Meth Price Change Use Patterns

8/14/2019 Meth Price Change Use Patterns

1/16

Contemporary Issues in Crime and Justice Number 134

September 2009

This bulletin has been independently peer reviewed.

CRIME AND JUSTICE

B ul le t in NSW Bureau of CrimeStatistics and Research

How do methamphetamine users respond tochanges in methamphetamine price?Jenny Chalmers 1, Deborah Bradford 2 & Craig Jones 2

One of the core objectives of supply-side drug law enforcement is to reduce drug use by raising the cost of buying drugs.The effectiveness of this strategy depends on how illicit drug users respond to the rise in costs. The aim of the current study

was to estimate how methamphetamine users would respond to changes in the prices of methamphetamine and heroin,using hypothetical drug purchasing scenarios. A sample of 101 people who reported using methamphetamine in the past month was recruited from primary healthcare facilities in Sydney, Wollongong and Newcastle. Participants were given ahypothetical xed drug budget, presented with a range of drug pricelists and asked how many units of each drug on the

pricelist they would buy with their drug budget. The prices of methamphetamine and heroin were varied independently across successive trials and the quantity of each drug purchased at each methamphetamine and heroin price was recorded.Results revealed that methamphetamine purchases decreased signi cantly as the price of methamphetamine increased (a 10% price increase led to an 18%-19% fall), as did heroin purchases in response to heroin price increases (a 10% priceincrease led to a 16%-27% fall). Among methamphetamine users, increases in methamphetamine prices produced somesubstitution into heroin. Additionally, dependent methamphetamine users purchased more pharmaceutical opioids whilethe non-dependent group purchased more cocaine. Dependent and non-dependent heroin users also responded differently to changes in the price of heroin. Dependent heroin users reacted to increased heroin prices by signi cantly increasing their purchases of methamphetamine, pharmaceutical opioids and benzodiazepines. At the same time they purchased less cocaine. Non-dependent heroin users responded simply by increasing their purchases of pharmaceutical opioids. Inmost instances where substitution occurred, the fall in consumption of amphetamine (or heroin) was considerably greater than the increase in consumption of other drugs.

KEYWORDS: price elasticity, cross-price elasticity, supply-side drug law enforcement, methamphetamine, ice.

IntroductIon

Methamphetamine is one of the most

commonly used illicit drugs in Australia. 3 The most recent National Drug StrategyHousehold Survey estimated that morethan one million Australians or 6.3 per cent of the population aged 14 years andolder had used methamphetamine atleast once in their lifetime (AustralianInstitute of Health and Welfare 2008).While this represents a decrease fromthe 2004 gure of 9.1 per cent, someaggregate markers of methamphetamine-

related harm have been increasingin recent years. For example, therehave been substantial increases in thenumber of police-recorded incidents of

use/possess amphetamines (Degenhardtet al. 2008; Snowball et al. 2008) andin the number of methamphetamine-

related presentations to NSW emergencydepartments (Snowball et al. 2008).

Perhaps of paramount concern to policymakers is the high level of dependencefound among populations of frequentmethamphetamine users. Researchestimates that there were approximately28,000 dependent methamphetamineusers in NSW and over 72,000dependent users across Australia in

2003 (McKetin et al. 2005a). Heavy useof methamphetamine is known to resultin substantial harms to users, includingnancial stress, emotional and social

problems, physical and mental healthproblems (Darke et al. 2008; Degenhardt& Topp 2003; McKetin, McLaren & Kelly

2005b; McKetin et al. 2006; Vincent et al.1998; White, Breen & Degenhardt 2003),risky sexual and injecting behaviours(Darke et al. 2008), as well as posingrisks to the wider community through highrates of criminal activity (Degenhardt etal. 2008; McKetin et al. 2005a; 2005b).

Australias efforts to reduce these harmsare underpinned by a policy approachof harm minimisation, which comprises

three elements: demand reduction,harm reduction and supply reduction.Demand reduction measures are thosethat primarily focus on prevention of illicit


2/16

2

B U R E A U O F C R I M E S T A T I S T I C S A N D R E S E A R C H

drug use (e.g. school-based educationprograms) or treatment for dependentusers (e.g. cognitive behaviouralcounselling or pharmacotherapies).Not all demand reduction activities areeffective but there is some evidence in

support of programs such as school-based drug prevention (e.g. Caulkinset al. 2002). While pharmacotherapieshave been well established for heroindependence for many years, therehas typically been much less successin developing such treatments for methamphetamine and other psycho-stimulants. Nevertheless, recent clinicaltrials have provided some promising signsof progress (e.g. Shearer et al. 2009).

Harm reduction strategies are not intendedto reduce drug use per se but insteadaim to reduce the harms associatedwith substance abuse (e.g. needleand syringe programs and supervisedinjecting facilities are intended to reducethe transmission of blood borne viruses).Again, some harm reduction efforts willbe more effective than others but there iscertainly evidence that strategies such asneedle and syringe programs have beeneffective in limiting the harms associatedwith injecting use of illicit drugs suchas methamphetamine (CommonwealthDepartment of Health and Ageing 2002;Hurley, Jolley & Kaldor 1997).

Supply reduction efforts are thosethat aim to intercept illicit drugs beforethey get onto the market (e.g. throughcross-border interdiction or dismantlingdomestic drug laboratories). One of thecore rationales underpinning supply sidelaw enforcement is that, by increasingthe risks associated with trading in illicitdrugs, dealers will be forced to increasethe price of drugs to compensatethemselves for the risks they take. Someeconomists argue that, like most other commodities, an increase in the price of a drug should reduce consumption of thatdrug. Reductions in use should, in turn,bring reductions in drug-related harm.The relative ef cacy of supply reductionefforts therefore depends critically on

the extent to which consumption of illicitdrugs responds to changes in drugprices. Economists quantify the consumer reaction to price change in terms of price

elasticity of demand. For a particular drug,own-price elasticity of demand measuresthe change in consumption of that drugas a consequence of variation in its ownprice. The more consumption falls inresponse to a price increase, for example,

the more elastic is demand. Price elasticdemand describes the situation whereconsumption responds to price changesto the extent that an x per cent increasein price leads to at least an x per centreduction in consumption. In contrast,price inelastic demand is said to occur when an x per cent price increase resultsin a less than x per cent reduction inconsumption. If demand is price elastic,pushing up the price of an illicit drugwill produce a reduction in its use anda reduction in overall expenditure onthe drug. If demand is inelastic, pushingup the price of a drug may producesome reduction in its use but overallexpenditure on the drug will increase.

The way in which consumption of adrug changes in response to variationsin its price is only part of the picture.Methamphetamine users, like most drugusers, tend to use a range of substances.Some drugs are used in concert withmethamphetamine (i.e. as complementdrugs) while others may be used in placeof methamphetamine (i.e. as substitutedrugs). Consequently, increases inthe price of methamphetamine mayresult in decreased consumption of methamphetamine and some other (complement) drugs but could leadto increases in consumption of other

just as harmful (substitute) drugs. Amore complete understanding of the

effectiveness of an increase in the priceof methamphetamine requires knowledgeof how consumption of all drugs respondsto that price increase. For a particular drug, cross-price elasticity measures thechange in consumption of another drugin response to variations in the price of that drug. When the cross-price elasticityis positive the drugs are considered to besubstitutes and when it is negative theyare considered to be complements.

Economists have tended to estimate theresponsiveness of demand for illicit drugsusing secondary data sources. Price istypically either estimated from self-report

surveys or from under-cover police drugbuying operations. Consumption onthe other hand, is typically measuredeither directly via self-report surveys(e.g. Becker, Grossman & Murphy 1991;Grossman & Chaloupka 1998; Saffer

& Chaloupka 1999; see Grossman etal. 2002 for a review) or indirectly fromadministrative data sources such asdrug-related emergency departmentadmissions or the proportion of arresteestesting positive for drugs (Silverman &Spruill, 1977; Caulkins, 1995; 2001; Dave,2004; 2006). Manski, Pepper and Petrie(2001) reviewed a large number of studiesthat used both direct and indirect methodsand estimated that the own-price elasticityof demand for cocaine ranged from -.59 to-2.5 (i.e. ranging from relatively inelasticto highly elastic).

