Introduction
Tasneem Chipty is the Managing Principal of Analysis Group, Inc., a leading economic and business consulting firm in the U.S With expertise in industrial organization, she examines market functions, consumer choices, competitive firm interactions, and the impact of regulation Additionally, her specialization in econometrics involves applying statistical methods, such as regression models, to analyze marketplace behaviors Chipty has taught at prestigious institutions, including The Ohio State University, Brandeis University, and MIT, focusing on industrial organization, regulatory policy, and econometrics She has authored or coauthored numerous academic articles in peer-reviewed journals like the American Economic Review and the Review of Economics and Statistics, utilizing empirical methods to explore consumer choice and pricing strategies.
I have provided consulting services to diverse organizations, including the Government of Australia, the U.S Department of Justice, the Federal Communications Commission, and the Massachusetts Health Policy Commission.
I have provided testimony to the World Trade Organization representing Australia in trade disputes related to the Tobacco Plain Packaging Act (TPP Act) My research has focused on the impact of tobacco control policies, particularly plain packaging, on smoking rates and consumption in Australia.
3 I received my Ph.D in Economics from the Massachusetts Institute of
In 1993, I leveraged my B.A degree in Economics and Mathematics from Wellesley College, earned in 1989, to navigate the evolving landscape of technology My resume, included as Appendix A, outlines my educational background, publications, and experience in providing expert testimony.
I have been engaged by Australia’s Department of Health to evaluate the effects of plain packaging on smoking rates in Australia as an independent expert My analysis will focus on individual-level survey data collected from January 2001 to September 2015, provided by Roy Morgan.
A research entity has gathered nationally representative data on smoking behaviors among Australians aged 14 and older, allowing for an analysis of the early impacts of plain packaging on smoking prevalence in Australia This data encompasses time periods both prior to and following the implementation of plain packaging.
The TPP Act implemented plain packaging for tobacco products in Australia, coinciding with the introduction of larger graphic health warnings Due to the simultaneous nature of these changes, it is challenging to isolate the specific impact of plain packaging from that of the updated health warnings Therefore, the effects discussed regarding the TPP Act include both the plain packaging and the enhanced graphic warnings, collectively referred to as the 2012 packaging changes.
Summary of Opinions
Based on my expertise as an economist and analysis of the Roy Morgan survey data, I conclude that the TPP Act is effectively achieving its goals The data shows a significant reduction in smoking prevalence, with an estimated decline of 0.55 percentage points post-implementation, compared to expected rates without the packaging changes The confidence interval for this reduction ranges from -0.095 to -1.01 percentage points As plain packaging aims to discourage smoking initiation, encourage cessation, and prevent relapse, the positive impacts of these changes are expected to increase over time.
1 Roy Morgan Research, “Smoking Overview: Single Source,” July 23, 2014 (hereinafter “RMSS Smoking Overview”), p 2
The prevalence of smoking is assessed through individual responses to Roy Morgan survey questions, which inquire whether participants currently smoke factory-made cigarettes or have used other tobacco products, such as roll-your-own cigarettes, cigars, or pipes, within the past month.
3 Competition and Consumer (Tobacco) Information Standard 2011, §§ 1.5 and 2.2, and Part 9, Division 4
The interaction effect complicates the separation of impacts, as one way plain packaging may decrease smoking rates is by enhancing the effectiveness of graphic health warnings.
Timing and Objectives of Plain Packaging
7 The TPP Act went into effect nationally between October and December 2012 Manufacturers were required to manufacture only products in plain packaging by October 1,
2012, and retailers were required to sell only products in plain packaging by December 1, 2012 5 Given the manufacturer mandate, many retailers were already stocking plain packs before
December 1, 2012 6 Thus, October and November 2012 were transition months, and the
Australian market was fully converted to plain packs by December 2012
8 As set forth in the TPP Act, its purpose was to improve public health by: (a)
“discouraging people from taking up smoking, or using tobacco products” (“initiation”); (b)
“encouraging people to give up smoking, and to stop using tobacco products” (“cessation”); (c)
Efforts to prevent smoking relapse and reduce exposure to tobacco smoke are crucial in supporting individuals who have quit smoking or stopped using tobacco products These targeted behaviors significantly influence personal decisions regarding smoking, and the long-term benefits of helping someone quit or remain smoke-free can positively impact their health and quality of life for decades.
The TPP Act is designed to gradually reduce smoking prevalence and tobacco consumption over time Its impact on smoking behaviors—such as initiation, cessation, and relapse—will manifest slowly, affecting only a portion of current and future smokers For instance, upon implementation, the policy is expected to decrease youth initiation while simultaneously promoting higher rates of youth cessation.
