CKGSB website

CKGSB Knowledge

The Real Cost of Air Pollution in China

by Liu Sha

April 13, 2018

a smoggy morning in Beijing
A smoggy morning in Beijing. Photo: Liu Sha/CKGSB Knowledge

Brian Viard, Associate Professor of Strategy and Economics at CKGSB, explains how air pollution is as bad for the economy as it is for public health

We all know that air pollution is bad for our health. But what is often overlooked is that high pollution levels also cause significant harm to our economic well being. Brian Viard, Associate Professor of Strategy and Economics at CKGSB, has been researching the economic effects of pollution for much of the past few years. His team has found persuasive evidence that the costs of air pollution are greater and more wide-ranging than most people realize. In this interview with CKGSB Knowledge, he explains how tackling the pollution crisis could actually make the Chinese economy more productive. 

Brian Viard, economist who has researched the economic effects of pollution

Q. What inspired you to begin researching the effect of pollution on the economy?

A. I have to give the credit to my co-author, Fu Shihe, who brought the idea up and we started talking about how to approach it. The reason we wanted to do this was, there had been various papers talking about pollution and productivity, but they were narrowly focused on just one occupation in a single firm or a few firms.

We wanted to make this more useful for broad policy-making, and to do so you have to estimate at a broader level than just one occupation. This is what motivated us to do the project. To set policy, you need something that is ideally nationwide, or at least something that covers a large region.

Q. Could you briefly introduce your research?

A. The objective is to relate pollution and output and get estimates at the national level. The simple way to do that is to directly see how firm output varies with pollution. But the problem with doing this is that it will tend to make pollution look like it is either good or less bad than it actually is. The reason is that, if you look across China, an area with high output will have high pollution levels, simply because firms there are producing a lot. So when you start doing the comparison, you might see that output is higher when you have a lot of pollution! It tends to make pollution look like a good thing.

So, that’s the main problem we faced. This is called “reverse causality.” Output causes more pollution, making it difficult to measure how pollution levels affect output. We solved this using the idea of “thermal inversions,” a meteorological phenomenon, which is when warm air ends up above colder air, trapping the colder air and air pollution below it.

Inversions are useful for us because they are caused purely by weather and are not affected by output. If we want to break the reverse causality, we want something that breaks the link between output and pollution. Inversions do this—they independently raise pollution levels and are unrelated to output. So, the strategy is, when inversions increase, to see how output responds to that.

Q. What are the major findings?

A. There are different ways to quantify the results. If you want to use this for policy purposes, there are two main ways to present the numbers. One is just based on the measure of pollution itself. The two pollutants, PM2.5 and SO2, are measured by density in micrograms per cubic meter. If we reduce the PM2.5 density by one microgram per cubic meter, manufacturing labor productivity increases by 0.011 percent. If you do the same for SO2, labor productivity goes up by 0.036 percent.

The second way is to translate this into monetary terms. If you lower PM2.5 by one percent (0.53 micrograms per cubic meter) through means other than lowering manufacturing output, the average manufacturing firm’s output would increase by RMB 74,000 ($11,100) annually. Across all firms, the increase would be RMB 11.8 billion ($1.77 billion) annually or 0.079% of China’s annual GDP. The same numbers for a one percent reduction in SO2 (0.15 micrograms per cubic meter) are RMB 70,000 annually and across all firms RMB 11.1 billion or 0.075% of GDP.

Q. Could you explain the mechanisms behind how air pollution affects labor productivity?

A. There are several ways that air pollution can affect labor productivity. Air pollution can affect people’s physical health, making them work more slowly. If it’s physical labor, the effect could be direct. When pollution is more extreme, workers may become sick and have to miss work. They may also have to miss work to stay home and take care of a family member, particularly young children or the elderly who are most affected by pollution. Another possible effect, sadly, is shorter life expectancy. People can die prematurely from the accumulated effects of pollution. If a worker who dies is replaced by one with less experience, productivity falls.

These are ways the pollution can affect physical performance. There is also evidence that air pollution can slow people’s thinking. This means even knowledge workers could be affected by high pollution levels. If you think about a factory manager who has to make decisions, air pollution could slow their decision-making.

Q. How do your estimates identify these mechanisms?

A. Unfortunately, the answer is that we cannot. The reason is that we can only measure revenue per employee at the firm level. We don’t observe individual employees or individual output, so we don’t know how many hours an individual works in a given year or how many days they miss work. This is an important question, but we cannot answer it, because our goal is to get something that’s comprehensive and nationwide. So, we must use data that is aggregated at a higher level.

Q. Why do you think it is important to conduct this research in China?

A. The straightforward answer is that pollution levels in China are fairly high. China is developing quickly and output has increased a lot. Some of the costs of this pollution, like health costs, have been identified fairly clearly. But there hasn’t been much evidence collected on the cost in terms of manufacturing output. If the government is considering doing something about pollution from a policy standpoint, the effects on output should be factored in so that they take a comprehensive look. Of course, there are other countries that have similar issues, such as India. We cannot use our data to estimate conditions in India, but our technique could be used. 