There is strong evidence from studies thatuse direct measures of use that own-priceelasticity of demand for heroin is relativelyelastic. Saffer and Chaloupka (1999), for example, estimated the price of cocaineand heroin from the US Department of Justices database of undercover drugpurchases and measured consumptionbased on the pooled 1988, 1990 and1991 national household drug usesurveys. Saffer and Chaloupka estimatedthat a 10 per cent increase in the priceof cocaine and heroin would result in2.8 and a 9.4 per cent decrease inparticipation in cocaine and heroin,respectively. While the studies based ondirect measures of use provide importantinsights into the elasticity of demand for drugs, the self-reported drug use datatend to be collected via household or

telephone surveys which by their verynature exclude marginalised populations,including those who might be at greatestrisk of being dependent on illicit drugs.This can be a problem when estimatingprice elasticity because dependent usersmight be less able than non-dependentusers to curb their consumption inresponse to price increases.

Indirect measures of drug use mightbetter re ect the experiences of

dependent users because those who usedrugs most frequently are also most likelyto come into contact with public agenciessuch as hospitals or police. While


3/16

3


most well-conducted studies that haveemployed indirect measures of use tendto also nd some responsiveness to pricechanges, the elasticity estimates havevaried considerably. For example, usingdata derived from arrestees and from

hospital emergency room admissions,Caulkins (1995; 2001) estimated the own-price elasticity of demand for cocaine tobe between -1.3 and -2.5 and own-priceelasticity of demand for heroin/morphineto be between -.84 and -1.5. On theother hand, Dave (2004; 2006) usedsimilar data but a different analyticalmethodology and estimated much moremodest short-run elasticities. In fact, Daveconcluded that demand falls within theinelastic range (-.15 to -.27 for cocaineand -.08 to -.10 for heroin).

Most of the studies reviewed aboveexamined the impact of the decliningheroin prices in post-Vietnam war Americaand, in particular, the crack cocaineepidemic that took hold in America in the1980s. In Australia, the heroin shortagethat manifested itself around Christmas2000 provided compelling support for the notion that demand for heroin mightbe relatively elastic. When the real priceof heroin more than doubled in a veryshort space of time, consumption of thedrug reduced dramatically, as did other indicators of drug related harm suchas fatal and non-fatal overdoses anddrug-related property offences (Moffatt,Weatherburn & Donnelly 2005; Smithsonet al. 2004; Weatherburn et al. 2003).While speci c price elasticities werenot estimated in relation to the heroinshortage, Weatherburn and colleagues

(2003) estimated from their survey of 167injecting drug users that a one per centincrease in the price per pure gram of heroin resulted in a 0.32 per cent fall inexpenditure on the drug.

In summary, the prevailing view, basedon direct and indirect observations of drug consumption measures, is thatconsumption of illicit drugs such asheroin and cocaine is responsive tochanges in price. However, the actual

elasticity estimates vary dependingon the populations, data sources andmethodologies employed. The economicresearch community is far less certain

about the implications of price changesin one drug for the consumption of other drugs. Analyses of drug use amongthe general population tend to nd thatalcohol is a complement for illicit drugs(e.g. Saffer and Chaloupka, 1999;

Pacula 1998). Furthermore, Saffer andChaloupka (1999) found complementaritybetween illicit drugs such as heroin,cocaine and cannabis. Studies usingindirect measures of illicit drug use alsotend to nd complementarity betweenheroin and cocaine use (e.g. Dave 2004;2006). Research into the effects of theAustralian heroin shortage found that aproportion of heroin users substitutedtheir use at least in the short term

with other drugs such as cocaine andmethamphetamine (Maher et al. 2007;Weatherburn et al. 2003). It is importantto point out, however, that these studiesobserved drug substitution among thoseusers who were still by de nition inthe market and this substitution doesnot appear to have occurred at a moreaggregate level (Snowball et al. 2008).Nevertheless, this research does suggestthat heroin and psycho-stimulant drugsmight act as short-term substitutes for some frequent substance users.

The literature reviewed above clearlyshows that estimates of price elasticityvary markedly depending on thepopulation under study, the nature of thedrug in question and the methodologyemployed to correlate measures of priceand consumption. Manski, Pepper andPetrie (2001) outline several reasons for these varying estimates, including thelack of uniform price data, the variation inprices charged (even by the same dealer)and the heterogeneity of drug users. For this reason, a growing body of overseasliterature has employed behaviouraleconomics techniques to exploreresponsiveness of demand in a morecontrolled environment (Cole et al. 2008;Goudie et al. 2007; Petry 2000; Petry &Bickel 1998; Sumnall et al. 2004). Thesestudies make use of experimental data,whereby current or former drug users

typically make hypothetical purchases of a range of drugs using imitation moneyprovided by the researcher. Participantsare given a price list outlining several

different drugs and are subsequentlyasked how much of each drug they wouldpurchase given a xed drug budget. Theresearcher then changes the prices of thevarious drugs over successive trials andrecords the participants drug purchases

at each of the drug prices. The resultingpurchase data is used to determine own-and cross-price elasticities. The theorybehind this methodological approach isthat, by virtue of making these purchasesin a controlled environment, any observedchanges in hypothetical consumptionshould be attributed entirely to changesin price. When purchased in a naturalsetting the quality of the drug can, andoften does, vary with price.

In a classic study of this kind conductedin the United States, Petry and Bickel(1998) found that the decrease inconsumption of heroin was roughlyproportional to the increase in its priceamong a group of current and former opioid-dependent participants. In contrastto the results of studies using secondarydata sources (Dave, 2004; 2006), Petryand Bickel found evidence of substitutionbetween illicit drugs. Increases in the

price of heroin induced relatively strongsubstitution into valium and cocaine(cross-price elasticities of 1.02 and0.8, respectively) with more moderatesubstitution into cannabis and alcohol(elasticities for both were 0.5). 4 In a morerecent study, Jofre-Bonet and Petry(2008) also found that the own-priceelasticities of demand for cocaine andheroin were close to one, regardless of levels of dependence on one or other of the drugs. Heroin-dependent people werefound to complement their heroin use withcocaine, alcohol and cannabis. The onlysubstitute for heroin was valium. Cocaine-dependent people complemented their cocaine use with heroin and alcohol butsubstituted into cannabis and valium. Inthe only study to consider amphetamineuse, Sumnall and colleagues (2004) foundthat demand for amphetamine, cocaineand ecstasy was quite elastic amongpoly-substance recreational users in the

UK. That is, purchases of each of thesedrugs decreased at a proportionatelygreater rate than the associated increasein price. Further, this study revealed


4/16

4


that for this population, alcohol was asubstitute for amphetamine, while ecstasyand cocaine purchases were independentof amphetamine prices.

There is clearly still much to be learnedabout how methamphetamine usersrespond when faced with signi cantchanges in price. We cannot be surethat samples of Australian drug userswould respond in the same way as thosestudied internationally. Furthermore, itmay be the case that responsivenessto changes in price varies accordingto the characteristics of those whouse drugs (e.g. contingent on their levels of dependence on heroin and/or methamphetamine) and the drug market

inhabited by the drug user. This is animportant knowledge gap in light of theconsiderable investment Australian lawenforcement agencies make in attemptingto curb illicit drug supply. The currentstudy used behavioural economicstechniques to examine the elasticity of demand for methamphetamine among asample of Australian methamphetamineusers. Because it is well establishedthat there is a high degree of poly-drug

use among populations of injectingdrug users (Darke & Hall 1995), wealso examined how this group wouldrespond to changes in the price of heroin. Participants were presented withhypothetical drug buying scenarios andthe price of methamphetamine and heroinwere systematically varied to determinehow consumption would likely respond tochanges in drug price.

The speci c research questions that the

current study aimed to address were:(a) How much would consumption

of methamphetamine change inresponse to changes in its price?

(b) How much would consumptionof heroin change in response tochanges in its price?

(c) How much would consumption of other drugs change in responseto changes in the price of methamphetamine or heroin?

(d) Does responsiveness of consumption vary according towhether the participant is dependenton methamphetamine or heroin?

Method

Procedure

The methodology, in large part,follows that of Petry and Bickel (1998).

