Modelling the Impact of Tariffs

Economic modelling suggests that the US tariffs can have quite significant adverse outcomes on both US output and wages.

Thank you for reading our work! Nominal News is an email newsletter read by over 4,000 readers that focuses on the application of economic research on current issues. Subscribe for free to stay-up-to-date with Nominal News directly in your inbox:

Our updated goal for 2025 is to hit 10,000 subscribers:

If you would like to support us further with reaching our subscriber goal, please consider sharing and liking this article!

---

Photo by Markus Winkler on Unsplash

The recent imposition of tariffs by the US has created significant market turmoil and debate about the impacts on the economy. Naturally, several economists have quickly produced research papers to estimate the potential impact of these tariffs. We will discuss two papers.

Quick Note – Predictions vs Causality

Before we dive into the papers, I’d like to emphasize that economics is generally a study that focuses on ‘causality’ type questions. In our current context, this would be focusing on the question of what impact will an X% tariff have on either GDP (Gross Domestic Output) or wages. These questions are a bit different from predictions, which focus on what will the GDP be next quarter or what will wages be next month. Predicting the value of a particular variable would require far more data, while the models used for such prediction often do not lend themselves to easy interpretation.¹

The focus on causality by economists means that conclusions need to be interpreted very specifically. That is, if economists establish that a policy will cause GDP to fall by 1%, but actual GDP rises by 2%, that does not mean that economists were wrong. The economist's conclusion implies that had this policy not been implemented, GDP would have grown by 3%. 

New Keynesian Model

Kalemli-Ozcan, Soylu and Yıldırım (2025) (“KSY”) went about modeling what the impact of the US tariffs would be on the US and other countries. KSY used a New-Keynesian model for this purpose. At Nominal News, we have discussed this model often, as it is a ‘workhorse’ (foundational) model in economics. The key tenet of the New-Keynesian model is the assumption that prices (this includes wages) are ‘sticky’ – i.e. nominal prices (stated prices) do not adjust instantaneously. This assumption implies, for example, that firms cannot immediately lower prices during a recession, resulting in lower sales. Similarly, workers during a recession are also unable to work for lower wages (wages are ‘stuck’), which means that as demand for a firm's product falls, the firm will lay people off. 

Model ‘Complexity’ and Economics

KSY also added global trade to their model, allowing firms to buy intermediate inputs from other countries. The KSY model is quite computationally complex, as it has 24 equations that need to be solved. The model contains variables such as how much labor each worker chooses to provide, how much each worker chooses to save, how much labor each firm should hire and how many goods each firm should import. Moreover, a central bank is also modeled as an entity that sets interest rates in order to ensure inflation remains stable (usually assumed to be 2%). Each equation also needs to hold for all countries.

This makes the model difficult to solve, or more precisely, very ‘expensive’ to solve. By expensive, I mean that it would require a lot of time and computer processing power to solve this model. Typically, these types of models would be solved by ‘guessing’ an answer – i.e. we guess how much each worker works and saves, how much each firm imports and hires, etc. The solution is a guess in which all the above 24 equations hold.

To simplify the solution finding process, KSY used certain modeling assumptions and mathematical approaches that allowed for simpler solution methods. It is worth pointing out that critics of economics papers often point out that certain modeling ‘assumptions’ are unrealistic, which suggests that the whole model (if not the academic field) is worthless. The truth is that often these assumptions are necessary to have any result at all, as without them, solving such models would be currently impossible.

Moreover, the more elements a model has that aim to capture the real world, the more assumptions economists need to make. Thus, economists often have a trade-off – model more components of the world (for example, whether to allow workers to choose how much labor to supply; whether to allow individuals to save; whether to allow trade between countries; or whether to allow for monetary policy), or model each specific component more accurately (i.e. focus only on worker’s labor decisions such as whether to accept a job offer or wait for a better one). 

This trade off is always on the mind of economists, which is why it is often important to understand what the economists were trying to specifically answer with a model they designed.

Testing the Model

Turning back to the KSY model, KSY looked at the previous time the US imposed significant tariffs, which was in 2018 on China. Empirical studies of the 2018 tariffs found that the tariffs increased Personal Consumption Expenditure (PCE) measured inflation by 0.1-0.2 percentage points (Barbiero and Stein, 2025). At the same time, Gross Domestic Product (GDP) (US output) was estimated to be 0.4% lower (Fajgelbaum, Goldberg, Kennedy and Khandelwal, 2020). 

KSY’s model found similar results when modelling the US tariffs, albeit with a slightly lower impact – PCE inflation was only 0.07 percentage points higher, while output (GDP) fell by 0.2%. However, these results suggest that the model appears to give us a reasonable estimate of the impacts of tariffs. 

Results

To test the impacts of the tariffs announced on April 2, 2025 (Note: Since that date, the tariff schedule has been altered with some countries receiving lower tariffs, while others having much higher tariffs), KSY ran the model with the following tariffs:

• 20% on Europe;

• 34% on China;

• 25% on Mexico;

• 25% on Canada;

• 10% on the Rest of the World.

The main outcomes of the increase in tariffs (pp stands for percentage points):

The column on the left shows the outcomes to the US economy if other countries do not ‘retaliate’ by raising their own import tariffs, while the right column shows if they do.

