Periodic Reporting for period 4 - FORENSICS (Illicit Markets, Unobserved Competitors, and Illegal Behavior)
Okres sprawozdawczy: 2022-03-01 do 2023-08-31
Many markets are characterized by unobserved behavior by agents. This can take the form of unobserved consumer behavior, for example the purchase of illicit products; unobserved firm behavior, such as illegal use of otherwise legal instruments; or unobserved market agents, as in competition in markets with illegal producers. To the extent that the empirical economic framework does not incorporate these unobserved actions or control for them in estimation, the resulting models are likely to be mis-specified. Naturally, if the models do not contain all elements relevant for decision making, then predictions based on the estimates will be misleading, which could result in incorrect policy recommendations.
This project addresses three situations (themes) in which unobserved behavior plays a crucial role. It does so by (i) developing state-of-the-art empirical models that incorporate illicit behaviors and (ii) proposing novel estimation methods that can be used to detect illicit behavior. In this sense, the methodologies put forward are "forensic" in nature as they deal with uncovering illicit or illegal behavior.
The first theme concerns markets where consumers engage in illicit behavior. These markets are prevalent in society as they constitute the market for illegal drugs, which is estimated at more than $300 billion per year (UN, 2012). Marijuana is the world's most widely used illicit substance (ONDCP, 2004) with vocal advocates for legalization. Indeed, legalization is quickly becoming the norm. In this theme we consider the impact of legalization on use of other licit and illicit products. We also propose a novel way to control for unobserved prices, which is crucial for measuring price sensitivity. Substitutability between illicit and licit products is critical information, as is the extent to which users respond to prices, for these together allow the authorities to gauge how, for example, an increase in marijuana taxes today will impact use across a portfolio of products in the future.
The second theme concerns markets where firms must make strategic decisions in the presence of an unidentified competitor. Many empirical models assume that firms observe (at least some) characteristics of their rivals. However, there are a number of markets where firms possess little information about their potential competitors. One such example is the market for legal drugs - legal marijuana dispensaries co-exist with illegal dealers, Due to the changing regulatory environment legalization is becoming commonplace,and we understand little about what policy tools can be used to curb the influence of the illegal producers.
The third theme examines the impact of firms using legal tools for illegal purposes, where our focus is on quantifying the impact of the illegal use of exclusionary deals, advertising, and market power. For example, upstream manufacturers often impose exclusive dealing contracts on their retailers, which may result in foreclosure of a competing brand. One of the challenging tasks has been to empirically identify the the (illegal) impact of vertical restraints imposed by one upstream firm on a competing firm.
In summary, this project addresses important challenges by developing methods to control for unobserved behaviors. Incorporating and estimating unobserved behavior in a variety of settings is an ambitious undertaking. However, it is vital as a key objective of the proposal is to provide policy makers with tangible tools that accurately reflect the unobserved nature of these markets.
This project addresses three situations (themes) in which unobserved behavior plays a crucial role. It does so by (i) developing state-of-the-art empirical models that incorporate illicit behaviors and (ii) proposing novel estimation methods that can be used to detect illicit behavior. In this sense, the methodologies put forward are "forensic" in nature as they deal with uncovering illicit or illegal behavior.
The first theme concerns markets where consumers engage in illicit behavior. These markets are prevalent in society as they constitute the market for illegal drugs, which is estimated at more than $300 billion per year (UN, 2012). Marijuana is the world's most widely used illicit substance (ONDCP, 2004) with vocal advocates for legalization. Indeed, legalization is quickly becoming the norm. In this theme we consider the impact of legalization on use of other licit and illicit products. We also propose a novel way to control for unobserved prices, which is crucial for measuring price sensitivity. Substitutability between illicit and licit products is critical information, as is the extent to which users respond to prices, for these together allow the authorities to gauge how, for example, an increase in marijuana taxes today will impact use across a portfolio of products in the future.
The second theme concerns markets where firms must make strategic decisions in the presence of an unidentified competitor. Many empirical models assume that firms observe (at least some) characteristics of their rivals. However, there are a number of markets where firms possess little information about their potential competitors. One such example is the market for legal drugs - legal marijuana dispensaries co-exist with illegal dealers, Due to the changing regulatory environment legalization is becoming commonplace,and we understand little about what policy tools can be used to curb the influence of the illegal producers.
The third theme examines the impact of firms using legal tools for illegal purposes, where our focus is on quantifying the impact of the illegal use of exclusionary deals, advertising, and market power. For example, upstream manufacturers often impose exclusive dealing contracts on their retailers, which may result in foreclosure of a competing brand. One of the challenging tasks has been to empirically identify the the (illegal) impact of vertical restraints imposed by one upstream firm on a competing firm.