Interviews were conducted betweenAugust and November 2008 by sixtrained interviewers. One hundredand one people who were at least 18years of age and who reported usingmethamphetamine in the previousmonth were recruited into the study.Potential participants were identi edthrough their attendance at one of four cooperating agencies in Sydney,Newcastle or Wollongong. Each of the agencies included a needle andsyringe programme. In the rst instance,participants were directly approached bythe researchers and invited to undertakea face-to-face interview. Snowballsampling was also used, wherebyparticipants who had completed thesurvey informed their friends about thestudy and these potential participantsapproached the researchers. Therewas no intent to sample a representativegroup of methamphetamine users.

Prior to participating in the interview,the nature of the study was explainedand participants were advised that anyinformation provided would be treatedcon dentially. All survey participants thensigned written consent forms.

Most interviews were conducted on site atone of the four agencies, although a smallnumber of interviews were conducted ata public housing estate where the localneedle and syringe program provided a

mobile service. During each interview,the interviewer read out all questionsand recorded participant responseson a paper copy of the questionnaire.Each interview ran for 30 to 45 minutesand participants were reimbursed $30as compensation for their time at thecompletion of the survey.

QuestIonnaIre

The critical part of the questionnaire

involved a series of questions abouthypothetical drug purchases across arange of drug prices. The interviewer read aloud the following instructions:

Think of a typical week. We are goingto use a price list and some pretendmoney to play a sort of game. Assumethat you have $200 that you can use tobuy drugs for that week or the lengthof time it would normally take you tospend $200 on drugs. The drugs that

are available to you and the price of those drugs are listed on this sheet.

The amount of each participants drugbudget was held constant at $200 duringthe experiment. 5 The size of the drugbudget approximates the weekly incomesupport received by single unemployedAustralians at the time of the interviews.As the majority of the study participantsrelied on income support as their primarysource of income, this budget was

intended to provide a realistic frame of reference for the hypothetical purchases.Participants were told that they did nothave to spend all of the $200 on the listeddrugs and were advised to hypotheticallypurchase drugs at the rate they normallywould. As a result, the timeframe over which participants could spend the $200budget was allowed to vary. For someparticipants, this budget was suf cientto cover their actual weekly usage, while

it would be clearly insuf cient for other respondents.

Each price list included the followingdrugs: heroin, methamphetamine (base,powder or crystal), cannabis, cocaine,non-prescription benzodiazepines,alcohol, and non-prescriptionpharmaceutical opioids (i.e. oxycodoneand morphine). The price of each drugre ects the average price from the 2008Illicit Drug Reporting System (IDRS)

survey (Phillips and Burns 2009). Thebaseline prices were $50 per cap of heroin, $50 per point of base, powder or crystal methamphetamine, $20 per gramof cannabis, $50 per cap of cocaine,$2 per benzodiazepine pill, $30 per pharmaceutical opioid pill 6 and $5 per unitof alcohol. Only one quantity per drugwas employed to avoid issues of quantitydiscounting.

The interviewer then explained the rules

of the game, as follows:You may buy any drugs that you like withthis money, but you can only spend $200on drugs. No other drugs are available


5/16


to you aside from those you buy withthis money. For example, you cannotuse drugs given to you by a friend andyou cannot buy any more drugs onceyou have spent the $200. Even if youhave drugs stashed away, you cannotuse them. You cant bargain with your

dealer or buy in bulk to reduce the price.Also, you cannot use the $200 to buy adrug not listed. The drugs that you buycan only be used by you and cannot begiven away or sold. Finally, the prices of some of the drugs will change acrossthe different price lists but the quality of the drugs does not change with pricechanges. The drugs are of the samequality that youd expect to get fromyour regular supplier and when the drugprices change I want you to imaginethat the prices will stay that way for the

foreseeable future. Please purchasethe drugs you would like from the list.

Illicit drug market adjustments do notonly manifest as changes in the price of a drug. More often than not it is the purityor quality of the drug that uctuates andprice changes tend to be accompaniedby variations in the purity or quality of the drug (Caulkins 2007). Hence, it wasimportant to impress upon participantsthat the quality did not vary with price.

Following the presentation of the baseprice list, eight further price lists werepresented to each participant in arandomised order and the prices of methamphetamine and heroin werevaried one at a time with each successivepresentation. In addition to the baselineprices of $50 per cap of heroin or pointof methamphetamine the following priceswere presented for both drugs: $10, $20,$75, and $100. This range of prices was

selected because it was within the boundsof what street-based samples of illicitdrug users in Sydney reported paying for heroin and methamphetamine (Phillipsand Burns 2009).

A number of other measures werealso collected from participantsboth to contextualise the patterns of drug use among this sample and toobserve how these characteristicsinteracted with willingness to purchase

methamphetamine and other drugs.These characteristics included a rangeof demographic characteristics (e.g. age,sex), patterns of drug use, severity of

dependence on methamphetamine andseverity of dependence on heroin (asindexed by the Severity of DependenceScale; Gossop et al. 1995).

analysIs

The behavioural economics literaturehas typically measured price elasticityin terms of the bivariate relationshipbetween hypothetical purchases(or consumption) and drug prices,operationalised as the percentage changein hypothetical consumption in relation tothe percentage change in price. The own-price elasticities for methamphetamineand heroin and cross-price elasticitiesfor all the drugs in relation to the pricesof both methamphetamine and heroinare measured by the slope of the plotof the price and consumption on log-log coordinates calculated by linear regression (Petry and Bickel 1998; Petry2000; Goudie et al. 2007). We follow thismethodology by estimating the impact of methamphetamine price on the amount of each drug purchased.

The data obtained from this surveycontain two features that complicate theanalysis of changes in the quantity of drugs purchased. Firstly, all of the 101participants provided multiple purchaseamounts in response to the nine uniqueprice lists. It is problematic to analysethe data as though the hypotheticalpurchases represent a simple randomsample because responses are correlatedwithin individuals. In other words,the amount any individual is willingto purchase at a given price is highlyrelated to the amount they are willingto purchase at another drug price. Thesecond complicating feature of thesedata is that the hypothetical units of a drug purchased are non-negativeintegers, or counts (e.g. 0, 1, 2, 3, 20points of methamphetamine). Becausethese count data are bounded by zeroand concentrated around a few discretevalues often close to zero thedistribution is highly skewed. The discrete,

skewed and non-independent natureof the data violates the assumptionsthat underpin ordinary least squaresregression and alternative regression

techniques were employed to estimatethe elasticities reported in this bulletin.Several models were considered butultimately zero-in ated negative binomialand zero-in ated poisson models werefound to provide the best t to the data. 7

All analyses were undertaken usingStata/SE 10.1. Statas cluster estimator of the variance covariance matrix was usedto account for the non-independent natureof the outcome variable (Baum 2006, pp.138-139).

calculatIng PrIce elastIcItyfroM count data analysIs

The estimated coef cient for a continuousexplanatory variable (such as price)represents the proportionate change inhypothetical purchases brought about bya $1 change in price, with all explanatoryvariables set to their mean values. Theestimated elasticity with respect to priceis simply the mean price ($51) multipliedby the estimated coef cient (Cameron etal. 1998).

results

PartIcIPants

The characteristics of the 101 participantsare summarised in Table 1. Mostparticipants were men (n=58). Themean age of participants was 38.6 years(range=18-58) and 84 had not completedany formal education beyond years 10or 11 of school. A little over half of theparticipants (n=52) reported that their primary source of income was income

support. All but ve of the respondentsindicated that they usually purchased theillicit drugs that they used. On average,participants reported that they were 16years of age when they rst used anyillicit drug and 21 years old when they rstused methamphetamine.

Heroin was the most often cited maindrug of choice at the time of the interview(n=42) with a third of respondentsciting methamphetamine as their main

drug of choice (n=33). One quarter (n=26) of the respondents reported thatmethamphetamine was the last drug usedwhile 39 participants last used heroin.