20 percent, holding everything else constant 8 Given estimated rates of initiation and cessation based on actual data, these effects would only lead to a 0.07 percentage point decline in overall
5 Tobacco Plain Packaging Act 2011, No 148, §§ 2 and 31-39
Research indicates that many smokers in Australia had already transitioned to plain packaging prior to December 2012, as supported by the findings of Melanie A Wakefield, Linda Hayes, Sarah Durkin, and Ron Borland.
“Introduction Effects of the Australian Plain Packaging Policy on Adult Smokers: a Cross-Sectional Study,”
BMJ Open, Vol 3, 2013, pp 1-4; and Michelle Scollo, Kylie Lindorff, Kerri Coomber, Megan Bayly, and
Melanie Wakefield, “Standardised Packaging and New Enlarged Graphic Health Warnings for Tobacco
The Tobacco Plain Packaging Act 2011 and the Competition and Consumer (Tobacco) Information Standard 2011 establish essential legislative requirements for tobacco products in Australia These regulations aim to enhance public health by mandating standardized packaging and clear information on tobacco products, thereby reducing their appeal and consumption The implementation of these laws reflects Australia's commitment to tobacco control and consumer protection, significantly impacting the marketing and sale of tobacco in the country.
7 Tobacco Plain Packaging Act 2011, No 148, § 3
The policy primarily influences youth behavior, resulting in a notable reduction in smoking prevalence One year after its implementation, there was a decrease of 0.18 percentage points in overall smoking rates, with even greater declines observed three years later The impact could be more significant if the policy also prompted adult cessation or led to higher cigarette prices.
Methodology: Before-After Regression Analysis of Smoking Prevalence
10 Smoking prevalence is the proportion of individuals in a population that smoke
In a group of 100 individuals, if 17 are smokers, the smoking prevalence is 17 percent, indicating a 17 percent chance that a randomly selected person is a smoker Implementing policies that discourage smoking can effectively lower this prevalence For instance, if the number of smokers decreases from 17 to 16, the smoking prevalence would drop by one percentage point, changing from 17 percent to 16 percent, thereby reducing the likelihood of selecting a smoker from the group.
11 To measure the effect of the packaging changes on smoking prevalence, I adopt a widely-used approach in policy analysis often referred to as “before-after” regression analysis
This analysis examines how an individual's decision to smoke is influenced by various sociodemographic factors and tobacco control policies, such as plain packaging and graphic health warnings It highlights two key aspects: first, it separates the impact of multiple factors affecting smoking behavior, and second, it evaluates the effects of packaging changes by comparing smoking rates before and after the policy implementation A significant reduction in smoking prevalence linked to these packaging changes would support the idea that such policies are effective Additionally, the results can help estimate the smoking rates that would have existed without these packaging modifications.
Regression analysis is a well-established method in academic literature and is extensively utilized by policymakers globally to assess the effects of various policies For a comprehensive understanding of the multiple regression model and statistical inference, please refer to Appendix C.
Roy Morgan Data
My analysis is based on data from the Roy Morgan Single Source Survey (RMSS) spanning January 2001 to September 2015 This nationally representative survey collects monthly responses from approximately 4,500 participants aged 14 and older, focusing on various smoking-related questions, including the use of different tobacco products such as factory-made cigarettes, roll-your-own cigarettes, pipes, and cigars Additionally, the RMSS data includes a range of demographic and socioeconomic details, such as age, gender, marital status, immigration status, educational attainment, employment status, income level, and state or territory of residence.
When choosing the time frame for analysis, I adhere to the principle that more data is preferable unless there are valid reasons to exclude it The sample design must meet two essential criteria: the pre-policy period should effectively predict smoking prevalence in the absence of the intervention, and the post-policy period must accurately capture the intervention's impact, if any exists.
Starting from January 2001, the Australian market underwent various tobacco control policies over the first decade of the sample period, highlighting the significance of this timeframe for analysis.
9 See, for example, Daniel L Rubinfeld, “Reference Guide on Multiple Regression,” Reference Manual on
The "Reference Guide on Multiple Regression," Third Edition, published by The National Academies Press in Washington, D.C in 2011, is accessible online at http://www.fjc.gov/public/pdf.nsf/lookup/SciMan3D01.pdf/$file/SciMan3D01.pdf, with the last access date recorded as October 25, 2015.
Roy Morgan employs a meticulous sampling procedure to guarantee a representative sample of the Australian population, ensuring national geographic coverage The sampling process incorporates sample weights that account for gender, age, geography, and household size These weighting targets are derived from the Labor Force Survey, a monthly publication by the Australian Bureau of Statistics that provides population estimates categorized by geography, age, and gender For more details, refer to the RMSS Smoking Overview, pages 2, 3, and 6.
