'Pushing buttons': an evaluation of the effect of Aboriginal income management on commercial gambling expenditure.
Abstract: The Northern Territory National Emergency Response Act 2007 was a radical intervention into the lives of Aboriginal residents of the Northern Territory, Australia. One of the intervention's key measures was income management--a scheme designed to limit the range of goods and services that may be purchased with social security payments. The aim of income management was to curb 'anti-social behaviours' such as excessive gambling and alcohol consumption. In this paper, we specifically test the efficacy of income management in reducing the amount spent on commercial gambling. To achieve this we conduct an interrupted time series analysis with deflated monthly electronic gaming machine (EGM) expenditure data from July 2002 to July 2010 for hotels and clubs in the towns of Alice Springs and Katherine. We find a negative association between income management and EGM revenues for only one gambling venue in each town. However, local complexity in the form of segregated markets along temporal, spatial and racial lines, along with other policy confounders, may obscure the effects of the macro-policy intervention. We conclude by making suggestions for locally-based responses to problematic forms of risky consumption that may be more sensitive to local geographies.

Keywords: Indigenous communities, Northern Territory Emergency Response, income management, gambling, electronic gaming machines, interrupted time series
Article Type: Report
Subject: Australian aborigines (Social aspects)
Australian aborigines (Alcohol use)
Australian aborigines (Finance)
Domestic policy (Analysis)
Government programs (Analysis)
Gambling (Prevention)
Alcoholism (Prevention)
Financial management (Methods)
Financial management (Laws, regulations and rules)
Authors: Lamb, David
Young, Martin
Pub Date: 12/22/2011
Publication: Name: Australian Journal of Social Issues Publisher: Australian Council of Social Service Audience: Academic Format: Magazine/Journal Subject: Sociology and social work Copyright: COPYRIGHT 2011 Australian Council of Social Service ISSN: 0157-6321
Issue: Date: Winter, 2011 Source Volume: 46 Source Issue: 2
Topic: Event Code: 290 Public affairs; 250 Financial management; 970 Government domestic functions; 930 Government regulation; 940 Government regulation (cont); 980 Legal issues & crime Advertising Code: 94 Legal/Government Regulation Computer Subject: Company financing; Government regulation
Product: Product Code: 7760000 Legal Gambling; 9915100 Financial Management; 9915500 Financial Systems & Controls NAICS Code: 7132 Gambling Industries
Geographic: Geographic Scope: Australia Geographic Name: Northern Territory; Northern Territory Geographic Code: 8AUNT Northern Territory; 8AUST Australia
Legal: Statute: Australia. Northern Territory National Emergency Response Act 2007
Accession Number: 276807853
Full Text: Introduction

In June 2007, the Northern Territory Government released a watershed report on child abuse in Australian Aboriginal communities titled Ampe Akelyernemane Meke Mekarle 'Little Children are Sacred' (Wild & Anderson 2007). This report documented a legacy of policy failures towards Aboriginal Australians by successive Commonwealth, State and Territory governments that were reflected in widespread child abuse, poor education, poor health, rampant alcohol abuse, and problem gambling (Wild & Anderson 2007; Gulson & Parkes 2010). Consequently, the report recommended that child abuse in Aboriginal communities '... be designated as an issue of urgent national significance by both the Australian and Northern Territory [NT] Governments ...' (Wild & Anderson 2007: 22).

The enormous media and political attention generated by the report was used by the Howard Government as a political platform from which to introduce a series of reforms, more commonly known as 'the intervention' (Commonwealth of Australia 2007a). This involved the introduction of the powerful Northern Territory National Emergency Response (NTER) Act 2007 and amendments to the Social Security and Other Legislation Amendment (Welfare Payments Reform) Act 2007, as well as exclusion of sections of the Racial Discrimination Act 1975 (RDA) that protect citizens from racial discrimination (Commonwealth of Australia 1976; Commonwealth of Australia 2007b, Commonwealth of Australia 2008). It represented a far-reaching, dramatic and controversial attempt to regulate the lives of Aboriginal people in the Northern Territory (Aboriginal and Torres Strait Islander Social Justice Commissioner 2007; Commonwealth of Australia 2008; Gulson & Parkes 2010).

A cornerstone of the intervention was the Commonwealth's income management legislation. Under this legislation, 50 per cent of the value of welfare payments received by Aboriginal residents living in prescribed areas of the Northern Territory were to be solely used for what the Commonwealth designated as 'priority needs' (including food, clothing, housing, healthcare and education related expenses, among others) (Commonwealth of Australia 2007a). Various methods have been introduced to manage quarantined funds, including the allocation of funds to retailers (where the funds could only be used to purchase priority needs), store cards (where funds could only be directed to a specific store or grocery chains to purchase priority needs), and, most recently, the 'BasicsCard' (where Electronic Funds Transfer Point of Sale (EFTPOS) cards are restricted to purchases of priority needs).

The overt aim of income management was to 'protect children and make communities safer' by reducing the impact of 'anti-social behaviours' (AIHW 2009: iii). The idea was to divert a greater proportion of welfare payments to essential items like food and clothing rather than more risky or socially damaging forms of consumption. To this end, four 'excluded items' were identified as 'anti-social behaviours' under the auspices of the Act. These were listed in the Social Security and Other Legislation Amendment (Welfare Payments Reform) Act 2007 under two categories: goods (including alcohol, tobacco products and pornographic material), and services (specifically gambling). Gambling was the only service listed as an excluded item.