6/16


Thirty- ve per cent of the respondentsreported consuming methamphetamineon seven or fewer days in the monthpreceding the interview and a similar number reported using methamphetaminedaily. Over one third (n=36) of the

participants reported using heroin dailywhile 23 had not used heroin in the pastmonth. Crystalline methamphetamine(commonly known as ice) was the formof methamphetamine most recently usedby 71 participants. Over 90 per cent of participants injected methamphetamineand seven per cent smoked it on their last occasion of use. Two-thirds of therespondents (n=65) met the criteria for methamphetamine dependence, which

re ects the highly addictive nature of the drug, particularly when smoked or injected (Cho and Melega 2002; McKetinet al. 2008). Seventy per cent of therespondents met the criteria for heroindependence. Almost all respondentswho had used heroin in the past monthmet the criteria for heroin dependence.Only 10 of the respondents were notdependent on one or both of these drugs.Seventy per cent of those dependent onmethamphetamine were also dependenton heroin. Less than half (46%) of thosedependent on methamphetamine reportedthat methamphetamine was their drugof choice while just over half of thosedependent on heroin reported heroin astheir drug of choice.

descrIPtIve overvIew ofdrug PurchasIng behavIour

There were 909 hypothetical purchasesfor each drug listed (i.e. 9 lists x 101participants). Following the behaviouraleconomics literature we assessed thereliability of the hypothetical purchasesby calculating the strength of therelationships between the total amountsof each drug purchased over the nineprice lists and self-reported days of useof that drug in the past month. Spearmanrank order correlation coef cientsbetween hypothetical purchases andactual use were signi cant at the one

per cent level for methamphetamine,cocaine, heroin, cannabis and alcohol,and signi cant at the ve per centlevel for benzodiazepines. We were

Table 1. Descriptive characteristics of the sample of methamphetamine users (n=101)

Characteristic % Mean (SD)Men 57.4 -

Age (years) - 38.6 (7.9)

Education Tertiary/TAFE/Trade 16.8 -Year 10/11 or secondary 34.7 -Less than year 10/11 48.5 -

Primary source of income Employment 11.9 -Income support 55.4 -Dealing/scoring 8.9 -Other crime/sex work 14.9 -Other 12.9 -

I usually purchase the drugs I use 95.1 -

Age at rst use of an illicit drug - 15.6 (4.8)

Age rst used methamphetamine - 21.4 (7.9)

Main drug of choice Methamphetamine 32.7 -Heroin 41.6 -Other 25.7 -

Last drug used Methamphetamine 25.7 -Heroin 38.6 -Other 35.7 -

Lifetime use Methamphetamine 100 -Heroin 96 -Cocaine 87.1 -Alcohol 96 -Cannabis 100 -

Benzodiazepines 72.1 -Other opioids 90.1 -

Used in past month Methamphetamine 100 -Heroin 77.2 -Cocaine 43.6 -Alcohol 69.3 -Cannabis 87.3 -Benzodiazepines 65.3 -Other opioids 52.5 -

Daily use* Methamphetamine 34.7 -Heroin 35.6 -Cocaine 5.9 -Alcohol 15.8 -Cannabis 58.4 -

Methamphetamine form mostrecently used

Powder 14.9 -Base 13.9 -Ice/crystal 70.3 -Other 1 -

Route of administration of mostrecent use

Inject 91.1 -Smoke 6.9 -Swallow 1 -Other 1 -

Dependent Methamphetamine** 64.4 -

Heroin*** 70.3 -* Daily use is at least 28 days per month

** Using a cut-off of 4 on severity of dependence scale (Topp and Mattick, 1997)

*** Using a cut-off of 3 on severity of dependence scale (Gonzlez-Siz et al., 2009)


7/16


8/16


Figure 2b. Proportion of respondents reporting that they would buyheroin and amount purchased if bought in relation toprice of heroin by level of dependence

Price of heroin ($/cap)

(%) (caps)

0

10

20

30

40

50

60

70

8090

10 20 50 75 1000

12

3

4

56

7

89

10Buys heroin(LHS)Buys (d)

Buys (n-d)

Caps of heroinbought (RHS)

Caps bought(d)

Caps bought(n-d)

Figure 3a. Average hypothetical purchases of alternate drugs as theprice of methamphetamine was varied across the trials

0

0.5

1.0

1.5

2.0

2.5

10 20 50 75 100

Price of methamphetamine ($/point)

Units of drug

heroin

alcohol

benzodiazepines

cannabis

cocaine

pharmaceuticalopioids

Figure 3b. Average hypothetical purchases of alternate drugs as theprice of methamphetamine was varied across the trials

methamphetamine dependent

0

0.5

1.0

1.5

2.0

2.5

10 20 50 75 100


Units of drug

heroin

alcohol

benzodiazepines

cannabis

cocaine


(averaged across only those who wouldbe willing to buy at that price). Figure2b shows the analogous responses for heroin purchases. Those dependent onmethamphetamine and particularly thosewho were dependent on heroin were more

likely than non-dependent participants topurchase those drugs at any price. Whilethose who were dependent on herointended to buy slightly more of the drug atprices of $50 and lower, dependent andnon-dependent methamphetamine userspurchased roughly the same number of units of methamphetamine at each price.

For participants who did not meet thecriteria for methamphetamine or heroindependence, it is clear that willingness to

purchase that drug is inversely related tothe price, since the proportion of peoplewho purchase each drug falls with everyprice increase. Among those dependenton the drug in question the tendency topurchase the drug varied little with pricechanges between $10 and $20 but theproportion willing to buy decreased athigher drug prices. Interestingly, increasesin drug price beyond the $50 baseline didnot greatly affect the amount purchased

but decreases in the price of heroin andmethamphetamine below the baselineprice resulted in large increases in theamount purchased.

Figures 3a, 3b and 3c show howaverage purchases of the other drugsavailable on the pricelist responded tomovements in the methamphetamineprice for all respondents, those dependenton methamphetamine and those notdependent on methamphetamine,

respectively. Visually, the most markedrelationships to emerge from thesegraphs were the positive correlationsbetween both heroin and cocainepurchases, and methamphetamineprice, for the dependent and non-dependent groups. Pharmaceuticalopioid purchases appeared also to risewith methamphetamine price, but lessmarkedly, as did purchases of cannabis inthe non-dependent group.

Figures 4a, 4b and 4c show howaverage purchases of these other drugschanged in response to movementsin the heroin price. Responses tended


9/16

9


A ten per cent increase in the price of methamphetamine was estimated tolead to a 2.00 per cent increase in heroinconsumption and a 3.33 per cent increasein pharmaceutical opioid purchases.

For non-dependent methamphetamineusers, the own-price elasticity of demand was again estimated to beelastic (-1.866), suggesting that aten per cent increase in the price of methamphetamine would result in an18.66 per cent reduction in the quantityof methamphetamine purchased. Therewas evidence of substitution into heroin,cannabis, cocaine and pharmaceuticalopioids, although this substitution wasonly statistically signi cant for heroin and

cocaine. A ten per cent increase in theprice of methamphetamine was estimatedto lead to a 0.54 per cent increase inheroin consumption and a 7.05 per centincrease in cocaine consumption. Therewas some evidence that alcohol andbenzodiazepines were complements for methamphetamine in this group but theelasticities were not statistically signi cant(-0.046 and -0.152, respectively).

Turning next to responsiveness to

changes in heroin price among heroin-dependent users, the own-price elasticitywas also estimated to be in the elasticrange (-1.553). A 10 per cent increasein the price of heroin was estimatedto result in a 15.53 per cent reductionin the amount of heroin purchased.Contrary to the results for changes inmethamphetamine price, cocaine(-0.538) was found to be a complementfor heroin while benzodiazepines (0.308),

pharmaceutical opioids (0.395) and

Figure 3c. Average hypothetical purchases of alternate drugs as theprice of methamphetamine was varied across the trials

not methamphetamine dependent

0

0.5

1.0

1.5

2.0

2.5

10 20 50 75 100


Units of drug

heroin

alcohol

benzodiazepines

cannabis

cocaine


Table 2. Own-price and cross-price elasticity of demand when methamphetamine and heroin price change

Meth. Heroin Alcohol Cann. CocainePharm.opioids Benzos

Methamphetamine price

Methamphetamine dependent -1.766*** 0.200*** -0.011 0.050 0.121 0.333*** 0.070

Not methamphetamine dependent -1.866*** 0.054** -0.046 0.036 0.705*** 0.137 -0.152

Heroin price

Heroin dependent 0.255*** -1.553*** 0.099 -0.094 -0.538*** 0.395*** 0.308***

Not heroin dependent 0.013 -2.674*** -0.009 0.038 -0.202 0.764* 0.054

*** p


10/16

10


methamphetamine (0.255) were found tobe substitutes. A ten per cent increase inthe price of heroin was estimated to leadto a 5.38 per cent decrease in cocaineconsumption, a 3.08 per cent increase inbenzodiazepine consumption, a 3.95 per

cent increase in pharmaceutical opioidpurchases and a 2.55 per cent increase inmethamphetamine consumption.

