Violating trust: a study of court verdicts in cases of gambling-driven economic crimes in the workplace

ABSTRACT

This study investigated trust-violating economic crimes committed in businesses and at public and non-profit organizations by employees with problem gambling. Verdicts delivered by the Swedish general courts over a five-year period (N = 283,884) were subject to a keyword search, and verdicts matching the search criteria (n = 1,232) were examined in detail, identifying 62 such cases. A notable finding was that middle-aged women with no previous criminal record were overrepresented compared with women in national statistics on crimes in general. Non-profit organizations and public services were relatively ineffective in detecting gambling-driven trust-violating crimes. The implications of our findings for criminological theory are discussed; the results lend support to Donald Cressey’s theory of trust-violating crime as driven by a non-shareable problem and agree with the ‘fraud triangle’ conceptualization. We conclude that although these gambling-driven crimes are relatively rare, the risk of them occurring should not be ignored by businesses and organizations as the consequences can be severe and, for small businesses, sometimes devastating.

KEYWORDS:

• Crime
• problem gambling
• workplace
• gender
• embezzlement
• fraud

Introduction

Gambling and crime have a long history of association (Banks & Waugh, 2019; Spapens et al., 2008). Gambling has in many countries – for religious, moral, and social reasons – been outlawed entirely or in some of its forms (e.g. Brenner & Brenner, 1990); to gamble was thus a crime. Gambling has a strong tendency to attract swindlers, card sharks, con artists, match and horse race fixers, money launderers, and organized crime (Ferentzy & Turner, 2009). Gambling is, in most countries, subject to specific legislation, regulation, and control to prevent and penalize unlawful behaviour of providers of gambling and gamblers (Spapens et al., 2008).

Gambling and crime also connect psychologically on the individual level because sensation seeking, impaired impulse control, and other psychological traits predispose both to crime and high involvement in gambling (e.g. Mestre-Bach et al., 2018). They are also sociologically and culturally associated, as people with a pro-criminal identity tend to be attracted to the world of gambling with its fast and conspicuous handovers of big money (Ferentzy & Turner, 2009). Finally, people with problem gambling (PG), who have difficulties controlling the extent of their gambling, may resort to crime to finance their gambling or solve acute economic problems caused by gambling losses (Lesieur, 1984).

Although there are clear connections between gambling/PG and crime, no specific criminological theory seems to have been formulated about how these are interrelated. In the USA, gambling was up to the 1960s illegal in most of the country. In academic studies of that time, gambling was mainly conceptualized as social deviance, an activity practiced foremost by marginalized and criminal individuals (Rosecrance, 1985). Later, North American studies investigated if crime rates were elevated in the vicinity of casinos (Barthe & Stitt, 2009) and estimated the societal costs of PG, including costs related to crime and incarceration (Walker & Kelly, 2011).

However, Donald Cressey’s (1973) theory of trust-violating crime as driven by a non-shareable problem (Robin, 1970), believed to be solvable with the help of money, is relevant. Trust violation refers to a situation where an individual violates a position of financial trust by committing various crimes (e.g. embezzlement, theft, and forgery); here, we interpret ‘financial trust’ as including having been entrusted with valuable goods. A problem is non-shareable because the individual is ashamed of having it or because others’ knowledge of it would cause great trouble in personal or professional life. This theory inspired the widely accepted ‘fraud triangle’ conceptualization in criminology (Dellaportas, 2013; Dorminey et al., 2010), in which opportunity, rationalization, and pressure together create the motivation to commit fraud and similar crimes; we argue that this applies to PG-driven crimes.

This article presents a study of PG-driven crimes committed in a workplace or non-profit organization (NPO). Thus, the study concerns criminogenic PG – PG leads to crime. In the cases analysed, an employee or trusted colleague has embezzled or stolen money for gambling or covering a pressing need for money caused by excessive gambling.

PG, in the context of this study, should be understood as an individual’s impaired ability to control the extent of gambling, which causes significant economic, emotional, or social harm to the individual and often also harm to relatives, significant others and society at large (Gambling Research Australia, 2005). The psychiatric diagnosis Disordered Gambling is included in the Diagnostic and Statistical Manual of Mental Disorders (DSM-5, American Psychiatric Association, 2013). Psychometric studies have shown that the commitment of criminal acts to finance gambling is highly correlated with the DSM-5 criteria for Disordered Gambling (Stinchfield et al., 2016) and indicates the highest level of PG severity (Strong & Kahler, 2007).