According to Jeffrey M Wooldridge in "Introductory Econometrics: A Modern Approach," the statistical significance of a regression coefficient tends to improve with larger sample sizes, as noted by Professor Rubinfeld in the "Reference Guide on Multiple Regression." Analyzing national tobacco control policies from January 2001 to February 2006, the implementation of graphic health warnings on tobacco packaging in March 2006, and a 25 percent increase in tobacco excise tax in April 2010, allows for a reasonable estimation of smoking prevalence post-December 2012 without the 2012 packaging changes For the analysis of the after-period, utilizing data up to September 2015 is recommended, considering the mechanisms by which plain packaging is expected to influence smoking behavior.
From January 2001 to September 2015, a total of 794,750 respondents participated in the study, averaging approximately 4,500 respondents each month This period includes 143 months of observational data prior to December 2012 and 34 months of data following the complete implementation of packaging changes.
Empirical Model
The empirical model aims to assess the impact of the 2012 packaging changes on smoking prevalence In this model, the dependent variable represents individual smoking status, defined as one for smokers and zero for non-smokers By averaging this variable across all individuals within a specific month, we can estimate the overall smoking prevalence for that population during that time.
Smoking prevalence in Australia is influenced by various factors, such as tobacco control measures, the population's sociodemographic makeup, and cultural attitudes towards smoking Key explanatory variables include the 2012 packaging changes, other tobacco control policies, sociodemographic factors, and a timeline of these influences.
Over the years, Australia has implemented a range of sub-national tobacco control policies aimed at reducing smoking rates and protecting public health Notable legislation includes the ban on smoking in public places, as detailed in the publication "Tobacco in Australia: Facts and Issues," edited by Michelle Scollo and Margaret H Winstanley, which highlights the effectiveness of these measures in promoting a smoke-free environment For further information, visit the Tobacco in Australia website.
Tobacco control policies significantly influence smoking prevalence trends, and analyzing data over an extended period allows for the assessment of secular trends that are affected by factors beyond these policies Conversely, a shorter time frame may skew the estimated trends by incorporating the effects of specific policy changes, such as the 2012 packaging regulations.
14 See for example, Australia Bureau of Statistics, “Tobacco Smoking,” 4338.0 - Profiles of Health in Australia, 2011-13, available at http://www.abs.gov.au/ausstats/abs@.nsf/Lookup/by%20Subject/4338.0~2011-
Economic theory and previous research indicate a relationship between various factors and smoking prevalence, which can help evaluate the validity of estimation results Relying on predictions from models that align with established theories and past findings enhances confidence in their accuracy.
I describe each of these variables, including findings from prior research, in greater detail here.
Smoking Status
An individual's smoking status is assessed through a series of questions regarding their current and recent tobacco use, specifically focusing on whether they smoke factory-made cigarettes.
A respondent is classified as a "smoker" if they answer "yes" to any of the following questions: "In the last month, have you smoked any roll-your-own cigarettes?", "In the last month, have you smoked any cigars?", or "In the last month, have you smoked a pipe?" Conversely, individuals who respond "no" to all inquiries regarding tobacco products are categorized as "non-smokers."
Policy Variables
The regression model accounts for the implementation of various national tobacco control policies in Australia during the studied period, which are expected to correlate with a decrease in smoking prevalence These policies aim to reduce the appeal of smoking by altering packaging attractiveness and increasing tobacco prices Five specific policy variables are analyzed to assess their impact on smoking rates.
In March 2006, graphic health warnings were introduced on cigarette packs, featuring a written message such as "smoking causes lung cancer" alongside one of fourteen mandated color images These warnings were required to cover 30% of the front and 90% of the back of cigarette packaging, and 30% of the front and 50% of the back of pipe and loose tobacco packaging This initiative replaced a smaller, text-based warning system that had been in place since 1995.
16 Tobacco in Australia, § A12.1.1, “History of Health Warnings in Australia,” available at http://www.tobaccoinaustralia.org.au/a12-1-1-history-health-warnings, visited on January 15, 2016
This article discusses three specific increases in tobacco excise taxes: a 25 percent increase implemented on April 30, 2010, a 12.5 percent increase on December 1, 2013, and another 12.5 percent increase on September 1, 2014 Each of these increases is represented by an indicator variable, highlighting their significance in tobacco taxation policy.
In 2012, an indicator variable was established to mark the simultaneous implementation of plain packaging and enhanced graphic health warnings for tobacco products Manufacturers were mandated to transition to plain packaging starting October 1, 2012, while retailers were required to exclusively sell these products in plain packaging from December 1, 2012.
In 2012, many retailers began stocking plain packs ahead of the December 1 mandate, making October and November transitional months that do not fit neatly into the before or after periods Therefore, my preferred method focuses on the impact of the packaging changes using a December 2012 indicator variable, while alternative analyses include October and November 2012 to investigate different policy start dates.
