How Economists and Non-Economists Talk Past Each Other
One particular assumption – ceteris paribus (holding all else constant) – is the culprit.
Thank you for reading our work! If you are not a subscriber, you can subscribe to our newsletter for free below to stay-up-to-date with Nominal News directly in your inbox:
If you would like to support us further with reaching our 1,000 subscriber goal by year end, please consider sharing and liking this article!
People related to economics often, unfortunately, like to ‘dunk’ on other people’s views and arguments. Often, however, this disagreement in views is caused by simply talking about different things caused by one assumption – ceteris paribus, or holding all else constant.
What is the Point of Economics
Economics, as a subject matter, focuses mainly on causality – that is, how will X impact Y. Many economic questions follow this pattern – if we build a subway line, how will this subway line impact home values; if we ban non-competes, how will wages respond; if we expand the child tax credit, how will children’s educational attainment change; if we reduce the speed limit on a road, how will the number of accidents change.
These causality questions differ from predictive questions, which are what will Y be in the future (and X is just one of many data points). For example, in causality questions, we will determine that reducing the speed limit will reduce the number of accidents on a road by 10. In predictive questions, we will want to predict the number of accidents. In our example, after the change in speed limit, we would determine that in the next 12 months there will be 60 accidents.
There are many things that will impact the accident number over the course of 12 months. Some of the change in accidents will be attributed to the implementation of the new speed limit (the above causality question), and some of it will be impacted by other factors, such as the overall driving culture, population growth or road quality.
Establishing Causality
Causality questions are focused on isolating the impact of only the change in X on Y. To isolate the impact of X, economists use research methods that have to control for all other factors that may be influencing both X and Y simultaneously over a period of time. For example, in the speed limit question, in order to determine the impact of changing the speed limit, we would also need to adjust for changes in the quality of the road or whether the police patrols more prior to and after the change in the speed limit, as these changes can influence the number of accidents significantly.
Once we control for all other factors, we are able to establish causality of X on Y – the impact of a changing the speed limit on the number of accidents. When presenting this impact, economists often add the phrase “ceteris paribus” or “holding all else constant”. For example, reducing the speed limit will reduce accidents by 4%, holding all else constant This assumption is critical. The “holding all else constant” assumption means that all other variables, such as road quality and police patrols (as mentioned above) stay at the same level.
This assumption is needed, because without it, we might be capturing the effect of another variable.1 Suppose, changing the speed limit also increases the number of police patrols. If police patrols also reduce the number of accidents, let's say by 2%, then if we do not hold police patrols constant, the impact of changing the speed limit might seem greater, at 6% speed reduction (4% + 2%). But this would be an overstatement of the impact of changing speed limits, even though it would be a better prediction of what might happen!
What if the opposite happens? That is with the change of speed limits, the number of police patrols decreases, causing accidents to go up by 5%. Then, when taking everything into account, the reduction in the speed limit might increase accidents, even though ceteris paribus, it decreases them.
So if one were to say that reducing speed limits increases accidents in this hypothetical example, although from a technical sense they are incorrect, they are giving the correct answer, as it captures the total effect of what will happen. Both statements can be seen as correct, depending on how we interpret the statement “does reducing speed limits increase accidents”. This whole confusion could be avoided, however, if we openly state all the assumptions we're making in the question (especially the ceteris paribus one).
Example of Housing Prices
This issue of talking past each other often occurs in questions where there are many variables that influence the outcome. One such case (discussed previously here at Nominal News) is the commonly cited example of the impact of new construction on house prices. Below is a chart from the Financial Times piece showing that people incorrectly predict that house prices will rise with more construction:
Now, ceteris paribus, housing costs will fall when housing supply rises. But what if we relax the ceteris paribus assumption and assume other things might change also? If the reason for additional housing construction is because of local economic improvement, incomes may be rising, as well as the number of people that want to live in the area. This will push housing prices up. If the latter price increase is larger than the price decrease due to new supply, then a person that sees new construction could be correctly predicting that housing prices will go up.
To reiterate, housing construction did not cause the increase in prices. But housing construction acts as a signal about the local economy. Since the local economy is doing better, this causes housing prices to go up. Moreover, new housing construction could also make the local region more productive, which further improves income and again increases housing prices.
The discrepancy discussed by the Financial Times thus can be explained by how people interpret the question they are being asked. If you think you are being asked about causality (i.e. ceteris paribus) then you should answer that housing prices will fall. If you think you are being asked about what housing costs will look like in the next 12-24 months, then using the fact that a city is seeing major development (since the housing supply is increasing), it appears reasonable to assume housing costs will go up in the booming economy.
