module 1

48(1) 23–40, January 2011

0042-0980 Print/1360-063X Online © 2011 Urban Studies Journal Limited

DOI: 10.1177/0042098009360688

Stef Adriaenssens and Jef Hendrickx are in the HUB—University College Brussels, Stormstraat 2, Brussels, 1000, Belgium. E-mail: [email protected] and [email protected].

Street-level Informal Economic Activities: Estimating the Yield of Begging in Brussels Stef Adriaenssens and Jef Hendrickx

[Paper first received, September 2008; in final form, October 2009]

Abstract

This article develops and applies a method to estimate the revenues of beggars in Brussels. This is relevant for three reasons. First, in the literature on the informal economy, we lack reliable empirical knowledge of informal street-level activities like begging, substantiating the expectation that beggars’ income will be low. Secondly, popular representation of beggars often depicts them as criminal and wealthy. Finally, recent legislation builds on the idea of criminal organisations behind beggars. Building on an analysis of existing attempts to measure beggars’ income, we aim for a triangulation with data from three different sources: observation, self-reports and quasi-experimental observations. This triangulation allows for more reliable and valid conclusions. Hypotheses based upon popular images and the criminalisation of begging are dismissed. The evidence does support the hypothesis based upon the literature on informal activities.

aspects of their lives that appeal to the public’s imagination.

Our starting-point is that a theory of action is valid if it is able to reconstruct the reasons of the actor. That’s where most popular theories of begging fail. Many everyday judgements about begging build upon assumptions referring to the ‘traditions’ of certain ethnic groups, detrimental effects on the safety feel- ings of the public and so forth. Nevertheless, both the public and policy-makers seem to be

1. Problem and Hypotheses

In many western European cities, begging is receiving growing attention from the public, policy-makers and social scientists (case stud- ies in Donovan, 2008; Fitzpatrick and Jones, 2005; Mitchell, 2005). Thereby, the wildest claims about the nature, the motivations and the income of beggars are made. This paper attempts to elucidate the earnings of people who beg. If one bears in mind that there is a huge gap between the definite assertions and the lack of reliable evidence, this is one of the

24 STEF ADRIAENSSENS AND JEF HENDRICKX

attached to this point of view, as expressed in everyday discourse as much as in discussions in political bodies. Related with this, a large proportion of the public seems to be con- vinced that begging is connected with deceit, fraud and organised crime.

What then are the reasons why people beg? A consistent starting-point for this question is that begging serves the same manifest function working in general has: to yield an income. To be clear: this starting-point does not preclude the existence of deceit, fraud or organised crime in begging; it just situates begging among the meaningful and purposive actions of real people. Whether begging as an income-generating activity is chosen or imposed, is part of the problem investigated in this contribution. The basic starting-point, however, is that the dominant motivation of begging is acquiring means.

We define begging as informal work in a public space, consisting of a receiver asking for a non-reciprocated gift. Begging is informal work in the sense that it is part of

those economic activities that circumvent the costs and are excluded from the benefits and rights ... of formal society (Feige, 1990, p. 992).

In the case of begging, this implies that it takes place within the public space. Like many other street-level informal activities (Dean and Gale, 1999), it has a discordant relation- ship to the formal and mainstream users of this space (other examples in Donovan, 2008; Venkatesh, 2006). The informal character of begging mainly refers to the second part of Feige’s definition: the work of begging nor- mally does not entitle people to formal rights and benefits. Their work does not benefit beggars with legal protection or access to, for example, health insurance, as normally results from formal jobs.

In some countries, regions or cities, beg- ging is explicitly prohibited—for example, in England and Wales (Fitzpatrick and Jones, 2005). In other places, begging is allowed and

sometimes the right to beg is even warranted, as is the case in Belgium and in the US (Hershkoff and Cohen, 1991). In Belgium, the legislator abolished the penalisation of begging and vagrancy in 1994 (Jamar and Herbots, 2006) and since then courts have acknowledged the right to beg (Fierens, 2004). Even so, begging continues to be an informal activity because of its insecure and unregu- lated nature. In fiscal and social security matters, for example, beggars find themselves in a legal no-man’s-land. They derive no rights in terms of social benefits from their activities. Furthermore, security forces often actively suppress begging activities in reality. This has a lot to do with the fact that beggars are users of a fiercely competed public space. Within the informal economy, begging is part of the sub-type of so-called survival activities, denoting that people are immersed in off-the- books transactions because of the destitute economic position they find themselves in (Portes and Haller, 2005).

In this article, we present and apply a method in order to estimate the revenues of people begging. There are three good grounds legiti- mating such an endeavour

(1) Uncovering the earnings from begging improves our understanding of the underground economy and, more specifi- cally, of survival activities.

(2) Common sense about begging is caught in a web of unsubstantiated presuppositions.

(3) Recent legislation in Western countries is based on strong assumptions about the nature of begging. Some of them relate to the income of beggars and can be tested with the help of these estimates.

It is important to stress that the first ground leads to expectations and hypotheses contra- dictory to grounds 2 and 3. We elaborate on all three grounds consecutively.

The first inspiration for this research lies in a general interest in the income from informal activities. It is important to measure

THE YIELD OF BEGGING IN BRUSSELS 25

the living standard of people thrown back on the underground economy for their survival. ‘Underground economy’ is the broader concept, denoting all off-the-books transactions and institutions, including both the informal and the illegal economy (Feige, 1990). The former refers to licit activities that take place off the books; ‘illegal’ activities are explicitly forbidden by law. Begging may both be informal as well as illegal, depend- ing on a country’s legislation. In Belgium, it is informal because the legislator has stated that begging cannot be prohibited.

Under the surface of the huge diversity in approaches of the underground economy, a consensus seems to exist that informal economies are strongly segmented (Pahl, 1987). Informal activities therefore take place within two distinct markets: a top end of relatively affluent workers, often simultane- ously employed in formal jobs, and a bot- tom end with informal work performed by marginalised groups (Slack, 2007; Williams, 2004). For the latter, the main determinant of informal activity is that their access to formal activities is blocked and they lack alternative income-generating opportunities (van Eck and Kazemier, 1988; Williams and Windebank, 2002). This subsistence motive is conceptu- alised as an informal economy of survival (for example, Portes et al., 1989) and correlates with low-quality occupations, low produc- tivity and income (Rosser et al., 2000; Trejos Solorzano and Del Cid, 2003). The statement that poverty and the corresponding lack of opportunities lead to a higher supply of infor- mal work is corroborated by data at a national level—for example, in eastern Europe, where the decline of national income in the 1990’s went hand-in-hand with the growth of infor- mal activities (Renooy et al., 2004). Although most economists studying the informal sector with macroeconomic comparative material hardly pay attention to the subsistence motive (for example, Friedman et al., 2000; Schneider, 2007), they generally do acknowledge the

relevance of national income and the preva- lence of poverty as a determinant.

