Figure 1: Average Tariff Rate and Tariff Exposure
Notes: This figure shows the monthly average tariff rate
(left panel) and tariff exposure (right panel) over the sample period
for products in the Numerator panel. For each year-month, we calculate
the mean across all observations. The tariff rate is the ratio of duties
collected to import value (Equation 1). Tariff
exposure multiplies the tariff rate by import penetration (Equation 3). Both
measures are in percent. The red dashed vertical line marks February
2025, when tariffs were announced.
Two-panel line chart showing monthly averages from January 2024 through December 2025. The left panel plots the tariff rate in percent; the right panel plots tariff exposure in percent. A red dashed vertical line marks February 2025 in both panels.
Left panel (Tariff Rate): The series is flat at approximately 2 percent from January 2024 through January 2025. It rises sharply beginning in February 2025, reaching roughly 10 percent by mid-2025, with some month-to-month variation. The series peaks at approximately 12 percent around mid-2025 before settling near 10 percent by December 2025.
Right panel (Tariff Exposure): The series is flat at approximately 0.011 percent from January 2024 through January 2025. It rises sharply after February 2025, climbing to roughly 0.050 percent by April 2025 and continuing to rise to approximately 0.075 percent by mid-2025. The series fluctuates between 0.050 and 0.080 percent through December 2025.
Figure 2: Event Study: Price, Spending, and Quantity Responses to Tariffs
Notes: The figure plots coefficients from event study
versions of Equations (4) and (5) run on Numerator
panel data, interacting the treatment variable with year-month
indicators (January 2025 omitted as reference). The top row uses the
tariff rate; the bottom row uses the tariff exposure measure. Left
panels show price (log Fisher Price Index); right panels show quantity
and spending (both in logs and spending is real spending). All
specifications include year-month and panel-unit-by-calendar-month
interaction fixed effects, with panel-specific linear pre-trends in the
price equation. Shaded areas indicate 95% confidence intervals based on
standard errors clustered at the HS4 code level. The vertical dashed
line marks February 2025, the tariff implementation date.
Four-panel chart arranged in a 2x2 grid. The top row uses the tariff rate specification; the bottom row uses the tariff exposure specification. Left panels show price responses (log Fisher Price Index); right panels show spending and quantity responses (both in logs). A vertical dashed line marks February 2025 in all panels. Shaded areas indicate 95 percent confidence intervals.
Top-left (Price, Tariff Rate): Coefficients fluctuate around zero from January 2024 through January 2025 with wide confidence intervals. After February 2025, coefficients turn positive and rise gradually, reaching approximately 0.2 to 0.3 percentage points by late 2025.
Top-right (Spending and Quantity, Tariff Rate): Pre-period coefficients for both spending and quantity are centered near zero. After February 2025, both series shift sharply negative beginning around June-July 2025, reaching approximately negative 0.7 to negative 0.8 percentage points by late 2025. Spending and quantity track each other closely.
Bottom-left (Price, Tariff Exposure): Similar pattern to the top-left but with larger magnitudes and wider confidence intervals. Pre-period coefficients fluctuate around zero. Post-period coefficients rise to approximately 1 percentage point by late 2025.
Bottom-right (Spending and Quantity, Tariff Exposure): Pre-period coefficients are near zero with wide confidence intervals. Post-period coefficients shift sharply negative, with spending and quantity both reaching approximately negative 1.5 percentage points by late 2025.
Figure 3: Average Spending and Quantity Responses: Stockpiling versus Sustained
Response
Notes: The figure reports average monthly responses of log
real spending and log quantity to tariffs, separately for a three-month
transition window (February–April 2025) and a steady-state window
covering the remainder of the sample period (May–December 2025).
Estimates are constructed as linear combinations of event-study
coefficients from Equations (4) and (5), averaging the
coefficients within each window. Standard errors are computed from the
full variance-covariance matrix of the estimated event-study
coefficients to account for correlation across event-time dummies, and
are clustered at the HS4 code level. The transition window captures any
anticipatory or stockpiling behavior that would unwind shortly after
tariff implementation; the steady-state window captures the sustained
response once such dynamics have dissipated. Horizontal bars report 95%
confidence intervals. Rows correspond to outcomes (spending, quantity);
columns correspond to treatment definitions (tariff rate, tariff
exposure).
Forest plot with two columns (Tariff Rate on the left, Tariff Exposure on the right) and two rows of outcomes (Spending on top, Quantity on bottom). Each panel shows two horizontal point-and-interval estimates: one for the Transition window (February-April 2025) and one for the Steady-state window (May-December 2025). Horizontal bars report 95 percent confidence intervals.