Indicator variables are employed to assess the impact of policies, assigning a value of one for the months when the policy was active and zero otherwise This method is widely recognized in literature as it provides flexibility in estimating policy effects without making assumptions about the relative impact of different policies For instance, using specific excise tax levels instead of a series of indicator variables suggests that the influence of tax increases on smoking behavior is directly proportional to the amount of the increase Consequently, effective tobacco control measures should lead to a decreased likelihood of tobacco use.
The Australian Government Department of Health outlines the historical evolution of tobacco excise arrangements in Australia since 1901, detailing the taxation policies implemented over the years to regulate tobacco use and generate revenue For more information, visit their official page.
18 The model does not control for the 12.5 percent increase in tobacco excise tax that was implemented on
September 1, 2015 because the data sample contains only one month of data after this tax increase
In 2012, graphic health warnings were significantly enlarged alongside the introduction of plain packaging, making it challenging to isolate the effects of plain packaging from those of the enlarged warnings without imposing restrictive assumptions.
20 See, for example, R.C Hill, W.E Griffith, and G.G Judge, Undergraduate Econometrics, Second Edition, Hoboken, New Jersey: John Wiley & Sons, 2000, pp 207-208
Proportionality dictates that a 25 percent tax increase on tobacco should lead to a smoking prevalence reduction that is double the impact of a 12.5 percent tax increase If research indicates a negative and statistically significant effect of a specific tobacco control measure, it would reinforce this perspective on the effectiveness of such policies.
Sociodemographic Characteristics
The regression model incorporates various demographic and socioeconomic characteristics from the RMSS data that are thought to influence smoking behaviors among Australians These included variables serve as controls to enhance the predictive accuracy of the model.
Gender An indicator variable that equals one if the respondent is female, and zero otherwise Prior research has shown that women are less likely to smoke than men 22
Marital status An indicator variable that equals one if the respondent is married, and zero otherwise Prior research has shown that married people smoke less than unmarried people 23
Foreign-born status serves as an indicator variable, assigning a value of one to respondents who are foreign-born and zero to those who are not Research indicates that individuals born outside of Australia exhibit distinct smoking behaviors compared to their Australian-born counterparts, likely due to varying attitudes towards smoking.
Age plays a significant role in smoking behavior, with studies showing that older individuals tend to smoke more than their younger counterparts However, this trend can reverse at advanced ages, indicating a complex relationship between age and smoking habits.
According to the "Profiles of Health in Australia" report, the prevalence of smoking among adults remains a significant public health concern For detailed statistics and insights on this issue, refer to the Tobacco in Australia resource, specifically section 1.3, which discusses adult smoking rates This information is crucial for understanding the impact of tobacco use in Australia.
23 See, for example, Liane McDermott, Annette Dobson, and Neville Owen, “Determinants of Continuity and Change over 10 Years in Young Women’s Smoking,” Addiction, Vol 104, 2009, pp 478-487
A comprehensive analysis of smoking trends in Australia highlights the prevalence of tobacco use among different birth countries, illustrating significant variations in smoking habits For detailed insights, refer to the Tobacco in Australia resource, specifically section 1.8, which discusses these trends extensively.
The prevalence of smoking among young adults and middle-aged to older adults in Australia is documented in various sources, including the Tobacco in Australia reports, which provide detailed statistics and insights into smoking trends For more information, refer to the sections on young adults and older adults available at Tobacco in Australia’s official website, alongside findings from the 1998 National Drug Strategy Household Survey by Pramod Adhikari and Amber Summerill.
2000, Chapter 2, available at http://www.aihw.gov.au/publication-detail/?idd42467215, visited on January
Education A set of indicator variables reflecting the education level of the respondent
Prior research has shown that more educated people smoke less 26
The work status of individuals is categorized into various indicators, including full-time employment, part-time work, unemployment, home duties, or non-participation in the workforce Previous studies have shown a correlation between employment status and smoking habits, revealing that employed individuals tend to smoke less compared to those who are unemployed.
Income A set of indicator variables reflecting the income bracket of the respondent
Prior research has shown that economically disadvantaged smoke more, meaning that higher income people smoke less 28
The article discusses the use of indicator variables representing the state or territory of respondents during the survey These variables are essential for accounting for varying norms and influences, such as regional tobacco control policies, which may contribute to differing smoking prevalence rates across various regions.
Time Trend
Descriptive Statistics
25 Figure 1 below presents a plot of the monthly overall smoking prevalence rate observed in the RMSS data, with two separate trend lines for the before and after periods As seen in this chart, there has been an overall decline in smoking prevalence in Australia for the past fifteen years There is also some indication that this decline in prevalence has accelerated in recent years
Note: Data are weighted using the population weights in the RMSS data
Source: RMSS data (January 2001 – September 2015)
26 Table 1 presents the same information through a different lens Specifically, Column (1) shows average smoking prevalence in the full sample, from January 2001 to
In September 2015, an analysis of smoking prevalence revealed a significant decline following the 2012 packaging changes Specifically, data from a symmetric 34-month period before and after the changes, excluding October and November 2012, showed a decrease of 2.2 percentage points in smoking rates The average prevalence dropped from 19.4 percent in the 34 months leading up to September 2012 to 17.2 percent in the subsequent 34 months after December 2012, indicating the effectiveness of the new packaging regulations.