In short, the literature on informal work and begging tends to classify begging as a survival activity. The overall conclusion would be that the income from begging is considerably lower than the income from formal work. Given the unattractive and harsh nature of begging (Smith, 2005), one can assume that it will even be lower than most other informal activities.

The second ground for estimating the income from begging has to do with ubiqui- tous and recurrent everyday judgements of begging. The representation of people who beg throughout the centuries is consistently built upon three associated images (Erskine and McIntosh, 1999): fraudulent beggars— for example, using children or shamming dis- abilities to evoke pity; ‘professional’ impostors working in an organised criminal network; and beggars acquiring great wealth. These three images reappear from the 16th century on (Geremek, 1980; Woodbridge, 2002). This second ground gives rise to expectations contrary to the first inspiration: the income from begging is comparable with profits from criminal activities and will be higher if more indicators of fraud are observable.

Thirdly, there exists a policy rationale for this research. The second ground refers to collective mentalities that inspire collective behaviour, such as state regulation. This has recently happened in quite a few Western countries, both in North America (Ellickson, 1996; Hopkins Burke, 2000; Mitchell, 2005) and in Europe. Similar penal laws were recently adopted in France1 and Belgium2 criminalising so-called organised begging. Thereby the explicit parallel is drawn with organised criminal activities such as pros- titution and human trafficking. The argu- mentation for the law assumes that begging is as rewarding as these criminal activities. Criminal organisations are also assumed to employ profit-maximising strategies, for

26 STEF ADRIAENSSENS AND JEF HENDRICKX

example, forcing children to accompany adult beggars, as children supposedly increase gifts substantially. The explanatory memorandum of the Belgian enactment3 arguing for the necessity of the law provides illustrations

This (act) does not have as a goal to criminalize the offence of begging again, but to punish those who exploit the begging of others, analogous to the legislation existing in prostitution (p. 4).

After the example of the exploitation of prostitution, the exploitation of mendicancy can be looked upon from the angle of human trafficking (p. 16).

The law decrees that begging with a minor is an aggravating circumstance of organised begging (article 433quater of the penal code). Article 433ter assumes that the assistance of children is inspired by the intention to “evoke pity of the passer-by” thus increasing the income of the beggar. The close resemblance between the legislative logic and popular images should be clear.

Summarising, the motivation to research the income of beggars is grounded in the social-scientific line of research on informal economic activities and in the popular images concerning begging and beggars, also leading to formal rule-making. The basic hypothesis refers to the income of beggars. Literature on informal work leads to expectations con- tradicting the popular judgements and law- making. The former hypothesises that begging generates a low income. From the latter, we infer that beggars yield a higher income from their activities, similar to, for example, human trafficking or the exploitation of prostitution. The criminalisation of (the exploitation of ) begging is based on strong assumptions with little empirical foundation. Therefore, we build hypotheses that allow us to falsify these assumptions. As a direct measure of ‘exploita- tion’ and ‘criminal organisation of begging’ is impossible, we use a detour to falsify the assumptions underlying recent legislation.

We therefore hypothesise that: begging yields profits comparable with human trafficking and prostitution; and, yields are higher if begging is linked to human trafficking and if there are indications of fraudulent strategies.

Four hypotheses will be tested with respect to the exploitation of begging: two referring to the overall profitability of begging activi- ties; one referring to the difference between the profitability of begging for indigenous persons relative to migrant east European beggars; and finally one referring to the sur- plus income by the assumed properties of exploitation, in particular by begging with children. The hypotheses are stated as follows

(1) Begging generates an income under or around the poverty line.

(2) The profitability of begging is comparable with other criminal activities.

(3) Beggars who have migrated yield a higher return.

(4) Begging with children yields a higher return.

The first and the second hypotheses refer to the discussion as to whether begging is able to generate high profits. There is a contradic- tion here between the expectation based on the social-scientific literature (hypothesis 1) and the second hypothesis inspired by the prevailing collective images (as a motivator for contemporary law-making). Provided the yields of begging are so low that it hardly allows beggars to earn an income above the poverty line, the informal-sector research is confirmed. If the yields are higher, this may lay an empirical foundation for the image about the high profitability of begging. The confirmation of the second hypothesis would be consistent with the existence of criminal organisations behind begging.

The third hypothesis refers to the assump- tions of recent legislation that there exists a close connection between human traffick- ing and ‘organised’ begging. Therefore, one expects that the yields of beggars who have

THE YIELD OF BEGGING IN BRUSSELS 27

migrated are higher in comparison with those of indigenous beggars. The final hypothesis tests the assumption in recent Belgian and French penal laws that children are brought in with the intent to increase revenues.

Before we build our estimates that allow us to test these hypotheses, we first give an overview of previous attempts to measure earnings from begging, sketch the context and features of begging in Brussels, and describe the different sources of data for our estimates.

2. Previous Attempts to Measure the Income of Beggars

The serious studies of the life of beggars are often based on qualitative data (for example, Danczuk, 2000; Fitzpatrick and Kennedy, 2000; Lankenau, 1999; Wardhaugh and Jones, 1999). These studies were helpful to reveal the experiences and perceptions of people who beg, but they cannot provide reliable estimates of their income. Therefore, we concentrate ourselves on a quantitative research strategy. Reports of systematised attempts to estimate

the beggars’ income are scarce. Moreover, the existing approaches have fundamental weaknesses. Given the concealed nature of the activities and the hard-to-reach population of beggars, these shortfalls are partly inevitable. Notwithstanding, an overview of the existing methods will be a useful start in order to list feasible approaches and the shortcomings to avoid. Roughly speaking, data are either based on standardised questionnaires or on observations. Although observations and self-reports have serious biases, most studies only use one source of data. We will argue that a mixed-method approach is to be recom- mended. An overview of the reviewed studies is presented in Table 1.