It is worth adding that the tariffs also impact other countries. However, as you can see in the table below, the adverse effects on the US are greater:

As we have discussed before, tariffs can end up both reducing output and increasing inflation. This is because tariffs on intermediate goods increase the marginal cost of production, which pushes inflation up, while at the same time reducing overall production efficiency. This creates the unique situation of falling output and increasing inflation. Typically, these two variables move in the same direction.

Tariff Threats

KSY also undertook a hypothetical scenario where instead of imposing tariffs, the US only threatens to impose them on a specific date. In this case, the US announces that tariffs will be imposed at some future date, but on the actual date, the tariffs are cancelled. In this situation, all economic agents act as if tariffs will be introduced. In anticipation of the tariff announcements, individuals expect to consume fewer goods in the future due to higher prices. This means that individuals need to save more today in order to be able to consume more in the future when prices are higher, resulting in a fall in spending prior to the tariff implementation date.

Now, if the tariffs are not implemented on the day, people re-adjust their spending-saving decision, as imported goods will now no longer be more expensive. Overall, in the case that tariffs do not get implemented, the economy has a lower output, by 0.7% of GDP, but also a 0.6 percentage point fall in inflation (i.e. deflation). The reason for this is due to the anticipatory behavior of individuals – the announcements of tariffs results in people cutting back on consumption.

Trade Impacts

Ignatenko, Macedoni, Lashkaripour and Simonovska (2025) (“IMCS”) also looked at US tariffs to predict the economic outcomes. Unlike the KSY model, IMCS focused on modeling trade between 123 countries, whereas KSY focused only on several countries (with most grouped as ‘Rest of World’). This required IMCS to simplify certain parts of the model – it includes only one production sector in each country, and there is no ability to save by workers (workers consume all their income). Using cross-border trade data, IMCS simulate what will happen once the US imposes tariffs. 

Retaliation vs non-Retaliation

If no country retaliates by increasing their own tariffs, IMCS find that the trade deficit, as a percentage of GDP, falls. However, this is due to the fact that trade collapses, as both imports and exports fall by about 30-35%. As we mentioned in our previous article, the fact that imports and exports fall by a similar amount isn’t surprising – via the Lerner Symmetry, import tariffs are more or less equivalent to export taxes. 

Regarding overall welfare and wages, interestingly IMCS find different results from KSY. By imposing the tariffs, when other countries do not retaliate, it appears the US benefits with higher real wages and higher welfare by 1.1%. In this model, the reason for this outcome is that tariffs strengthen the bargaining position of US workers (as there is higher demand for US domestic labor) and weaken the position of foreign workers (as there is less demand from the US for products, workers abroad must settle for lower wages). Foreign countries see a 0.26% decline in welfare. 

However, these results change if foreign countries retaliate or, more accurately, respond optimally to US imposed tariffs. If foreign countries retaliate by imposing optimal tariffs from their perspective, US welfare falls by 1%, while foreign countries only see a 0.05% welfare decline. This tells us that under the ICMS model, imposing retaliatory tariffs is an optimal strategy.

Lastly, IMCS also looked at the revenues the tariffs could generate for the US government. If other countries do not impose their own tariffs, the tariffs can at most generate 1.5% of US GDP, which is approximately $400bln, or approximately 6% of the US Federal budget. If retaliation occurs, this number is cut in half.

Limitations, Modeling and Economics

So which model is right? Since both models have somewhat opposing outcomes and even opposing policy prescriptions (KSY would tell foreign countries to not retaliate as they are better off without retaliation², while IMCS suggests that foreign countries should impose their own tariffs on the US), this is an important question.

I will preface this by stating that I do not have 100% certainty on which model is better, but I will elaborate on what I look at when ‘picking’ models. Firstly, as mentioned earlier, economists are limited by computational power and cannot build a model of everything. Thus, each of the two models we discussed today focused on different issues. The KSY model was built to study inflation dynamics in the world with multiple sectors and trade. On the other hand, the IMCS model focused more on capturing realistic trade flows between countries, and did not include any direct role for monetary policy. 

Thus, in terms of output and inflation outcomes, I would assume that the KSY model will more closely reflect the true tariff impacts. For changes in import and export volumes, the IMCS model is more likely to be accurate. Interestingly, the KSY authors state in their paper that not modeling for multiple sectors (the IMCS paper did not have multiple sectors) can result in underestimating the negative impact of tariffs on GDP. 

Ultimately, whichever way the tariff policy is looked at, the costs of this policy are significant, far outweighing any potential benefits. Any goals tariff proponents desire, can be achieved in far better ways. 

P.S. Economic Modeling 

One element I wanted to elaborate on with today's article is how economics approaches modeling. The two papers showed how the same broad policy (tariffs) analyzed from disparate angles can give differing perspectives. Moreover, these differences are often a direct effect modeling assumptions which can generate specific dynamics. For example, if you would like a model that gives you realistic (i.e. similar to data) behavior of inflation you may need to have the New Keynesian assumption of sticky prices. 

As one works more and more with economics models, one can often quickly know the outcomes that come from a model based on the assumptions the model has. Having witnessed this as a graduate student, it can be quite impressive to see an economist at a seminar immediately asking valid questions about model outcomes while the presenter is still on their first slide.

1. For example, machine learning algorithms can be good at predictions but interpreting what specific variable caused what outcome is very complex, if not impossible.

2. For example, without retaliating, GDP in Europe would actually increase by 0.1% rather than falling by -0.3%. Similarly for China, GDP without retaliation falls only by 0.07% instead of falling by 0.12%.