Table 1: Average Smoking Prevalence, Before and After the Packaging Changes
The average values were determined using sample weights from RMSS, with statistical significance indicated by asterisks: ***, **, and * denote differences between the pre- and post-periods that are significant at the 1%, 5%, and 10% levels, respectively.
Source: RMSS Data (January 2001 – September 2015)
27 Table 2 presents the sociodemographic factors included in the model using the same approach: it shows average values in the full sample, from January 2001 to September
2015, and it shows changes over a symmetric 34-month window before and after the December
Since 2012, Australia has experienced significant demographic shifts, including a decline in the number of married individuals, an increase in the foreign-born population, and an aging demographic with a growing percentage of residents over 65 Additionally, there has been a decrease in full-time employment, coupled with rising affluence among the population To accurately assess the impact of packaging changes on smoking prevalence, it is essential to account for these sociodemographic transformations and other policy changes, as they may contribute to the observed decline in smoking rates or obscure it.
Table 2: Mean Values of Demographic Variables, Before and After the Packaging Changes
Averages in this report were calculated using sample weights from RMSS, and due to rounding, displayed calculations may not exactly match presented numbers Statistical significance is indicated by asterisks: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively For clarity, categorical variables have been aggregated, although they can be accessed in greater detail Specifically, the Age variable is segmented into 12 categories, ranging from ages 14-17 to over 64; the Education variable consists of 4 categories, including Tertiary and Less Education; the Work Status variable is divided into 5 segments, such as Employed Full Time and Unemployed; and the Income variable is categorized into 19 segments, starting from income less than $6,000 to various increments.
The regression analysis incorporates a detailed categorization of variables, focusing on loan amounts ranging from $5,000 to over $130,000, with increments of $10,000 up to $130,000 A comprehensive list of these variables can be found in Appendix D.
Source: RMSS Data (January 2001 – September 2015).
Regression Results
28 Table 3 presents a summary of the regression results For conciseness, the table shows the estimated “coefficient” of only the 2012 packaging changes (e.g., -0.0237), along with p-values (e.g., 0.017) that describe the significance level at which the estimated coefficients are statistically different from zero 29 A negative coefficient estimate indicates that the policy has reduced smoking prevalence For the other explanatory variables or groups of variables, the table indicates whether each of their estimated effects is statistically significant at conventional thresholds (displayed by asterisks) The full set of results, including all coefficient estimates, is shown in Appendix D
29 The table columns present results associated with different constructions of the before and after periods As previously explained, because October and November are transition months in which some smokers had plain packs while others did not, those months do not properly belong in either the before or the after period Column (1) presents the results of clean before and after periods by excluding October and November 2012 from the analysis As a robustness check, Columns (2) to (4) present results based on the December, November, and October policy start dates, using data from the entire sample period These alternative analyses retain data from the months of October and November 2012, and include them in either the after- period or the before-period, though neither belongs completely in the one or the other
30 The results show that the 2012 packaging changes are associated with a statistically significant decline in smoking prevalence in all constructions of the before and after periods All of the estimated coefficients are negative and statistically different from zero at the 2.9 percent significance level or lower The estimated coefficient in the preferred model, shown in Column (1), is -0.0237, and it is statistically significant at the 1.7 percent significance level These results support the conclusion that the 2012 packaging changes have reduced smoking prevalence beyond trend, over the period December 2012 to September 2015, relative to what prevalence would have been otherwise
The p-value represents the minimum significance level at which an estimated coefficient is considered statistically different from zero For instance, a coefficient with a p-value of 0.10 indicates statistical significance at the 10 percent level, while a p-value of 0.017 signifies significance at the 1.7 percent level Although there is no strict threshold for determining statistical difference from zero, generally, smaller p-values increase confidence that the true parameter is indeed different from zero Practically, many economists accept p-values below 0.10 as indicative of statistical significance.