There is one peculiar older illustration of questionnaire-based self-reports: in 1932–33, two sociology students conducted a survey of people who beg in Shanghai (cited in Lu, 1999). Respondents were asked to list their families’ monthly income at the time of the interview and from their previous occupa- tion. Lu’s discussion does not mention the sampling strategy used. Some contemporary

Table 1. Studies with estimates of beggars’ income

Study Method Sample Yield measured

Jiang and Wu; in Lu Questionnaire Sampling unclear Self-reported (1999) Unit: people who beg n = 700 monthly income in Shangai

(Murdoch (1994) Questionnaire Sampling unclear Self-reported daily Unit: people who beg n = 145 income from begging in central London

Bose and Hwang (2002) Questionnaire Sampling: systematic Self-reported Unit: people who beg search of public places income in Toronto n = 54 Lowest payment for interview with high response rate

O’Flaherty (1996) Questionnaire Sampling: systematic Self-reported Unit: ‘daytime search at well-known maximum and streetpeople’ in locations minimum daily Manhattan n = 209 earnings

Butovskaya et al. (2004) Observation of people Observations of Number of gifts in who beg during beggars in Moscow trains 2 minutes. 2 minutes n = 178

28 STEF ADRIAENSSENS AND JEF HENDRICKX

studies also provide little information about the crucial issue of sampling. For instance, Alison Murdoch’s research report (1994) only mentions where respondents were inter- viewed, but remains unclear about the criteria for their selection.

Others do document the sampling strategy. Bose and Hwang (2002) researched begging with the help of a standardised questionnaire in Toronto. The researchers “located panhan- dlers by systematically searching major streets and subway stations” (p. 477). They estimated the income of beggars through self-reported hourly, daily and monthly yields. The general conclusion of this research was that begging is the main source of the respondents’ income, but that it brings in rather meagre revenues. The Toronto study attempted to test the reli- ability of self-reported income of people who beg through offering different amounts of compensation for co-operation in the inter- view. According to the researchers, the estab- lishment of the lowest amount with a high response rate would serve as an indication of their income. This interesting approach could hardly be tested because the number of respondents was rather low (n = 54).

In the same vein, O’Flaherty (1996, p. 82) systematically interviewed the “daytime street- people” in New York’s Manhattan by cruising the well-known locations during several week- ends. The author does not communicate the proportion of beggars within the sample, but he did question the survival strategies, with begging as one of seven categories. The general conclusion is consistent with the study by Bose and Hwang: low earnings for long and hard work (O’Flaherty, 1996, pp. 84–85).

The basic disadvantage of self-reported measures obviously has to do with mem- ory effects and socially desirable answer- ing. Many studies on people who beg (for example, Melrose, 1999) and other excluded groups or hidden populations (Sifaneck and Neaigus, 2001) report the distrust towards outsiders. Beggars often confuse interviewers with officials, fearing that telling the truth

about their income may lead to sanctions. In our fieldwork, we also noticed that respon- dents had problems with questions about their average income, probably due to the lack of registering of their income; usually the yields are immediately consumed. These drawbacks made us decide only to use self- reported measures for information that can- not be attained otherwise.

From the observational method, we found one example: Butovskaya et al. (2004) used it to compare the amount of gifts received in a fixed time-span by people who beg in Moscow. Basically this is an interesting approach that may be able to overcome some of the weak- nesses of self-reported income measures. However, the linear relation between the num- ber of gifts observed in a fixed time-interval, on the one hand, and the income of a beggar, rests on two strong assumptions.

First, it assumes that the alms received have the same mean value for each (type of) beggar. The researchers have no data supporting this assumption and neither do they propound convincing arguments for this a priori (in fact, the problem is not addressed). Other research does indicate that most of the alms received are rather small, mainly consisting of coins, literally ‘spare money’ (Adler et al., 2000; McIntosh and Erskine, 1999). This does not exclude the possibility that the mean gift var- ies considerably between beggars. Alms-givers may give different sums to different types of beggars, or a certain specialisation of alms- givers may exist, related to their perceptions of ‘deserving poor’.

Secondly, one should be aware that the frequency of observed gifts in a given time- period is an indicator of productivity, not of income. The use of productivity as an indicator of income passes over the probable differences of working time between beggars.

Summarising, data based on observations avoid some of the drawbacks of self-reported data. Observing beggars and their alms gives access to reliable data on the frequency of gifts, but information on the working time is lacking.

THE YIELD OF BEGGING IN BRUSSELS 29

3. The Context: Poverty and Begging in Brussels

The Brussels Capital Region is the part of Belgium with a high concentration of extreme poverty. The most recent study available esti- mates the number of homeless people in the region at 1200 (Réa, 2001).

One important background regulatory fea- ture is the legal status of begging in Belgium. Since the start of begging regulation, different measures have been used: periods of penali- sation, assistance and institutionalisation alternated and sometimes even occurred in the same period (Depreeuw, 1988). In 1891, the most recent law (until further notice), prohibiting vagrancy and begging, was pro- mulgated. It lasted until 1993. By that time, there was an overall political consensus on the inhumanity of criminalising people who beg, leading to the abolition of this criminal law.

What does the begging population look like in Brussels? The survey we conducted taught us that the great majority (85.4 per cent) of Brussels beggars fall into three types: male indigenous beggars and female Roma beggars alone or accompanied by children.4 Indigenous beggars are those born in Belgium or with an official language of the Brussels capital region as a mother tongue (French or Dutch). Members of this group are often homeless and have a history of drug or alco- hol addiction. This is similar to the profile of people who beg in Britain, according to Danczuk’s study (2000).

The background and issues of Roma beg- gars are fundamentally different. The Roma in Brussels originate from Romania, the larg- est Roma population in central and eastern Europe (Ringold et al., 2003, p. 89). They pre- dominantly migrated recently, definitely after the fall of the Iron Curtain and the implosion of the communist regimes there. Important push factors are the economic backwardness of the region in comparison with western Europe, high unemployment and poverty, in particular for Roma (UNDP, 2003). Around

three-quarters of the interviewed Roma indicated that they were unemployed (UNDP, 2003). They also suffer from discrimination and racial violence, exacerbating the hopeless- ness of their situation (OSCE, 2000).

4. Methods and Ethical Aspects

In order to estimate the beggars’ income as precisely as possible, we constructed a design based on the conclusion that self- reports and observations have distinct limi- tations. Therefore, we used both methods, complemented with a quasi-experimental version of participant observation. These three distinct sources correspond to the different data we need in order to estimate the income of beggars.

The begging time is assessed through stan- dardised interviews with beggars in the Brussels capital region, conducted in the autumn of 2005 and the spring of 2006 (n = 268). Three typical problems in questioning beggars arose: the absence of a register, the volatility of beg- ging and the difficulties of questioning beggars.

First, no register of beggars exists. As earn- ings are dependent upon traffic, access to the population of beggars was constructed with the help of a detour through the places where people beg. A register of 255 possible begging locations was constructed with the help of volunteer reports, police reports and a list of public marketplaces, subway stations and supermarkets of the major chains.