Q. A problem the government faces is that firms claim extra effort to reduce air pollution will negatively affect their profits. How can the government use your findings to respond to such complaints?

A. I have to say that there is a way they can and there is a way that they cannot. When a factory produces pollution, it doesn’t just affect itself but also the other factories around it. No individual firm probably has sufficient incentive to solve the problem. If a factory reduces its pollution, it will help the firm but by such a small amount that they will choose not to do this. But the pollution reduction objectively speaking is a good thing, because the firm is helping out other firms around it who benefit from the reduction. So I don’t think the government will ever win this argument with an individual firm. It’s up to the government to make this happen because it has to be coordinated.

However, what we show is that a coordinated effort to solve this problem has significant benefits. If the government is considering pollution reduction policies, these benefits should be taken into consideration.

There is a way in which the government could convince firms to do something about this that would be in their interest. If a firm takes steps to reduce their employees’ exposure to pollution, that will likely help their productivity. For example, if the firm installs filters to clean the air inside a factory, this will increase the firm’s productivity. Similarly, if workers are missing work because their children are getting sick from pollution exposure, then cleaning the air in the school could also help productivity. That’s the argument you could win because it’s in the firm’s interest. Unfortunately, however, these steps won’t reduce the pollution itself.

I hope our estimates can help the government make the case to society as a whole that cleaning up pollution offers not only health benefits but benefits in manufacturing output as well.

Q. Given that automation has been displacing human labor from assembly lines and AI may further replace white-collar workers in offices, do you think the costs of air pollution will be lower in the future?

A. If a factory could be automated, its output of course wouldn’t be affected much by air pollution. There would be very little labor to be affected. That is true. But then the question you have to ask yourself is: these workers who used to be in the factory, what are they doing now? For example, if they all go to service industries, because manufacturing is all automated, this raises the question of what effect air pollution will have on service industries. We didn’t estimate that because we only have data from manufacturing firms. If service industries are affected, there is still going to be productivity loss. It will just be shifted to the service sector.

If you’re asking whether we expect to find effects on the service sector, I would guess that we would. Whether these effects are higher or lower than for manufacturing, I would not want to speculate. But certainly I would expect to find effects, because a lot of service jobs are done outdoors or in indoor settings that are not protected from pollutants.

Another alternative for the displaced workers is that they will still be involved in manufacturing but doing jobs like controlling the machines and coordinating things. So, the question is how they will be affected. That’s a harder question. We do compare high-tech and low-tech manufacturing firms and find similar outcomes, which suggests that there would be effects.

Q. The data used in the study is from 2007. What would change and what would not if you used more recent data?

A. Obviously it would be good to have more recent data but we don’t have it. To answer your question, I don’t think the fundamental relationship that we found between pollution and productivity would change dramatically. The main way I guess it might change somewhat would be, if firms have taken a lot of steps since 2007 to reduce their employees’ exposure, you might see the effects get smaller.

Q. Based on your study, what do you think we should do about air pollution?

A. Obviously one direct way to reduce air pollution is to shut down factories. This will of course lead to a productivity boost due to less pollution. But we’ll also have less output because we shut the factories down. And obviously the effects of the latter will outweigh the former. So, I wouldn’t recommend the government shut down factories just to reduce pollution. Of course, if a factory is to be shut down for other reasons, like violating environmental regulations or being inefficient, there will be a productivity boost. My point is that, in the cost-benefit analysis of whether these actions should be taken, this productivity boost should be considered.

There are other ways to reduce pollution, such as installing abatement equipment that extract pollution before it leaks into the air. That’s a way to get a productivity boost. There are also non-factory sources of pollution like automobiles, so if you have cleaner automobiles that will reduce pollution and boost productivity. Road dust actually produces a lot of pollution. If you see these trucks that spray the streets of Beijing with water, it’s not just cleaning the street but also reducing dust and therefore particulates. These are the main ways of reducing air pollution. Of course, all of these cost money, so that doesn’t necessarily mean they should all be done. But the productivity benefits should be considered in these calculations.

You may also like

The Road to Recovery

Economist and author George Magnus, looks at China’s falling birth rate and the future of the economy post COVID-19 George.

by Mable-Ann Chang | Oct. 27 2021

Machine Learning Shows Us the True Value of Data

Sun Baohong, Professor of Marketing at CKGSB, looks at the vital importance of machine learning to brand positioning and understanding consumers Baohong.

by | Oct. 27 2021

A Counter Narrative

Kevin Rudd, former Prime Minister of Australia, discusses the cost of protectionism and foresees trade as the major battlefield of the future.

by Mable-Ann Chang | Oct. 11 2021

Coming of Age

Key Opinion Leaders are becoming the vital link for brands to reach online consumers in China.

by SHI Weijun | Oct. 11 2021