Studies have found that PG is a relatively common reason for committing embezzlement and similar crimes in the workplace. For instance, a study of fraud committed in financial institutions in Australia found that more than half of the cases were motivated by PG (Warfield, 2013). An American investigation of employee thefts in cases where sums exceeded $100,000, and where the motivating factor was known, concluded that about a third of these cases were related to PG (Marquet International, 2013). Studies have also consistently found that such crimes are often committed over several years and tend to escalate in severity. In other words, they are committed more frequently, more habitually, and add to an increasingly large sum of money wrongfully appropriated (Adolphe et al., 2019; Binde, 2016a; Lind et al., 2015). The sums misappropriated because of PG can be very large – in the order of millions of USD or Euro (Marquet International, 2013; Warfield, 2013).

Previous studies have shown that perpetrators typically have appeared to be well-adjusted citizens and could, therefore, be offered positions of trust in companies and in NPOs. They often held such positions for many years before they started to gamble excessively and got the idea to ‘borrow’ money from their workplace. This is a typical element in these cases; individuals are often caught spiralling downward after escalating gambling losses, leading them to commit crimes, which eventually leads to their convictions (Binde, 2016a; Crofts, 2003; Lesieur, 1984). Henceforth, this scenario is in the present article referred to as the ‘typical’ scenario of trust-violating gambling-driven criminality.

To give a comprehensive picture of the commitment and consequences of criminogenic PG in Swedish workplaces and NPOs, this study explored the following seven topics: (a) the prevalence of this kind of criminality, (b) the characteristics of offenders, such as gender and age; (c) their previous criminal convictions, (d) the characteristics of the crimes, that is, the types of crimes committed, (e) the crime scenario and its duration, (f) the types of victims and the consequences experienced in terms of harm to employers and NPOs, and (g) the penalties imposed on those found guilty. Where possible, we compared our results with national crime statistics to highlight differences between trust-violating PG-driven criminality and criminality in general. We paid particular attention to gender because we wished to explore the gender gap in crime (Estrada et al., 2016) in this group of offenders and because concerns have been raised about the ‘feminization’ of PG (Volberg, 2003).

Our study renders a picture of those who commit these kinds of PG-driven crimes. These results are valuable for understanding, preventing, and detecting this manifestation of PG in workplaces and NPOs. The study has a bearing on the criminological issue of whether there are types of crimes that are driven by configurations of specific factors, rather than an individual proclivity for committing crimes in general (i.e. the general theory of crime, Hirschi & Gottfredson, 1988). Our results are of relevance to criminologists with an interest in white-collar crime (Clarkson & Darjee, 2022; Hirschi & Gottfredson, 1987; Steffensmeier, 1989) and, more generally, the motivating factors for committing economic crimes.

Method

Materials

This article presents a study based on a subset of data collected in a larger investigation of criminogenic PG in Sweden (Binde et al., 2021). The investigation was based on Portable Document Format (PDF) files of all the verdicts (N = 283,884) passed over the five years 2014‒2018 by Swedish general courts. These verdicts, which are typically between 15 and 40 pages in length, include information on the defendant, the victims, the particulars of the crimes committed, the harm and damages inflicted, and the courts’ reasoning regarding the severity of the crime and its appropriate sanctions. Verdicts were available via the Internet in the Swedish JUNO database. Statistics on crimes in general were obtained online from the Swedish National Council for Crime Prevention (Brå) (https://www.bra.se/) database. The investigation was approved by the Ethical Review Board in Stockholm (No. 2018/2021–31/5).

A keyword search for verdicts that included the Swedish term for ‘problem gambling’ and similar terms yielded 1,232 hits. These verdicts were inspected manually by the three authors of this article with the objective of finding undisputable cases of criminogenic PG, that is, cases in which the court explicitly considered PG to have been the driving force of criminality (Figure 1). The investigation identified 282 such cases, while the remaining verdicts were either irrelevant to the study (for example, it was stated that PG was not present) or PG was mentioned but not explicitly as being the cause of the crime committed. For example, the victim had PG, or the perpetrator had earlier suffered from PG (for details on this procedure, see: Binde et al., 2021).

Figure 1. Flowchart of the identification of 62 cases of PG-driven crimes committed in workplaces or NPOs. (Note: the shift from 321 verdicts to 282 cases is because some cases had been judged in both a district court and a court of appeal; for our study, we combined the information from these verdicts.).

