	.011 (.022)
Age unknown	.057 (.019)**	.043 (.019)*
Gender (ref.: female)
Male	-.052 (.015)***	-.051 (.015)**
Other	-.122 (.030)***	-.127 (.031)***
Nationality/ethnicity (ref.: Belgian)
Roma	.266 (.023)***	.463 (.031)***
Romanian (excl. Roma)	-.060 (.031)	-.008 (.031)
Polish	-.098 (.023)***	-.059 (.024)**
Maghreb countries	-.202 (.022)***	-.164 (.022)***
Other (excl. Roma)	-.061 (.022)**	-.015 (.022)
Unknown	-.107 (.024)***	-.074 (.024)**
Drug or alcohol addiction?	.185 (.014)***	.236 (.025)***
Mental health problems?	.083 (.014)***	.120 (.016)***
Rough sleepers?	.026 (.018)	.020 (.023)
Income (ref.: no income)
Work	-.447 (.026)***	-.327 (.037)***
Unspecified income	-.279 (.035)***	-.208 (.044)***
Unknown income	-.232 (021)***	-.163 (032)***
Benefits	-.315 (016)***	-.235 (032)***
Interaction terms
Roma X no income		.362 (.037)***
Addiction X no income		.082 (.030)**
Mental health problems X no income		-.107 (.029)***
Rough sleeper X no income		-0,010 (.026)
Interaction terms?	No	Yes
R²	.179	.199
Prob > F	<0.001	<0.001

Effects of a Linear Probability Model with robust standard errors (SE). n=5 634
* p <0.05; ** p <0.01; *** p <0.001
Model 1 estimates all main effects; model 2 adds interaction terms.

Conclusion and discussion

33People who live on the street are often fairly visible in the urban landscape. When it comes to social-scientific evidence, however, they seem to be practically invisible. This leads to street people often being the subject of public discussion, political talks and debates, but these discussions are rarely based on data beyond the anecdotal. The problem is not only the possible preference for a fact-free debate or one in which evidence is scarce. Even for those who would prefer things to be different, the problem remains that reliable data and analyses concerning this excluded group are only available in limited quantities. The statistical sources that aim to document the lives of Belgians or Brussels residents hardly represent this group, if at all. They are uncounted, which adds an extra dimension to social exclusion [Cobham, 2020]. People who live on the streets share this statistical invisibility with many other marginalised groups. This often forces governments and institutions to implement policies in blind ignorance.

34There are few quick wins to be found in this need for data on hidden groups. Much of the research that does try to map hidden groups does so directly, for example by means of questionnaires. Furthermore, researchers today often try to use increasing use of data that is collected by institutions anyway, for example based on government administrative procedures. For highly marginalised groups, this approach often fails to make much of a contribution, precisely because these groups are poorly or incompletely documented by administrative sources, due among other things to the lower uptake of rights.

35This article shows that there may be a rich source of information available from organisations in the non-profit sector who focus specifically on vulnerable groups. Some collect highly valuable data in the context of outreach or for other reasons, which is very relevant to gaining better understanding of the poorest members of society and the policy that affects them. Those with the ambition to pursue effective policies will find a rich source of information in this type of data. By way of a case study, we might mention Bruss’Help’s rich data collection on homeless people [Pierre, 2025]. This article illustrates the usability and feasibility of this strategy, and immediately also demonstrates the type of policy-relevant conclusions a decent analysis can lead to. The data collected by organisations such as Diogenes vzw does not merely offer a rich picture of the world in which street people live. It also makes it possible to analyse trends and differences within this group. In general, it is impossible to rule out the possibility of data being misused. Nevertheless, recent experiences suggest that stigmatisation and repression are at least as likely to stem from a lack of knowledge than from the abuse of data.

36At the same time, there are of course quality requirements that must be considered. As is the case with administrative data, it is possible in some cases that an issue or group is not documented or insufficiently documented. This risk is also present in the data obtained from outreach work. The data set we have analysed here only applies to the street people supported by Diogenes, so it is not necessarily representative for everyone who begs in Brussels. All the same, this data set offers a rare and valuable insight into a group that is difficult to reach. Moreover, the fact that Diogenes is a relatively large organisation, and that their social workers have a fairly large diversity of knowledge about specific linguistic and ethnic groups, reduces this risk considerably. Triangulation with other sources, for example survey-based research among people who beg with a different sample design (such as [Adriaenssens and Soare, 2023; Soare et al., 2025]), or more qualitative approaches such as ethnographic research, remain important to refine the image.

37In any event, the data reported here has the significant advantage that it enables us to compare the street people who beg with members of the same group who do not beg. An important observation based on this analysis is that begging functions as a survival strategy. Indicators of vulnerability and deprivation predict whether someone will resort, or will be forced to resort, to begging as an income-generating activity. Finding and keeping a formal job is extremely difficult for street people anyway, and there are very few observations of this happening. These findings are in line with existing research into informal activities among people in highly precarious living situations but often clash with social myths about the nature of begging and people who beg. This contribution therefore confirms or supports the idea that people who have alternative sources of income are less likely to resort to begging. If begging is a “last resort”, it does of course compete with other options that exist, such as street prostitution or forms of petty crime. In all probability, these are riskier, more harmful and more problematic to society and to the people in question. For assistance providers, this will probably not come as a surprising conclusion, but for policymakers and the general public, it may be an important correction to stubborn myths about beggars. It is also important information to inform future policy.