August 2021

This month, our podcast features a conversation with Leila Bengali, Nora Pankratz, and Jisung Park regarding their recent work on how climate change and weather affects labor supply. Below is an edited transcript of their conversation. Click here for audio and video.

Leila Bengali: Today our topic is how climate change and weather directly affect the economy, focusing on labor supply and businesses. We have with us Jisung Park and Nora Pankratz, of UCLA. I’d like to start with the definition of what it means for climate change and weather to directly affect the economy.

Jisung Park: To some extent, this is a matter of semantics, what you call direct versus indirect effects. Direct effects are things like temperature stress on human beings and what that does to health outcomes, mortality and morbidity, cognitive outcomes, which may affect learning and human capital formation, or labor productivity, labor supply, and other workplace outcomes directly. That’s what we were referring to by direct effects.

Nora Pankratz: I think in the context of what we’re going to discuss today, what’s really interesting is that there have been studies at the aggregate economic level on these micro-channels but then also at the level in between, at how firms are affected by a combination of possible mechanisms that lead to direct reductions in performance through these various macroeconomic channels.

Leila Bengali: Let’s start broadly with some examples of direct effects. Could you talk in more detail about what you’ve found?

Jisung Park: Just to contextualize this, the intuition that temperature might affect human behavior is not new. We’ve actually known based on laboratory experiments dating back to the mid-20th century that both hotter and colder temperatures affect human task productivity and performance, whether it’s on cognitive or physical tasks. What is new in the recent literature is taking real-world data and applying tools that allow us to infer causal links between temperature and human economic outcomes of interest. To give you a couple of snippets from the literature, there’s some path-breaking using individual time-use data and local weather information to show that on hotter days, workers in exposed industries (construction, agriculture, manufacturing) appear to reduce their labor hours. There are also studies that look at the effect of temperature on total output and per-worker productivity. There’s a great paper looking at Indian manufacturing firms: garment manufacturing, cloth-weaving, steel, diamond-processing. This paper finds pretty significant impacts of hotter temperature – somewhere on the order of a 2%-4% decline in productivity for every 2 degrees above 70 degrees Fahrenheit. That means that on a 90-degree day, you’re looking at daily productivity losses per worker on the order of 30%-40%. Those estimates vary considerably by context. If you look in other settings in the United States those effects, at least in some contexts, are more muted. But they’re nevertheless non-zero.

Leila Bengali: Nora, you’ve done some recent research on how firms’ buyer and supplier relationships respond to climate risk in the form of weather shocks. Can you walk us through a general idea you have in mind, for example, firms buy inputs from other firms, and those other firms may be hit by a weather shock. Tell us what you find in your research.

Nora Pankratz: Recently there have been a couple of studies focusing particularly on the firm level. These studies have outlined how firms could be affected, not only through work loss on hot days, but also disruptions in transportation, or potentially additional costs incurred if more cooling is necessary for facilities on really hot days. If we know that there could be negative effects on a single firm, what are the effects on a supply chain? What we hypothesize is that if there is a small shock on productivity at the supplier level, then this shock could be meaningful to the customer. If there’s a shortage in the desired quantity of inputs that this customer firm would like to purchase, and isn’t able to purchase as initially planned, that could then translate to shocks to operating performance. That’s exactly what we test in the study. We use methodologies in order to isolate causal effects, using quasi-random sources of variation in the appearance of heat waves and floods. What we find is that if a supplier firm is hit or is affected by a prolonged period of temperatures above roughly 98 degrees Fahrenheit, that doesn’t only affect the supplier firm, but it also affects the remotely located customer. This evidence is consistent with the idea that there might be a knock-on effect in supply chains, which means that the cost of these shocks are essentially shared among customers and suppliers.

Leila Bengali: So if there is a climate shock in one part of the country, that could have a big effect on another part of the country, even if that other part of the country didn’t have a climate shock. Is that the right way to think about your work?

Nora Pankratz: Yes. The scope goes one step further. Indeed, 40% of our observations in the study (which encompasses a total of 80,000 connections between customers and suppliers) are tied to US firms. However, we also have global connections of supply chains. What that means is that even if firms in the US directly are not affected in the sense of exposure to temperatures, there could be knock-on effects through supply chains, through the exposure elsewhere in the world, where the effects of climate change may be more impactful.

Leila Bengali: Let’s get back to workers. I think one possible mechanism for how weather and temperature could affect labor supply and/or production is fatigue caused by heat. You both have done some recent work on this in a new paper. Jisung, maybe you could start by telling us the research question and what you find.