Despite a change in government at the Commonwealth level, the intervention and its associated income management procedures have remained a policy platform. Indeed, legislation was passed on 10 July 2010 that introduced a new model of income management (NIM) which was implemented in Alice Springs and Katherine from 30 August 2010 (SPRC 2010: 7). NIM is based on wider criteria for inclusion, which transcend racial boundaries to include 'disengaged youths' (aged 15-24) who have received social security payments for three out of the previous six months, long-term welfare recipients aged 25 and above, parents and carers referred by child protection authorities, those deemed as vulnerable by a Centrelink social worker (this may include some individuals on aged pension or disability support pension for example), and a volunteer stream of people on income support (SPRC 2010: 7-9). This series of manoeuvres allowed NIM to be consistent with the RDA and, depending on the results of an evaluation to be concluded at the end of 2011, will be progressively rolled out across the country. Surprisingly, given the governmental support for income management and the possibility of nation-wide implementation, there is very little empirical evidence of its effect in changing consumption patterns. On the contrary, there is a small but growing body Of research that questions its effectiveness in harm reduction, even suggesting that it could be psychologically damaging to Aboriginal people because of its application along racially-defined lines (Brimblecome et al. 2010; Commonwealth of Australia 2008).

In this paper we contribute to this research by examining the relationship between income management and commercial gambling. Among Aboriginal people in the Northern Territory, gambling takes two primary forms. The first includes 'non-commercial' card-games that take place largely on remote Aboriginal communities (Young et al. 2007; Stevens & Young 2009). The second is electronic gaming machine (EGM or 'poker machine') gambling, a form of gambling described in the vernacular as 'pressing buttons'. This form of gambling occurs in sporting and recreation clubs, public hotels/bars and casinos in the urban centres of the Northern Territory. Since income management is designed to reduce the amount of money available for EGM gambling, if the scheme is working as intended, a reduction of EGM expenditures directly following its roll out may be anticipated in urban areas that service Aboriginal populations. This is the scenario we test in the current paper. In other words, has the pushing of a powerful and interventionist policy 'button' translated into Aboriginal people spending less money at the level of the EGM venue?

To answer this question we conducted an interrupted time series analysis of EGM expenditure data in two remote urban centres, Alice Springs and Katherine, Northern Territory, to test whether there had been a decline in gambling revenues at the level of the individual venue following the introduction of income-management. (1)

Geography of the Northern Territory Intervention

Income management was introduced between September 2007 and October 2008 into areas defined by the NTER Act as 'prescribed areas' (AIHW 2009). These included all lands under the remit of the Aboriginal Land Rights Act (NT) 1976 (traditional lands that make up approximately 36 per cent of the Northern Territory), community living areas (areas of land smaller than four square kilometres that lie within pastoral leases), and town camps (described more fully below) (Northern Land Council 2003). Due to the range of land-tenures listed under the Act, prescribed areas vary significantly in size, from towns with populations of a few hundred to small outstations with only a handful of inhabitants. The primary prescribed areas incorporated 73 communities across the Northern Territory (AIHW 2009; Taylor 2008).

While there is a clear spatial distinction between the remote prescribed communities directly affected by the intervention and the urban centres indirectly affected, town camps literally bridge the gap between these two realms. Town camps are Aboriginal communities that are located proximate to, or directly within, urban areas. They offer well-established and strong chain-migration links between urban areas and the more remote communities (Taylor 1989, 1990). In 1982, the House of Representatives Standing Committee on Aboriginal Committees (HRSCAA 1982: xii) officially defined a town camp as:

Town camps are particularly important in the context of our current study as they provide easy access to urban facilities and EGM venues, yet their populations are still subject to income management. In all prescribed areas, the excluded items of pornographic material and alcohol were banned completely. However, some exceptions were made to cater for non-Aboriginal recreational use within prescribed areas so that some items (mainly alcohol) could be transported or consumed within them (such as in Kakadu National Park) (FaHCSIA 2009a). The overall result is a fragmented administrative landscape of remote prescribed areas that are controlled directly by the Commonwealth Government. This control is exercised through a five-year lease term introduced by the NTER Act for over 60 communities. These leases afforded the Commonwealth 'exclusive possession and quiet enjoyment of the land while the lease is in force' (Commonwealth of Australia 2007c). While Aboriginal land owners still maintain ownership of the land, the leases provided temporary tenure to the Commonwealth. The leases were justified by the Commonwealth on the grounds that they allowed 'prompt access for the delivery of services, repair of buildings and upgrade of infrastructure in communities' (FaHCSIA 2009b).

Previous evaluations of the impact of income management

At the time of writing, we are only aware of a handful of studies that comment on the effectiveness of income management. In 2008 the Central Land Council (CLC) reported perspectives on the NTER from 141 residents of six remote communities in central Australia (CLC 2008). One identified advantage of income management was a perceived reduction in gambling and drinking, although gambling in this context referred to community card games rather than EGMs. However, the report added the qualifier that:

A similar opinion-based survey of Government Business Managers (GBMs) in 62 of the 73 prescribed communities was conducted by FaHCSIA (2009c). Almost 80 per cent of GBMs perceived a positive effect on 'shopping and budgeting patterns', and while this is somewhat vague, we could infer a perception that less money was spent on excluded services. Third, the Australian Institute of Health and Welfare (AIHW) released a report on the progress of the income management scheme (AIHW 2009). This study involved face-to-face interviews with 76 individuals from four of the remote prescribed areas. The authors found that two-thirds of interview participants (a total of 42 individuals) reported spending less on gambling. The AIHW outlined some important limitations of their findings, particularly in a geographic sense, pointing out that their four selected study areas were not representative of all prescribed areas. Given this small sample size and the fact that 40 per cent of respondents were from one community, we would argue that the results of the AIHW report as well as those of the CLC and FaHCSIA reports are an inadequate basis for rigorously judging the effectiveness of income management (see also Hunt 2010; Sanders 2010). A more empirically robust study was conducted by Brimblecombe and colleagues (2010) who used regression analysis to examine the impact of income management on food and tobacco sales within 10 remote communities. This study found no associated increase in spending on food items and no permanent decrease in the sale of tobacco. The study also reported no increase in spending on healthy food items such as fruit and vegetables. Unfortunately, this study was limited to 10 of 73 prescribed areas. Given the depth of this intervention into the lives of Aboriginal people, and the plans for extension to all disadvantaged people on social security benefits across the country, we need more convincing evidence not only about the effect on the purchase of banned goods, but also on the excluded service--commercial gambling.