In summary, this project addresses important challenges by developing methods to control for unobserved behaviors. Incorporating and estimating unobserved behavior in a variety of settings is an ambitious undertaking. However, it is vital as a key objective of the proposal is to provide policy makers with tangible tools that accurately reflect the unobserved nature of these markets.
Through our papers on themes 1 and 2, we learned about substitution between legal and illegal products, including alcohol, cigarettes, illegal recreational marijuana, legal medical marijuana, and opioids. The result of these studies inform the current policy debate on marijuana legalization taking place globally. In particular, we added to the knowledge regarding the effectiveness of taxes on marijuana and other sin goods on controlling youth use; the impact of legalizing medical or recreational marijuana on opioid use; and we are exploring how a combination of taxes and subsidies for legal producers can be used to steer use to the legal market away from the illicit drug dealing market. These are first order policy questions.
Through our work on theme 1 on the importance of unobserved factors that influences choices, we examined internet access decisions. We have learned about how effective (or not) public policies (such as subsidizing prices, increasing education, providing hardware) are at increasing internet take up, particularly among the most vulnerable populations. Our findings inform the ongoing policy debate regarding how to connect the over 3 billion people who are currently offline today. This is particularly timely given the United Nations goal to provide every person with access to the internet by 2030.
Through our work on theme 3, we have learned that restraints imposed by wholesalers on their retailers regarding who they sell their item to (exclusive restrictions) can impact retailers who are not directly involved in the exclusive relationship. This is the first empirical evidence of interrelated effects of exclusive deals. Within the context of the CPU industry, our findings indicate that the ability to innovate and offer superior value is not necessarily the only, or even the most important factor driving market share growth in the x86 industry. Rather, Intel's ability to strike exclusive deals, emerges as an important factors underlying its market share leadership.
The results of this grant have been disseminated through more than 100 presentations including policy institutes (e.g. the Directorate General for Competition of the European Commission); . policy output (e.g. VoxEU, Economic Policy, and CEPR Insights); and press coverage (e.g. The Economist) highlighting their relevance for policy.
Through our work on theme 1 on the importance of unobserved factors that influences choices, we examined internet access decisions. We have learned about how effective (or not) public policies (such as subsidizing prices, increasing education, providing hardware) are at increasing internet take up, particularly among the most vulnerable populations. Our findings inform the ongoing policy debate regarding how to connect the over 3 billion people who are currently offline today. This is particularly timely given the United Nations goal to provide every person with access to the internet by 2030.
Through our work on theme 3, we have learned that restraints imposed by wholesalers on their retailers regarding who they sell their item to (exclusive restrictions) can impact retailers who are not directly involved in the exclusive relationship. This is the first empirical evidence of interrelated effects of exclusive deals. Within the context of the CPU industry, our findings indicate that the ability to innovate and offer superior value is not necessarily the only, or even the most important factor driving market share growth in the x86 industry. Rather, Intel's ability to strike exclusive deals, emerges as an important factors underlying its market share leadership.
The results of this grant have been disseminated through more than 100 presentations including policy institutes (e.g. the Directorate General for Competition of the European Commission); . policy output (e.g. VoxEU, Economic Policy, and CEPR Insights); and press coverage (e.g. The Economist) highlighting their relevance for policy.
The work on the interaction of illegal and legal markets has provided novel results, where the estimation methods have advanced the field beyond the state of the art.
In particular, we developed a novel empirical model to assess the impact of marijuana legalization on the consumption of "sin" goods - marijuana, alcohol, cigarettes, and combinations of these products. We estimated a dynamic model of multi-substance use that involved both illegal and legal products while controlling for persistence in use and restricted (endogenous) access to illegal products. The estimation methodology developed a new way to construct panel data from cross-sectional data. Our results inform the policy debate regarding the impact of marijuana legalization on both the long-term use of sin goods and the resulting tax revenues.
We extended the notion of endogenous access to recreational marijuana into the context of multi-substance use. To estimate this model, we extended the shared-factor approach used in the Bayesian literature to a bundle-choice setting. The latter allows for parsimonious modelling of the correlation structure. This results in a novel estimation technique, which is widely applicable.
In particular, we developed a novel empirical model to assess the impact of marijuana legalization on the consumption of "sin" goods - marijuana, alcohol, cigarettes, and combinations of these products. We estimated a dynamic model of multi-substance use that involved both illegal and legal products while controlling for persistence in use and restricted (endogenous) access to illegal products. The estimation methodology developed a new way to construct panel data from cross-sectional data. Our results inform the policy debate regarding the impact of marijuana legalization on both the long-term use of sin goods and the resulting tax revenues.
We extended the notion of endogenous access to recreational marijuana into the context of multi-substance use. To estimate this model, we extended the shared-factor approach used in the Bayesian literature to a bundle-choice setting. The latter allows for parsimonious modelling of the correlation structure. This results in a novel estimation technique, which is widely applicable.