Turning lastly to the results for non-dependent heroin users, the own-priceresponsiveness was highly elastic(-2.674), which indicates a greater than26 per cent reduction in consumption for heroin with each 10 per cent increasein its price. There was evidence of substitution into pharmaceutical opioid

consumption (cross-price elasticity= 0.764). While cocaine appeared tocomplement heroin use among thisgroup (cross-price elasticity = -0.202),this complementarity was not statisticallysigni cant.

IMPact of PrIcechanges on totalconsuMPtIon

While the elasticities shown in Table 2indicate that there is some substitutioninto other drugs when the price of methamphetamine and heroin rise,the net effect of changes in the priceof these drugs cannot be assessedwithout accounting for the base ratesof consumption of these drugs. 10 For example, while there is strong evidenceof substitution into cocaine use when theprice of methamphetamine increased,

the base rate of consumption of cocainewas quite low among this group. Evenif cocaine use doubled, therefore,it would not be enough to offset thebene ts associated with the decrease inmethamphetamine use.

Table 3a shows the total number of units of each drug purchased at eachmethamphetamine price. The table hasthree sections: the top section reportsdrug purchases for all participants;

the second section reports purchasesfor the subset of 65 participants whomet the criteria for methamphetaminedependence and the third section

Figure 4a. Average hypothetical purchases of alternate drugsas the price of heroin was varied across the trials

Units of drug

methamphetamine

alcohol

benzodiazepines

cannabis

cocaine


0

0.5

1.0

1.5

2.0

2.5

3.0

10 20 50 75 100


Figure 4b. Average hypothetical purchases of alternate drugsas the price of heroin was varied across the trials

heroin dependentUnits of drug

methamphetamine

alcohol

benzodiazepines

cannabis

cocaine


0

0.5

1.0

1.5

2.0

2.5

3.0


10 20 50 75 100

Figure 4c. Average hypothetical purchases of alternate drugsas the price of heroin was varied across the trials

not heroin dependentUnits of drug

methamphetamine

alcohol

benzodiazepines

cannabis

cocaine


0

0.5

1.0

1.5

2.0

2.5

3.0


10 20 50 75 100


11/16

11


Table 3a. Drugs bought at each methamphetamine price withpercentage difference from baseline in brackets*

Price of methamphetamine (point)

10 20 50 75 100

Methamphetamine (points) 710(487) 362(199) 121 68(-44) 42(-65)

Heroin (caps) 108(-5) 110(-4) 114 122(7) 134(18)

Alcohol (drinks) 133(-6) 130(-8) 142 131(-8) 140(-1)

Benzodiazepines (pills) 136(-18) 194(17) 166 177(7) 164(-1)

Cannabis (grams) 122(-10) 123(-10) 136 126(-7) 132(-3)

Cocaine (caps) 34(-8) 34(-8) 37 55(49) 55(49)

Pharmaceutical opioids (pills) 38(-31) 39(-29) 55 45(-18) 58(5)

Methamphetamine dependent (n= )


Heroin (caps) 51(-9) 51(-9) 56 61(9) 72(29)

Alcohol (drinks) 88(1) 92(6) 87 102(17) 97(11)

Benzodiazepines (pills) 94(-14) 114(5) 109 115(6) 112(3)

Cannabis (grams) 84(-10) 84(-10) 93 83(-11) 90(-3)

Cocaine (caps) 27(-44) 30(7) 28 37(32) 37(32)

Pharmaceutical opioids (pills) 24(-38) 24(-38) 39 25(-36) 38(-3)

Not methamphetamine dependent (n=3 )


Heroin (caps) 57(-2) 59(2) 58 61(5) 62(7)

Alcohol (drinks) 45(-18) 38(-31) 55 29(-47) 43(-22)

Benzodiazepines (pills) 42(-26) 80(40) 57 62(9) 52(-9)

Cannabis (grams) 38(-12) 39(-9) 43 43(0) 42(-2)

Cocaine (caps) 7(-22) 4(-56) 9 18(100) 18(100)

Pharmaceutical opioids (pills) 14(-13) 15(-6) 16 20(25) 20(25)

reports purchases for the 36 participantswho did not meet the criteria for methamphetamine dependence.

Looking rst at the results for thewhole sample, the most strikingnding is the large increase in units of methamphetamine purchased at lower methamphetamine prices. At baseline,participants purchased 121 points of methamphetamine. This increased byalmost 200 per cent when the price was$20 (to 362 units) and by almost 500per cent (to 710 units) when the pricewas $10. It is also clear that there wassome substitution into heroin at higher methamphetamine prices (by 18% to 134caps at the highest methamphetamine

price). While there was also evidenceof substitution into cocaine andpharmaceutical opioids, the base rate of use of both of these substitutes was quitelow. Participants only bought 37 units of cocaine and 55 units of pharmaceuticalopioids at baseline prices.

Consistent with the results shown inTable 2, there was some evidencethat dependence played a smallrole in determining responsesto price reductions. Dependentmethamphetamine users tended to beslightly less responsive to changes inmethamphetamine price increases thannon-dependent users. For example,dependent users purchased 38 per centless methamphetamine by volume whenthe price rose to $75, compared with a 57per cent reduction among non-dependentusers. Dependent users also tended toshow greater overall increases in heroin

consumption at higher methamphetamineprices than non-dependent users. Table3a also reveals that the strong substitutioninto cocaine use among non-dependentusers comes off a low base rate of use of nine caps at the baseline price of $50.

Table 3b shows the analogous resultswhen heroin prices were varied. Amongthe whole sample, there was again a largeincrease in units of heroin purchasedat lower heroin prices. At baseline,

participants purchased 114 caps of heroin.This increased by approximately 200 per cent when the price was $20 (to 344 caps)and by more than 500 per cent (to 692

caps) when the price was $10. Among allrespondents there was evidence of strongsubstitution into benzodiazepines and

methamphetamine and these increasescame off high base rates of use (166 pillsand 121 caps purchased at the baselineheroin price of $50). The substitution intopharmaceutical opioids shown in Table2, on the other hand, comes off a lowbase rate of use of these drugs (55 pillsat the baseline heroin price). Similarly, thecomplementarity observed among cocainepurchases related to low base rates of use(37 caps at the baseline heroin price).

The disparity between dependent andnon-dependent heroin users observedin Table 2 is also apparent in Table 3b.However, while non-dependent heroin

users were indeed much more responsiveto changes in heroin price than dependentusers (increasing their purchases by

1600% at the lowest price), baselineheroin consumption was very low amongthe non-dependent group. These 30participants only purchased four caps of heroin at the baseline price of $50. Asa result, the largest changes in overallconsumption were observed amongthe dependent group. The large overallsubstitution into benzodiazepines wasmanifest only among heroin-dependentusers who purchased 152 of the 166

pills purchased at the baseline heroinprice. Similarly, the substitution intomethamphetamine was only observedamong heroin-dependent users.


12/16

12


dIscussIon

The results of the current study can besummarised as follows:

Demand for both methamphetamineand heroin was estimated to berelatively elastic (a 10% increasein methamphetamine price wasestimated to result in a 18 - 19%decrease in the quantity of methamphetamine purchased and acorresponding increase in the priceof heroin would lead to a 16% - 27%reduction in the quantity of heroin

purchased);Own-price elasticity of demand for methamphetamine did not varyaccording to methamphetamine

dependence but non-dependentheroin users were found to besigni cantly more responsive to own-

price changes than dependent users.This greater responsiveness amongnon-dependent heroin users waslargely attributed to the low baselinerates of heroin use among this group;

Increases in the price of methamphetamine were estimatedto lead to signi cant substitutioninto heroin (cross-price elasticity[CPE]=0.200 among dependent usersand 0.054 among non-dependent

users), pharmaceutical opioids(CPE=0.333 among dependent users)and cocaine (CPE=0.705 among non-dependent users). Substitution was

of most concern in relation to heroinamong methamphetamine dependentparticipants given the relatively lowbase rates of use of the other twodrugs;

No drugs were found to be signi cantcomplements for methamphetamineuse (i.e. there were no signi cantfalls in other drug use as the price of methamphetamine increased);

Increases in the price of heroinwere estimated to lead to signi cantsubstitution into methamphetamine(CPE=0.255 among dependentusers), benzodiazepines (CPE=0.308among dependent users) andpharmaceutical opioids (CPE=0.395among dependent users and0.764 among non-dependentusers). Substitution was of mostconcern among heroin-dependentparticipants in relation to their use of methamphetamine and, in particular,their use of benzodiazepines;

Cocaine was found to be acomplement for heroin use amongdependent heroin users (CPE=-0.538)but, again, off a relatively low baserate of use among this group.

