{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

Decriminalization

Decriminalization - '9(om-qmmnozro—L Jars-r...

Info iconThis preview shows pages 1–14. Sign up to view the full content.

View Full Document Right Arrow Icon
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Background image of page 2
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Background image of page 4
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Background image of page 6
Background image of page 7

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Background image of page 8
Background image of page 9

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Background image of page 10
Background image of page 11

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Background image of page 12
Background image of page 13

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Background image of page 14
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: '9? (om-qmmnozro—L . Jars-r:- nan‘nmmmmmmmmmmm 5|mmimm—Lorom-srmmno:mdomg38$§$%gga$33ai§afiija mmmmmmmaa U1 54 _ H591 seal 59 i it: the HM tional Illicit Drugs Strategy since 1993‘; Decriminalization and Marijuana Smoking Prevalence: Evidence From Australia Kannika DAMRONGPLASIT Department of Health Services, University of California, Los An ECWOWIICS; Nana/mg Technological University Chang Hser Department of Economics, University of Southern California, Department of Economics and Finance, City University of Hong Kong, and WISE, Xiamen University ([email protected]) Xueyan ZHAo Department of Econometrics and Business Statistics, and Centre for Health Economics, Monash University“ Australia- This article uses the 2001 National Drug Strategy Household Survey to assess the impact of marijuana decriminalization policy on marijuana smoking prevalence in Australia. Both parametric and nonparamet- ric methods are used. The parametric approach includes endogenous probit switching, two-part, sample selection, and standard dummy variable models, while the nonparametric approach uses propensity score stratification matching. Specification analyses are also conducted. A nonparametric kemel—bascd test is constructed to select between parametric and nonparametric models, and the likelihood ratio testis used to choose among parametric models. Our analyses favor the endogenous switching model where decrim— inalization increases the probability of smoking by 16.2%. KEY WORDS: Average treatment effect; Bootstrapping; Endogenous prohit switching; Parametric and nonparametric specification analysis; Prepensity score matching. 1. INTRODUCTION Illicit drug usage is widespread around the world, posing significant social and economic costs to the health care, jus— tice, and social welfare systems in both1 eveloped and devel- oping countries. According to the July , 2001 issue of the Economist, the global retail sale of illegal drugs is estimatedes? US$150 billion a year, which is in the same league as world— wide sale of tobacco and alcohol and about half the size of the pharmaceutical industry; Significant amounts of public funds have been spent by governments worldwide to deal with the consequences f bstance buse'andeucational programs; for examl, the tts’ drugs policyoss 0-“ ater US$35—$40 billion a year, while theLAustralian govern— ment has committed more than AUS$1 billn twd its N— r 7 r——i W V W , .WWiWu—m Amoiit d, marijuana is by far the most widely used. It is commonly considered a “softer” drug compared with “harder” drugs, such as cocaine, heroin, or amphetamines. The prevalence of hydroponic cultivation in recent years has signif- icantly improved the productivity of covert production. While there is more support for using marijuana for medical purposes in treating patients with nausea, glaucoma, spasm, and pain, much conthersy has surrounded the detrimental health ef~ fects of recreational use of marijuana. Some suggest that mari- j name use is linked to lung cancer, immune system deterioration, harmful effects on blood circulation, and short-term memory loss. For heavy users, there is also the problem of drug depen- dency and the related withdrawal symptoms, such as anxiety and loss of appetite. At the center of the controversy is whether legal sanctioning is the best approach to reduce the use and the associated harm ‘ U n fired Era-les’ drags gritty casts (ti/film? gimme/y 9’5 f semen/3M Major zine). whirl e rm as-tmir‘m, .«avé’i’flmé‘flf Hirer} a?er en‘penda‘fwe its $117174? £2 affirm for ill/LE year we zcbZ/fi (Moan: inns-)5 of the drug. The ongoing debate on marijuana decriminalization concentrates on the potential benefits and costs ofpol— icy. A major supporting argument for decriminalization is that a criminal charge is too severe a penalty relative to the crime itself. A criminal record can have many negative consequences on the subsequent life of an otherwise law-abiding person; for example, an offender may lose out in future employment oppor— tunities or face problems in international travel. Furthermore, decrirninalization would allow the government to separate the market of marijuana from the market of other, harder drugs, thereby permitting the authorities to redirect their resources used in law