Among the 282 criminogenic PG cases over the five-year period, 67 were cases of criminogenic PG in workplaces and NPOs, wherein the perpetrator had abused a position of trust in an enterprise or organization and was not the only owner of the business. In 62 cases, the primary victim (see below) had been subject to trust violation; these cases are analysed in this study (Figure 1).

Analysis

In the larger investigation, data on all relevant variables that could be extracted from the verdicts – such as the perpetrator’s gender and age, type of crime, sanctions, and type of victim – were coded, entered in an MS Excel spreadsheet, and analysed quantitatively. The coding of most variables was uncomplicated as they were facts stated in the verdicts. However, the classification of cases in which the court stated that PG was the main driving cause of the crimes was occasionally intricate and necessitated discussion among the three coders (the authors of this article) until consensus was reached. Our approach was conservative, including only the cases in which there could be no doubt that the court had concluded that the crimes were PG-driven (for details on the coding, see: Binde et al., 2021). Cases with victims and PG-driven crimes of different types were coded as primary, secondary, and tertiary according to the extent to which victims had been subjected to the criminal activity of the perpetrator and the proportion of various crime types committed. This coding was based on the summary included in all verdicts, listing the damage in economic terms suffered by each victim and the number of crimes of various kinds.

Several analyses in this study concern the same variables as in our larger investigation of criminogenic PG in Sweden, which allows for a comparison between the results. However, analyses carried out only in this study concerned the duration of the PG-driven criminal acts (in months); the dates of the first and last crimes in a series are always stated in a court verdict. If only one criminal act was committed, or if several were committed over a few days, we coded the duration as one month. We also coded the type of trust-breaking crime scenario, whether the perpetrator had an accomplice in crime, and the perpetrator’s versatility in committing several forms of crime.

Descriptive statistics include frequencies, average values, and median values. Although the number of cases is small, we performed, when possible, chi-squared tests, t-tests, and Fisher’s exact test to gain knowledge on to what degree differences between men and women in our cases, and between our cases compared to national crime statistics, might result from random variation (the null-hypothesis) or having been created by real-world phenomena. Notably, however, our 62 cases were a total sample. Thus, our descriptive statistics show the characteristics of these crimes during the five-year period.

Results

Prevalence

As shown in Figure 1, we found 62 clear cases of gambling-driven crimes in workplaces or NPOs in the five-year period covered by the study.

These 62 cases were a small fraction of Swedish economic criminality in general. For example, over the five-year period, we found 37 clear cases of PG-driven embezzlement (see below). This amounts to 4.4% of the 841 embezzlement cases committed for all kinds of reasons subject to legal proceedings during the same period, according to national crime statistics. Because fraud, theft, and handling of stolen goods are very common crimes committed in many other contexts than the workplace or NPO, the proportion of these crimes in this study, compared with national statistics, was even lower (at 0.26%, 0.01%, and 0,03%, respectively).

Perpetrator characteristics

In the 62 cases analysed in this study, 63% of the perpetrators were men and 37% were women. The proportion of women in this study was notably higher than the proportion of women in national crime statistics on criminal convictions of all kinds over the same five-year period, which was 16%. This difference was statistically significant (X² (1, N = 294,692) = 21.80, p = 0.000).

The mean age of the perpetrators was 42 years. However, the mean age of women was 48 years, which was markedly higher than that of men, which was 38 years. The difference was statistically significant, t(60) = −3.50, p= 0.001.

The youngest and oldest were 21 and 74 years old, respectively. The age distribution of cases in this study compared to national statistics on crimes in general is shown in Figure 2.

Figure 2. Age distribution, men and women, cases in this study compared with national statistics on convictions in district courts, for all kinds of crimes, 2014–2018.

A notable observation was that women in the age range of 40–59 years were greatly overrepresented in our study, amounting to 69% of convicted women, compared with men in that age range in our study (39%) and women in that age range in national crime statistics covering all kinds of crimes (32%). The first of these differences was statistically significant (X² (1, n = 80,056) = 23.12, p < .001), while the second difference could not be tested against the null hypothesis because of the low number of cases in our study.

We also noted that, although few in this study, women aged 60 years and above were also overrepresented when compared to men in our study and women in national statistics. There were no cases where the convicted perpetrator was below 21 years of age in our study, while approximately 16% of the convicts in national crime statistics, during the same time period, were at that age.