A second potential problem was the assumed short-term variation in the begging popula- tion. The precarious judicial status of many people who beg, the possible transience of migratory beggars and the irregular approach by the police force all support the assumption that begging is a volatile phenomenon. The choice for a register of begging places also bears the risk that respondents were not beg- ging at the time the location was observed. Therefore, each location was visited three times at different moments of the day and in the week. Furthermore, the researchers chose

30 STEF ADRIAENSSENS AND JEF HENDRICKX

to conduct two waves of interviews: one in the autumn of 2005, one in the spring of 2006, preventing our data from being overinflu- enced by seasonal coincidences. Finally, we took measures to overcome the inaccessibility of people who beg. This group is hard to reach due to general distrust and the linguistic and ethnic diversity. There was an expectation of a significant proportion of analphabetic respondents (afterwards confirmed by the data). Face-to-face interviews guaranteed the participation of illiterate respondents. Linguistic diversity was overcome by a ques- tionnaire in four languages (French, Dutch, English and Romanian) and interviewers mastering these languages. In general, this proved to be an effective method: 85.8 per cent of the respondents agreed to be interviewed.

In order to estimate the mean frequency of gifts, data were collected through observa- tions of the three types of beggar.5 People begging were randomly selected in an area in central Brussels. The observations were made in crowded places such as the central station or the Rue Neuve. Observers posi- tioned themselves at a fair distance from the beggars, preventing interaction. Because of the crowded nature of the spaces, this hardly received attention. The researchers recorded the exact time of alms collected by beggars during 60 sessions of 36 hours in total. The duration of the observed period was quite uneven, as the researchers had no control over the beggars or their context. During these ses- sions, 225 gifts were recorded. The duration of the observations was divided evenly for each type of beggar.

The data from this second source were meant to estimate the mean begging time beggars of all three types needed to get a gift in kind or a gift in coins. However, it proved possible to determine the value of the gifts in kind or notes through these observations. Therefore, the income in a given period of begging time from gifts in kind and notes was estimable on this source of data alone. As the observation

of real beggars is a more reliable and therefore superior source of information than data of test subjects simulating begging, we preferred to rely on the former as much as possible. The reason why gifts in kind and in notes are taken together thus is based on a methodological rather than an intrinsic communality: both were measured by observations of beggars.

The third source of data was necessary in order to estimate the mean value of the gifts in coins. For the estimate of the value of gifts in coins, a quasi-experimental use of observation was set up.6 We are well aware that the denotation of ‘quasi-experimental use of observation’ is a rather inelegant formula- tion. The reference to observation was added because the design has no causal ambitions whatsoever. On the other hand, as one element is actively manipulated—the exposure of the public to a certain kind of beggar –legitimises the reference to quasi-experimentation.

Begging activities were simulated in and in the vicinity of the Rue Neuve, an important commercial area in central Brussels with frequent begging activities. Six experimental subjects engaged in begging activities during sessions of two hours. Thereby, Roma female and indigenous male beggars were simulated. The third type of beggar, female Roma accom- panied by a child or children, has not been included in the design for ethical reasons. The six test subjects consisted of four male and two female test subjects. They begged during three sessions of two hours each. For every gift in coins, the test subjects recorded the value, the time and some background variables of the alms-giver with the help of a small hidden microphone. The test subjects were watched by an observer, for support and to have a backup record of the timing of the gift and the characteristics of the alms-giver. Because they were mainly a backup for security reasons and in order not to arouse suspicion, the observers posted themselves at quite a distance. In total, 149 gifts in coins were recorded during these sessions, allowing us to estimate mean and

THE YIELD OF BEGGING IN BRUSSELS 31

the distribution of the value of alms in coins for the main types of beggar.

The combination of the data of the second and the third sources allows us to estimate the return of begging activities in a given time-period. This provided us with the necessary information to measure the mean and distribution of the frequency and values of the distinct types of gifts (coins, notes, in kind), itemised per type of beggar. The data obtained from the questionnaire are used to estimate the mean working time our respondents ‘work’. This allows us to make the inference from productivity of begging to estimated income.

Research into informal and underground phenomena poses difficulties with regard to measuring and method, but it is also cause for obvious ethical concerns. This is all the more the case for a vulnerable group such as homeless people or people who beg (Melrose, 1999; Williams and Cheal, 2002). Our main ethical concerns were twofold: to prevent adverse effects on people who beg and to avoid insecure situations for the researchers.

The aim not to divert earnings from people who beg was achieved through two interven- tions. When interviewing the begging popu- lation, respondents were offered a payment of 5 € in exchange for their collaboration because the interview took working time. Secondly, the alms received during the quasi- experimental observations obviously were diverted from earnings of people begging in the vicinity of our researchers. Therefore, the takings were redistributed to people begging in the direct vicinity of the places where we begged.

The second ethical concern affects the safety of the researchers and in particular those imi- tating beggars. During the sessions, the test sub- jects were watched all the time by an observer. This allowed for support in case of problems. Although the police were informed in advance of the research, the test subjects behaved like other people who beg when chased of by the

police or private security companies. In case someone was arrested, the observers did carry a letter from the chief of police clarifying the aim of the begging activities.

5. Estimates

The calculation of the income of begging in a given time-period is possible with data measuring the value of the alms and their frequency. In order to estimate the income from begging, one also needs information about the begging time. Evidence for the first is mainly collected through observation, the second by the quasi-experimental observation and, for the final information, we made use of the results of the survey.

The calculation was complicated because of the variety of alms beggars receive. Basically they receive gifts in money, mainly coins and sometimes notes and in kind. The latter type consists of a wide variety: cigarettes, food, soft drinks, and sometimes even utensils. The income of beggars in a given time-period t (Y(t)) thus equals the sum of the value of gifts in coins (YC(t)), in notes (YN(t)) and in kind (YK(t))

Y(t) = YC(t) + YN(t) + YK(t)

The mean income for every term is the mean value of the respective gifts multiplied by the mean frequency of gifts in a time-period t

mY(t) = mC . mNC(t) + mN . mNN(t) + mK . mNK(t)

where, mC, mN and mK denote the mean value of gifts in coins, in notes and in kind; and mNC(t), mNN(t) and mNK(t) denote the mean number of gifts in coins, in notes and in kind in a time-period t.

These estimates build on observations and quasi-experimental participant observation. The income of begging equals the value of the alms received in a fixed time-period (for example, per hour) multiplied by the begging time.