Tariff Rate column, Spending: The transition-window average is approximately negative 0.3 with a wide confidence interval spanning roughly negative 0.6 to positive 0.2. The steady-state average is approximately negative 0.5 with a confidence interval that excludes zero (roughly negative 0.7 to negative 0.3).
Tariff Rate column, Quantity: The transition-window average is near zero (approximately 0.1) with a wide confidence interval. The steady-state average is approximately negative 0.4 with a confidence interval that excludes zero.
Tariff Exposure column, Spending: The transition-window average is approximately negative 0.2 with a very wide confidence interval that includes zero (roughly negative 2 to positive 1). The steady-state average is approximately negative 0.8 with a confidence interval that excludes zero.
Tariff Exposure column, Quantity: The transition-window average is near zero with a wide confidence interval. The steady-state average is approximately negative 0.8 with a confidence interval that excludes zero.
Figure 4: Welfare Losses as Share of Approximate Income
Notes: The figure presents three welfare measures: net
spending effect, deadweight loss, and Sato-Vartia cost-of-living index,
as a percentage of approximate household income. See Equations (9)–(15) for measure
definitions. Income is approximated using bracket midpoints: $20,000
(low), $82,500 (middle), $187,500 (high). The high-income midpoint is a
lower bound on mean income in this group. The deadweight loss and
Sato-Vartia bars are visually almost indistinguishable, reflecting the
close agreement between the two price-anchored welfare measures.
Grouped bar chart showing three welfare measures (Net Spending Effect, Deadweight Loss, and Sato-Vartia cost-of-living index) as a percentage of approximate household income, for three income groups (Low, Middle, High). Bars extend downward from zero, indicating welfare losses.
Low-income group: The Net Spending Effect bar reaches negative 0.37 percent. The Deadweight Loss bar reaches negative 0.19 percent. The Sato-Vartia bar reaches negative 0.18 percent. The Deadweight Loss and Sato-Vartia bars are visually almost indistinguishable.
Middle-income group: The Net Spending Effect bar reaches negative 0.13 percent. The Deadweight Loss bar reaches negative 0.05 percent. The Sato-Vartia bar reaches negative 0.05 percent.
High-income group: The Net Spending Effect bar reaches negative 0.08 percent. The Deadweight Loss bar reaches negative 0.02 percent. The Sato-Vartia bar reaches negative 0.02 percent.
The chart shows that the welfare burden is sharply regressive across all three measures, with low-income households bearing the largest cost as a share of income.
Figure 5: Household Sentiment and Intentions from the Survey
Notes: Panel A reports the share of respondents in the
Numerator survey selecting each Q5 concern item in each of the six
biweekly waves. Panel B reports the share selecting each Q7
anticipated-behavior item.
Two-panel grouped bar chart showing survey responses across six biweekly waves (weeks of 4/21, 5/5, 5/19, 6/2, 6/16, and 6/30).
Panel A (Tariff Concerns, Q5): Shows the share of respondents selecting each concern item. "Higher prices on everyday goods" is the most common concern at approximately 55-60 percent across all waves. "General inflation/Rising cost of goods" is second at approximately 40-45 percent. "Higher prices on non-essential items" is third at approximately 25-30 percent. Other concerns (limited availability, impact on job/industry, stock market, healthcare costs, slower economic growth) range from approximately 5 to 20 percent. "No concerns" is selected by approximately 5 percent. Responses are stable across all six waves.
Panel B (Anticipated Reactions, Q7): Shows the share selecting each anticipated behavioral reaction. "Cut back spending on non-essentials" is the most common at approximately 40-50 percent. "Look for sales or coupons" is second at approximately 35-45 percent. "Delay non-essential/big-ticket purchases" is third at approximately 30-35 percent. "Switch to lower-priced retailers" is at approximately 25-30 percent. "Buy fewer imported goods," "Delay purchases until prices stabilize," "Switch to U.S.-made alternatives," and "Buy/stock up earlier" each range from approximately 15 to 25 percent. "Reallocate budget to essentials" and "Increase budget for certain items" are at approximately 10-15 percent. "I don't expect to make any changes" is at approximately 10 percent. Responses are stable across all six waves.
Figure A.1: Numerator Panel Demographics
Eight-panel grouped bar chart comparing three series: Raw (Pre-Weighting) Numerator panelists in blue, Weighted Numerator panelists in red, and American Community Survey (ACS) targets in yellow, across eight demographic dimensions as of March 2025. Values are shown as percentages.
Age: Six bars each for age groups 18-24, 25-34, 35-44, 45-54, 55-64, and 65+. The weighted panel closely matches ACS targets. The 65+ group is the largest at roughly 29 percent (weighted) versus 21 percent (ACS), while 18-24 is smallest at roughly 4 percent.