There was a notable difference between men and women with regards to the number of inhabitants in their towns of residence. For men, the average population size was 273,417 inhabitants, and for women, it was almost half of that at 156,012 inhabitants (median values: 96,952 and 54,180, respectively). However, this difference did not reach the level of statistical significance. A close inspection of the data revealed that the difference between men and women was caused mainly by a higher proportion of men living in one of Sweden’s three major cities (men 28%, women 17%) and a higher proportion of women living in communities with less than 10,000 inhabitants (women 30%, men 18%).

Previous criminality

The majority (85%) of the perpetrators were first-time offenders. As Figure 3 shows, none of the women in this study had a criminal record. In national statistics on all kinds of crimes judged in general courts during the same five-year period, 49% of the women had no previous convictions in the past ten years and therefore most of these women were probably first-time offenders. As for the men in this study, 77% were first-time offenders, 15% had minor previous criminality (i.e. one to three sentences; none of them for having committed many separate offences), and 8% had extensive previous criminality (i.e. more than three previous sentences or at least one sentence for having committed many offences). According to national crime statistics, 36% of the men had no previous convictions in the past ten years, 26% had one or two previous convictions, and 38% had three or more previous convictions. Although our coding of ‘previous criminality’ is not exactly the same as the number of ‘previous convictions’ in national crime statistics, it is safe to say that the proportion of first-time and few-time offenders was notably greater in our study than among those convicted for crimes in general. There was a significant difference between the previous criminality of perpetrators in national statistics and those in our sample, according to Fisher’s exact test, p < 0.001.

Figure 3. Previous criminality among the cases in this study, compared with national crime statistics of the same five-year period (2014–2018), on number of previous convictions in the past ten years for all kinds of crimes.

Crime characteristics

There were no cases in which the perpetrator had an accomplice. In only three cases, the perpetrator was sentenced not only for committing a PG-driven trust-violating crime, but also for crimes in settings other than the workplace or at NPOs.

The most common type of crime coded as ‘primary’ was embezzlement and larceny by servant for 60% of the cases; these two crimes are very similar in practice and will henceforth simply be called ‘embezzlement’. Fraud accounted for 21% of the cases, whereas theft accounted for 18%. The remaining 1% was a case of handling stolen goods. In 13% of the cases, the perpetrator had committed trust-violating property offences that were classified as being of different kinds by the court. This did not necessarily reflect criminal flexibility but was often the consequence of juridical technicalities in describing wrongdoings that in practice were similar, that is, wrongfully appropriating money from the employer or an NPO, such as by means of embezzlement, breach of trust, theft, or fraud.

The crime scenario and its duration

Of the 62 cases, 37% were characterized by the typical scenario of escalating PG-driven trust-violating criminality, as found in previous studies and outlined in the introduction: the perpetrator had long been a trusted staff member at a company or organization, began gambling excessively, depleted their own economic funds, and began to misappropriate money from the business or organization. As said, a recurring element in the typical scenario is that the perpetrator intends to temporarily and secretly ‘borrow’ money from the company or organization, which would be paid back when gambling luck change for the better. The ‘borrow’ idea was present in nearly all the typical scenario cases in this study. Several perpetrators paid back money, but none did it in full.

In 29% of the cases, the verdicts only revealed that crimes had been committed over an extended period, that is, several months or years, but not whether the gambling and economic troubles of the perpetrator had escalated. In 19% of the cases, only a few criminal acts were committed over a short period before detection (i.e. some days or weeks), and in 6% of the cases, only one criminal act had been committed. In a few of these cases, the court stated that criminality had been preceded by escalating PG.

From among the cases, 8% were of other types. In one case, criminality and excessive gambling were exacerbated by attention deficit hyperactivity disorder (ADHD). In another case, a sudden onset of PG was caused by dopamine agonist (pramipexole) medication for Parkinson’s disease, which is a well-known adverse reaction to such drugs (e.g. Dodd et al., 2005). In a third case, an individual with chronic PG got a job as a salesman and then began to commit crimes to finance his gambling. In the remaining two cases, the perpetrators denied having done anything criminal or having PG, despite overwhelming evidence to the contrary.

In summary, the ‘typical scenario’ of escalating PG-driven trust-violating criminality was the single most common scenario, identified clearly in approximately one-third of the 62 cases. Another third of the cases might also have been of this type; however, we only know from the verdicts that the crimes were committed over an extended period. The remaining one-third of the cases followed other scenarios.