32 STEF ADRIAENSSENS AND JEF HENDRICKX

The most recurrent type of gifts consists of coins (81.8 per cent). We calculated con- fidence intervals for the mean gift in coins respectively in kind and in notes (Table 2). This way it is possible to control for whether there is a significant difference in the mean value of gifts between the groups. First, we check whether the mean value of gifts dif- fers significantly between the three groups (Roma alone, Roma with children and indig- enous beggars). This is done with the help of ANOVA tests and two sample t-tests. This is not the case for coin gifts between Roma alone and indigenous beggars (two sample t-test p-value = 0.84) or for the mean value in kind or notes between the three types of beg- gar (ANOVA-test f = 0.073, p-value = 0.93). The normal distribution cannot be rejected on the basis of a Kolmogorov–Smirnov test. Combining these results, a 95 per cent confi- dence interval for the mean value of the gift in coins is estimated [0.66, 0.89]; for the gifts in kind or notes it is [0.95, 2.03].

Next, we estimate the frequency of gifts, analysing the interval times between the gifts (Table 3), modelled by an exponential dis- tribution.7 A 95 per cent confidence interval for the mean of an exponential distribution is calculated (Festinger, 1943)

, nX nX2 2

, . , .n n 2 2 0 975

2 2 0 025| |

; E

where, X −

is the sample mean (in minutes) and c22n,0.025 (respectively c22n,0.975) denotes the 2.5 per cent (respectively 97.5 per cent) percentile of a chi-squared distribution with 2n degrees of freedom. The results of these estimates are shown in Table 3.

To compare the mean interval time of Roma with child(ren) with Roma alone, we use the following test-statistic for two independent exponential distributions

F X

X

2

1 =

Under the null-hypothesis that the population mean of the two distributions are equal, F has an F-distribution with degrees of freedom 2n1 and 2n2. X

− 1 and X

− 2 denote the sample

mean of the first and second sample; n1 and n2 represent the sample size of the first and second sample.

There is no significant difference in the average value of the interval times between Roma with child(ren) and Roma alone.8 From the F-test comparing the mean interval time, we infer a significant difference between the Roma beggars and the indigenous beggars.

Table 2. Confidence intervals for the value of gifts (in €)

95 per cent confidence interval for mean

N Mean S.D. Lower bound Upper bound

Gifts in coins Roma alone 55 0.76 0.69 0.58 0.95 Indigenous 94 0.79 0.72 0.64 0.93

Gifts in kind or notes Roma with child(ren) 12 1.55 0.35 0.78 2.33 Roma alone 8 1.58 0.61 0.14 3.01 Indigenous 9 1.34 0.51 0.16 2.51

Gifts overall Gift in coins 194 0.78 0.71 0.66 0.89 Gifts in kind or notes 29 1.49 1.41 0.95 2.03

Sources: quasi-experimental observation (gifts in coins) and observation of beggars (gifts in kind and notes).

THE YIELD OF BEGGING IN BRUSSELS 33

The p-value is 1.2 × 10−11 for gifts in coins and 0.034 for gifts in kind or notes. This allows us to start from confidence intervals for Roma begging alone or with children on the one hand and for the indigenous on the other hand.

The mean income per hour (Table 4) is a result of the average inter val time between gifts and the average value of a gift combined. The confidence interval of the ratio of two means is calculated with the help of the method of Fieller (Fieller, 1940; Motulsky, 1995). The confidence interval for the total income per hour is

measured through a combined estimation for the income in coins and the income in kind or notes. The method to calculate a confidence interval for the sum of two population means is similar to the formula for the confidence interval for a difference between two population means.

Finally, the income per day is estimated with the help of the survey data of the mean begging time per day (Table 5). In order to combine this self-reported working time with the other data, we start from the assumption that 90 per cent of the reported begging time is productive working time; the remaining

Table 3. Interval times (in minutes) for the gifts

95 per cent confidence interval for mean

N Mean S.D. Lower bound Upper bound

Gifts in coins Roma with child(ren) 38 18.32 21.15 13.65 25.88 Roma alone 38 17.53 16.65 13.06 24.77 Roma overall 76 17.92 18.91 14.49 22.75 Indigenous 109 6.56 7.72 5.49 7.99

Gifts in kind or notes Roma with child(ren) 13 53.54 61.01 33.20 100.55 Roma alone 8 88.00 61.42 48.81 203.83 Roma overall 21 66.67 62.03 46.70 116.49 Indigenous beggars 21 34.36 31.27 23.36 55.50

Source: observation of beggars.

Table 4. Income (in €) per hour

95 per cent confidence interval for mean

Mean S.E. df Lower Bound Upper Bound

Gifts in coins Roma 2.60 0.356 223 2.01 3.47 Indigenous 7.10 0.863 256 5.61 9.11

Gifts in kind or notes Roma 1.34 0.377 48 0.76 2.57 Indigenous 2.61 0.731 48 1.48 4.98

Total income Roma 3.94 0.583 2.77 5.12 Indigenous 9.71 1.243 7.21 12.21

34 STEF ADRIAENSSENS AND JEF HENDRICKX

10 per cent is invested in organisation and preparation. To find a confidence interval for the income per day, we use a confidence interval for a product of two population means (Wold, 1974).

In order to be able to compare the daily income of beggars with the poverty line in Belgium, we estimated the necessary income per working day in order to avoid poverty. According to research by the OECD (2004), a ‘typical’ employee in Belgium works 200 days a year. Therefore, a daily income of 49.32 € is necessary in order to evade poverty. As beg- ging is often hampered by rain, the cold or police actions, there is no proof that the beg- gars in fact work the same amount of days.

The estimates in Table 5 provide us with the basis to test the four hypotheses we for- mulated earlier.

Hypothesis 1: Begging generates an income under or around the poverty line.

On the basis of the informal work literature, we expected that begging is a survival activity. Therefore, the yields should not exceed the lower revenues of formal work, or even stay below them. We chose as a basis of compari- son the poverty line that refers to the overall income distribution, often applied in the European Union. A person is poor whenever his income is lower than 60 per cent of the median income (Boarini and d’Ercole, 2006; Ruggeri Laderchi et al., 2003).9 Table 6 indi- cates that Roma beggars stay far below the poverty line. The depth of their poverty is even more serious than illustrated here, as we based our comparison on the poverty line of a single-person household. This is a realistic

Table 5. Self-reported average begging time and income per day

95 per cent confidence interval for mean

N Mean S.D. Lower bound Upper bound

Self-reported average begging time (in hours)a Roma with child(ren) 49 4.43 1.79 3.91 4.94 Roma alone 40 4.76 2.31 4.02 5.50 Indigenous 45 5.99 3.57 4.92 7.06

Estimated income per day (in €) Roma 16.26 2.523 11.18 21.33 Indigenous 52.35 8.183 35.90 68.81

a Source: survey.

