Amid the deluge of data, speculations, and commiserations about the ongoing COVID-19 (coronavirus) pandemic, here’s a modeling exercise mixing epidemiology and economics that I haven’t seen done, and that I think is important. Stated in four parts:
1. Closing schools and businesses saves lives by slowing the spread of disease, facilitating the treatment of infected people, and buying time to develop medicines and vaccines. How many lives it saves is uncertain, but one can make a range of estimates based on a range of assumptions.
2. Closing schools and businesses induces economic recession. The massive drop in consumption has already led to a staggering number of layoffs, further reducing consumption, … a feedback loop. This is probably hard to model, but since we already have data on the ongoing economic impacts, we should at least be able to make crude estimates.
3. Recessions harm people, not just financially but mentally and physically via impacts on crime, suicide rates, etc. This is very hard to model and the uncertainties involved are enormous. Still, one would hope that the social sciences have given us at least a rough sense of the relationships involved. (More on this below.)
4. Therefore, we should be able to estimate, as a function of time spent under social isolation measures, the gain and loss of lives due to both direct disease impacts and indirect economic impacts. Which is greater? Do they cross? (I.e. at some point, is tanking the economy more deadly than the COVID-19 virus? Note that this doesn’t mean we shouldn’t fight the spread of the virus, but that if we do, we’d better have a solid plan to shift the economic curve!)
Which of the graphs above — which will have giant error bars, of course — would our models look like? What are the scales on the axes? What if, rather than lives, we calculated expected future years of life? After all, I’d much rather that I die than my children. (Of course, there’s more to living than simply not dying; minimizing our odds of death all the time would lead to a stultifying and inhumane existence! But things like quality of life are even harder to assess.)
It’s perhaps morbid to think about this, but it’s important if we’re to minimize harm. In the past few days, I’ve been struck by how extremely sheltered my “bubble” of contacts is from people directly harmed by financial concerns. On the other hand, I’ve been dismayed by how poorly argued the claims I’ve read are that our coronavirus response is overdone. How can we reconcile the various camps? By actually making our claims concrete, well-defined, and quantitative.
I’m incapable of tackling this exercise myself, but perhaps other people will. Or perhaps it’s already been done — I’d love to read about it. Or perhaps it’s impossible — even for order-of-magnitude estimates — but clearly stating the reasons for this might highlight directions to which we should be devoting future research efforts.
I’ve looked just a little bit into the literature on the social impact of recessions (in the US), which is a fascinating topic with some aspects that surprised me. I’ll briefly comment on some of this.
Do recessions make people healthier?
I wondered about the link between economic recessions and crime, and between recessions and health. Naively, I would have expected that a higher unemployment rate leads to more crime, both violent and nonviolent. However, the correlation is apparently weak, masked by larger shifts in culture, incarceration rates, and other factors (link, “Crime and the Great Recession”). A stronger link emerges if one looks at young people: “the average arrest rate for a cohort entering the labour market during a recession is 10.2% higher than for an otherwise similar cohort entering a more buoyant labour market” — from Brian Bell, “Do recessions increase crime?” (2015). Continuing, “the key impact of unemployment on crime is the early experience of unemployment rather than the average unemployment experienced over the cohort’s life cycle.” The effects are long term: “Recessions not only lead to short-term negative outcomes on the labour market but can indeed produce career criminals.” Again, this is just a dip into this topic; I’m sure there’s more out there.
About health, I was quite surprised to learn that recessions lower the death rate! The reasons are many, including less driving and fewer workplace accidents. Here’s a nice NPR piece and a Nature article . Here are articles specifically on the Great Depression (#1, #2). Before we get too cheerful about recessions, however, it’s worth noting that suicide rates increase when the economy falters, and suicides and drug overdoses are increasingly major causes of death in the US.
(Before concluding, I’ll insert a note encouraging you to read my post about a book I’m writing, if you haven’t already. The topic: Life. All of it!)
We’re awash in uncertainty these days. As I write this, my university is essentially closed, all of next term’s classes will be held on-line, and my research group is working from home on data analysis, writing, and other things we can do without a lab. I don’t know what the next few weeks will bring. Perhaps the exercise I’ve outlined above is pointless, unlikely to influence any decision making. Perhaps it’s impossible, and we’re stuck with politicians and punditry rather than quantitative analysis. But I am optimistic that we can do better at assessing options, perhaps not for this crisis, but at least the next one!
I encourage everyone to take care of themselves. I also recommend going outside. My lovely town, Eugene, Oregon, is in many places eerily quiet, but the weather is beautiful and it’s great to see many people walking or biking along the paths, watching the turtles and waterbirds.
Another drawing of the CsgG channel protein. I took a photo as I was working on it. I’ve since finished the illustration, but it’s not very good, and I may end up scrapping it.
— Raghuveer Parthasarathy, March 20, 2020