Types of victims

Table 1 shows the types of employers or organizations that had become victims, the sums unlawfully appropriated over the five-year period 2014-2018, and the duration of trust-violating criminality. This typology was developed for a previous Swedish study on this topic (Binde, 2016b) to form groups, large enough to be compared statistically, of similar victims. It should be noted that some perpetrators paid back the money to the company or organization when they won at gambling; we calculated the economic harm to the victim as the sum unlawfully appropriated and not paid back, representing the harm inflicted in economic terms.

Table 1. Type of employer or similar, sums appropriated in SEK and the duration of trust-violating criminality, 2014-2018

Percent (N)	Employer or similar	Men/women (N)	Sum appropriated, total	Sum appropriated, average	Sum appropriated, median	Highest sum	Duration in months (average)	Duration in months (median)	Longest duration in months
32% (N = 20)	Private company, other than bank, store or restaurant	12/8	56,901,420	2,845,071	879,409	17,900,000	19	16	81
27% (N = 17)	Store or restaurant	12/5	7,708,729	453,455	65,853	4,870,000	8	2	64
23% (N = 14)	Non-profit organization	8/6	16,965,156	1,211,797	533,925	6,377,032	29	13	84
10% (N = 6)	Public services	3/3	13,111,178	2,185,196	194,822	7,161,000	30	16	112
8% (N = 5)	Bank	4/1	39,908,494	7,981,699	6,415,542	21,256,699	20	14	52
100% (N = 62)	Total	39/23	134,594,977	2,170,887	443,400	21,256,699	19	11	112

In one-third of the cases, a private business – other than a bank, store, or restaurant – was a victim of the crime. The sums misappropriated were relatively large, with the median sum being close to Swedish Krona (SEK) 900,000 (SEK 1 million is approximately USD 114,000 or EUR 99,000). The highest sum, nearly SEK 18 million, was embezzled during 17 months from a private business by a member of its executive team. He unlawfully appropriated around SEK 26 million, a part of which he repaid before his trial in court.

Stores and restaurants were the second most common types of victims. The sums wrongfully appropriated were notably smaller than those appropriated from other types of businesses, and the duration of criminality was also shorter.

There were 14 NPOs among the victims, such as housing cooperatives, sports clubs, political parties, and charities. The median amount of money misappropriated was almost the same as that in the company cases. The highest sum embezzled was SEK 6.4 million from a charity supporting the education of children and youth in poor countries. This crime extended over a period of more than five years. Since the charity received donations of about SEK 1.5 million per year, most of that money during these years went to the embezzlers’ gambling habit rather than charitable purposes.

Public service organizations – schools, public administrations, social welfare services, and a museum – were victims in six cases. The median sum unlawfully appropriated was lower than that from companies, but higher than that from stores and restaurants. It should be noted that the duration of criminality in this domain was among the longest, with a median of 16 months. In one case, the duration was more than nine years. In this case, the head of a unit at a social assistance centre unlawfully paid out social security benefits to several vulnerable and socially disadvantaged clients, and diverted the payments to herself or manipulated the clients to give her the aid money. The money misappropriated – SEK 5.4 million – was spent not only on gambling but also on alcohol.

Not surprisingly, the highest sums unlawfully appropriated were from banks. There were five such cases, all middle managers, with the median sum being SEK 6.4 million. The highest sum was over SEK 21 million, embezzled by a woman who worked as a customer advisor and misappropriated money from customers’ accounts for 14 months.

Penalties

A prison sentence was imposed in 24% of the cases, as that was the sanction stipulated by the Penal Code for such severe crimes. Fines or conditional sentences, including outpatient treatment for PG, were imposed by the court for 32% of the offenders. In 34% of the cases, the perpetrators received similar non-prison penalties but no treatment because they no longer suffered from PG, having quit gambling on their own or with external help. In the remaining 10% of the cases, some individuals received non-prison penalties but no PG treatment because they were already enrolled in such treatment; other individuals were sentenced to PG treatment in a closed residential care institution. One individual ended up in prison, instead of serving a conditional sentence and PG treatment, because he had previously such PG treatment but had relapsed into PG-driven criminality.

Discussion

This study focused on PG-driven crimes committed by individuals who had been given a position of trust in their places of work or in an NPO. When possible, we compared our results with national statistics on crime in general.

We found 62 cases of gambling-driven crimes in the workplace or NPOs over a five-year period, which constituted a very small part of economic criminality, in general, during the same period, although the embezzlement cases amounted to 4.4% of all such crimes committed in Sweden.