Table 6. Estimated income of beggars as percentage of income of prostitution and of the poverty line

95 per cent confidence interval for mean

Mean Lower Bound Upper Bound

Percentage of the minimum income of prostitution Roma 4.09 2.82 5.37 Indigenous 13.18 9.04 17.33

Percentage of the 60 per cent poverty line Roma 32.97 22.67 43.25 Indigenous 106.14 72.79 139.52

THE YIELD OF BEGGING IN BRUSSELS 35

option for the indigenous beggars. The Roma, on the other hand, often have young children to take care of (65.5 per cent of the female Roma respondents in our survey) and they usually depend on the begging revenues as their only income.

The situation of the indigenous beggars looks somehow different. It is impossible to rule out that indigenous people who beg might avoid poverty due to their income from begging. In addition, a majority of this latter group (72.3 per cent) enjoys some welfare benefit in addition to their income from begging.

Hypothesis 2: The profitability of begging is comparable with that of other criminal activities.

A direct verification of this assumption would require considerable police resources. An indirect test is possible, however. From the assumption that criminal entrepreneurs seek to maximise profit, we expect that the exploitation of people who beg will yield considerable gross revenues in order to be an attractive activity. Therefore, we attempt to compare the gross revenues of begging with those of illegal or semi-illegal activities. After an analysis of the available literature, it was clear that only few reliable estimates exist of the revenues of illegal or semi-legal activities. Moffat and Peters (2004) published a detailed, recent and reliable estimate of the gross revenues of prostitution in the UK. The mean price of an encounter was £55 in 1999 (£60.38 or 87.61 € in 2006 prices); the mean time was 30 minutes. Prostitutes have an average of 21 (window prostitution) to 25 (streetwalkers) encounters a week. The comparison with begging shows a massive difference, lending no support whatsoever for the assertion that begging is a serious candi- date for a criminal entrepreneurial strategy. On the basis of the Brussels data, there is a rather strong support for the opposite asser- tion: gross earnings from begging are so low

that they probably attract little attention from criminal organisations.

Hypothesis 3: Beggars who are migrants yield a higher return.

This hypothesis is inspired by the legislator’s vision that ‘organised begging’ is closely linked to human trafficking. This assumption is not supported either. The migrants who beg, pre- dominantly female Roma, have a consistently lower productivity than indigenous beggars. Both in frequency and in mean value of the alms they receive, Roma come off worse. In general, they also seem to invest less time in their begging activity than indigenous beggars.

Hypothesis 4: Begging with children yields a higher return.

Our estimates do not seem to indicate a sur- plus value of begging with children. Because begging with children was not simulated in our quasi-experimental observations, this statement is based on begging time and frequency of gifts. There is no significant difference in the frequency of gifts between female Roma beggars with children and Roma women begging alone; the value of gifts in kind and in notes is not higher. Finally, the working time of people who beg with children is not longer than those working alone.

6. Conclusions and Discussion

First and foremost, it is important to stress the limited possibilities to generalise this research. Due to the impressive variety of beg- ging contexts, and because of the general lack of knowledge, it is not possible to generalise the findings of this study straightforwardly to other places and times. Applying these conclusions to other cities and contexts there- fore cannot be done without the necessary reservations. The relevance of this research rather is inductive: it provides a consistent and empirically based starting-point for research elsewhere.

36 STEF ADRIAENSSENS AND JEF HENDRICKX

This paper primarily develops a method to measure the yields of begging. Basically, the strategy is built upon a careful assessment of the available methods in the literature. Two strategies were found: observation and self- reports. Self-reports lack reliability because of their sensitivity to sampling errors, memory effects, socially desirable answering and non- response. Observation may overcome these weaknesses. However, because only the fre- quency of gifts can be observed, this approach is limited in the scope of data that one can collect. Therefore, we used a combination of observations, self-reports and a third source of data: quasi-experimental observation, simulating a person who begs in order to reconstruct the value of gifts. Triangulating these data, we estimated the income of three groups in the population of beggars: indig- enous male beggars, Roma migrant women alone and with children. These three groups constitute the large majority of beggars in Brussels.

The question is whether the applied method leads to valid and useful results. In addition to the consistency of the method, the conclusion that the results are consistent with the social-scientific theories and that the estimates lead to significant differences between the groups, do seem to support the supposition of validity and usefulness. The development of a multimethod approach to tackle the problem seems to survive the empirical confrontation well. As expected in the design, the combination of data from different sources in a careful design allows for more insight. In general, this approach is an extension of recent pleas to measure infor- mal economic activity with approaches that are close to the problem at hand (Alderslade et al., 2006) and arguing against an approach that builds upon macro-economic or macro- sociological heroic assumptions (for exam- ple, Thomas, 1990).

The hypotheses are based on three sources: the literature on informal work as a survival

activity, the popular social myths about the nature of begging and the assumptions underpinning recent legislation that crimi- nalises some forms of begging. The informal economy literature expects that the yields of begging will be rather low, probably under or around the poverty line. These hypotheses based upon the social-scientific literature are antithetical to the popular images and the discourse legitimating legislation. The latter assume that begging often is a fraudulent activity, organised by criminal groups, with high profits. Recent legislation in France and Belgium builds upon the assumptions that some criminal groups coerce people into begging, often trafficked migrants, and that these criminal groups also make use of children accompanying the beggars in order to evoke pity. Based on these popular beliefs and the rationale for legislation, we hypothesised that beggars generate returns comparable with those of other criminal activities, and that migrant and child beggars yield higher returns.

Our data do not support the hypotheses inspired by popular beliefs and legislation. The estimates seem to indicate that they can be categorised as myths. On the other hand, our findings are consistent with the hypoth- esis based on social-scientific literature on informal work, at least for what concerns the Roma who beg. For the indigenous people who beg, the results are inconclusive: their earnings from begging surpass those of Roma and it is not certain whether it is impossible for them to evade the poverty line by begging.

The most striking element for policy issues probably is the conclusion that the produc- tivity of Roma with children is comparable with those who beg without children. The question is how this can be explained. Observations and public discussions indicate that the presence of children arouses intense feelings. We argue that these intense feelings lead to two types of behaviour with more or

THE YIELD OF BEGGING IN BRUSSELS 37

less compensating effects. Some passers-by may give more frequently to begging people who beg with children, while others may refrain from giving because of the presence of children.

The second question is what determines the significant gap in income between Roma and indigenous beggars. At first sight, this seems to indicate a negative view of the public towards Roma. There are indeed plenty of case reports and observations of negative feelings towards this group. Passers-by may have an unfavour- able image of Roma beggars, due to the per- sistent stories about ‘organised begging’ and exploitation, or even to a general xenophobic distrust (as argued in Butovskaya et al., 2004). However, the difference may also be caused by the mere difference in quantity. The fact that there are more Roma than indigenous beggars may lead to a lower income, assuming that more or less equal parts of the public are willing to give alms to indigenous people and to Roma people begging.

