Mapping Big Tobacco's Global Expansion

Passages Per Year Chart
Frequent Authors (click to show)
Distinctive Terms (click to show)
Topic Model (click to show)
Text Passages

Globalizing Death

Mapping Big Tobacco's Worldwide Expansion

Tobacco exports should be expanded aggressively because Americans are smoking less.

Big Tobacco’s success over the last half century has depended more and more on exporting cigarettes all over the world. As public health efforts found success at reducing smoking rates in the United States and Europe, tobacco companies expanded their business into developing countries. There, unencumbered by high taxes, advertising restrictions, or youth smoking regulations, they could fall back on marketing strategies.

On this page, we can explore how Big Tobacco captured the globe by visualizing when and how often individual countries were discussed in the tobacco documents. The basic assumption behind this map is that rising or falling number of mentions of a country gives us a proxy to track a company’s interest in that country over time. We will freely admit, of course, that it is an imprecise proxy--there are many reasons besides corporate expansion that might lead to mentions of a country. However, we think that this approach is still successful at capturing general trends. Any time you click on a country, the frequency chart and document table automatically conduct a search for the selected country, giving you a way to track its history over time.

To assure that we track the industry’s worldwide expansion rather than world events, we have limited this analysis to internal letters and memos as well as marketing documents while excluding, most importantly, news reports. Of course, that doesn’t mean that world events are fully excluded. Search hits for “vietnam,” for example, show a clear spike around the Vietnam War. By default, the data from American Tobacco, British American Tobacco, Brown & Williamson, Lorillard, Philip Morris, and RJ Reynolds is shown in aggregate. However, you can also limit the analysis to one specific company by unselecting all of the other ones.

By default, the map adjusts for population numbers and colors countries based on the absolute number of search hits rather than relative frequencies. Population adjustment means that the color of a country is determined relative to its total population, i.e. how many documents mention the country per million inhabitants. At the moment, we are using population numbers from 2018. We may try to compile more accurate yearly population data in the future. This assures that smaller but still important markets like Vietnam aren’t overshadowed by larger neighbors like China.

Using absolute counts most clearly shows Big Tobacco’s expansion from the United States and Europe first to Latin America and then onto East Asia and Africa. However, the rising number of documents mentioning, for example, Thailand in the 1980s and ‘90s is partially caused by the fact that we have about 8.4 million documents from 1980 to but only 2.2 million documents from 1950 to 1979. In other words, given more documents, any term, including country names, is likely to get mentioned more often. To account for this in increase in the number of document, you can switch the display to “Frequencies,” which normalizes the absolute counts by total number of words in the tobacco documents for that year. More precisely: it normalizes it by the number of words that are a) in the currently selected collections and b) belong to the document types group “internal communication” and “marketing documents.”

Dataset Notes and Caveats

Country borders and names aren’t static. The period covered by this map, 1950 to 1998, covers the decolonization of most of Africa and the collapse of the Soviet Union, leading to new countries and country borders. These political changes entailed a number of compromises when creating this map.

First off, In some cases, we aggregated the totals for multiple names of the same country: Iran aggregates the search results for “iran” and “persia.” The same is true for Russia (“russia” and “soviet union”) or Myanmar (“myanmar” and “burma”). However, if a former country didn’t have a clear successor state, search results for the former name won’t show up in the map totals. The 9000 hits for “yugoslavia,” for example, aren’t display on the map because Yugoslavia didn’t have one main successor state.

Secondly, in three cases, we “merged” countries, meaning that we treated them as if they were one country with a shared search term and population total. These case are: South Sudan and Sudan, both of which share the search term “sudan;” Somalia and Somaliland (search term: “somalia” “Somaliland” appears less than a 100 times in the tobacco documents and cannot be searched for through Tobacco Analytics because we only added terms to our search vocabulary that appeared at least 100 times.); and the Democratic Republic of the Congo and the Republic of the Congo (search term “congo” Both “democratic republic of the congo” and “republic of the congo” are very rare terms in the tobacco documents because either country was often referred to simply as “congo.” Additionally, “democratic republic of the congo” contains “republic of the congo,” meaning that any hit for the “democratic republic of the congo” would also count as a hit for the “republic of the congo.”).