Assessing tariff pass-through to consumer prices in Canada: Lessons from 2018

Introduction

In response to recent tariffs imposed on Canada by the United States, the Canadian government has introduced counter-tariffs on certain goods imported from the United States.1 A key issue for inflation is the need to understand how these counter-tariffs may impact final consumer prices.

Pinning down the precise timing and magnitude of any tariff pass-through to consumer prices is difficult. This is because:

• the tariff may be absorbed—in part or entirely—by the importer
• historical precedent is limited
• the aggregate consumer price index (CPI) typically used to study consumer prices is not well suited to isolating the impact of tariffs from other economic drivers

To help fill this knowledge gap, we examine a previous tariff episode and use detailed consumer price microdata. In May 2018, the United States imposed tariffs on Canadian steel and aluminum. In response, in July 2018, Canada imposed counter-tariffs on US steel, aluminum and some final goods.2 The list of tariffed items was wide-ranging—from grocery products and household items to major appliances—but comprised less than 1% of the overall CPI basket. The tariff rate on steel was 25% and the rate on aluminum and final goods was 10%. This trade dispute lasted just less than a year. In May 2019, the United States lifted its tariffs, and Canada lifted its counter-tariffs in turn.

Using the synthetic control method (SCM), we find that Canada’s 2018 counter-tariffs resulted in, on average, high but incomplete pass-through of tariffs to consumer goods prices. That said, the magnitude and speed of the price increases varied considerably across tariffed consumer goods.  

Insights from this analysis helped inform the Bank of Canada’s projection scenarios in the April 2025 Monetary Policy Report. These scenarios assume that 75% of the increased costs from tariffs is passed on to consumer prices within six quarters. While many factors are different now compared with 2018—including the scope and magnitude of tariffs—our methodology nonetheless provides a useful anchor for evaluating the impact of tariffs on consumer prices. Going forward, this methodology will be leveraged to monitor the impact of new tariffs on consumer prices in real time.3

Data and methodology

We obtain our estimates of tariff pass-through in three steps:

1. First, we use consumer price microdata to construct price indexes for tariffed and non-tariffed items.
2. We then use SCM to estimate the impact of tariffs on the prices of tariffed items.
3. Finally, to obtain estimates of tariff pass-through for each item, we scale these impacts by the 2018 10% tariff rate and by estimates of the share of Canadian final consumption goods that are directly imported from the United States.

Disaggregated price indexes of representative products

We use consumer price microdata from Statistics Canada in our analysis. These data include prices and metadata for a sample of goods and services of unchanged quantity and quality used to construct the CPI.4 The data consist of around 100,000 product price quotes each month, beginning in February 1998. For our analysis, we use a subset of the microdata that comprises only goods (excluding energy) for which monthly price data are available from January 2010 to December 2019.

In the consumer price microdata, we identify tariffed and non-tariffed goods during the 2018 trade war at the representative product (RP) level. To do this, we manually construct a mapping of tariffs (which are applied to imported goods using Harmonized System [HS] codes) to RPs in the CPI.

In the microdata, each RP contains from at least 10 to hundreds of monthly individual product price quotes, which we use to construct item price indexes. Our final dataset consists of price indexes for more than 300 RPs, of which 37 were tariffed. A list of these RPs is available in Appendix A.

Estimating price impacts using the synthetic control method

To estimate the impact of tariffs on prices, we use SCM—a data-driven approach commonly employed to evaluate the effects of major policy interventions over time.5 In this context, the goal is to assess what would have happened to the prices of tariffed RPs had the tariffs not been applied.

Since this counterfactual cannot be directly observed, SCM constructs a “synthetic” version of each tariffed RP—a weighted combination of similar but non-tariffed RPs—such that their price trends closely match those of the tariffed RP in the pre-tariff period. If this synthetic control successfully mirrors the pre-tariff price path, then any post-tariff divergence between the actual and synthetic prices provides an estimate of the tariff's effect.