An earlier study of this type of PG-driven criminality in Sweden examined a newspaper article database, using the same keyword search criteria as in this study (Binde, 2016b). The study found 55 cases over the five-year period from 2009 to 2014. This corresponds with the prevalence found in this study, which reports 62 cases in a five-year period. Thus, the prevalence of PG-driven trust-violating criminality seems to have remained stable over time in Sweden. This is consistent with the prevalence of PG (Problem Gambling Severity Index, score of 8+ or equivalent, Ferris & Wynne, 2001) oscillating around 0.5% in Sweden for the past 20 years (Folkhälsomyndigheten, 2019; Statens folkhälsoinstitut, 2010, 2012). This level of the prevalence of PG is similar to the levels in many other Western countries (Williams et al., 2012), which is consistent with the findings of the earlier Swedish study that the occurrence of known PG-driven trust-violating offences in the workplace in Sweden, Australia, and the United States is remarkably similar – in the range of one to two cases per million inhabitants per year (Binde, 2016b).

The above-mentioned previous Swedish study (Binde, 2016b) estimated that the real number of cases of trust-violating PG-driven criminality was at least ten times higher than those covered by newspaper reports. There were many non-detected cases, cases not reported to the police, and cases that were reported, but for various reasons not judged in court. There are no reasons to assume differently regarding this study.

A main finding is the overrepresentation of middle-aged women without a criminal record, compared to men in this study as well as to female convicts in national crime statistics. Notably, all women in this study were apparently first-time offenders. We also found that the women in our study lived in places (villages and towns) that were half as large as the places where the men lived, in terms of population size. Among the convicted, 37% were women. Thus, the proportion of women seems to have been larger than that of women among people with PG in Sweden in 2021, which was 20% (Folkhälsomyndigheten, 2022).

Our results are likely to reflect the increasing number of middle-aged women with PG in recent years, of which the majority have problems with online slots and online casino gambling (Folkhälsomyndigheten, 2019). These are particularly risky form of gambling (Allami et al., 2021; Håkanson et al., 2017), which have been heavily promoted in Sweden in the past years (Håkanson & Widinghoff, 2019). Our study suggests that the typical female perpetrator in these PG-driven crimes is middle-aged, often living in a small community with no opportunities for physical casino gambling, who in recent years has been exposed to casino and slot machine gambling via the Internet, and therefore have had a high risk for developing PG. As a result, these women depleted their own economic resources by gambling excessively, and then began unlawfully appropriating money at their workplaces or at NPOs, where they held positions of trust.

There are several additional explanations for the overrepresentation of middle-aged women in this study compared with national statistics on crime in general. In general, women develop PG later in life than men (Sundqvist & Rosendahl, 2019). The crime committed in many of the cases in this study is embezzlement, which has about an equal number of male and female convicts in Sweden, according to national crime statistics (Binde et al., 2021, p. 10). This is because many women work as cashiers, bookkeepers, and in other similar roles and professions, which are conducive to acquiring money through embezzlement (Steffensmeier, 1989).

As for the men in our study, they were also older and had less previous criminality compared with men in national statistics on crime in general; however, this difference was not as pronounced as it was for women. The reason for the non-existence of perpetrators below 21 years of age in our study may be that it usually takes a few years to develop severe PG and that young people are less likely than others to be offered positions where they are trusted to handle substantial sums of money.

Thus, the gender gap between men and women in this study was notably smaller than for crime in general. This has been found in other studies on this kind of criminality (Warfield, 2013). An American study observed a reversed gender gap, that 62% of PG-driven embezzlers were women (Marquet International, 2013).

Some studies (Lind et al., 2021) have found a link between criminal offending, PG, and social disadvantage – unemployment, low education, low income, and receiving social welfare assistance. We did not find this to any notable extent among the cases in this study. There are two reasons for this.

First, while the study by Lind et al. (2021) and similar studies (see, Lind et al., 2021, for a review of the literature) have investigated the co-occurrence of PG and criminal offending, our study includes only cases in which, according to a court, PG has caused crime. Thus, the association between PG and offending among our cases is, by definition, causally unidirectional from PG to crime, although criminal behaviour might still have been affected in its intensity and character by confounding factors such as personality disorders, psychiatric conditions, and social contexts (Dennison et al., 2021; Lind et al., 2015).