Electronic gaming machines in the Northern Territory

In 1995, the NT Government enacted the Gaming Machine Act 1995 No. 50/1995, permitting the installation of EGMs in public hotels (pubs) and sporting or recreational clubs. Prior to this, the casinos in Darwin and Alice Springs had held exclusive rights to EGMs. A cap on the number of machines was set for the entire Northern Territory, with 500 machines permitted in venues north of Tennant Creek and 180 machines in venues to the south (although this cap has since been relaxed). Caps were also set for pubs (10 EGMs per pub) and clubs (45 EGMs per club). Pubs and clubs were permitted EGM licenses only if they already held a liquor license, a move that tied gambling to places where alcohol is consumed. Despite the capping limits on individual venues, the introduction of EGMs has continued steadily over the past decade through both an increase in the number of EGMs per venue (as individual venues move towards their capped limit) and the opening of new EGM venues.

The limited evidence available to date suggests that Aboriginal participation in EGM-based gambling has been increasing (Young et al. 2007). However, this participation is stratified by venue type. Research by Young and colleagues (2011) that involved visits to a number of venues in remote towns in the Northern Territory, including Katherine, Tennant Creek, and Nhulunbuy, found that certain pubs were more likely to cater specifically to an Aboriginal clientele than the clubs (see also Brady 1988; Merlan 1998). Moreover, pub usage by Aboriginal patrons is temporally patterned, with people using the venue during the day until 2pm, at which time the restrictions on the purchase of takeaway alcohol are lifted and people leave to drink elsewhere. In the words of Young and colleagues (2011: 64):

This spatio-temporal patterning is reinforced by the gate-keeping practices of different venues, with clubs being more heavily regulated. These include regulations which permit access only to those meeting set standards of clothing, footwear and personal hygiene--restrictions that allow clubs to regulate access at the level of the individual. A particular sort of space is produced by the clubs, one that accepts Aboriginal patrons to the extent that they conform to existing 'white' norms. Nonetheless, a significant proportion of EGM players are Aboriginal, particularly in the larger clubs. Pubs tend to create a very different kind of space, one that is less reflective of these 'white' norms and more inclusionary of socioeconomically disadvantaged Aboriginal people. Therefore, we would expect that these practices make entry by welfare recipients into clubs more difficult than for pubs (Young et al. 2011). The implication of this market segregation for our present study is that a time series analysis of EGM revenues for clubs is unlikely to show any effect of income management simply because there will be too few Aboriginal welfare recipients playing EGMs in clubs to significantly affect the total revenue of these venues. We would expect significant reductions, if evident at all, to appear in those pubs that are most frequented by Aboriginal people.

Study areas

Alice Springs, located in Australia's central Simpson Desert (Figure 1), is the largest town in central Australia, with a usually resident population of just over 21,600, 16 per cent of whom identify as Aboriginal or Torres Strait Islander (ABS 2007a). The town is bisected by the Todd River and Stuart Highway (Figure 2). Given Alice Springs is a desert town, the Todd River is usually empty and forms a sandy axis to the town that is used by Aboriginal people not only for sleeping and camping (locally called 'sleeping rough') but also as a major meeting point (Foster et al. 2005). There are more than 20 town camps (official and unofficial) in the vicinity of Alice Springs, with an estimated combined population in 2006 of nearly 1,200 (ABS 2007b) (Figure 1). This population estimate likely under-represents the true population at any given time. Tangentyere Council, the organisation responsible for the provision of social services to the town camps, has estimated a population of 1,700 (Foster et al. 2005). This higher estimate reflects the fact that the population of the town camps is highly mobile and incorporates local residents (who are themselves mobile), visitors from other town camps, and visitors from the remote hinterland. Indeed, the hinterland or service area of Alice Springs is truly vast, and encompasses the entire centre of Australia, stretching across from the Western Australian border to the west, the South Australian border to the south, and to Queensland in the east.

Alice Springs had a total of nine pubs and clubs with EGMs, as well as one casino, in July 2010. These venues are the Alice Springs Golf Club (six EGMs), Alice Springs Memorial Club (45 EGMs), Alice Springs RSL Club (24 EGMs), Federal Sports Club (10 EGMs), Gapview Resort Hotel (10 EGMs), Gillen Club (45 EGMs), Heavitree Gap Tavern (10 EGMs), Todd Tavern (10 EGMs), Town & Country Tavern (10 EGMs), and Lasseters Casino (~250 EGMs) (see Figure 2 for their locations). We excluded the Town & Country Tavern because it only installed EGMs in 2009 after the intervention began. We also excluded the casino, as time series expenditure data are only available from the NT government at the aggregated level for both the Darwin and Alice Springs casinos.

Katherine is located at approximately 300km to the south of Darwin on the Stuart Highway (Figure 1). Its usual population in 2006 was just under 5,900 (place of usual residence) of whom 28 per cent identified as Aboriginal or Tortes Strait Islander (ABS 2007c). Three town camps are located in or near the township limits (only two are identified on Figure 3, the third is located beyond the map to the southwest). These town camps had a population in 2006 of 435 persons, of whom 95 per cent identified as Aboriginal or Torres Strait Islander, with roughly half over the age of 18 (ABS 2007d). Again, this number is likely an underestimation. The Katherine town camps have well-established chain-migration links with rural communities in the Katherine region, and other urban centres such as Darwin and Alice Springs (Taylor 1989; Merlan 1998). Taylor (1989) has suggested that nearly one-quarter of the Aboriginal population of Katherine are 'temporary residents involved in a series of repetitive short-term movements many of which emanate from communities in the Katherine region.' A total of nine communities in the Katherine region, including the three Katherine town camps, were subject to income management (AIHW 2009). There are five venues within Katherine that possess EGM licenses. These are the Katherine Hotel (10 EGMs), Crossways Hotel (since renamed the Stuart Hotel) (10 EGMs), Katherine Club (45 EGMs), Katherine Country Club (35 EGMs), and Katherine Sports and Recreation Club (19 EGMs).