Table 3: Summary Estimation Results Using Individual-Level RMSS Data,
Start Date for Packaging Changes:
Dec 2012, Excluding Oct and Nov 2012 Dec 2012 Nov 2012 Oct 2012
Excise Tax 2010 YES*** YES*** YES*** YES***
Excise Tax 2013 YES* YES* YES* YES*
Excise Tax 2014 YES* YES* YES* YES*
GHW 2006 YES YES YES YES
Time Trend YES* YES* YES** YES*
Female YES*** YES*** YES*** YES***
Married YES*** YES*** YES*** YES***
Foreign YES*** YES*** YES*** YES***
Age Groups YES*** YES*** YES*** YES***
Education Groups YES*** YES*** YES*** YES***
Work Status YES*** YES*** YES*** YES***
Income Groups YES*** YES*** YES*** YES***
Indicators YES*** YES*** YES*** YES***
Constant YES*** YES*** YES*** YES***
P-values are presented in parentheses, with asterisks denoting statistical significance at the 1% (***), 5% (**), and 10% (*) levels The inclusion of a variable or group of variables in the regression model is indicated by "YES."
Source: RMSS Data (January 2001 - September 2015)
31 To interpret the magnitude of the estimated effect of the 2012 packaging changes,
I translated the estimated coefficients shown in Table 3 to reflect the implied percentage point reduction in smoking prevalence attributable to the packaging changes 30 These calculations are
The translation of coefficient estimates from a probit model is essential, as the sign of an explanatory variable's effect, like the 2012 packaging changes, is indicated by its coefficient However, the actual impact on smoking prevalence, while controlling for other variables, is influenced by a complex function of all explanatory variables in the regression model.
New York, New York: McGraw Hill, 2004 (hereinafter “Gujarati (2004)”), pp 613-614; and J.M Woolridge,
Econometric Analysis of Cross Section and Panel Data, Cambridge, Massachusetts: MIT Press, 2002
In this study, I analyze the impact of the 2012 packaging changes on smoking prevalence by comparing the model's predictions for the actual world with those for a counterfactual scenario without these changes Specifically, I focus on the post-policy period from December 2012 to September 2015, calculating the average smoking prevalence by averaging the predicted smoking probabilities for all individuals during this timeframe The findings are detailed in Table 4, as referenced in Woolridge (2002), pages 458-459.
Table 4: Predicted Effects of the 2012 Packaging Changes on Smoking Prevalence,
Start Date for Packaging Changes:
Dec 2012, Excluding Oct and Nov 2012 Dec 2012 Nov 2012 Oct 2012
Note: Due to rounding, calculations based on displayed precision may not replicate the numbers presented Source: Table 3
32 My discussion focuses on the predicted effects associated with the estimation results of Column (1) of Table 3, with estimated coefficient of -0.0237 As seen in Tables 1 and
4, actual and predicted smoking prevalence in the post-implementation period are about 17.21 percent The model predicts that smoking prevalence in the counterfactual world without the
To assess the impact of tobacco control policies, I utilize model estimates to predict smoking probabilities for individuals in the post-policy sample, taking into account actual policies, the sociodemographic composition of the Australian population, and overall trends For comparison, I replicate this process while excluding the 2012 packaging changes by setting the corresponding indicator variable to zero This approach allows for an analysis of the effects stemming from the implementation of these packaging changes.
2012 packaging change indicator is to undo the effect of the packaging changes
The 2012 packaging changes are estimated to have reduced smoking prevalence by approximately 0.55 percentage points, reflecting a 17.77 percent decrease The confidence interval for this estimate ranges from -1.01 to -0.095 percentage points, providing strong evidence that the impact of these packaging changes is statistically significant and different from zero.
33 Put differently, as shown in Table 1 smoking prevalence in Australia declined from an average of 19.4 percent in the 34 months before the 2012 packaging changes to an average of 17.2 percent in the 34 months after the 2012 packaging changes Without the 2012 packaging changes, the model predicts that smoking prevalence would have still declined, but only to 17.77 percent Thus, the packaging changes should be credited with about 0.55 percentage points (or about 25 percent) of the 2.2 percentage points of actual decline over this period Similar effects are seen across all of the specifications shown in Table 4
34 A detailed look at the full set of results (presented in Appendix D) shows that the vast majority of other explanatory variables included in the regression model are statistically significant and have the expected signs, meaning that the estimated effects are directionally consistent with prior research For example, excise tax policies are all associated with declines in smoking prevalence The largest tax effect, both in terms of magnitude and statistical significance, is associated with the tax increase in April 2010, which was the first and largest excise tax increase during this period In addition, as expected, the model finds that: (a) married people are less likely to smoke than unmarried people; (b) employed people are less likely to smoke than unemployed people; and (c) educated people are less likely to smoke than uneducated people Each of these effects is statistically significant at conventional thresholds The model is not able to measure a statistically significant effect of the 2006 graphic health warnings 33
32 Note, the difference between 0.55 percentage points and 0.56 percentage points, based on a difference between 17.77 percent and 17.21 percent, is due to rounding The actual difference is 0.55 percentage points
Despite a noticeable decline in smoking prevalence in the area following the policy change, measuring the specific impact of the policy remains challenging The presence of a trend line can obscure the effects of the policy, potentially leading to a situation where the policy's influence is indistinguishable from the underlying trend Consequently, it becomes difficult to accurately assess the separate effects of the policy and the existing trend.