In short, the results suggest that demandfor both methamphetamine and heroinappears to be in the elastic range. Weestimated that the own-price elasticitiesfor these two drugs were greater thanone, which suggests that any increase inthe price of these drugs would result in adecrease in consumption and a decreasein overall expenditure on these drugs.

However, there was also evidence of substitution into other drugs when theprice of heroin and methamphetaminerose. This substitution must be viewedwith concern. Here, it is important toacknowledge that these ndings arespeci c to the study participants, whoshould not be taken as representative of methamphetamine users as a whole, andwho operate in particular drug markets.Cocaine, for example, has limitedavailability outside of Sydney.

Weighing up the net effect of these pricechanges in light of this substitution isa dif cult task. Because we were not

Table 3b. Drugs bought at each heroin price with percentagedifference from baseline in brackets*

Price of heroin ($ per cap)

10 20 50 75 100

Methamphetamine (points) 102(-16) 118(-2) 121 124(2) 136(12)

Heroin (caps) 692(507) 344(202) 114 66(-42) 54(-53)

Alcohol (drinks) 112(-21) 114(-20) 142 130(-8) 120(-15)

Benzodiazepines (pills) 99(-40) 122(-27) 166 219(32) 217(31)

Cannabis (grams) 132(-3) 101(-26) 136 114(-16) 113(-17)

Cocaine (caps) 67(81) 52(41) 37 40(8) 27(-27)


Heroin dependent (n= 1)

Methamphetamine (points) 40(-30) 52(-9) 57 61(7) 70(23)

Heroin (caps) 624(467) 315(186) 110 63(-43) 53(-52)

Alcohol (drinks) 45(-20) 43(-23) 56 59(5) 55(-2)

Benzodiazepines (pills) 87(-43) 110(-28) 152 207(36) 200(32)

Cannabis (grams) 87(1) 60(-30) 86 67(-22) 68(-21)

Cocaine (caps) 58(100) 44(52) 29 34(17) 20(-31)


Not heroin dependent (n=30)

Methamphetamine (points) 62(-3) 66(3) 64 63(-2) 66(3)

Heroin (caps) 68(1600) 29(625) 4 3(-25) 1(-75)

Alcohol (drinks) 67(-22) 71(-17) 86 71(-17) 65(-24)

Benzodiazepines (pills) 12(-14) 12(-14) 14 12(-14) 17(21)

Cannabis (grams) 45(-10) 41(-18) 50 47(-6) 45(-10)

Cocaine (caps) 9(13) 8(0) 8 6(-25) 7(-13)



13/16

13


interested in testing the income elasticityof demand, the study design involvedgiving participants a xed budget and avariable amount of time in which to spendthat budget. As a result, most participantsspent their entire drug budget on each

trial and overall expenditure varied littlewith drug prices. Based on the elasticityestimates and baseline purchases of thevarious drugs, we suspect that the overalldecrease in heroin and methamphetaminepurchases would outweigh the observedsubstitution into other drugs. Whenmethamphetamine prices were varied, of most concern was substitution into heroinuse among those who met the criteriafor methamphetamine dependence.

However, in response to a $10 increasein the price of methamphetamine from$50 (20 per cent) the estimated increasein heroin purchases from baseline wouldbe in the order of two caps of heroin(from 56 caps) for the entire group,compared with a decrease of 30 pointsof methamphetamine (from 84 points). 11 For heroin dependent respondents, thebiggest concerns were substitution intobenzodiazepines and methamphetamineamong heroin-dependent participants.Again, however, the analogous priceincrease from base rate, would translateinto nine additional benzodiazepine pills(from 152 pills) and three extra points of methamphetamine (from 57 points). Thiscompares with a decrease of 34 caps of heroin (from 110 caps).

The current ndings support the ndingsof similar behavioural-economics studiesand economic studies of secondary datathat demand for an illicit drug is relativelyelastic with respect to its own-price (e.g.Goudie et al. 2007; Grossman et al.,2002). They are also consistent withobservations from the Australian heroinshortage that took hold around Christmas2000. As the real price of heroin doubledin a short space of time, indicatorsof heroin use declined rapidly. Whileethnographic research revealed thatmany primary heroin users supplementedtheir heroin use with methamphetamine

while heroin supply was low (Maher etal., 2007), the supplementation does notappear to have been large enough tomanifest itself in population-level data

prices was also consistent with therange reported among illicit drug usersin the Sydney market (Phillips andBurns 2009). Ajzen and Fishbein (1980)proposed in their often-cited theory of reasoned action framework that, in such

circumstances, peoples behaviouralintentions usually closely re ect their behaviour. Indeed, a good example of the close alignment of intention andbehaviour can be found from observingelectoral opinion polls. If conducted withmethodological rigour, telephone pollsusually very closely predict electoraloutcomes. Nevertheless, that there is aclose association between intentions andbehaviour remains an assumption of the

methodology employed here that we arenot able to test.

While every effort was made to ensurethat the scenarios re ected real-worlddrug market characteristics, the complexcharacteristics of real world drug-marketscannot be perfectly simulated in anexperimental setting. The experimentaldesign provided respondents with a xedbudget, which is often not the case inreality. Income can vary depending on

availability of legitimate employment or through willingness to commit criminalactivity to raise income to purchaseillicit drugs. While the majority of therespondents were reliant on a xedincome income support a notableproportion received most of their incomefrom criminal activity or sex work, whichare more exible income ows. As theprice of a drug increases, one real worldresponse would be to undertake more

crime (although there is, of course, anupper limit to the amount of crime or sexwork one person can do).

A further limitation of the currentndings is that we did not attemptto obtain a representative sample of methamphetamine users, nor did wemake any attempt to control for thecharacteristics of our sample other than their level of dependence. Whilewe cannot, therefore, generalise

the results of the current study to allmethamphetamine users, it was not our intention to do so here. Our primary aimwas to determine how responsive sentinel

over the longer term (Snowball et al.2008).

Like Saffers and Chapoulkas (1999)study, we found complementarity betweenheroin, cocaine and cannabis, at leastamong the heroin dependent group.On the other hand, the current ndingson cross-price responsiveness are notconsistent with the only behaviouraleconomics study to have exploredpoly-drug use involving amphetamines.While Sumnall et al. (2004) establishedthat demand for amphetamine wasrelatively elastic, increases in the priceof amphetamine were found to have noimpact on purchases of cocaine andecstasy although they did result in somesubstitution into alcohol. These discrepantndings are likely to be a function of the populations of drug users sampled.Sumnall et al. (2004) investigated ayounger group of drug users that reporteda preference for ecstasy, whereas thecurrent study focused on a comparativelyolder population of drug users, of whicha signi cant portion regarded heroin or methamphetamine as their drug of choice.Furthermore, in the United Kingdom,

there is some evidence that amphetamineuse is decreasing amongst poly-drugusers (Sumnall et al. 2004), whereas inAustralia, as reviewed in the introduction,methamphetamines remain one of themost commonly used illicit drugs.

Based as they are on hypothetical drugpurchases, the applicability of the resultsdepends on the degree to which statedpreferences would translate into actualbehaviour. However, there are primafacie reasons to have con dence thatpeople would behave in the way theysay they would. We followed a well-respected experimental approach. Thescenario conditions were de ned so thatthey re ected real-world drug marketsas closely as possible. The xed incomewas set at a level that approximatedweekly levels of income support onwhich most people in the sample receivedas their primary source of income and

drug prices were set at the median levelsestimated from sentinel groups of illicitdrug users such as the sample obtainedhere. The range of methamphetamine


14/16

14


groups of frequent methamphetamineusers might be to changes in its price andthat of heroin. Nevertheless, it is possiblethat other groups of methamphetamineusers might respond differently than thegroup studied here.