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

Given the role of both Alice Springs and Katherine as service centres for a range of communities that are income managed, they presented a logical choice for assessment of the effects of income management on EGM expenditures. As these centres were subjected to income management relatively early they also provided sufficient data points for reliable time series analysis. In Alice Springs, the implementation of income management was preceded by a community information program (from October 2007 to December 2007) and income management officially commenced in early January 2008. Katherine's income quarantining began a year earlier in December 2007. These intervention points are marked for all charts in Figures 4 and 5.

[FIGURE 3 OMITTED]

Methods

Dataset

For this study, we analysed EGM monthly expenditure data from July 2002 to July 2010, disaggregated by EGM venue and month. Expenditure values were adjusted for inflation using the Consumer Price Index for Capital Cities to 1996 AUD (ABS 2009).We used expenditure per machine (EPM) as the outcome measure as this allows direct comparison across different venues which have differing numbers of machines (clubs having up to 45 and pubs up to 10). While there is the potential for the number of machines within a venue (particularly clubs) to vary over time, hence introducing the risk of conflating supply with market characteristics, none of the venues in question changed their number of EGMs during or after the implementation of income management. In this sense, EPM remains a pure market demand measure. Further, there is a very strong positive correlation between EPM and number of EGMs in all venues across the NT to the point that marginal demand does not decrease with an increase in EGM numbers (Young et al. 2006; 2008). In terms of other potential confounding effects, in January 2010 the NT Government implemented a smoking ban in venues across the Territory. Consistent with the effect of such bans in other parts of Australia (Lal & Siahpush 2008), the smoking ban reduced all the pub revenues, particularly those of the Todd Tavern. Therefore, in the final analysis, we excluded EPM values from January to July 2010 to avoid a potential confounding effect.

Interrupted time series analysis

Interrupted time series analysis, also referred to as intervention or impact analysis, is a well-established methodology designed to estimate the effects of an event (in this case the implementation of income management) on a particular outcome variable measured over time (EGM expenditure) (Box & Tiao 1975; McCleary & Hay 1980). The technique provides answers to two major questions: 1) was there an impact on EGM expenditure and what was the magnitude of this impact? 2) how long did the impact last? We used Autoregressive Integrated Moving Average (ARIMA) and Seasonal Autoregressive Integrated Moving Average (SARIMA) techniques, specifically the Box-Jenkins strategy, to develop a parsimonious ARIMA model of the EPM time series that accounts for trend, seasonality and autocorrelation inherent in the time series data (Chatfield 1980; Box & Jenkins 1976). For this modelling process, we used the pre-intervention series to develop the model and then tested the significance of each parameter estimate along with the Ljung-Box Q statistic to verify the selected model. We refer the reader to the numerous texts available for a more in-depth discussion of the ARIMA modelling process (for example, Box & Jenkins 1976; Chatfield 1980; Cryer & Chan 2008; McCleary & Hay 1980).

The impact of the interruption is modelled by estimating a transfer function, which may take one of several forms (Cryer & Chan 2008). Following McDowall and colleagues (1980), we selected two plausible transfer functions: first, the impact of income management began immediately and remained constant through the study period (abrupt permanent) and second, the impact began immediately but declined through the study period (abrupt temporary). The abrupt permanent pattern suggests a level shift in the post-intervention series. This may be estimated using a step function, where [[omega].sub.0] estimates the magnitude of a constant change in EPM after the introduction of income management (McDowall et al. 1980). This transfer function aligns with the goals of the intervention in the NT, as one would expect that if there was an impact of income management on EPM the onset would be immediate and permanent. The abrupt temporary pattern uses a first-order transfer function where the impact of the magnitude [[omega].sub.0] applies only during the first month of the intervention and decreases by the rate [delta] until returning to pre- intervention levels. This transfer function would apply if the NT intervention caused a temporary change in gambling behaviour, gradually reverting to pre-intervention levels over time, as has been suggested for alcohol consumption (AIDA & CHETRE 2010). Significance for the intervention model is tested through traditional means.

Results

The results of the interrupted time series are presented in Table 1 for Alice Springs and Table 2 for Katherine. Each row contains the estimated values and significance for the initial magnitude of the intervention of EPM ([[omega].sub.0]) and rate of decline of that effect ([delta]). These values are grouped by the venue and the type of pattern used for the transfer function. Of the venues in the two study areas, only the Todd Tavern (Alice Springs) and Crossways Hotel (Katherine) showed significant and permanent reduction in EPM after the intervention (see Figures 4 and 5). Specifically, the abrupt permanent impact pattern had a significant negative estimate for [[omega].sub.0]. According to this estimate, monthly EPM after December 2007 fell by approximately $800 for the Crossways Hotel and $450 for the Todd Tavern (adjusted to 1996 values). A visual interpretation of the EPM series suggests that these changes were seasonally patterned, with a potential reduction in the winter peak.

The only other venue to show a significant effect was the Katherine Hotel. While neither of the permanent impact patterns showed significant results, the abrupt temporary estimate for magnitude and rate did. Surprisingly, [[omega].sub.0] had a positive rather than negative estimate, showing an increase in the level from pre-to post-intervention for December 2007. The rate of 0.776 indicates a gradual return to pre-intervention levels rather than a rapid return. In other words, the Katherine Hotel experienced a temporary spike in EGM expenditure at the start of income management.

As anticipated, none of the clubs displayed an effect. We suggest that this is because Aboriginal patrons are more highly represented in the public hotels rather than the membership-based clubs (Young et al. 2011). That is not to say that EGM expenditure in the clubs was not affected to some degree by income management, rather that Aboriginal patrons are unlikely to be gambling in sufficient proportions in the clubs for this to be reflected in aggregate monthly figures at statistically significant levels. Muddying the waters somewhat is the high natural variability in EGM expenditures in remote towns (Young et al. 2009). We have attempted to control for this by using EPM as the outcome measure but this is potentially affected by a range of other variables that may mask the effect of income management.