Income: Six brackets from Under $20K through $125k+. The weighted panel aligns closely with ACS targets across all brackets.
Ethnicity: Five categories. White/Caucasian is the largest group at approximately 64 percent (weighted) versus 68 percent (ACS). Black or African American is approximately 14 percent, Hispanic/Latino approximately 15 percent, Asian approximately 6 percent.
Gender: Female is approximately 73 percent (weighted) versus 81 percent (ACS raw); Male approximately 27 percent.
Children Under 17: Has Child approximately 27 percent, Has No Child approximately 73 percent.
Household Size: Five groups from 1 through 5+. Two-person households are the largest at approximately 35 percent.
Urbanicity: Three categories. Suburban is largest at approximately 38 percent, Urban at approximately 35 percent, Rural at approximately 27 percent.
Census Division: Nine divisions shown. South Atlantic is the largest at approximately 21 percent. The weighted panel matches ACS targets within 2 percentage points for all dimensions.
Figure A.2: Retail sales, total excluding motor vehicles and parts
(seasonally adjusted) constructed using Census Bureau estimates and
Numerator panel data
Line chart with three series from January 2018 through December 2025, indexed to 100 in 2019.
The red solid line is the Census Bureau's seasonally-adjusted retail sales series. The blue dashed line is the Numerator-constructed series. The gray dash-dotted line is the Census non-seasonally-adjusted series with authors' seasonal adjustment applied.
All three series move together closely. They begin at approximately 95 in early 2018, rise to 100 by mid-2019, and remain roughly flat through early 2020. There is a sharp dip to approximately 85 in April 2020, followed by a rapid recovery and overshoot to approximately 110-115 by early 2021. The series then fluctuate between approximately 120 and 125 through 2025, ending near 125.
Figure A.3: Numerator Fisher Price Index versus the PCE deflator for
food
Two-panel chart.
Panel A (Numerator FPI): Line chart from January 2018 through December 2025, indexed to January 2018 = 100. The solid blue line (Numerator FPI) and the dotted green line (PCE Food) move in close alignment. Both are stable near 100 through 2019, rise modestly in 2020, then climb sharply from 2021, reaching approximately 130 by mid-2025. The Numerator FPI ends at 130.7.
Panel B (12-month Percent Change): Line chart from January 2019 through November 2025 showing year-over-year percent changes. Both series show near-zero inflation in 2019-2020, rising sharply to approximately 13 percent in 2022, then decelerating to approximately 1-2 percent by late 2023. Both series show a modest pickup in 2025, ending at 1.9 percent.
Figure A.4: Income-Specific Fisher Price Indices
Notes: The figure plots the Fisher Price Index constructed
separately for each income group (Low, Middle, High) from January 2024
through December 2025, expressed as cumulative percentage change from
January 2024. The vertical dashed line marks February 2025, when tariffs
were implemented. Income groups are defined as in Section 3.3.
Line chart showing cumulative percentage change from January 2024 for three income groups (Low in red, Middle in blue, High in green) from January 2024 through December 2025. A vertical dashed line marks February 2025.
All three series begin at zero in January 2024. Through 2024, they gradually diverge: by January 2025, the Low-income series is at approximately 2 percent, Middle at approximately 1.5 percent, and High at approximately 1 percent. After February 2025, the gap widens further. By December 2025, the Low-income series reaches approximately 4 percent cumulative price growth, Middle approximately 4 percent, and High approximately 3.5 percent.
Low-income households consistently experience the highest cumulative price growth throughout the sample period.
Figure B.1: Validation: Actual vs. Proxy Tariff Measures
Notes: Each point represents one HS4 code (\(N = 101\)). Panel A shows tariff rates: the
\(x\)-axis is the proxy tariff rate,
computed as the duties paid over total imports for each HS4 code in 2024
using trade data; the \(y\)-axis is the
actual tariff rate from products with verified country of origin using
spending-weighted origin shares. Panel B shows tariff exposure: the
\(x\)-axis multiplies the proxy tariff
rate by import penetration (Equation 3); the \(y\)-axis multiplies the actual rate by
import penetration (Equation B.1). The
dashed line is the 45-degree line; the solid line is the OLS fit. Panel
A: Pearson = 0.63, Spearman = 0.66. Panel B: Pearson = 0.79, Spearman =
0.85.
Two-panel scatter plot with 101 HS4 codes.