In practice, we determine the weights of the synthetic controls using least squares. We regress the log price of each tariffed RP on the log prices of its five most highly correlated non-tariffed RPs over a pre-tariff sample (January 2010 to June 2018).6 We then use the estimated regression coefficients to generate a predicted synthetic counterfactual price path for each tariffed RP over the post-tariff period (July 2018 to December 2019).7

The impact of tariffs on prices is therefore the actual price minus the synthetic price:

\(\displaystyle\,Price\,impact_{i,t}\) \(\displaystyle=\,log(P_{actual\, i,t})\) \(\displaystyle-\,log(P_{synthetic\, i,t})\)

for representative product \(i\) and time period \(t\).

Crucially, we are unable to identify the country of origin of specific products in the consumer price microdata; we know only their representative product name. Therefore, we cannot implement SCM to compare the price path of a product imported from the United States and subject to counter-tariffs with the price path of the same product imported from elsewhere. As an example, orange juice is an RP in the CPI data that was subject to counter-tariffs, but within this narrow category we cannot distinguish tariffed (i.e., made in the United States) versus non-tariffed (i.e., made elsewhere) orange juice. However, we can distinguish between orange juice and apple juice, another RP in the CPI data that was not subject to tariffs regardless of its country of origin. This means our estimated price impacts could include some indirect effects (e.g., demand substitution or opportunistic price increases on orange juice not made in the United States).

Assessing tariff pass-through

Finally, to convert our estimated price impacts for each RP into tariff pass-through, we need to scale the price impacts by the 10% tariff rate in 2018 and by the share of Canadian final consumption goods that are imported from the United States:

\(\displaystyle\, Passthrough_{i,t}\) \(\displaystyle=\, \frac{Price\,impact_{i,t}} {Tariff\,\,rate * (\frac{US\,imports} {Consumption})_{i,\,2017}}\) \(\displaystyle\, .\)

Scaling by the US import share of consumption is necessary because our price impact estimates for RPs that are affected by tariffs include impacts for both US- and non-US-made individual products within those RPs.

Due to data limitations, measuring the US import share of consumption at the disaggregated level is an imprecise exercise that requires manual mapping of tariffed RPs between several different goods classification systems. We take the 2017 US import share of total imports (based on 6-digit HS codes) from Statistics Canada’s merchandise trade data and multiply it by the share of total supply in the Canadian economy that is directly imported from abroad (based on detailed North American Product Classification System [NAPCS] codes) from Statistics Canada’s 2017 input-output tables. We assume that the import share of total supply and the import share of consumption are equivalent.

When presenting aggregated results, we show a range of estimates of tariff pass-through. We do so because in some cases, issues with imperfect concordance between RPs, HS codes and NAPCS codes or potential confounding factors may be causing our pass-through estimates to be unreasonably large at the RP level. Our range of estimates is based on various treatments of outlier estimates (in addition to the unadjusted estimates). These treatments include removing tariffed RPs with outlier pass-through estimates and capping RP-level pass-through estimates at thresholds that we deem to be unrealistic (e.g., pass-through greater than 100% and greater than 200%).

Results: Price impacts

Estimated price impacts vary substantially at the RP level. That said, we can broadly categorize tariffed RPs as having experienced temporary, persistent or no price impacts. Chart 1 gives an example of each, showing the RPs’ actual and synthetic (counterfactual) price paths.

• In around 30% of cases, we find that RP prices respond quickly but only temporarily to tariffs, and that price impacts reversed after the tariffs were lifted (as seen with plastic wrap in Chart 1, panel a). These cases are most consistent with the economic theory that the imposition and removal of tariffs should each constitute one-time offsetting price shocks.
• In around 40% of cases, we find a more persistent impact that does not fully reverse following the end of tariffs (as seen with facial tissues in Chart 1, panel b). These results point to the possible presence of downward price rigidities, though confounding factors that are difficult to control for could also be at play. Confounding factors are especially likely for specific RPs such as carbonated soft drinks, which were subject to direct tariffs as well as tariffs on aluminum on both sides of the border. Moreover, many countries from which Canada imports soft drinks imposed sugar taxes in the 2018–19 period.
• In the remaining 30% of cases, we find no significant impact of tariffs on RP prices (as seen with pickles in Chart 1, panel c).

Appendix B presents the full set of charts showing the synthetic control estimates for each RP.