[FIGURE 4 OMITTED]

[FIGURE 5 OMITTED]

Discussion

Of all the venues in Alice Springs, the Todd Tavern was the only venue to display an immediate and significant drop in EGM revenue following the start of the intervention. The results suggest that income management may have been responsible for this reduction. However, the non-significant results for the other two pubs (the Gapview and the Heavitree Gap) require careful consideration because evidence suggests that pubs in remote urban centres tend to be more socially inclusive of a range of Aboriginal people than the membership-based clubs (Brady 1988; Merlan 1998; Young et al. 2011). Given that we know all pubs in Alice Springs are frequented by Aboriginal people, the key question is why the Todd Tavern alone would display a significant result?

The answer to this question, we suggest, lies at least in part in the local- level variation in markets. There are a number of reasons to suggest the Todd Tavern is the pub most frequented by Aboriginal people. First, the Todd Tavern has a separate bar that is used only by Aboriginal people. This is the Riverside Bar, also informally referred to as the 'Aboriginal bar', 'animal bar', 'monkey bar' or 'peanut bar' (by Aboriginal and non-Aboriginal alike). The Todd Tavern has been featured specifically in the national media for the large number of Aboriginal drinkers it serves between 10am and 2pm (after which time the bar closes) (Rebgetz 2010; Sleath 2010). Indeed, the tavern's license was suspended in 2010 for having over 236 patrons in the Riverside Bar when it was only permitted to hold 100 (Northern Territory Licensing Commission 2010), suggesting that the bar is generally at capacity during its opening hours.

The Todd Tavern is also the most spatially accessible pub in town. As Figure 2 shows, the pub is located adjacent to the central business district, at the end of the Todd Mall, the main formal thoroughfare in town. Importantly, it also intersects with the Todd River bed, which is an important thoroughfare for Aboriginal people. Thus, the Todd Tavern is effectively located at the intersection of two of the major Aboriginal routes through town. In comparison, the Heavitree Gap and Gap View are located further from centre of town (Figure 2). Furthermore, EGMs in the Todd Tavern are located in their own lounge bar that faces directly onto the Todd River. This means that the venue may be more appealing to those who just wish to gamble and avoid the heavy throng of drinkers.

In Katherine, EGM expenditure in the Crossways (now Stuart) hotel was negatively affected by income management, the Katherine Hotel was positively and temporarily affected, and the clubs remained unaffected. We suggest the result for the Crossways Hotel is significant in that it indicates a potential link to income management. This makes sense given we know that this particular hotel has a large Aboriginal clientele (Young et al. 2011). The reverse pattern for the Katherine Hotel is more difficult to explain. From a simple visual examination of Figure 5 it is difficult to identify this increase, which is obscured by the distinct seasonality in EGM expenditure in the Northern Territory (Young et al. 2009). To aid interpretation, the mean EPM for the Katherine Hotel is displayed in Figure 5 as a grey bar for July through December and as a black bar for January through June. These bars indicate a general pattern of increased spending from July to December and less spending from January to June. They also illustrate the increase in EGM spend immediately following the intervention and the gradual return to lower levels the following year.

However, it is difficult to reconcile income management, a policy designed to reduce disposable income on certain goods and services, with the increase in EGM revenues for this venue. We know that the Aboriginal population in these two regions is mobile (as are other groups in the Northern Territory, such as fly-in-fly out workers and tourists). This migration is intermittent and could conceivably account for the temporary increase in expenditure at the Katherine Hotel (Taylor 1989). It is also possible that the weak or non-existent effect across venues other than the Todd Tavern and Crossways Hotel may be influenced by other, unintended policy consequences of the intervention. Social service providers in Alice Springs have reported an increase in the number of remote Aboriginal people spending time in town since the intervention, certainly since the last mobility study in town camps conducted by Foster and colleagues (2005). Any increase is potentially associated with the alcohol bans on all Aboriginal communities in the Northern Territory (introduced as part of the intervention) which may have caused more people to come into town to drink alcohol, an aspect of rural-urban temporary migration in the Northern Territory which has been noted for some time (Taylor 1989).There may be an inter-play between temporary migration, income management, macro-policy, venue usage and EGM gambling that the current analysis is unable to disentangle. Therefore, as suggested by the SPRC (2010: 24), effective evaluation of income management demands an analysis at the local geographical level:

Indeed, it is clear that the strength of the conclusions we can draw are limited both by the correlative nature of our analysis, by other potential confounding policies and by the local variability in EGM markets. In terms of our analysis we have a limited number of observations and the significant outcomes are correlated rather than causally linked to the treatment variable (in this case income management). Thrown into this mix is the effect of other policy interventions. In the words of the SPRC (2010:20):

As a case in point we refer to the school enrolment and attendance through welfare reform measure (SEAM) that conditionally linked social security payment to school attendance. SEAM was trialled in Katherine and its town camps from the beginning of the 2009 school year (SPRC 2010: 34) and may have affected social security payments from this time onwards. In addition, local gambling markets are dynamic and localised. While the pubs in Alice Springs and Katherine have a significant Aboriginal clientele, this visitation tends to peak during the middle of the day (Young et al. 2011). As we have pointed out above, visitation patterns are closely tied to alcohol management regimes. This means that the pubs are most likely to have Aboriginal patronage (and by implication Aboriginal gamblers) before 2pm, but a different clientele after that time (for example tradesmen and local residents in the late afternoon and evening). For example, the Riverside Bar of the Todd Tavern in Alice Springs closes at 2pm, but the rest of the venue remains open, including the EGM area. The possibility exists that the effects of the intervention could be diffused by this temporal and motivational market segregation in that there is not a large enough proportion of Aboriginal gamblers in the other pubs to significantly affect revenues. As a final point, there are a number of gambling options other than EGMs, including community card games, TAB betting and internet gambling, that may compete for disposable income. For example, in an environment of restricted cash availability, card games may become more important as a distributor of scarce resources. In addition, given the intent of the NT government to improve broadband internet access to Aboriginal communities (Rennie et al. 2011), the opportunity to gamble off-premises may increase. The effect of income management on these product substitutes may similarly confound our analysis.

