An Interview with Chris Bystroff
A couple of years ago I read a paper titled “Footprints to Singularity: A Global Population Model Explains Late 20th Century Slow-Down and Predicts Peak Within Ten Years.” The banner finding of the paper—that world population is very close to peaking—seemed intriguing, but the population statistics then available seemed to show continued growth, at about 80 million net additions per year. I was unfamiliar with the author, Chris Bystroff, but made a mental note to keep track of how his modeling squared with developing demographic trends. Since then, significant population declines have been reported in China, South Korea, Japan, and many other countries. Official world population growth forecasts are being revised downward rapidly. I recently revisited Bystroff’s paper and was impressed by its methodology as much as its conclusions. I contacted him and invited him to engage in an interview by email; he graciously agreed.
Chris Bystroff is a San Francisco Bay Area native who studied biochemistry at the University of California San Diego and earned a doctorate studying the three-dimensional structures of protein molecules. He completed postdoctoral stints at the University of California San Francisco and the University of Washington, and in the 1990s was a visiting professor at the National University of Engineering (UNI) in Nicaragua. In 1999 he took a faculty position at Rensselaer Polytechnic Institute (RPI) in upstate New York, where he still works. At RPI he rose to full professor and maintains a research laboratory funded by the National Science Foundation, the National Institutes of Health, and the Grantham Foundation. His lab is currently working on a contraceptive vaccine and a contraceptive transgene for temporary, non-hormonal birth control in men and women. Bystroff teaches courses in protein structure, protein computational modeling, and, since 2016, a course in modeling human population. He has 75 peer-reviewed published papers to his credit, along with one book and three patents.
RH: Thank you, Chris, for taking time to answer a few questions. Let’s start with some basic information. How are population statistics typically compiled? And how do demographers make forecasts about population growth? Do you have a critique of these methods?
CB: I have had the pleasure of being interviewed for a podcast along with demographer Wolfgang Lutz, who is the lead author on the publications that the UN uses for their published population projections. Lutz makes forecasts using trends in birth and death rates, their first and second derivatives, and using a stochastic model for immigration. I have critiqued this method saying that it ignores global limits. When I said this during our interview, Lutz countered that (to paraphrase his response) there are no physical limits to population growth. His view was that climate change and other environmental effects have had a negligible effect on the birth/death rates. He thinks the decline in the birth rates is due to Demographic Transition Theory (DTT), which states that economic advancement leads to decreased family size. DTT is encoded into the population projections through an assumption that Total Fertility Rate will converge on replacement value (TFR=2.1); this despite recent TFR values well below 2 and going down in many countries. I tried to convert him to an alternative view, in which the Earth has a voice, because in his model there is no Earth, no limits, and no carrying capacity (Lutz dismissed the notion that there is a carrying capacity for humans). To convince him, I pointed to the projected effect of climate change on agriculture. I pointed to declining freshwater aquifers. But changing an ingrained model takes more than just data and arguments.
Population statistics can be census data or vital statistics. Both have shortcomings. Census is regarded as more accurate, but census numbers can be politicized. China essentially admitted to overstating their population. India projects pride in having overtaken China in the count. In the US, states are motivated to overstate population since this could give them additional seats in Congress. Census numbers are not all direct counts. Projections, vital statistics, and demographic statistics are used to estimate the non-response rate. Then the estimated non-responses are added to the responses to get the reported count. I have not looked into the non-response rate, but I think it must be rather large, perhaps 10%. If so, then the reported census numbers may be contaminated with expectations. This was the case with China as reported by China-watcher Yi Fuxian. China overstated 2020 population, along with crop production. A paper from 2010 showed that China has done this in the past, significantly underestimating mortality. People ask why my predictions were off and I think, well, the predictions are only as good as the data going in.
RH: Your analysis leads you to a conclusion that few demographers would likely accept. How is your methodology different from theirs? How would you defend it against likely critiques?
CB: In discussions with the few demographers I have met (I am not a demographer, as you may know, I’m a biochemist), I have put them on the defensive and have not heard good defenses. The typical defense is to say, “This is what everybody is saying.” This is a fallacy. Just because people are saying it doesn’t mean it is true. I had an interaction with John Sterman (MIT) at the System Dynamic Society meeting in Albuquerque in 2020. Sterman is the lead author on En-Roads, a comprehensive model for climate change. I’m sure you know of it and have probably played with it. The population model in En-Roads is taken from the UN population model. I complained that there is no feedback from climate change to population, as we know there should be. A truer population model should allow for a crash (using the dramatic term). He stopped talking to me at that point.
