Older Cars Lost Share in Georgia’s New Uncleared Listings in Late June
The share of newly listed, uncleared cars from model year 2019 or earlier fell from 39.3% before 22 June to 30.5% over the next four weeks. A comparable series for customs-cleared listings was almost unchanged. The break is clear in this listing dataset; its cause is not.
A visible break in the uncleared-listing series
This analysis follows cars when their advertisements first appear, not when they cross the border. Listings marked as not customs-cleared are used as a proxy for vehicles in the import or auction channel.
Across the sixteen weeks from 2 March through 21 June, 39.3% of 15,254 newly listed uncleared cars were from model year 2019 or earlier. Across the four weeks from 22 June through 19 July, the share was 30.5% of 4,106 listings. The difference is 8.8 percentage points.
The weekly observations immediately before the break were 36.1%, 43.1% and 35.4%. The next four were 33.1%, 29.5%, 28.7% and 30.2%. None of the four post-break observations overlaps the range of the preceding sixteen weeks. That makes the change more substantial than a single unusual week.
Weekly sample sizes also remained within the earlier range. The data therefore show a change in composition, not simply a week in which very few cars were listed.
A visible break in the uncleared-listing series
This analysis follows cars when their advertisements first appear, not when they cross the border. Listings marked as not customs-cleared are used as a proxy for vehicles in the import or auction channel.
Across the sixteen weeks from 2 March through 21 June, 39.3% of 15,254 newly listed uncleared cars were from model year 2019 or earlier. Across the four weeks from 22 June through 19 July, the share was 30.5% of 4,106 listings. The difference is 8.8 percentage points.
The weekly observations immediately before the break were 36.1%, 43.1% and 35.4%. The next four were 33.1%, 29.5%, 28.7% and 30.2%. None of the four post-break observations overlaps the range of the preceding sixteen weeks. That makes the change more substantial than a single unusual week.
Weekly sample sizes also remained within the earlier range. The data therefore show a change in composition, not simply a week in which very few cars were listed.
Older-car share in uncleared and cleared new listings
The two channels have different baseline age profiles. The comparison concerns their timing and shape, not equality of the populations.
The cleared-listing comparison did not show the same move
For newly listed cars already marked as customs-cleared, the share from model year 2019 or earlier was 68.4% during the sixteen pre-break weeks and 68.8% during the following four. Its monthly readings from March through partial July remained close to 68–69%.
This is a useful comparison, but it is not a controlled experiment. Cleared and uncleared listings represent structurally different markets: their age profiles, sellers, time in stock and reasons for being listed can differ. The stable cleared series argues against a broad change affecting every listing in the database at the same time. It does not by itself prove that the excise caused the break in the uncleared series.
The cleared-listing comparison did not show the same move
For newly listed cars already marked as customs-cleared, the share from model year 2019 or earlier was 68.4% during the sixteen pre-break weeks and 68.8% during the following four. Its monthly readings from March through partial July remained close to 68–69%.
This is a useful comparison, but it is not a controlled experiment. Cleared and uncleared listings represent structurally different markets: their age profiles, sellers, time in stock and reasons for being listed can differ. The stable cleared series argues against a broad change affecting every listing in the database at the same time. It does not by itself prove that the excise caused the break in the uncleared series.
Older-car share in uncleared and cleared new listings
The two channels have different baseline age profiles. The comparison concerns their timing and shape, not equality of the populations.
The timing is compatible with the tax transition
The revised passenger-car excise rules entered into force on 2 April 2026. Cars aged 0–6 years are taxed at GEL 1.5 per cubic centimetre and cars older than six years at GEL 4.5. Transitional rules preserve the previous treatment for vehicles imported, or already in transportation, before 2 April, subject to additional conditions for overland transport.
A delayed change in listings is therefore plausible: vehicles already purchased or moving under the transitional provision could continue to appear after the law changed. But the dataset does not record purchase date, shipping start, arrival date or the excise actually paid. It cannot measure the lag directly.
The responsible conclusion is that the late-June break is consistent with a delayed response to the new age-based schedule. It is not an estimate of the policy’s causal effect.
Model-year mix of new uncleared listings: March and partial July
March n = 3,786; July n = 2,771 through 20 July. Shares describe advertised supply, not purchases or border crossings.
The change was concentrated in specific model-year bands
Comparing March with the partial period from 1–20 July, the 2015–2019 band fell from 32.6% to 23.3% of newly listed uncleared cars. The 2014-and-earlier band changed less, from 7.6% to 6.2%. The 2020–2022 band was also close to stable, at 42.0% in March and 40.7% in partial July.