Conclusion

In summary, income management is associated with a reduction in EGM revenue for one pub in each of the towns studied. We have suggested that the reason this effect only appears in the expenditure data for a single venue in each may relate to the local variation in EGM markets, where Aboriginal patrons tend to use specific pubs, although our conclusions are limited by the correlative analysis and other confounding policy effects. We tentatively conclude that the intervention has appeared to reduce the cash available for EGM gambling, although this effect is difficult to separate from a range of localised spatial, social and temporal variables. The core issue here is one of underlying complexity--there exist heavily localised and nuanced relationships between gambling behaviour, Aboriginal patronage of venues, culturally influenced patterns of engagement and government policy. In a place like the Northern Territory, the results of the intervention will play out in different ways in different local contexts and these variations are difficult to isolate in a correlative empirical analysis such as the one presented here. Our mixed results, along with the findings of Brimblecombe et al. (2010), make the case that the Australian Government should not rush into another blanket policy of extending income management until we have more convincing evidence about the positive local effects of these policies. Implementation of the evaluative framework developed by the SPRC (2010) is crucial in this regard. Without such an evidence base, we are in no position to identify the local consequences of Federal 'button pushing.' For example, while we may expect the current extension of NIM to other disadvantaged social groups to reduce further gambling revenues across a wider range of venues (possibly including clubs) from August 2011 onwards, we would require a repeat of the current analysis in several years' time to test this hypothesis empirically.

In this context of complexity and variation, we argue that attention could be usefully directed towards policy options at the local level. Local complexity means we need more sensitive measures to understand, assess and respond to local issues such as alcohol and gambling consumption. For example, in August 2010, the Northern Territory Government announced the buy-back of alcohol licences of three takeaway outlets in Alice Springs, including the Heavitree Gap Store. We would expect this initiative to reduce visitation of the pub next door. In this sense, local-level spatial research techniques such as the measurement and modelling of venue catchments may have a lot to offer in evaluating the local impacts of macro policy interventions (Doran & Young 2010). Macro policy 'button-pressing', if it is to be used, requires a concurrent process for rigorous evaluation (SPRC, 2010), as well as a raft of local-level measures sensitive to local geographies. Here we support the words of Fred Chaney (former deputy chairman of the Australian Native Title Tribunal) when asked about failed government measures towards Aboriginal people: '... you can't solve these things by centralised bureaucratic direction' (O'Brien 2007). It is unlikely that macro-policies alone, especially those that are not supported by a convincing evidence base (as is the case for the NTER), will solve the problems faced by Aboriginal Australians (see also the similar conclusions by Hunt 2010 & Sanders 2010). We also require locally-sensitive policies formulated jointly with Aboriginal interest groups. Wild & Anderson (2007) made similar recommendations, but these appear to have been drowned out in the wake of the intervention.

Acknowledgements

This research was supported by the Australian Research Council Linkage Grant LP0990584 and the Community Benefit Fund of the Northern Territory Government. The authors thank two anonymous reviewers for critical comment on an earlier draft and Francis Markham, Menzies School of Health Research, for his valuable critique, editorial assistance and contributions to graphic production.

References

Aboriginal and Torres Strait Islander Social Justice Commissioner (2007) Social Justice Report 2007, Sydney, Human Rights and Equal Opportunity Commission.

ABS (Australian Bureau of Statistics) (2007a) 2006 Census QuickStats: Alice Springs (Urban Centre/Locality), http://www.censusdata.abs.gov.au/ ABSNavigation/prenav/ProductSelect? newproducttype=QuickStats&collection= Census&period=2006&areacode=UCL700200&action=401.

--(2007b) 2006 Census QuickStats: Alice Springs (T)--Town Camps (Indigenous Area), http://www.censusdata.abs.gov.au/ABSNavigation/ prenav/ProductSelect?newproducttype=QuickStats&collection=Census& period=2006 &areacode=IARE30001&action=401.

--(2007c) Census QuickStats: Katherine (Urban Centre/Locality), http://www.censusdata.abs.gov.au/ABSNavigation/prenav/ProductSelect ?newproducttype=QuickStats&collection=Census&period=2006&areacode--UCL7036 00&action=401.

--(2007d) Population Distribution, Aboriginal and Torres Strait Islander Australians, 2006, Cat. no. 4705.0.

--(2009) Consumer Price Index, Australia, Groups, Index Numbers by Capital City, Cat. no. 6401.0.

AIDA (Australian Indigenous Doctors' Association) & CHETRE (Centre for Health Equity Training, Research and Evaluation) (2010) Health impact assessment of the Northern Territory Emergency Response, Canberra, Australian Indigenous Doctors' Association.

AIHW (Australian Institute of Health and Welfare) (2009) Report on the Evaluation of Income Management in the Northern Territory, Canberra, Australian Institute of Health and Welfare.

Box, G.E. & Jenkins, G.M. (1976) Time Series Analysis Forecasting and Control, Oakland, Holden-Day.

Box G.E. & Tiao, G. (1975) 'Intervention analysis with applications to economic and environmental problems', Journal of the American Statistical Association, 70 (349), 70-79.

Brady, M. (1988) Where the beer truck stopped: drinking in a northern Australian town, Darwin, North Australia Research Unit, Australian National University.

Brimblecome, J.K., McDonnell, J., Barnes, A., Dhurrkay, J.G., Thomas, D.P. & Bailie, R.S. (2010) 'Impact of income management on store sales in the Northern Territory', Medical Journal of Australia, 192 (10), 549-554.