enforcement and criminal justice system from the “softer” cannabis to “harder” drugs like cocaine, heroin, and amphetamines. For instance, amid-03' report by, Jeffrey Mitch-,- “The Budgetary Implications of Marijuana Prohibition,” noted that legalizing marijuana could reduce the cost of enforcement in the United States by US$717 billion per year. Moreover, sup— porters also argue that when marijuana is illegal, young mari- juana users are unnecessarily exposed to harder drug dealers, making it easier for them to move on to consume harder drugs. For those who argue against decriminalization, their first claim is that decriminalization inevitably lowers both the legal and social costs associated with the use of marijuana, thus send— ing a signal that it is acceptable to smoke marijuana, which may encourage higher consumption of the drug. Another ar— gument cited against decriminalization is the gateway theory; that is, there is a growing concern that exposure to marijuana . © 2009 American Statistical Association deumal of Business 8: Economic Statistics Accepted for publication DOI: 10.1198/jbes.2009.06129 35-$MJ é’J-HIM Ct. gear awarding geles andJRand Corporation 1, and Divisfim “C 99 trio 7 V so 61 62 as 64 as as 67 as as 70 71 72 73 74 75 75 7? re 79 so at 82 as 84 as as 87 as as so 9: s2 93 94 95 96 9'7 93 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 115 117 118 to fire same issue n" has ricer: {Miran-led QDOJ‘JQU'I-p-LDN—t 0‘1 (:1 U! U1 U1 Ln 01 01 U! 0‘1 4: 4’:- 4: «L A J:- 45 41 4 ~13- (J 03 DO (a m 0) 03 \5 0 l“ -‘ -* —‘ "‘ -‘ —" I C») CD 0 \J \J \3 U M —‘ --" —‘ M- 2 by youths may lead to their subsequent consumption of other, harder drugs. Given the foregoing debates, empirical evidence of the impact of marijuana decriminalization on marijuana us- age is crucial. In particular, if decriminalization has little or no impact on smoking prevalence, then this is a strong argument in favor of the policy. On the other hand, if there is evidence that decriminalization significantly stimulates more marijuana smoking, then a liberal approach toward marijuana may not be as beneficial as is advocated by its supporters. Empirical results using various data sources from the United States are mixed. Saffer and Chaloupka (1995, 1998), and Pac- ula, Chriqui, and King (2003) found the impact of decriminal— ization on marijuana smoking prevalence to be positive and significant. In contrast, DiNardo and Lemieux (2001), Pacula (1998), and Thies and Register (1993) reported insignificant ef— fects of marijuana policy reform on individual smoking deci- sions. There have been three empirical studies of the Australian experience. Cameron and Williams (2001) and Zhao and Har- ris (2004) both found ,a positive and significant marginal effect of decriminalization on prevalence of about 2%, while Williams (2004) found it to be significant only for the subsample of males age 25 years old and older. Typically, binary probit models have been used in these studies, with the decriminalization dummy variable treated as an exogenous explanatory variable. In this work, we used the 2001 Australian National Drug Strategy Household Surveys (NDSHS) to study the‘impact of marijuana decriminalization on marijuana usage. Australia con— sists of six states and two territories. As of 2001, South Aus- tralia, Australia Capital Territory, and Northern Territory had already decriminalized the possession and cultivation of small quantities of marijuana for personal consumption. Under this regulation, while supplying and cultivating commercial quan- _ tities of marijuana still attract severe criminal charges, an “on the spot” fine has replaced criminal charges for minor users. An individual caught using or growing marijuana must pay a fine (usually AUS$150—$200) within a specified period, usually within 60 days, to be eligible for the reduced penalty involving no criminal record or imprisonment. If the person fails to pay the fine, however, a criminal proceeding may follow, possibly leading to a jail sentence. The Australian policy of decriminal— ization iscommonly known. as the Cannabis Expiation Notice system (CEN). Finally, for those states that have not decrirn- inalized marijuana, criminal offenses for possessing, consum- ing, and cultivating the drug are retained. The present study aimed to assess the impact of decriminal— ization on marijuana smoking prevalence. There are three major differences between this study and earlier studies. First, existing studies usually treat decriminalization as an exogenous dummy variable when performing regression analysis. This is also the case here, because our data are based on the 2001 survey. As of 2001, three states in Australia had already decriminalized mar- ijuana use: South Australia in 1987, Australia Capital Territory in 1992, and Northern Territory in 1996. Thus, in analyzing this set of data, we take a state’s decriminalization decision as pre- determined and focus on another source of joint dependence: marijuana smoking behavior and residential choice. At the in- dividual level, the decision of living in a particular state may not be random, and the decisions of where to live and whether to smoke may be related. Individuals may not be randomly se— lected to different states, and there may be selection bias arising Journal of Business 8: Economic Statistics, '2'??? 