Perhaps the most critical limitation of the approach used here was that illicitdrug market adjustments are more likelyto manifest as changes in the purityand/or quantity of drug sold for a givenprice (Caulkins 2007). Concerned aboutthe dif culties associated with varyingquality in an experimental setting we, ina sense, proxy such quality changes asprice changes. Again we impressed uponthe respondents the need to assume that

quality, purity and quantity were constantover the life of the experiment. However,some respondents were reluctant to buymethamphetamine when its price was $10on the basis that this price would normallybe associated with poor quality product.Gaining an understanding of the complexinterplay between drug price, drug purityand drug consumption should be at theforefront of future attempts to understandconsumer responsiveness to changes in

drug market characteristics.

conclusIon

In summary, the current study foundthat, for this group of methamphetamineusers, consumption of heroin andmethamphetamine fell substantially inresponse to increases in their price.While there was evidence of substitutioninto other drugs, this substitution provednowhere near large enough to offset thereductions in methamphetamine andheroin use. In determining the net effecton drug use this analysis illustrates theimportance of distinguishing betweenthe behavioural responses of thosedependent on the drug in question andthose who are not dependent, and takingaccount of the rates of drug use prior toany price change. These ndings lendsupport to one of the primary levers bywhich supply side drug law enforcement

policy aims to limit drug-related harm,which is to put upward pressure on drugprices and, in turn, downward pressure ondrug consumption.

acknowledgeMents

This research was funded by NSWHealth with in-kind support from the DrugPolicy Modelling Program (DPMP). Theauthors would like to thank the agency

management and staff for facilitatingrecruitment and interviews, RebeccaMcketin for sharing her wisdom about allthings methamphetamine, Erin Kellyand Natasha Sindicich for assistance inrecruiting agencies, Colleen Faes andother DPMP colleagues for undertakingthe interviews, Tim Slade and BarbaraToson for statistical advice, MichaelLodge for helpful comments and DonWeatherburn for valuable feedback.We also thank Florence Sin for desktoppublishing this report. But most of all weowe a debt of gratitude to the participantsfor their patience and enthusiasm for thegame.

notes

Drug Policy Modelling Program, NationalDrug and Alcohol Research Centre, UNSW

NSW Bureau of Crime Statistics andResearch

In this paper methamphetamine includesamphetamine and methamphetamine butexcludes ecstasy.

However, purchases of heroin wereindependent of changes in the price of valium, a nding later con rmed by Petry(2001) in a study examining drug useamong alcoholics.

Behavioural economic analyses andeconomic analyses have both found that

purchases of a drug are contingent onavailable income (e.g. Petry and Bickel1998; Bretteville-Jensen 2006) and incomeelasticity of demand was not the focus of this study.

One tablet represented 100mg of morphineand 80mg of oxycodone.

Count data is typically analysed with aPoisson regression model where theexpected hypothetical units purchased ismodelled as a function of the explanatoryvariables. However, the Poisson modelmakes the strong assumption that theexpected count (mean of units purchased)is identical to its variance. Often, as wasthe case here, the variance exceeds the

1.

2.

3.

4.

5.

6.

7.

mean and the negative binomial modelis more appropriate since, by allowing for heteroskedasticity, the variance can differ from the mean (Greene 2008; Kennedy2008). However, the standard negativebinomial model was also found to providea poor t to the data here due the largenumber of zero counts. Zero-in atedregression models (poisson and negativebinomial) were adopted to account for thehigh number of zero responses. The zero-in ated models differentiate between twogroups of respondents - those who wouldnot purchase the drug at any price andthose who would be willing to purchase thedrug at some price (Greene 2008, pp.922-924).

The Friedman two-way analysis of variance

by ranks test was applied to determinewhether the median of the six groupingsof hypothetical purchases was equal for allprices (all participants, methamphetaminedependent and not methamphetaminedependent vis--vis methamphetamineprice and all participants, heroin dependentand not heroin dependent vis--vis heroinprice) (Siegel and Castellan, 1988). Thetests indicated that in each of the groupingsthere was a signi cant difference betweenthe median purchases for at least twoof the prices at the 1 per cent level of signi cance. The Page test, a distributionfree rank-test for ordered alternativessuitable for repeated measure data, wasthen used to determine whether price wasinversely related to hypothetical purchases(Siegel and Castellan, 1988). For eachgrouping the Page test suggests thatthere is a statistically signi cant inverserelationship between purchases and therelevant price; at the one per cent levelof signi cance for the three groupingspertaining to methamphetamine price, at

the one percent level for all participants andheroin dependent participants in relationto heroin price, but only at the ve per centlevel for non-heroin dependent participantsin relation to heroin price.

The Friedman tests revealed that purchasesof cannabis and pharmaceutical opioidswere signi cantly related to the heroin priceat the one per cent level in both groups.In the dependent group, so too werecocaine purchases. Methamphetamine andbenzodiazepine purchases also showeda signi cant relationship at the ve per cent level. In the non-dependent groupalcohol was signi cantly related to price atthe ve per-cent level. Visual observation

8.

9.


15/16

1


suggests that there is substitution intopharmaceutical opioids in both groups withthe dependent group also substituting intomethamphetamine and benzodiazepines.Members of this group also appear topurchase cocaine with heroin. Page testsrevealed that when all participants wereconsidered there was a signi cant inverserelationship between the heroin priceand cocaine at the 10 per cent level andsigni cant positive relationships betweenheroin price and both methamphetamineand pharmaceutical opioids at the ve per cent and one per cent levels respectively.However, when grouped by level of dependence only the relationship betweenheroin price and methamphetamine in thedependent group maintained signi cance.

We thank our anonymous reviewer for bringing this to our attention.

These estimates are based on theelasticities shown in Table 2.

references

Ajzen, I & Fishbein, M (Eds) 1980,Understanding attitudes and predicting social behaviour , Prentice-Hall, NewJersey.

Australian Institute of Health and Welfare2008, 2007 National drug strategyhousehold survey: detailed ndings, Drug Statistics Series no. 22, cat. no. PHE 107,AIHW, Canberra.

Baum, CF 2006, An introduction tomodern econometrics using Stata , StataPress, College Station, Texas.

Becker, GS, Grossman, M & Murphy, KM1991, Rational addiction and the effect

of price on consumption, Economics of Drugs , vol. 81, no. 2, pp. 237-241.

Bretteville-Jensen, AL 2006, Drugdemand - initiation, continuation andquitting, De Economist , vol. 154, pp.491-516.

Cameron, A, Trivedi, P, Hammond, P &Holly, A 1998, Regression analysis of count data , Cambridge University Press,Cambridge.

Caulkins, JP 1995, Estimating elasticitiesof demand for cocaine and heroin withdata from the Drug Use Forecasting System , Final Report, United States

10.

11.

Department of Justice, National Instituteof Justice, NCJ 173823, Washington DC.

Caulkins, JP 2001, Drug prices andemergency department mentions for cocaine and heroin, American Journal of

Public Health , vol. 91, no. 9, 1446-1448.Caulkins, JP 2007, Price and purityanalysis for illicit drug: data andconceptual issues, Drug and Alcohol Dependence , vol. 90, pp. S61-S68

Caulkins, JP, Pacula, RL, Paddock, S& Chiesa, J 2002, School-based drug

prevention: what kind of drug use does it prevent? Drug Police Research Center,Rand Corporation, Santa Monica,California.

Cho, AK & Melega, WP 2002, Patternsof methamphetamine abuse and their consequences, Journal of AddictiveDisorders , vol. 21, pp. 21-34.

Cole, JC, Goudie, AJ, Field, M,Loverseed, A, Charton, S 2008, Theeffects of perceived quality on thebehavioural economics of alcohol,methamphetamine, cannabis, cocaine,and ecstasy purchases, Drug and Alcohol

Dependence , vol. 94, pp. 183-190.

Commonwealth Department of Health andAgeing 2002, Return on investment inneedle and syringe programs in Australia:report , Commonwealth Department of Health and Ageing, Canberra.

Darke, S & Hall, W 1995, Levels andcorrelates of polydrug use among heroinusers and regular methamphetamineusers, Drug and Alcohol Dependence ,

vol. 39, pp. 231-235.Darke, S, Kaye, S, McKetin, R & Du ou,J 2008, Major physical and psychologicalharms of methamphetamine use, Drug and Alcohol Review , vol. 27, pp. 253-262.