35 Finally, to probe the robustness of the empirical finding that the 2012 packaging changes are associated with a statistically significant reduction in smoking prevalence, I explored various alternative model specifications, including:
Seasonality significantly influences smoking behavior, as various factors like weather and holidays can impact consumption patterns throughout the year However, smoking prevalence is expected to be more stable due to the addictive nature of smoking, making it less likely for individuals to alternate between smoking and abstaining monthly To account for seasonal variations in my analysis, I incorporate a set of calendar month indicators into the regression model.
Replacing tax policy indicator variables with the excise tax amount offers a more adaptable approach to understanding how price increases influence smoking prevalence This method avoids the assumption that the impact of tax hikes on smoking rates is directly proportional to the tax increase size If this proportionality assumption is valid, using a single variable for tax amounts could enhance the precision of the estimated coefficients.
Incorporating more disaggregated demographic groups offers a flexible approach to understanding the impact of sociodemographic characteristics on smoking prevalence This method avoids assumptions about the effects across various groups, such as differing income categories By aggregating demographic groups with similar smoking prevalence, the precision of estimated coefficients can be enhanced, leading to more accurate insights into smoking behaviors.
34 See, for example, D Momperousse, C.D Delveno, and M.J Lewis, “Exploring the Seasonality of Cigarette- Smoking Behaviour,” Tobacco Control, Vol 16(1), 2007, pp 69-70
35 The more restrictive model would produce biased parameter estimates if the imposed assumption does not hold
In this case, statistical testing cannot reject the restriction that the effect of tax increases on smoking prevalence is proportional to the size of the tax increase
Conclusion
36 The evidence shows that 2012 packaging changes are succeeding in reducing smoking prevalence beyond trend In terms of order of magnitude, smoking prevalence is 0.55 percentage points lower over the period December 2012 to September 2015 than it would have been without the packaging changes For reasons I have explained, this effect is likely understated and is expected to grow over time This evidence supports the conclusion that the TPP Act is having its intended effect
36 Woolridge (2002), pp 453-461; and Gujarati (2004), pp 595-615
Cell: (617) 697 6826 Fourteenth Floor tasneem.chipty@analysisgroup.com Boston, MA 02199
Dr Chipty is a renowned expert in industrial organization, antitrust economics, and econometrics, providing valuable insights on competitive issues and the impact of firm behavior on market outcomes Her extensive research spans various industries, including airlines, healthcare, and pharmaceuticals, where she analyzes the effects of government policies on consumer behavior With a strong background in litigation, Dr Chipty has offered testimony and expert analyses before key regulatory bodies such as the Federal Trade Commission and the World Trade Organization, showcasing her expertise in economic and econometric evaluations.
Copyright Board She is co-editor of forthcoming next edition of the American Bar Association’s book
Dr Chipty is an expert in antitrust damages with a strong academic background, having published research on vertical integration's strategic role in market foreclosure, the influence of firm size and network effects on business negotiations, and regulatory impacts on corporate behavior Before joining Analysis Group, she served as Vice President at Charles River Associates and has taught antitrust, regulation, industrial organization, and econometrics at prestigious institutions such as Ohio State University, Brandeis University, and MIT Dr Chipty earned her Ph.D in economics from MIT and holds an undergraduate degree in economics and mathematics from Wellesley College.
Ph.D Economics, Massachusetts Institute of Technology
B.A Mathematics and Economics, with honors, Wellesley College
2010 – Present Analysis Group, Inc Boston, MA
1999 – 2010 Charles River Associates, Inc Boston, MA
Vice President (2005-2010) Principal (2002-2005) Senior Associate (1999-2002)
2005 Massachusetts Institute of Technology, Boston, MA
Visiting Associate Professor of Economics
1997 – 1999 Brandeis University, Graduate School of International Economics and Finance, Waltham,
MA Visiting Assistant Professor of Economics
1993 – 1999 Ohio State University, Columbus, OH
In the Matter of the Statement of Proposed Royalties for the Retransmission of Distant Television Signals in Canada for the years 2014 to 2018, testimony was submitted on October 2, 2015, with a trial testimony occurring on January 25-26, 2016 This was on behalf of major telecommunications companies including Bell Canada, Cogeco Cable Inc., Rogers Communications Inc., Shaw Communications Inc., Videotron G.P., and Telus Communications Company, represented by attorneys from Fasken Martineau DuMoulin, LLP, specifically Jay Kerr-Wilson.