Chatfield, C. (1989) The Analysis of Time Series: An Introduction (4th ed.), New York, Chapman and Hall.

Commonwealth of Australia (1976) Aboriginal Land Rights (Northern Territory) Act 1976, http://www.austlii.edu.au/au/legis/cth/consol_act/ alrta1976444.

Commonwealth of Australia (2007a) Northern Territory National Emergency Response Act 2007 (no. 129) http://www.austlii.edu.au/au/legis/cth/consol_ act/ntnera2007531/.

Commonwealth of Australia (2007b) Social Security and Other Legislation Amendment (Welfare Payment Reform) (no. 130) http://www.austlii.edu. au/au/legis/cth/consol_act/ssaolapra2007674/notes.html.

Commonwealth of Australia (2007c) Northern Territory National Emergency Response Act 2007 (no. 129), Section 35 Terms and Conditions of Leases http://www.austlii.edu.au/au/legis/cth/consol_act/ntnera2007531/s35.html.

Commonwealth of Australia (2008) Northern Territory Emergency Response Review. Report to the Review Board http://www.nterreview.gov.au/docs/ report_nter review.pd f.

Cryer, J.D. & Chan, K.S. (2008) Time Series Analysis with Applications in R (2nd ed.), New York, Springer.

Doran, B. & Young, M. (2010) 'Predicting the spatial distribution of gambling vulnerability: an application of gravity modelling using ABS Mesh Blocks', Applied Geographer, 30 (1), 141-152.

FaHCSIA (Department of Families, Housing, Community Services and Indigenous Affairs) (2009a) 'About the Response' http://www.fahcsia.gov. au/sa/indigenous/progserv/ntresponse/about_response/Pages/default.aspx.

--(2009b) Closing the Gap in the Northern Territory Monitoring Report-July to December 2009--Part Two http://www.facs.gov.au/sa/indigenous/ pubs/nter_reports/closing_gap_NT_jul_dec_2009/Documents/Part2_ Monitoring_Report/sec5.htm.

--(2009c) 2009 Survey of Government Business Managers http://www. fahcsia.gov.au/sa/indigenous/pubs/nter_reports/2009_survey_gbm/ Documents/sec2.htm.

Foster, D., Mitchell, J., Ulrik, J. & Williams, R. (2005) Population and mobility in the town camps of Alice Springs, Alice Springs, Tangentyere Council.

Gulson, K.N. & Parkes, R.J. (2010) 'From the barrel of the gun: policy incursions, land, and Aboriginal peoples in Australia', Environment and Planning A, 42 (2), 300-313.

House of Representatives Standing Committee on Aboriginal Affairs (1982) Strategies to Help Overcome the Problems of Aboriginal Town Camps, Canberra, Australian Government Publishing Service.

Hunt, J. (2010) 'Missed opportunity: the NTER and sustainable development outcomes for Aboriginal people in the Northern Territory,' Australian Journal of Social Issues, 45 (3), 417-431.

Jenkins, G. (1979) 'Practical experiences', in O.D. Anderson (ed.) Forecasting, New York, North-Holland, 43-166.

Lal, A. & Siahpush, M. (2008) 'The effect of smoke-free policies on electronic gaming machine expenditure in Victoria, Australia', Journal of Epidemiology and Community Health, 62 (1), 11-15.

Macklin, J. (2010) 'Rent, payment for NTER five-year leases', Media release, Department of Families, Housing, Community Services and Indigenous Affairs, Canberra, 25 May, http://www.jennymacklin.fahcsia.gov.au/ mediareleases/2010/Pages/rent_nter_25may10.aspx.

McCleary, R. & Hay Jr., R. A. (1980) Applied Time Series Analysis for the Social Sciences, Beverly Hills, Sage Publications.

McDowall, D., McCleary, R., Meidinger, E.E. & Hay Jr, R.A. (1980) Interrupted Time Series Analysis, Thousand Oaks, Sage Publications.

Merlan, E (1998) Caging the Rainbow: Places, Politics, and Aborigines in a North Australian Town, Honolulu, University of Hawai'i Press.

Northern Land Council (2003) 'Land & sea rights: community living areas' http://www.nlc.org.au/html/land_community.html.

Northern Territory Licensing Commission (2010) 'Reasons for Decision' http://www.nt.gov. au/justice/commission/decisions/100702 %20Todd%20 Tavern%20110.pdf.

O'Brien, K. (2007) 'Chaney retires after 12 years', 7.30 Report, ABC Television 19 April http://www.abc.net.au/ 7.30/content/2007/s1901893.htm.

Productivity Commission (2010) Gambling: Productivity Commission Inquiry Report No. 50, Canberra, Productivity Commission.

Rebgetz, L. (2010) 'NT government takes aim at Alice 'animal bars", ABC News, 30 August http://www.abc.net.au/news/stories/2010/08/30/2997294. htm.

Rennie, E., Crouch, A., Wright, A. & Thomas, J. (2011) Home Internet for remote Indigenous communities, Sydney: Australian Communications Consumer Action Network.

Sanders, W. (2010) 'Ideology, evidence and competing principles in Australian Indigenous affairs: from Brough to Rudd via Pearson and the NTER', Australian Journal of Social Issues, 45 (3), 307-331.

Sleath, E. (2010) 'Alice Springs responds', ABC Alice Springs, 19 August http-//www.abc.net.au/local/stories/2010/08/19/2987669.htm.

SPRC (Social Policy Research Centre) (2010) Evaluation framework for New Income Management (NIM), Sydney, University of New South Wales.

Stevens, M. & Young, M. (2009) 'Betting on the evidence: reported gambling problems amongst the Indigenous population of the Northern Territory', Australian and New Zealand Journal of Public Health, 33 (9), 556-565.

Taylor, J. (1989) 'Public policy and Aboriginal population mobility: insights from the Katherine region, Northern Territory', Australian Geographer, 20 (1), 47-53.

