Tea drinking and mortality in pre-industrial England

[Update: I now have some doubts about this paper - see this follow-up post]

The importance of clean water for public health is thoroughly uncontroversial in modern times. In the temporary absence of a safe water source, one recommendation is to boil water for drinking, which will kill off most bacteria and other pathogens. However, prior to the acceptance of the germ theory of disease, the importance of clean water was relatively unknown. Water-borne diseases such as dysentery and cholera were relatively common (at least, compared with modern times).

In the late 17th Century, the English began drinking tea in large numbers (more on that in a moment). One of the important aspects of tea drinking is that it requires boiling of water. Did that lead to a reduction in mortality, especially from water-borne disease? That is the research question addressed in this forthcoming article by Francisca Antman (University of Colorado, Boulder), to be published in the journal Review of Economics and Statistics (ungated earlier version here, and relatively non-technical summary by the author here). Why investigate this? Antman notes that:

...several historians have suggested that the custom of tea drinking was instrumental in curbing deaths from water-borne diseases and thus sowing the seeds for economic growth.

Antman's research is the first to quantitatively attempt to assess these claims. She uses data on mortality rates at the parish level for 404 parishes from the mid-16th Century to the mid-19th Century, and a couple of different proxies for water quality:

The primary water quality measure used in the analysis is the number of water sources within 3 km of the parish, as calculated using data from the United Kingdom Environment Agency Statutory Main River Map of England overlaid on a map of historical parish boundaries... It is expected that parishes with a higher number of rivers proximate to the parish would have benefited from greater availability of running water, and thus would have benefited from relatively cleaner water compared with those parishes which were limited to only a few sources and thus suffered from a greater likelihood of contamination...

An alternative water quality proxy, the average elevation within a parish, is also offered to show that the relationship between tea and mortality is robust to alternative measures of water quality... Elevation is believed to be positively correlated with water quality because parishes at higher elevation would have been less likely to be subjected to water contamination from surrounding areas.

Antman applies two different strategies to identify the effect of tea drinking on mortality. In the first, she compares the decline in mortality between high (above the median) water quality parishes and low (below the median) water quality parishes over time. She particularly compares the difference between before and after the widespread adoption of tea drinking, which she dates as:

...the Tea and Windows Act of 1784 which reduced the tea tax from 119 to 12.5 percent at one stroke...

The second strategy uses lagged national-level tea imports as an indicator of when tea drinking increased in prevalence, but is otherwise similar. The results from the first strategy are nicely summarised in panel A1 of Figure 1 from the paper:

Notice that mortality declines in both high-water-quality parishes (dashed line with diamonds) and low-water-quality parishes (solid line with circles) after 1784, but that the decline was larger in low-water-quality parishes. And even though there is an up-tick in mortality towards the end of the time period (probably due to urbanisation, as urban areas had higher mortality than rural areas), the difference between high-water-quality and low-water-quality parishes continued to increase. Antman notes that:

With regard to the magnitudes of the impact of tea drinking on mortality, the estimates suggest that areas with worse water quality saw yearly mortality rates drops by about 18% by the end of the period, relative to parishes with better water quality...

The results are similar using the second strategy, although the size of the effect was smaller. Antman also shows that her results are robust to controlling for smallpox mortality, and to controlling for wages, as well as robust to including different proxies for parish-level population. So, it does seems that tea drinking, by promoting the boiling of water, did reduce mortality in pre-industrial England.

[HT: The Dangerous Economist, last year]