The final paragraphs of this discussion are reserved for the policy conclusions that can be drawn from this research. It is a return- ing theme to discuss the recent tendency to criminalise begging and other activities of the poor. Although one should point to the continuity of legislation penalising begging (Baker, 2009), legitimating criminalisation is built upon a different logic in different times and places. In the US and in Canada, for example, the dominant logic is planning and structuring public space, thus effectively circumventing campaigns that defend the rights of the urban poor (Blomley, 2007). In continental Europe, the new forms of legislation build their discourse upon the protection against exploitation of people who beg. The latter logic is built upon a number of heroic assumptions about the nature and motivations behind begging, linking up almost perfectly with the older myths about beggars and fraud. The paradox is that, despite the difference in logic, both types of

resulting regulations turn their weapons on the people who beg.

Assessments of these different types of regulations can be and should be based upon different logics also. While an appraisal of the internal (in)consistency or the hid- den logic behind the legislation often is an effective strategy (Fitzpatrick and Jones, 2005; Mitchell, 2005), this paper follows an alternative logic, confronting the empirical assumptions with the real life of people who beg (compare Fitzpatrick and Kennedy, 2000; Kennedy and Fitzpatrick, 2001). The overall policy conclusion is that the recent legislation tackles a problem that does not exist, or that is trifling and ephemeral at best. The evidence suggests that the most pressing problem of the begging population, and in particular of the Roma, is their astoundingly low standard of living. If the estimates of their earnings prove anything, it is that Roma people who beg are primarily in need of social support instead of criminal disciplining.

Notes

1. Article 706-55 of the French Code of Penal Procedure, adopted 12 February 2003.

2. Article 433ter and 433quater of the Belgian Penal Code, adopted 10 August 2005.

3. Chambre des Représentants de Belgique (14 January 2005), Projet de Loi modifiant diverses dispositions en vue de renforcer la lutte contre la traite et le trafic des êtres humains, Document parlementaire de la 51e législature, no. 1560/001.

4. The design of this survey and other data collected are explained in the methods section.

5. Data collected 9 November 2005 to 10 February 2006.

6. Data collected 17 October 2005 to 31 January 2006.

7. Confirmed by a Kolmogorov–Smirnov test. 8. P-value = 0.85 for gifts in coins and p-value =

0.25 for gifts in kind or notes. 9. At the time of our data collection, the 60 per

cent poverty line for an individual in Belgium equalled 822 € per month, or 9.864 € per year.

38 STEF ADRIAENSSENS AND JEF HENDRICKX

Acknowledgements

The authors would like to thank Ann Clé, Koen De Borgher, Tim Matthees, Rob Nijs and Annuska Rodrigues Bento for their excellent support in the fieldwork. There are also grateful for the comments

of the anonymous referees.

References

Adler, M., Bromley, C. and Rosie, M. (2000) Begging as a challenge to the welfare state, in: R. Jowell, J. Curtice, A. Park et al. (Eds) British Social Attitudes: The 17th Report: Focussing on Diversity, pp. 209–237. London: Sage.

Alderslade, J., Talmage, J. and Freeman, Y. (2006) Measuring the Informal Economy: One Neighborhood at a Time: Washington, DC: Brookings Institution Press.

Baker, D. J. (2009) A critical evaluation of the historical and contemporary justifications for criminalising begging, Journal of Criminal Law, 73(3), pp. 212–240.

Blomley, N. (2007) How to turn a beggar into a bus stop: law, traffic and the ‘function of the place’, Urban Studies, 44(9), pp. 1697–1712.

Boarini, R. and d’Ercole, M. M. (2006) Measures of Material Deprivation in OECD Countries, No. 37. Paris: OECD.

Bose, R. and Hwang, S. W. (2002) Income and spending patterns among panhandlers, Canadian Medical Association Journal, 167(5), pp. 477–479.

Butovskaya, M., Kemp, F., Diakonov, I. and Smirnov, A. (2004) Urban begging and ethnic nepotism in Russia: an ethological pilot study, in: F. Salter (Ed.) Welfare, Ethnicity and Altruism: New Findings and Evolutionary Theory, pp. 27–52. London: Frank Cass.

Danczuk, S. (2000) Walk on By: Begging, Street Drinking and the Giving Age. London: Crisis.

Dean, H. and Gale, K. (1999) Begging and the contradicitons of citizenship, in: H. Dean (Ed.) Begging Questions: Street-level Economic Activity and Social Policy Failure, pp. 13–26. Bristol: Policy Press.

Depreeuw, W. (1988) Landloperij, bedelarij en thuisloosheid: Een socio-historische analyse van repressie, bijstand en instellingen. Antwerp: Kluwer Rechtswetenschappen.

Donovan, M. G. (2008) Informal cities and the contestation of public space: the case of Bogota’s street vendors, 1988–2003, Urban Studies, 45(1), pp. 29–51.

Eck, R. van and Kazemier, B. (1988) Features of the hidden economy in the Netherlands, Review of Income and Wealth, 34(3), pp. 251–273.

Ellickson, R. C. (1996) Controlling chronic mis- conduct in city spaces: of panhandlers, skid rows, and public-space zoning, Yale Law Journal, 105, pp. 1165–1248.

Erskine, A. and McIntosh, I. (1999) Why begging offends: historical perspectives and continuities, in: H. Dean (Ed.) Begging Questions: Street-level Economic Activity and Social Policy Failure, pp. 27–42. Bristol: Policy Press.

Feige, E. L. (1990) Defining and estimating underground and informal economies: the new institutional economics approach, World Development, 18(7), pp. 989–1002.

Festinger, L. (1943) An exact test of significance for means of samples drawn from populations with an exponential frequency distribution, Psychometrika, 8(3), pp. 153–160.

Fieller, E. C. (1940) The biological standardiza- tion of insulin, Journal of the Royal Statistical Society, 8(1), pp. 1–64.

Fierens, J. (2004) La répression de la mendicité en 2004, Journal des tribunaux, pp. 543–544.

Fitzpatrick, S. and Jones, A. (2005) Pursuing social justice or social cohesion? Coercion in street homelessness policies in England, Journal of Social Policy, 34, pp. 389–406.

Fitzpatrick, S. and Kennedy, C. (2000) Getting By: Beg g ing, Rough Sleeping and The Big Issue in Glasgow and Edinburgh. Bristol: The Policy Press.

Friedman, E., Johnson, S., Kaufmann, D. and Zoido-Lobaton, P. (2000) Dodging the grabbing hand: the determinants of unofficial activity in 69 countries, Journal of Public Economics, 76, pp. 459–493.