2009 from those living in the decriminalized states versus those in the nondecriminalized states. In this article we attempt to address the potential endogeneity of marijuana smoking and the indi~ vidual’s decision to reside in a decriminalized versus a nonde— criminalized state by allowing an individual’s marijuana smok- ing behavior equation to be correlated with his or her residential choice. Second, we provide a more flexible marijuana smok— ing behavior equation by allowing individuals to respond dif~ ferently when the legal and institutional arrangement changes. Third, both parametric and nonparametric analyses are con- ducted, and their reliabilities are examined. Essentially, the advantages of the parametric approach are the disadvantages of the nonparametric approach, and the ad— vantages of the nonparametric approach are the disadvantages of the parametric approach. The advantages of the parametric approach are that it can simultaneously take into account selec— tion on observables and unobservables (provided that the model is correctly specified) and allows (efficient) estimation of the ef— fects of individual factors on the outcomes. The disadvantages of the parametric approach are that both the conditional mean functions of observable factors and the probability distributions of the effects of unobservable factors must be specified. The advantages of the nonparametric approach are that neither the conditional mean functions of obsarvable factors nor the proba— bility distributions of the effects of unobservable factors need to be specified. The disadvantages (of propensity score matching) are that it only takes into account selection on observables and only estimates the treatment effects. We discuss the pros and cons through our specification analyses. The article is organized as follows. Section 2 presents an en— dogenous probit switching model as the maintained hypothe— sis and treat the traditional dummy variable approach, sample selection model, and two-part model as its nested alternatives, called the binary probit, bivaiiate probit, and two-part models, respectively. Section 3 provides a description of our data. Sec~ tion 4 reports our estimation results and compares them with re— sults in the literature. Section 5 presents alternative measures of treatment effect from the propensity score stratification match— ing method. Section 6 presents specification analyses, and Sec- tion 7 concludes. 2. THE MODEL We assume that the utility for an individual’s residential choice (d m 1 if residing in decriminalized state and 0 if not) and marijuana consumption (M) is separable from the utility of consuming other goods. Similar to Cameiro, Hansen, and Heckman (2003), Keane and Wolpin (1997), and others, we assume that an individual‘s utility function for marijuana con— sumption and residential choice is statewdependent on the initial endowment and institutional arrangement of the two residential regimes. In particular, we let an individual’s utility function be U(M, diet, 52*) = dU1(M|a) + (1 —d)U0(M]a) + h(d!a*), where U1(v) and U0(-) are the utility functions of consuming mari- juana for an individual residing in a decriminalized state and a nondecriminalized state, respectively; h(-) denotes the utility of living in decriminalized state or nondecriminalized state; and a and a* are sociodemographic, institutional, and idiosyncratic components that affect M and (1, respectively. a and a"‘ may 60 6 1 62 63 64 65 66 57 68 69 70 71 72 73 7'4 .75 76 77 78 79 80 5'1 82 83 a4 35 36 87 88 as so 9‘! 92 93 94 95 96 Q7 98 99 ion ' 101 102 103 104 105 106 107 108 109 110 111 1i2 113 114 115 116 117 118 ates M . O mm-qmmewM-A ...L_L_A._.._n_n._s_n...u.4 (coo-atonmhwm—‘o awmwmmm WOJDDI'O ommwmma$m40w§flgt§§$§£ #1 --|-. mmmmmmmmmmh Damrongplasit, Hsiao, and Zhao: Decriminalization and Marijuana Smoking Prevalence contain overlapping elements. U1(-) and U0(-) are assumed to be different, because the same amount of marijuana consump- tion may lead to different levels of utility in decrirninalized and nondecriminalized states because of differing institutional se- tups; for example, the risk of smoking marijuana could be dif- ferent. Maximizing utility subject to budget constraint 1 yields M1(p,1la) 2 O for an individual residing in a decriminalized state and Mo(p,1la) 3 O for an individual residing in a nonde— criminalized state, where p denotes price of marijuana. Substh tuting M1(-) and Mg (-) into the utility function yields the condi- tional indirect utility V1(p, Ila, n*) for residing in a decriminal— ized state and V0(p, Ila, a*) for residing in a nondecriminalized state. Then d: 1 if d* = V1(p,1ln,a*) —— V009, Ila, 51*) > 0 and d = 0 if (1* 5 0. Approximating M1(-) and M0(-) by if :Otr +151x+€1 (2'1) and 323‘: dual-7301+ 80, (2-2) such that I M106) :1»? ifyi‘ > 0 and (2.3) M105): 0 Hit 5 0, and M000 = y3 MS > 0 and (2.4) Marx) = 0 ifyt s 0, where x denotes the observable factors of p, I, and a that affect the demand for marijuana and a) and so denote the effects of unobservablej Equations (2.l)—(2.4) imply Prob(M; (x) 2 0) = Probe? 