Jisung Park: We were curious what the effect of heat, or temperature generally, might be on workers and workplace safety. We got data from the universe of worker compensation claims in California. Importantly, we know the zip code of the injury itself and the daily weather on the date of the injury. We link this information. We find that on hotter days, injury risk seems to go up considerably. If you were to compare the risk on a day with high temperatures in the 90s or above relative to a more optimal day for workplace safety (in the 50s or 60s), we’re talking about an increase in injury risk anywhere between 5%-15%, depending on how high the temperature gets. To put this in perspective, looking at the US, on average 3 out of 100 FTE (full-time equivalent) workers will experience a serious injury on the job in any given year. But in industries like construction, agriculture, manufacturing, warehousing – these rates are much higher. The average injury rate in warehousing is roughly 18 times that of finance workers. One of the things we find is that the change in temperature-related injury risk is a far larger problem for workers in highly exposed industries and occupations, who also tend to have lower levels of education and are lower-income workers.

Leila Bengali: One hypothesis is that heat causes fatigue, fatigue causes injuries. If that’s the case, I imagine that workers and firms can take actions to mitigate this channel. Nora, could you talk to us about the general idea of adaptation, and whether you find evidence of adaptation in your work?

Nora Pankratz: Adaptation is really important here. What we have in mind is the need for the economy as a whole to think about how we’re going to live and work in a hotter world. Given that we find this evidence of temperatures increasing the risk to injury at the workplace, we should be thinking about what we can do to mitigate the risk. In that sense, California has been very forward-thinking. In 2006, California implemented the Heat Illness Prevention Standard. This standard prescribes or mandates that employers have to take certain measures to protect their workers at the workplace. These include straightforward measures such as providing workers with additional breaks, water, and access to shaded areas when temperatures exceed certain thresholds. We make use of this mandated adaptation and test whether the link between extreme temperatures and injuries in the workplace gets weaker once these measures are implemented in all kinds of workplaces in California. Our evidence is in line with the idea that the implementation of this standard was really helpful to decrease the link between extreme temperatures and additional injuries in the workplace.

Leila Bengali: So this is good news. Extreme weather does have this negative effect, but there are ways that companies, firms, and individuals can counteract these negative effects.

Jisung Park: On the bad news piece, though, there’s been a lot of attention on heat illnesses. We find that heat illnesses increase on hotter days, which isn’t surprising. But what was surprising was that the vast majority of the increase in injuries on hotter days appears to be for ostensibly unrelated claims. So, for example, things like a construction worker falling off a ladder or getting your hand stuck in a conveyor belt or being struck by a moving vehicle on a work site. The evidence is consistent with this story of fatigue or cognitive errors elevating risk on hotter days.

Leila Bengali: It seems like this is made even more important by projections and forecasts that a lot of areas will experience more days that tend to be hotter, on average. Now I want to put this work in some context: is there an estimate to take your work and extrapolate it out in terms of what it means for the economy?

Nora Pankratz: One specific problem with heat, that could potentially apply to other climate hazards, is that one additional day of heat may not appear extremely damaging to the economy or to any individual economic outcome. However, what’s really dangerous about heat is the combination of the small effects of one additional day of heat, and what you just mentioned, the projection of relatively large numbers of additional heat days that we may face in the U.S. and around the world. That makes it important to think about this point estimate. Think about an increase in the number of heat days (days with a temperature of more than 90 degrees Fahrenheit) of 16 days, which would be equivalent to one standard deviation. Sixteen days corresponds to something along the lines of three additional work weeks. What we find in terms of economic impact is that such an increase of an additional 16 heat days leads to a decrease in the operating performance of a variety of businesses of 4% over the year.

Jisung Park: In our injury work, we estimate that hotter temperatures cause anywhere between 15,000-25,000 additional workplace injuries per year, in California alone. If you take off-the-shelf estimates of the total social costs of those injuries in terms of medical bills, lost work time, lost productivity, that comes out to around $1 billion per year, in California alone.

Leila Bengali: I want to conclude by asking you both where you think more research needs to be done in this area or, based on your research, how would you recommend that macro-forecasters in general take these direct effects into account?

Jisung Park: The magnitudes we’re discussing seem large enough to merit at least considering the inclusion of temperature projections into models of macroeconomic performance. In terms of things we need to better understand, I think there’s still a lot of work to be done to get a handle on how adaptation decisions are made, whether for the household, firm, or individual worker. And importantly, the various constraints, whether that’s financial constraints or information constraints – what are the constraints to adaptation investment decisions? As a labor economist, I’m also interested in how the forces of climate change interact with broader forces of technological change that are fundamentally shifting the nature of work and having very different effects for individuals with higher and lower levels of education.

Nora Pankratz: I am personally interested in the effects on financial markets. I think some of the potential tools for adaptation are insurance-related solutions. I think that would be one really important field to discuss further, because insurance could provide a really valuable tool for risk-sharing. At the same time, some research, specifically in agricultural economics, has also indicated that there may be a downside to the existence of insurance if there are any effects through inefficient pricing, creating disincentives for firms to not invest as much as they potentially could. If we tie this back to the question about how this could be relevant for forecasting, I think one important point is that there is a variety of estimates of economic damages. So, incorporating those in forecasts is probably relatively straightforward. Then the piece where we know much less is indeed adaptation – about what the potential of adaptation is to mitigate damages.