Second, most perpetrators were trusted employees or long-time members of NPOs; they were apparently well-adjusted citizens. These characteristics of theirs is probably why virtually all the differences between the 282 cases in our larger investigation of PG-driven crime (Binde et al., 2021), which included people from all walks of life, and national crime statistics, are accentuated in the present study.

We found that the convictions for PG-driven trust-violating embezzlement in this study amounted to 4.4% of the convictions for embezzlement in general courts. This is a lower percentage than found in other studies with roughly comparable data material (e.g. Marquet International, 2013; Warfield, 2013). However, this study included only cases in which the court explicitly stated that PG was the driving force of crime. There were numerous cases in which PG could have played a major, or at least a significant role, for the commitment of crime, but from the statements in the verdicts, we could not be certain of this.

A little over a third of the cases in our study were clearly characterized by the ‘typical’ escalating scenario, which has been identified in previous studies of trust-violating PG-driven crimes (Binde, 2016a; Crofts, 2003; Lesieur, 1984). The initial theft of money is followed by repeated thefts; moreover, the motivation to illicitly procure money is increased by mounting gambling losses and the hope of scoring big wins and being able to return what has been stolen (‘borrowed’) without anyone having noticed; this particular idea was documented in almost all typical cases. Presumably, numerous other cases also belong to the escalating scenario, although this could not be deduced from the verdict. Finally, many of the singular or occasional embezzlements, frauds, and thefts in our study would most likely have been followed by a series of others, had they not been detected.

There were no cases in which the perpetrator had an accomplice. The crimes analysed here are typically not committed by people with a criminal identity, or influenced by peers with pro-criminal attitudes, but by people who are ashamed of what they do, and therefore do it alone without telling anyone about it. Versatility in committing crimes, that is, committing crimes of different kinds, was found to be low, suggesting that the opportunity to illicitly procure money is taken when possible and thereafter exploited without much motivation to extend the range of criminal acts. This tallies well with the typical escalating scenario, in which the perpetrator often perceives himself or herself as being an honest person who has gotten into deep trouble and tries to make everything right without bothering anybody else.

As in previous studies (e.g. Warfield, 2013), banks suffered the greatest economic harm per case of PG-driven trust-violating crimes (median: SEK 6.4 million), while stores and restaurants suffered the least (median: SEK 66,000). However, regarding the harm inflicted, these figures should be seen in the context of the turnover of these businesses, as for a small shop or restaurant, losing SEK 66,000 might be a significant sum. The figures on the duration of the crime show that crimes committed in stores and restaurants were typically detected in a few months, while NPOs and public services stand out as having been relatively poor at detecting PG-driven crimes among the staff.

Nearly all perpetrators who could be sentenced to a penalty other than prison received such penalties, and nearly all who needed PG treatment and could be offered it, were sentenced to treatment. This is a pleasing finding, suggesting that Swedish courts are not prejudiced against people with PG and consider that treatment might help. Internationally, there are a limited number of studies on how PG-driven crime is judged by courts in various jurisdictions (e.g. Brooks & Blaszczynski, 2011; Liberato, 2019). It appears that Swedish courts, without openly viewing PG as a mitigating factor, usually consider PG a reason for choosing a less restrictive sentence compared to cases without PG motivation.

In addition to direct economic harm, many of the victims suffered indirect economic harm and significant harm of other kinds, the character of which is only hinted at in the verdicts. As for companies, their bookkeeping might need to undergo revision, as they were distorted during embezzlement, which may necessitate additional tax payment. Other employees may have been shocked by the fact that one of their long-time colleagues had unlawfully appropriated large sums of money from the company, which negatively impacted their work performance for some time. The social reputation of businesses, in particular banks, might take a big hit if their susceptibility to crime becomes widely known. The same applies to public organizations and NPOs. Business activities might be disrupted or even ended, as in the case of a cooperative kindergarten that went bankrupt as a result of an SEK 2.7 million embezzlement.

We acknowledge that 62 cases might be few as the subject of the statistics presented here. However, these cases were not a random or other sample from a larger data set, the representability of which could be questioned. They were all cases that met our strict criteria for inclusion in the study, in other words cases resulting from an exhaustive survey. We believe that our description of these cases provides a good picture of the reality of trust-violating criminogenic PG in Sweden during the five-year period covered.

The principal strength of this study is the reliability of the data. The study was based on factual demographic information of the perpetrators of trust-violating PG-driven crimes, as well as other data on the crimes established to be true by general courts.