Panel A (Actual vs. Proxy Tariff Rate): The x-axis shows the proxy tariff rate (0 to about 0.4); the y-axis shows the actual tariff rate from origin-known products (0 to about 0.5). Points are scattered with moderate correlation. A dashed 45-degree line and a solid OLS fit line are shown. Most points cluster below 0.2 on both axes, with several outliers above 0.3. Pearson correlation is 0.630, Spearman is 0.659.
Panel B (Actual vs. Proxy Tariff Exposure): The x-axis shows the proxy tariff exposure (0 to about 0.2); the y-axis shows the actual tariff exposure (0 to about 0.25). Points show a tighter relationship than in Panel A, with most clustering along the 45-degree line. The OLS fit is close to the 45-degree line, indicating approximate unbiasedness. Pearson correlation is 0.789, Spearman is 0.846.
Figure C.1: Average Tariff Rate by Country
Four-panel line chart showing monthly average tariff rates from January 2024 through December 2025 for Canada, China, Mexico, and Rest of World (RoW).
Canada: The tariff rate is near 0 percent through January 2025, rises sharply to approximately 5 percent in February-March 2025, and fluctuates between approximately 3 and 5 percent through December 2025.
China: The tariff rate starts at approximately 20 percent in January 2024 (reflecting pre-existing tariffs from 2018-19). It rises sharply after February 2025, reaching approximately 45-50 percent by mid-2025, with considerable volatility.
Mexico: The tariff rate is approximately 0 percent through January 2025, rises to approximately 4 percent by March 2025, then fluctuates between approximately 1 and 4 percent through December 2025.
RoW (Rest of World): The tariff rate is approximately 2 percent through January 2025, rises to approximately 8-10 percent by mid-2025, reaching a peak of approximately 12 percent before settling near 10 percent.
Figure C.2: Average Tariff Rate and Exposure by Sector
Seventeen small-multiple panels, one for each sector, showing monthly average tariff rate (left axis, in red/orange) and tariff exposure (right axis, in blue) from January 2024 through December 2025.
Sectors with the largest tariff rate increases include: Apparel, Footwear, and Accessories (rising from approximately 15 to 30 percent), Baby (approximately 5 to 40 percent), Electronics (approximately 5 to 25 percent), Toys (approximately 0 to 25 percent), and Party & Occasions (approximately 0 to 40 percent).
Sectors with the largest tariff exposure increases include: Electronics, Baby, Toys, and Apparel, reflecting high import penetration in these categories. Grocery and Health & Beauty show more modest increases in both measures, consistent with higher domestic production shares.
All sectors show the common pattern of low and stable measures through January 2025, with sharp increases beginning in February 2025.
Figure C.3: Average Tariff Rate and Exposure by
Essentiality
Two-panel line chart showing monthly average tariff rate (left axis) and tariff exposure (right axis) from January 2024 through December 2025, separately for Essential and Non-essential goods.
Essential goods: The tariff rate rises from approximately 2 percent pre-tariff to approximately 8 percent post-tariff. Tariff exposure rises from near 0 to approximately 0.015 percent.
Non-essential goods: The tariff rate rises from approximately 3 percent pre-tariff to approximately 15 percent post-tariff. Tariff exposure rises from approximately 0.005 to approximately 0.060 percent.
Non-essential goods face substantially higher tariff rates and tariff exposure than essential goods throughout the post-tariff period, consistent with higher import penetration in non-essential categories.
Figure D.1: Leave-one-sector-out estimates of β
(a) Tariff Rate
(b) Tariff Exposure
Two sets of forest plots, Panel A for the Tariff Rate specification and Panel B for the Tariff Exposure specification. Each set contains three columns (Price, Spending, Quantity). Each row reports the benchmark tariff coefficient when the indicated sector is excluded from the estimation sample, with 95 percent confidence intervals. A red dashed vertical line marks the full-sample estimate. The row labeled (Full) reports the full-sample estimate.
Panel A (Tariff Rate): The full-sample price coefficient is approximately 0.15. Leave-one-out estimates range narrowly around this value, with no sector's removal changing the qualitative conclusion. The full-sample spending coefficient is approximately negative 0.46, and leave-one-out estimates range from approximately negative 0.35 to negative 0.55. The largest deviation occurs when Home & Garden is dropped. The quantity column shows similar stability, with estimates ranging from approximately negative 0.35 to negative 0.50.
Panel B (Tariff Exposure): The full-sample price coefficient is approximately 0.20, with leave-one-out estimates ranging from approximately 0.10 to 0.30. The spending coefficient is approximately negative 0.63, with leave-one-out estimates ranging from approximately negative 0.45 to negative 0.75. The quantity column similarly ranges around negative 0.45 to negative 0.65. No single sector's removal overturns the sign, statistical significance, or order of magnitude of any coefficient.