6 01x) = fjéflwmneu 0'81 and_Prob(Mo(x) = 0) = Probozg g le) = fj‘mw 0X) f (80) dag. We approximate d* by 00 a reduced—form specification, (2.5) such that d =_ 1 if d* > 0 and 0 otherwise, where z and 1) de— note the additional observable factors in a* that are not in a and d*=1/1x+rzz+v, the effects of unobservable factors in both a and (1* that affect ' residential choiCe.‘ ' ' ' " ' ' " ” ‘ Our data are in the form (y, d), where y = 1 indicates that an individual is a marijuana smoker and 0 otherwise, and d m 1 if an individual resides in a decriminaiized state and 0 otherwise. From (2.1)—(2.4), it follows that yr} ifdyT+(i—d)y3>0 and . (2.6) y = 0 otherwise. Equations (2.l)—(2.6) lead to an endogenous switching model. The model is in a limited information framework in which we have a structural form specification for the demand for mari- juana for individuals residing in a decriminalized state and a nondecriminalized' state and a reduced-form specification for the residentiai choice. The identification of the structural mar- ijuana use equation is achieved through the excluded vari« ables, 2:, that are important in predicting the residential choice (e.g., Hsiao 1983). Many of the conventional models have become special cases of this model. For instance, we could have the following: 7 it in 3 ‘6 ii to an , ,0“, = p01,. yfi 0, known as a restricted Switching model, where my and pm, denote the correlations between 81 and 1; and between 8.3 and 1), respectively. {by When ph, = pm, : O, the residential choice equation’s error term is uncorrelated with the smoking equationfiln this case, model (2.1)7(2.6) is a generalization to the fre- quently used two-part model (Duan et at. 1983, 1984) or the exogenous regime—switching model of Quandt (1972). (Q; When til 2 ,80 = ,6, and p1,, = pov # O, the model is anal- ogous to the sample selection model (Amemiya 1985) in which Mzct+fix+6d+a ifur+l8x+dd+s>0 (a); and (2.7) M = 0 otherwise, where or = Oe'o and 5 = or; m org. ti); When pm : p0,, = 0 and a1 = a}, modei/é.1)—(2.6) re— duces to dummy variable approach to ennui-nation the effect -. of decriminalization with M given by (2.7) and LUICOITB-s lated with (2.5). Wild 1) If 011,;31,oeo, fig, and density functions of 121 13-114 severe known, then Prob(y = llx) is known for parametric analy- sis. We assume that (31,30, 1)) are jointly normally distrib- uted with mean 0 and covariance We“) 2 Wang) 2 Wm) = 1, Govern. 801) = 1010, covtsn, v1) = 101:): and covtaor, v1) = pou- Therefore, wemay estimate the average treatment effect (ATE) by [@0351 + 5136) f CPOXO + .50x)1f(x) 056 (2-3) and the average treatment effect of the treated (ATET) by [@0361 ifirx) — (Mao + £301)if(xld= lldx (2-9) or their restricted version, where @(a) denotes the integrated standard normal from —00 to a. If the sampie is randomly drawn, then the ATE or ATET may be approximated by 1 “ A . a [we + ext) — cites + 50x11}, (2.10) i=1 where the summation is over the complete sample or over those residing in a decriminalized state. In what follows, we first present parametric analysis, then nonparametric analysis. We show that the results are sensitive to model specifications and discuss which model is likely to yield more accurate measurements of the effects of decriminal- ization. 3. DESCRIPTION OF VARIABLES AND DATA 3.1 Selection of Explanatory Variables Marijuana is an addictive recreational drug, and studies on recreational drugs have arisen from many disciplines, such as psychology, medicine, epidemiology, sociology, and eco- nomics. Maximizing the state~dependent utility function leads to demand for marijuana as a function of price, income, as well as some standard socioeconomic and demographic vari- ables that capture heterogeneity in demand, such as age, sex, evalua'iii‘i} s1 62 c 63 64 * 55 as 67 as so 70 71 72 73 74 75 re 4;": 7? ix- ra - 79 80 81 % 82 83 84 85 86 57 88 89 90 91 92 93 95 96 97 as 99 10b 101 102 103 104 105 106 107 105 109 110 111 112 113 114 115 116 117 11s CDCOKlmtJ'lvth-l mmmmmmmmmmhAh-ni: ##A-hwmm (.0030) ' 4 marital status, educational attainment, work status, and ethnic background for the Australian indigenous population, as com- monly postulated in empirical studies (e.g., Becker and Murphy 1988; Pacula 1998; Williams 2004; Zhao and Harris 2004). The impact of the legal risk of smoking via the decriminalization policy is captured by allowing the responses to the conditional variables to be different for the two types of states using the endogenous switching model described earlier. For residential choice, the literature (Feridhanusetyawan and Kilkenny 1996; Kittiprapas and McCann 1999) suggests that both personal and location characteristics are key determinants of an individual’s residential choice. Because we are taking a . limited-information approach, in addition to those variables den termining marijuana smoking, additional variables, including number of dependent children (which might influence residen- tial choice) and unemployment rate in each individual’s state of residence (a proxy for state-specific effects) are also used to predict residential choice as well as to provide exclusion restric...
View Full Document

{[ snackBarMessage ]}