Aggregating the heterogeneous price impacts from the RP level up, we find that the 2018 tariffs increased prices of affected RPs (which, as detailed in the previous section, may include both tariffed and non-tariffed products within affected RPs) by an average of 2.5% within six quarters (Chart 2).

Results: Assessing tariff pass-through

Based on the midpoint of the range of estimates discussed in the Data and methodology section, average tariff pass-through in 2018 is estimated to be around 60% after six quarters across all tariffed RPs (Chart 3).

Chart 3: Tariff pass-through by major category

Chart 3: Tariff pass-through by major category

Range and midpoint estimates per quarter

Note: Shaded areas represent the range of estimates; solid lines represent the estimated midpoints.

Pass-through was slightly higher for food in stores (around 70%), slightly lower for durable goods (around 50%) and around 60% for semi- and non-durable goods. Pass-through for food items also appeared to spike slightly earlier than that for non-food items, though all categories saw peak pass-through within six quarters.

• The range of estimates for pass-through to food is large. This is because many RPs in this category had unusually high pass-through estimates and were therefore trimmed or adjusted as outliers when we constructed the range of estimates on aggregate. That said, higher tariff pass-through for food items is consistent with generally high turnover of product, low margins and low inventory in this category relative to other goods.
• In contrast, there are fewer outlier pass-through estimates for items excluding food, and so the range of estimates for this category—and especially durable goods—is smaller.

However, tariff pass-through may vary relative to the tariff rate. For example, some producers may have been able to absorb a 10% tariff rate without adjusting prices but would not be able to absorb the 25% tariff rate that is presently in effect on some goods.

Considering this potential nonlinearity, we also estimate average tariff pass-through to consumer prices for only RPs that saw positive (i.e., greater than zero) and significant price impacts in 2018. In other words, for this conditional pass-through, we remove from our average pass-through calculations RPs that saw no significant price impacts.8

• Average pass-through for this subset of RPs is around 80%—around 20 percentage points higher than the midpoint of the unconditional pass-through estimates (when all RPs are included in the average pass-through calculations) (Chart 4).
• Conditional pass-through is around 100% for food, around 70% for durable goods and around 90% for semi- and non-durable goods. These pass-through estimates are all also around 20 to 30 percentage points higher than their unconditional counterpart estimates (see Appendix C).
  • As is the case with the unconditional estimates, the range of conditional estimates is far larger for food and smaller for durable goods.

Conclusion

Our analysis of Canadian counter-tariffs on the United States in 2018–19 shows that the impact of tariffs on consumer prices varied widely across different goods. On average, tariff pass-through was high, though incomplete, and relatively fast. These findings suggest that present and future tariffs could also lead to substantial upward pressure on consumer prices, especially given the potential for higher tariff rates.

This analysis is only one part of Bank staff’s efforts to fully understand the implications of trade policy changes for the Canadian economy. Our results are broadly in line with insights from other work, including ongoing consultations with businesses and households.9 Staff will continue to monitor trade policy developments and leverage the highly disaggregated consumer price microdata to monitor the impact of new tariffs on consumer prices.

Appendix

References

Abadie, A. 2021. “Using Synthetic Controls: Feasibility, Data Requirements, and Methodological Aspects.” Journal of Economic Literature 59 (2): 391–425.

Abadie, A., A. Diamond and J. Hainmueller. 2010. “Synthetic Control Methods for Comparative Case Studies: Estimating the Effect of California’s Tobacco Control Program.” Journal of the American Statistical Association 105 (490): 493–505.

Abadie, A., A. Diamond and J. Hainmueller. 2015. “Comparative Politics and the Synthetic Control Method.” American Journal of Political Science 59 (2): 495–510.

Abadie, A. and J. Gardeazabal. 2003. “The Economic Costs of Conflict: A Case Study of the Basque Country.” American Economic Review 93 (1): 113–132.

Cavallo, A., P. Llamas and F. Vazquez. 2025. “Tracking the Short-Run Price Impact of U.S. Tariffs.” Harvard Business School Working Paper.

Bilyk, O., M. Khan and O. Kostyshyna. 2024. “Pricing Behaviour and Inflation During the COVID-19 Pandemic: Insights from Consumer Prices Microdata.” Bank of Canada Staff Analytical Note No. 2024-6.

Endnotes