--(1990) 'Aboriginal intra-urban mobility in Katherine, Northern Territory', Urban Policy and Research, 8 (2), 76-80.

--(2008) 'Appendix 10--The spatial and temporal diffusion of NTER measures', in P. Yu, M. Duncan & B. Gray (eds.) Report of the NTER Review Board, Canberra, Commonwealth of Australia, http://www. nterreview.gov.au/docs/report_nter_review/appendix-10.htm.

Wild, R. & Anderson, P. (2007) Ampe Akelyernemane Meke Mekarle 'Little Children are Sacred', Darwin, Northern Territory Government Board of Inquiry into the Protection of Aboriginal Children from Sexual Abuse.

Young, M., Abu-Duhou, I., Barnes, T., Creed, E., Morris, M., Stevens, M. & Tyler, B. (2006) Northern Territory Gambling Prevalence Survey 2005, Darwin, School for Social and Policy Research, Charles Darwin University.

Young, M., Barnes, T., Stevens, M., Paterson, M. & Morris, M. (2007) 'The changing landscape of Indigenous gambling in Northern Australia: current knowledge and future directions', International Gambling Studies, 7 (3), 327-343.

Young, M., Tyler, W., Lamb, D. & Stevens, M. (2008) Expenditure on electronic gaming machines in the Northern Territory: a venues-based analysis, Darwin, School for Social and Policy Research, Charles Darwin University.

Young, M, Lamb, D.S. & Doran, B. (2009) 'Mountains and molehills: a spatiotemporal analysis of poker machine expenditure in the Northern Territory of Australia', Australian Geographer, 40 (3), 249-296.

--(2011) 'Gambling, resource distribution, and racial economy: an examination of poker machine expenditure in three remote Australian towns', Geographical Research, 49 (1), 59-71.

Endnotes

(1.) An interrupted time series analysis is commonly used to quantify the effect of regulatory changes (e.g. Lal & Siahpush 2008). It attempts to detect and estimate the magnitude of any statistically significant change in the level of a data series after an external shock, such as the introduction of a new regulation.
Any group of Aboriginals living at identified camp sites near
   or within towns or cities which form part of the socio-cultural
   structure of towns and cities, but which have a lifestyle that does
   not conform to that of the majority of non-Aboriginal residents
   and are not provided with essential services and housing on a
   basis comparable to the rest of the population.


... evidence of the impact of income management on gambling
   is inconclusive. Moreover, there were other people in the
   community who reported that income management has not
   resulted in any changes to gambling (CLC 2008: 25).


... the clientele of a number of pubs (e.g. Katherine Hotel,
   Crossways Hotel, and the Walkabout Tavern), particularly
   during busy mid-morning sessions, is almost entirely Aboriginal.
   This market appears highly diurnal with a marked peak in
   visitation from mid-morning to early-afternoon. In contrast with
   the pubs, remote clubs in the NT tend to have a more diverse
   clientele. The clubs are generally quieter, more controlled venues
   that cater to a wider range of social groups.


We recommend that a key component of the impact evaluation be an
   ecological analysis describing the association between the
   prevalence of income management and key outcome variables
   aggregated across small to medium geographic areas. Whether this
   analysis will be suitable for the examination of long-term outcomes
   will depend upon the extent to which people move between different
   locations.


Separating the impacts of NIM from those of other policies and
   programs is challenging. NIM is being implemented as part of a
   range of intersecting Commonwealth and Territory social policy
   initiatives which will have an impact on individuals and
   communities. Examples include policies related to alcohol
   restrictions, school nutrition and attendance programs and a range
   of other health initiatives. There may also be changes to other
   policies during the evaluation period.


Table 1. Estimated effect of income management on monthly
expenditure per machine  in AUD1996, Alice Springs.

                         Abrupt                Abrupt
                        Permanent             Temporary
                       Coefficient     P     Coefficient     P
Todd Tavern
[omega]O                  -457.47   0.028       -164.18    0.215
[sigma]                        -       -          -1.00   <0.001

Heavitree Gap
Tavern
[omega]O                  -406.55   0.150       -258.73    0.377
[sigma]                        -       -           0.00    1.000

Gapview Resort
Hotel
[omega]O                  -283.09   0.240       -240.51    0.290
[sigma]                        -       -           0.00    1.000

Alice Springs
Golf Club
[omega]O                  -173.89   0.314        205.24    0.359
[sigma]                        -       -           0.72    0.193

Alice Springs
Memorial Club
[omega]O                  -157.52   0.166       -280.78    0.375
[sigma]                        -       -           0.62    0.319

Gillen Club
[omega]O                    26.04   0.912       -291.57    0.182
[sigma]                        -       -             20    0.792

Federal Sports Club
[omega]O                   -12.12   0.761        195.19    0.408
[sigma]                        -       -           0.48    0.474

Alice Springs RSL
[omega]O                    63.43   0.760        -62.15    0.683
[sigma]                        -                  -0.65    0.567

Table 2. Estimated effect of income management on monthly
expenditure per machine  in AUD1996. Katherine.

                          Abrupt                 Abrupt
                         Permanent              Temporary
                       Coefficient     P      Coefficient     P
Crossways Hotel
(now Stuart Hotel)
[Omega]O                   -866.52   <0.001       -823.78    0.005
[sigma]                         -        -           1.00   <0.001

Katherine Hotel
[Omega]O                    111.07    0.311       1205.31    0.001
[sigma]                         -        -           0.78   <0.001

Gapview Resort Hotel
[Omega]O                     53.36    0.794        234.79    0.116
[sigma]                                             -0.89   <0.001

Katherine
Country Club
[Omega]O                   -299.58    0.518         82.55    0.513
[sigma]                         -        -          -0.98   <0.001

Katherine Sports
and Recreation Club
[Omega]O                     56.14    0.653       -215.30    0.419
[sigma]                         -        -          -0.47    0.601
Gale Copyright: Copyright 2011 Gale, Cengage Learning. All rights reserved.