Cadahia Pedro, Golpe Antonio, Martín-Álvarez Juan M, Asensio Eva
1Department of Economics, University of Huelva, Huelva, Spain.
2Department of Quantitative Methods for Economics and Business, Universidad Internacional de La Rioja, Logroño, Spain.
Tob Induc Dis. 2021 Dec 15;19:98. doi: 10.18332/tid/143321. eCollection 2021.
There is a literature that questions the veracity of the studies commissioned by transnational tobacco companies (TTCs) to measure the illicit tobacco trade. Furthermore, there are studies that have indicated that the empty pack surveys (EPSs) ordered by TTCs overestimate the size of this trade. This study simultaneously analyzed whether the EPSs established in each of the 47 Spanish provinces were accurate and measured anomalies observed in provinces where sales exceed expected values.
To achieve the objectives of this study, provincial data on cigarette sales, prices and GDP per capita were used. These data were modeled with machine learning techniques that are widely used to detect anomalies in other areas.
The magnitude of the average anomaly in provinces where sales are higher than their expected values exceeds 40%, while the average anomaly in provinces where sales are lower than their expected values (as detected by the EPSs) is <15%. Furthermore, the results reveal that there is a clear geographical pattern to the provinces in which sales below reasonable values are observed. In addition, the values provided by the EPSs in Spain, as indicated in the previous literature, are slightly overestimated. Finally, some regions bordering other countries or that are highly influenced by tourism have observed sales that are higher than their expected values.
Cooperation between countries in their tobacco control policies can have better effects than policies developed based on information from a single country. The lack of control over the transactions of tourists and the inhabitants of bordering countries can cause important anomalies that distort the understanding of tobacco consumption that governments have based on official data.
有文献对跨国烟草公司委托开展的用于衡量非法烟草贸易的研究的准确性提出质疑。此外，还有研究表明，跨国烟草公司开展的空包装调查高估了该贸易的规模。本研究同时分析了西班牙47个省份各自开展的空包装调查是否准确，并对销售超过预期值的省份中观察到的异常情况进行了衡量。
为实现本研究的目标，使用了各省份的香烟销售、价格及人均国内生产总值数据。这些数据采用机器学习技术进行建模，该技术在其他领域广泛用于检测异常情况。
销售高于预期值的省份的平均异常幅度超过40%，而销售低于预期值的省份（如通过空包装调查所检测到的）的平均异常幅度小于15%。此外，结果显示，观察到销售低于合理值的省份存在明显的地理模式。此外，正如之前文献中所指出的，西班牙空包装调查提供的值略有高估。最后，一些与其他国家接壤或受旅游业影响较大的地区观察到销售高于其预期值。
各国在烟草控制政策方面的合作可能比基于单个国家信息制定的政策产生更好的效果。对游客及边境国家居民交易缺乏管控可能导致重大异常情况，从而扭曲政府基于官方数据对烟草消费的理解。