I also put Wolfgang Lutz on the defensive. When the interviewer asked us at the end of the interview what we could take away from each other’s critiques, Lutz said he would consider putting global limits into his model. I in turn said I would start to consider economics more.
My model is different in that I focus on the carrying capacity (which is a non-starter for demographers like Lutz). I subscribe to the theory that a K-selected species is limited by carrying capacity and that we are a K-selected species. It’s Ecology 101. Also, my model uses the power of system dynamics, inspired as I was by The Limits to Growth. A systems approach allows me to set constraints in a way that demographics analysis does not. For instance, I put total world biocapacity into a two-stock sub-system with global hectare units. The sub-system can’t grow, just like the Earth can’t grow. We can only change the allocation of hectares (humansphere versus ecosphere), we can’t add hectares. This view is inspired by Bill Rees and Mathis Wackernagel. Also, my model is different in that it explicitly factors in technology and treats technology like a living thing, growing exponentially. Others have published about technology, but I am not aware of it being included in a formal model before. It is always discussed informally (i.e., non-numerically). Finally, I can’t neglect to mention the machine-learning aspect. My model was fit to population data using a simple machine-learning algorithm (called hyperfit in my paper). The program allowed me to optimize the parameters of the model even though they are interdependent. For instance, the total biocapacity of the Earth is an unknown that I can find by machine learning, but it correlates with another parameter, the sensitivity of the carrying capacity to ecosystem loss. Using hyperfit I could explore how parameters were related, and that was useful for the story-telling part of the paper.
RH: What evidence do you see that world population is in fact peaking right about now?
CB: The TFR of more than half of the world’s population is now below replacement value. That alone should tell us the population is sinking. Some think TFR tells you about future birth rates, but that’s not how it is calculated. There’s such a thing as “demographic inertia” if the age demographics are skewed (lots of young people), but they’re not very skewed, so TFR relates to birth rates now, and expresses that number in terms of average family size later. So, we can look at TFR as an indicator of the slope of the population. I haven’t done it yet, but if I took the TFR times the size of the country and summed it over all countries (and then divided by the sum of the populations), I would probably get a number below 2.1. That would be a decline. If you look at the TFR data publicized by the UN in 2022 and compare those to the TFR published by the OECD in 2021, the difference is night and day. In the latter, the vast majority of countries have TFR less than 2.1. In fact, the eyeball average is about 1.5. This includes China (TFR=1.16) and India (TFR=2.03). What does the OECD know that the UN doesn’t? So, I would say the world population is peaking right now or has already peaked. The uncertainty comes from the degree of demographic inertia and uncertainty in the statistics.
RH: How much of this has to do with declining birth rates, versus increasing death rates?
CB: It’s all about declining birth rates. But as climate change heats up, population decline will be about increasing death rates. If the IPCC “hot models” are correct, then we can expect a 5-6 degree change by 2100. Long before that, equatorial countries will be losing population due to heat, and/or migration to the poles.
RH: Do you think falling birth rates might be influenced somewhat by environmental hormone-mimicking toxins? What other factors?
CB: Fertility is falling everywhere. So, the explanation must be global. Environmental toxins may have a local effect, but are they everywhere? I think a simpler explanation is that decreased fertility is intentional. What is everywhere is social media. Everyone knows about climate change and the associated sea level rise, raging storms, heat waves, and droughts. The outlook for the future is gloomy, and if you are a woman of childbearing age, you might think twice about bringing a child into this world with this future. Also, everywhere the prices of things are going up—childcare and rent among them. When asked why they chose to have zero or one child, women invariably cite the cost of raising a child. So that’s what I blame, climate change and cost of childcare, not toxins.
RH: Pro-growth economists are fretting over falling birth rates, portraying them as a catastrophe. How do you see population decline?
CB: I think population decline will be a relief, like an escape valve in a pressure cooker. As we shrink in numbers we will grow in other areas, like resources per capita and free time. I don’t buy the argument that an aging population is a big problem. Climate change is a big problem. An aging population is a small problem, not worth mentioning by comparison. Pro-growth economists are motivated by what economic theory says about economic stability, namely, that you need growth for stability. I don’t think that is a law of nature. I believe that when the economy shrinks, we will invent an economy that is stable with a shrinking populace. Japan is already shrinking and it is not going through a depression. Doesn’t that mean the doomsaying economists are wrong about needing growth?
So, I believe we will Decline and Prosper (book by Vegard Skirbekk). But part of me says The Heat Will Kill You First (book by Jeff Goodell), so what happens to you may depend on where you live. In any case, the population Countdown (book by Alan Weisman) is happening now or soon, certainly not in 2100.