Dave, D 2004, Illicit drug use amongarrestees and drug prices, NBER Working Paper 10648 , National Bureauof Economic Research, Cambridge,Massachusetts.

Dave, D 2006, The effects of cocaine andheroin price on drug-related emergencydepartment visits, Journal of HealthEconomics , vol. 25, pp. 311-333.

Degenhardt, L, Roxburgh, A, Black,E, Bruno, R, Campbell, G, Kinner, S &Fetherston, J 2008, The epidemiologyof methamphetamine use and harm inAustralia, Drug and Alcohol Review , vol.27, pp. 243-252.

Degenhardt, L & Topp, L 2003, Crystalmeth use among polydrug users inSydneys dance party subculture:characteristics, use patterns andassociated harms, International Journal of Drug Policy , vol. 14, pp. 17-24.

Gonzlez-Siz, F, Domingo-Salvany, A,Barrio, G, Snchez-Niub, A, Brugal,MT, de la Fuente, L & Alonso, J 2009,Severity of Dependence Scale as a

diagnostic tool for heroin and cocainedependence, European AddictionResearch , vol. 15, pp. 87-93.

Gossop, M, Darke, S, Grif ths, P, Hando,J, Powis, B, Hall, W & Strang, J 1995,The Severity of Dependence Scale(SDS): psychometric properties of theSDS in English and Australian samples of heroin, cocaine and methamphetamineusers, Addiction , vol. 90, no. 5, pp. 607-614.

Goudie, AJ, Sumnall, HR, Field, M,Clayton, H & Cole, JC 2007, The effectsof price and perceived quality on thebehavioural economics of alcohol,methamphetamine, cannabis, cocaineand ecstasy purchases, Drug and Alcohol Dependence , vol. 89, pp. 107-115.

Greene, W. 2008, Econometric analysis ,6th edition, Prentice Hall, Upper SaddleRiver, NJ.

Grossman, M & Chaloupka, FJ 1998, Thedemand for cocaine by young adults: arational addiction approach, Journal of Health Economics , vol. 17, pp. 427-474.

Grossman, M, Chaloupka, F & Shim, K2002, Illegal drug use and public policy,Health Affairs , vol. 21, pp. 134-145.

Hurley, SF, Jolley, DJ & Kaldor, JM1997, Effectiveness of needle-exchangeprogrammes for prevention of HIVinfection, The Lancet , vol. 349, pp. 1797-1800.

Jofre-Bonet, M & Petry, NM 2008,Trading apples for oranges? Results of


16/16


NSW Bureau of Crime Statistics and Research - Level 8, St James Centre, 111 Elizabeth Street, Sydney 2000 [email protected] www.bocsar.nsw.gov.au Ph: (02) 9231 9190 Fax: (02) 9231 9187

ISSN 1030 - 1046 ISBN 978-1-921626-24-1 State of New South Wales through the Department of Justice and Attorney General NSW 2009 You may copy distribute display download and otherwise freely

an experiment on the effects of heroinand cocaine price changes on addictspolydrug use, Journal of Economic Behavior & Organization , vol. 66,pp. 281-311.

Kennedy, P 2008, A guide toeconometrics , 6th edition, MIT Press,Cambridge, Massachusetts.

McKetin, R, McLaren, J & Kelly, E 2005b,The Sydney methamphetamine market:patterns of supply, use, personal harmsand social consequences, National Drug Law Enforcement Research Fund Monograph no. 13, National Drug LawEnforcement Research Fund, Canberra.

McKetin, R, McLaren, J, Kelly, E, Hall,

W & Hickman, M 2005a, Estimatingthe number of regular and dependentmethamphetamine users in Australia,NDARC Technical Report no. 230,National Drug and Alcohol ResearchCentre, Sydney.

McKetin, R, McLaren, J, Lubman,DI & Hides, L 2006, The prevalenceof psychotic symptoms amongmethamphetamine users, Addiction , vol.101, pp. 1473-1478.

McKetin, R, Ross, J, Kelly, E, Baker, E,Lee, N, Lubman, D & Mattick, R 2008,A comparison of harms associated withinjecting vs smoking methamphetamineamong methamphetamine treatmententrants in Australia, Drug and Alcohol Review , vol. 27, pp. 277-285.

Maher, L, Li, J, Jalaludin, B, Wand, H,Jayasuriya, R, Dixon, D & Laldor, JM2007, Impact of a reduction in heroinavailability on patterns of drug use, riskbehaviour and incidence of hepatitis Cvirus infection in injecting drug users inNew South Wales, Australia, Drug and

Alcohol Dependence , vol. 89, pp. 244-250.

Manski, CF, Pepper, JV & Petrie, CV2001, Informing Americas policy onillegal drugs: what we dont know keepshurting us , National Academy Press,Washington DC.

Moffatt, S, Weatherburn, D & Donnelly,N 2005, What caused the recent dropin property crime?, Crime and JusticeBulletin no. 85, NSW Bureau of CrimeStatistics and Research, Sydney.

Pacula, LR 1998, Does increasing thebeer tax reduce marijuana consumption?Journal of Health Economics, vol. 17,no. 5, pp. 557-585.

Petry, NM 2000, Effects of increasingincome on polydrug use: a comparisonof heroin, cocaine and alcohol abusers,

Addiction , vol. 95, pp. 705-717.

Petry, NM 2001, A behavioural economicanalysis of polydrug abuse in alcoholics:asymmetrical substitution of alcohol andcocaine, Drug and Alcohol Dependence ,vol. 62, pp. 31-39.

Petry, NM & Bickel, WK 1998, Polydrugabuse in heroin addicts: a behaviouraleconomic analysis, Addiction , vol. 93,no. 3, pp. 321-335.

Phillips, B & Burns, L 2009, New SouthWales drug trends 2008: ndings fromthe Illicit Drug Reporting System (IDRS),

Australian Drug Trends Series , no. 20,

National Drug and Alcohol ResearchCentre, Sydney.

Saffer, H & Chaloupka, FJ 1999, Thedemand for illicit drugs, Economic Enquiry , vol. 37, no. 3, pp. 401-411.

Shearer, J, Darke, S, Rodgers, C,Slade, T, van Beek, I, Lewis, J, Brady,D, McKetin, R, Mattick, R & Wodak,A 2009, A double-blind, placebo-controlled trial of moda nil (200 ms/day)

for methamphetamine dependence, Addiction , vol. 104, no. 2, pp. 224-233.

Siegel, S and Castellan, N (1988)Nonparametric statistics for thebehavioural sciences , McGraw-Hill., NewYork.

Silverman, LP & Spruill, NL 1977, Urbancrime and the price of heroin, Journal of Urban Economics , vol. 4, pp. 80-103.

Smithson, M, McFadden, M, Mwesigye,S & Casey, T 2004, The impact of illicitdrug supply reduction on health andsocial outcomes: the heroin shortage inthe Australian Capital Territory, Addiction ,vol. 98, pp. 340-348.

Snowball, L, Moffatt, S, Weatherburn, D &Burgess, M 2008, Did the heroin shortageincrease methamphetamine use?,Crime and Justice Bulletin no. 114, NSWBureau of Crime Statistics and Research,Sydney.

Sumnall, HR, Tyler, E, Wagstaff, GF &Cole, JC 2004, A behavioural economicanalysis of alcohol, amphetamine,cocaine and ecstasy purchases bypolysubstance misusers, Drug and

Alcohol Dependence , vol. 76, pp. 93-99.

Topp, L & Mattick, RP 1997, Choosinga cut-off on the Severity of DependenceScale (SDS) for methamphetamineusers, Addiction , vol. 92, no. 7, pp. 839-845.

Vincent, N, Shoobridge, J, Ask, A, Allsop,S & Ali, R 1998, Physical and mentalhealth problems in methamphetamineusers from metropolitan Adelaide,

Australia, Drug and Alcohol Review , vol.17, pp. 187-195.

Weatherburn, D, Jones, C, Freeman, K &Makkai, T 2003, Supply control and harmreduction: lessons from the Australianheroin drought, Addiction , vol. 98, pp.83-91.

White, B, Breen, C & Degenhardt, L2003, NSW party drug trends 2003:ndings from the Party Drugs Initiative(PDI), NDARC Technical Report no. 182,National Drug and Alcohol ResearchCentre, Sydney.

Documents

Meth Price Change Use Patterns