Medicine – Another Case
A similar situation arose during the COVID-19 pandemic. At the height of the pandemic, hospitalization rates were quite high. Severe cases required patients to be put on ventilators (a machine that assisted with breathing). Unfortunately, many patients placed on ventilators did not survive. Soon, theories emerged that ventilators were causing the deaths of people with COVID-19, as the death of ventilated COVID-19 patients seemed higher than in other cases.
Naturally, the ventilators were not causing deaths. Ceteris paribus, they were reducing the number of deaths from COVID-19. However, it became clearer that once a patient was placed on a ventilator, the patient’s probability of survival was worse than first thought. Moreover, the ventilator, as a medical procedure, was not saving as many lives for COVID-19 patients, as it did for other diseases.
Doctors realized that, even though the ventilators were not causing bad patient outcomes, given the ventilators had a lower efficacy, it was better to try other medical interventions to improve health outcomes, before being placed on the ventilator.
Similarly to the other examples, the difference in opinion about what having a patient placed on a ventilator does to survivability occurs because, in one instance, we are thinking with the ceteris paribus assumption (i.e. ventilators improve health outcomes holding all else constant), while in another instance, we are thinking about predicting what will happen once a patient is placed on a ventilator. Both insights are valuable.
Crowd Wisdom – Don't Dismiss It
Crowd opinions are often ignored and outright dismissed by many economic commentators. Often the discrepancy between the ‘expert’ statement and the crowd statement can be attributed to simply thinking about the question differently. The ceteris paribus assumption is one such example. It is an assumption that is counterintuitive to natural human thinking, as humans think in patterns and correlations. These patterns that we notice can actually convey a lot of deep insights even if we might not realize them at first. The ventilator case above is one such pattern where doctors’ noticed that something was off in the approach to dealing with COVID-19.2
Crowd wisdom should be taken into account and looked into. A deeper dive can lead to many valuable insights. Crowd wisdom actually influenced my dissertation topic, as my advisor and I discussed why, in 2016, there was a strong anti-trade sentiment in the US, which influenced the elections. Today, economic research understands that there can be quite a lot of people that lose out from trade, while previously, it was nearly universally accepted by the economics profession that opening up to trade is welfare improving for everyone.
Interesting Reads from the Week
- discusses new software under development by Microsoft that has significant privacy concerns.
- talks about how studies (some of which are in progress) have shown that a social safety net does not discourage work and risk taking. On the contrary, it improves health outcomes and increases job opportunities.
- tackles some of the arguments against student loan forgiveness. A snippet: “Using lifetime earnings forecasts as an approach for deciding whether someone currently needs financial help is suspect, too: it’s like claiming JK Rowling never needed government assistance when she was penniless because someday she’d publish Harry Potter.”
If you enjoyed this article, you may also enjoy the following ones from Nominal News:
Community Civilian Interventions (June 12, 2023) – cities have always been interested in improving safety in neighborhoods. Civilian government interventions have been proposed as a solution instead of police. However, civilian interventions work best if local government presence is strong.
When Free Trade isn’t Free (February 20, 2023) – how trade issues extend beyond tariffs and quotas and into regulations.
Discrimination in Real Estate (September 17, 2023) – unlike explicit discrimination, statistical discrimination is less talked about. However, it has significant impacts on outcomes for people impacted by it. Real estate agents, often unaware of their own statistical discrimination, end up perpetuating outcomes for minorities.
A simpler example: the Law of Demand states that if prices rise, quantity demanded falls, ceteris paribus. This Law can fail without ceteris paribus. Suppose, a bar increases the price of beer. Accordingly, the number of beers sold should fall, ceteris paribus. However, if at the same time, the bar offers a discount on food items, more people might go to the bar and thus end up ordering more beers. That is because bar food and beers are complements. Without ceteris paribus, in this example, the sale of beer goes up when the price of beer goes up.
The ceteris paribus assumption is also something economists themselves challenge, as it can be an assumption that influences model outcomes significantly. Bierens and Swanson (2000) show that how we impose this assumption (i.e. what state of the world we hold constant – what GDP level, what inflation rate, etc.) can impact our understanding how economic variables interact with each other.
The trade opening was a perfect example of ceteris non paribus. Large fiscal deficits led to over valuation of the dollar and large trade deficits, especially of manufactured goods especially concentrated in the Midwest!