Geremek, B. (1980) Truands et misérables dans l’Europe moderne (1350–1600). Paris: Gallimard.

Hershkoff, H. and Cohen, A. S. (1991) Begging to differ: the First Amendment and the right to beg, Harvard Law Review, 104, pp. 896–916.

Hopkins Burke, R. (2000) The regulation of begging and vagrancy: a critical discussion, Crime Prevention and Community Safety: An International Journal, 2, pp. 43–52.

THE YIELD OF BEGGING IN BRUSSELS 39

Jamar, N. and Herbots, P. (2006) Bedelarij en exploitatie, Nieuw Juridisch Weekblad, 135, pp. 98–109.

Kennedy, C. and Fitzpatrick, S. (2001) Begging, rough sleeping and social exclusion: implica- tions for social policy, Urban Studies, 38(11), pp. 2001–2016.

Lankenau, S. E. (1999) Stronger than dir t: public humiliation and status enhancement among panhandlers, Journal of Contemporary Ethnography, 28(3), pp. 288–318.

Lu, H. (1999) Becoming urban: mendicancy and vagrants in modern Shanghai, Journal of Social History, 33(1), pp. 7–36.

McIntosh, I. and Erskine, A. (1999) ‘Feel rotten. I do, I feel rotten’: exploring the begging encounter, in: H. Dean (Ed.) Begging Questions. Street-level Economic Activity and Social Policy Failure, pp. 183–202. Bristol: Policy Press.

Melrose, M. (1999) Word from the street: the perils and pains of researching begging, in: H. Dean (Ed.) Begging Questions: Street-level Economic Activity and Social Policy Failure, pp. 143–160. Bristol: Policy Press.

Mitchell, D. (2005) The S.U.V. model of citizen- ship: bubbles, buffer zones, and the ‘purely atomic’ individual, Political Geography, 24, pp. 77–100.

Moffatt, P. G. and Peters, S. A. (2004) Pricing personal ser vices: an empirical study of earnings in the UK prostitution industry, Scottish Journal of Political Economy, 51(5), pp. 675–690.

Motulsky, H. (1995) Intuitive Biostatistics. Oxford: Oxford University Press.

Murdoch, A. (1994) We Are Human Too: A Study of People Who Beg. London: Crisis.

OECD (Organisation for Economic Co-operation and Development) (2004) Employment Outlook 2004. Paris: OECD.

O’Flaherty, B. (1996) Making Room: The Economics of Homelessness. Cambridge, MA: Harvard University Press.

OSCE (Organization for Security and Co-operation in Europe) (2000) Report on the Situation of Roma and Sinti in the OSCE area. The Hague: OSCE.

Pahl, R. E. (1987) Does jobless mean workless? Unemployment and informal work, Annals of the American Academy of Political and Social Science, 493, pp. 36–46.

Portes, A. and Haller, W. (2005) The informal economy, in: N. J. Smelser and R. Swedberg (Eds) The Handbook of Economic Sociology, 2nd edn, pp. 403–425. Princeton, NJ: Princeton University Press.

Portes, A., Castells, M. and Benton, L. A. (1989) Conclusion: the policy implications of infor- mality, in: A. Portes, M. Castells and L. A. Benton (Eds) The Informal Economy: Studies in Advanced and Less Developed Countries, pp. 298–311. Baltimore, MD: Johns Hopkins University Press.

Réa, A. (2001) La problematique des personnes sans-abri en Région de Bruxelles-Capitale. Bruxelles: ULB.

Renooy, P. H., Ivarsson, S., Wusten-Gritsai, O. van der and Meijer, R. (2004) Undeclared Work in an Enlarged Union: An Analysis of Undeclared Work: An In-depth Study of Specific Items. Brussels: European Commission.

Ringold, D., Orenstein, M. A. and Wilkens, E. (2003) Roma in an Expanding Europe: Breaking the Poverty Circle. Washington, DC: The World Bank.

Rosser, J. B. J., Rosser, M. V. and Ahmed, E. (2000) Income inequality and the informal economy in transition economies, Journal of Comparative Economics, 28(1), pp. 156–171.

Ruggeri Laderchi, C., Saith, R. and Stewart, F. (2003) Does it matter that we do not agree on the definition of poverty? A comparison of four approaches, Oxford Development Studies, 31(3), pp. 243–274.

Schneider, F. (2007) Shadow Economies and Corruption All Over the World: New Estimates for 145 Countries. Linz: Johannes Kepler University.

Sifaneck, S. J. and Neaigus, A. (2001) The eth- nographic accessing, sampling and screening of hidden population: heroin sniffers in New York City, Addiction Research and Theory, 9(6), pp. 519–543.

Slack, T. (2007) The contours and correlates of informal work in rural Pennsylvania, Rural Sociology, 72(1), pp. 69–89.

Smith, P. K. (2005) The economics of anti-begging regulations, American Journal of Economics and Sociology, 64(2), pp. 549–577.

Thomas, J. J. (1990) Measuring the underground economy: a suitable case for interdisciplinary treatment?, American Behavioral Scientist, 33(5), pp. 621–637.

40 STEF ADRIAENSSENS AND JEF HENDRICKX

Trejos Solorzano, J. D. and Del Cid, M. (2003) Decent work and the informal economy in Central America. International Labour Office, Geneva.

UNDP (United Nations Development Programme) (2003) Avoiding the dependency trap: the Roma in central and eastern Europe. UNDP, Bratislava.

Venkatesh, S. A. (2006) Off the Books: The Underground Economy of the Urban Poor. Cambridge, MA: Harvard University Press.

Wardhaugh, J. and Jones, J. (1999) Begging in time and space: ‘shadow work’ and the rural context, in: H. Dean (Ed.) Begging Questions: Street-level Economic Activity and Social Policy Failure, pp. 101–119. Bristol: Policy Press.

Williams, C. C. (2004) Tackling undeclared work in advanced economies: towards an

evidence-based public policy approach, Policy Studies, 25(4), pp. 243–258.

Williams, C. C. and Windebank, J. (2002) The uneven geographies of informal eco- nomic activities: a case study of two British cities, Work, Employment & Society, 16(2), pp. 231–250.

Williams, M. and Cheal, B. (2002) Can we mea- sure homelessness? A critical evaluation of the method of ‘capture-recapture’, Social Research Methodology, 5(4), pp. 315–331.

Wold, S. (1974) Confidence limits on the product of two uncertain numbers, Analytical Chemistry, 46(11), p. 1614.

Woodbridge, L. (2002) Impostors, monsters, and spies: what rogue literature can tell us about early modern subjectivity, Early Modern Literature Studies, 9(4), pp. 1–11.