The gain appeared in the newest bands. Model year 2023 rose from 9.7% to 16.1%, while 2024 or newer rose from 8.0% to 13.6%.
These are shares of listings, not counts of purchasing decisions. A lower share of 2015–2019 cars can reflect fewer such cars appearing, more newer cars appearing, or both.
The change was concentrated in specific model-year bands
Comparing March with the partial period from 1–20 July, the 2015–2019 band fell from 32.6% to 23.3% of newly listed uncleared cars. The 2014-and-earlier band changed less, from 7.6% to 6.2%. The 2020–2022 band was also close to stable, at 42.0% in March and 40.7% in partial July.
The gain appeared in the newest bands. Model year 2023 rose from 9.7% to 16.1%, while 2024 or newer rose from 8.0% to 13.6%.
These are shares of listings, not counts of purchasing decisions. A lower share of 2015–2019 cars can reflect fewer such cars appearing, more newer cars appearing, or both.
Model-year mix of new uncleared listings: March and partial July
March n = 3,786; July n = 2,771 through 20 July. Shares describe advertised supply, not purchases or border crossings.
Recorded engine-size mix: March and partial July
Listings with recorded displacement only. The median was 2.0 litres in both periods; this does not rule out model-level changes.
Recorded engine size did not shift in the same way
The median recorded engine displacement among newly listed uncleared cars was 2.0 litres in March and 2.0 litres in partial July. The share at or below 1.6 litres moved from 24.1% to 24.9%, while the share at or above 3.0 litres moved from 17.8% to 18.9%.
The listing series therefore contains no comparable step toward smaller recorded engines. That is a narrower statement than “importers did not downsize”: engine volume is missing for some listings, broad shares can conceal model-level changes, and July covers only twenty days.
Recorded engine size did not shift in the same way
The median recorded engine displacement among newly listed uncleared cars was 2.0 litres in March and 2.0 litres in partial July. The share at or below 1.6 litres moved from 24.1% to 24.9%, while the share at or above 3.0 litres moved from 17.8% to 18.9%.
The listing series therefore contains no comparable step toward smaller recorded engines. That is a narrower statement than “importers did not downsize”: engine volume is missing for some listings, broad shares can conceal model-level changes, and July covers only twenty days.
Recorded engine-size mix: March and partial July
Listings with recorded displacement only. The median was 2.0 litres in both periods; this does not rule out model-level changes.
What the dataset cannot establish
“Uncleared” is a proxy, not a shipment record. A car may remain uncleared for reasons unrelated to a recent import, and sellers can publish advertisements at different points in the logistics process.
Model year is also only a proxy for the excise age calculation. The law uses production year and the year of the taxable transaction or customs declaration. Cars near the boundary may therefore be placed on the wrong side by the listing field.
Finally, four post-break weeks are enough to show that the result is not a one-week anomaly, but not enough to call the new level permanent. The relevant follow-up is whether the share remains near 30% over a longer complete period.
No days-on-market, sell-through or removal metric is used. A 9 July maintenance pass changed the historical classification of dormant listings, making exit-based comparisons across that date unreliable; first-listing composition is not affected by that operation.
AutoBridge listing data under the canonical filter: passenger cars, not hidden, first listed on or after 1 March 2026, asking price at least $2,000. Weeks are based on first-listing date. July is partial.
Methodology
Data: AutoBridge listings dataset — passenger cars, listings publicly visible (not hidden by the seller), first listed from 1 March 2026, asking price of at least $2,000.
Customs status: each listing carries a cleared or not-cleared field as supplied to AutoBridge. Not-cleared listings are used here as a stand-in for cars arriving through the import and auction channel. This is not an independent customs check of each vehicle, and a car can remain not cleared for reasons unrelated to a recent import.
Model year as an age proxy: the excise depends on a car's age, which this dataset does not hold directly, so model year is used instead. The law counts age from the year of manufacture to the year of the customs declaration, so cars close to the six-year boundary can fall on the wrong side of the split used here.
Dates: every date is the date the advertisement first appeared — not the date the car was bought, shipped or cleared.
34,005 listings pass the filter. The weekly series runs from 2 March to 19 July 2026 and covers 15,254 not-cleared listings first published before 22 June and 4,106 published after it. Month-to-month comparisons set March against 1–20 July.
2 March — 20 July 2026. The weekly series ends on 19 July and the partial week beginning 20 July is excluded; July is not a full month. The timing fits a delayed response to the excise change of 2 April 2026, but listing data cannot prove what caused it. No measure of sales, sale speed or time on the market is used: a maintenance pass on 9 July 2026 rewrote the recorded history of dormant listings, which makes any such measure unreliable across that date.