Challenges to Australia’s Tobacco Plain Packaging Act, on behalf of Australia, in the World Trade Organization Submitted testimony on March 9, 2015, May 31, 2015, September 14, 2015, October
Caroline Behrend et al filed a civil action against Comcast Corporation in the United States District Court for the Eastern District of Pennsylvania, case number 03-6604 Testimonies were submitted on multiple dates, including April 10, May 6, and May 11 in 2009, as well as on August 21 and September 18, 2009, and later on May 22, 2012, and January 2013.
15, 2014; testified at deposition on May 22, 2009; testified at a class recertification hearing on
October 26, 2009, on behalf of Comcast Corporation Attorneys: Kasowitz Benson Torres &
Freidman (Michael Shuster and Sheron Korpus) and Davis Polk (David Toscano and Arthur Burke)
In the case of American Broadcasting Company Inc et al v Aereo, 12 Civ 1543, filed in the United States District Court for the Southern District of New York, testimony was submitted on December 20, 2013, in support of Aereo The legal representation for Aereo was provided by the law firm Fish and Richardson, with attorney David Hosp leading the case.
DISH Network LLC f/k/a Echostar Satellite LLC v ESPN, Inc., and ESPN Classic, Inc., No 09 CIV
In the United States District Court for the Southern District of New York, case number 6875 (JGK) (FM), testimony was submitted on July 29, 2011 Subsequent depositions were conducted on November 22, 2011, and in January 2013, with a trial testimony presented in February 2013 on behalf of DISH Network The legal representation included attorneys from Flemming Zulack Williamson Zauderer LLP, specifically Dean Nyciper, and from Simpson Thatcher, namely Barry Ostrager and Mary Kay Vyskocil.
In the case Echostar Satellite LLC v ESPN, Inc., et al., indexed as 08-600282 in the Supreme Court of New York County, a deposition was conducted on June 23, 2011, where testimony was provided on behalf of Echostar Satellite LLC The legal representation for Echostar includes the law firm Flemming Zulack Williamson Zauderer LLP, with Dean Nyciper serving as the attorney.
In the case of Casitas Municipal Water District v United States (Case No 05-168L) in the United States Court of Federal Claims, testimony was submitted on February 25, 2010, and February 8, 2007 Key depositions took place on March 10, 2010, followed by trial testimony on October 28, 2010, with representation from the U.S Justice Department, specifically attorneys James Gette and Barrett Atwood.
In Canada, the collection of royalties for the reproduction and public communication of musical or dramatic musical works by online music services will be managed by CSI and SOCAN These organizations are responsible for ensuring that creators receive fair compensation for their work in the digital landscape.
2007 to 2010, before the Canadian Copyright Board Submitted testimony on April 29, 2010 and on June 9, 2010, and testified at trial on June 28-9, 2010, on behalf of Apple Inc., Bell Canada
Enterprises Inc., Rogers Communications Inc., Telus Communications Company, and Videotron Ltd are involved in legal proceedings, with Goodmans LLP representing Apple Inc through attorney Michael Koch, while the remaining companies are represented by Fasken Martineau DuMoulin, LLP, with attorney Jay Kerr-Wilson.
In the case of United States of America v Daily Gazette Company and MediaNews Group, Inc., Civil Action No 2:07-0329, the U.S Justice Department submitted testimony on September 1, 2009, in the Southern District of West Virginia, represented by John Reed, Mark Merva, and Norm Familant.
In re ASARCO LLC, et al., Case No 05-21207 in the United States Bankruptcy Court for the
On August 1, 2008, testimony was submitted in the Southern District of Texas, Corpus Christi Division, and a deposition was given on August 7, 2008, representing Ready Mix USA, LLC The legal representation for this case was provided by the law firm Baker, Donelson, Bearman, Caldwell & Berkowitz P.C., with Gary Shockley as the attorney.
SOCAN Tariff No 16 pertains to the royalties collected by SOCAN for the public performance and communication of musical and dramatic works in Canada from 2007 to 2009 Testimonies were submitted on November 30, 2007, and a trial took place in January 2008, representing a consortium of Canadian background music users, including Bell ExpressVu, Chum Satellite Services, and DMX Canada The legal representation for this case was provided by Fasken Martineau DuMoulin, LLP, with attorneys Jay Kerr-Wilson and Aidan O’Neill involved.
In the case concerning the Digital Performance Right in Sound Recordings and Ephemeral Recordings for a New Subscription Service (CRB Proceeding 2005-5), testimonies were submitted on October 30, 2006, and July 24, 2007 Additionally, depositions were given on May 8, 2007, and trial testimonies occurred in June 2007, representing Sirius Satellite Radio and XM Satellite Radio The legal representation for Sirius was provided by Wiley Rein, LLP, with Bruce Joseph as the attorney, while Weil, Gotshal & Manges represented XM, with Ralph Miller as their attorney.
MERGERS AND ACQUISITIONS AND OTHER GOVERNMENT INVESTIGATIONS