However, while the strict definition of what cases to include is a strength of this study, it is also a limitation, because it excludes the analysis of cases in which PG only partially contributed to criminality, as well as cases with less pronounced and occasional gambling problems. Furthermore, this study is limited to a quantitative analysis of the material.

The qualitative nature of the suffering caused by the perpetrators to their families and significant others, such as psychological damage in terms of broken marriages, ruined economy, and imposed lifestyle change for the worse, are without a doubt significant in many cases and warrant specific study; however, we have found no international literature on this. Qualitative investigations on this topic have the potential to highlight how demographic, sociological, situational, and psychological factors interact systemically to cause severe PG and the commitment of crimes, an interaction that quantitative studies capture only broadly. The harm caused to businesses and NPOs, apart from economic loss, would be valuable to investigate in greater detail.

Conclusion: implications for criminology, gambling studies, and policy

The results of this study are not in line with the general theory of crime (Hirschi & Gottfredson, 1988). The convicted in our study differ markedly – in terms of gender, age, and previous criminality – from Swedish convicts in general (c.f., Steffensmeier, 1989). Rather, the results tally well with Cressey’s (1973) theory of trust-violating crime, which is driven by a non-sharable and hidden personal problem, in our cases PG, believed to be solvable by the means of money.

Cressey’s theory has, however, been criticized for not been strongly supported by the results of subsequent studies on fraud, showing that a non-sharable problem is far from always present (Klenowski & Copes, 2013). However, in our study, PG has without doubt been a problem for all perpetrators and there is no indication in the verdicts that the problem had been shared with anyone in the workplace or an NPO, and presumably very few others, if any, knew about the PG.

In gambling studies, it has been argued that thought patterns typical of PG, such as ‘chasing losses’ and the idea that a win is likely to occur after a series of losses, interact with the fraud triangle factors in a systematic way to generate PG-driven trust-violating crime (Binde, 2016a). This creates the ‘typical scenario’ that clearly characterized one-third of the cases in this study. The implication for gambling studies is that a systemic and processual perspective on the progression of PG in its severe phase is a valuable complement to the analysis of specific demographic, sociological, and psychological traits as being PG risk factors (Allami et al., 2021).

Previous research and practical experience show that relevant measures to prevent PG-related criminality in businesses and organizations are about establishing a substance use and gambling policy, having PG awareness, implementing effective control of economic transactions, paying attention to signs of excessive gambling among the staff and responding appropriately, and rehabilitating staff with PG (Binde, 2016c; Nower, 2003). This study shows that such preventive measures are especially needed in NPOs and public services, because the duration of PG-related criminality is longer in these places (i.e. detected later) than in other kinds of workplaces. More generally, a cornerstone in addressing grave and acute PG is to lower the threshold for seeking help, which can be done on the national level by, for example, offering internet-based PG treatment (Sagoe et al., 2021).

Gambling companies should proactively monitor the gambling habits and losses of their customers by being on the lookout for individuals who try to recoup their losses and spend more than they can afford, and initiate a dialogue about cutting down on or abstaining from gambling with those who meet these and other relevant PG-indicating criteria (Jonsson et al., 2019). Legislation should require gambling companies to do this as a compliance measure, and regulatory authorities should ensure that it is followed. There are many actors in the social care system, such as social workers, debt restructurers, and primary and occupational healthcare providers, who should look out for PG-driven need of money that potentially could, or already has, led to economic criminality.

This study has shown that middle-aged women living in small-sized communities in Sweden have an elevated risk level for developing severe PG and, consequently, are also at an elevated risk level for less severe gambling problems that have not yet escalated to a critical level. Although few at present, these women do not resemble the stereotypical picture that many people in social services and the treatment system have in mind when thinking of an individual with PG – typically male, focusing on sports and horse betting. Therefore, the demographic profile and characteristics of these women should be considered when such professionals encounter distressed women seeking help for economic and emotional problems, as the stigma of PG and fear of prosecution in case crimes have been committed, can make it difficult to admit the true cause of the problem.

Constraints on publishing

There were no constraints on publishing.

Ethical approval

The Ethical Review Board in Stockholm Protocol number: 2018/2021-31/5

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

Funding for this study was provided within the frame of the programme grant ‘Responding to and Reducing Gambling Problems – Studies in Help-seeking, Measurement, Comorbidity and Policy Impacts’ (REGAPS) financed by the Swedish Research Council for Health, Working Life and Welfare (Forte), grant number 2016-07091