Small purchases, big questions Interactive investigative report
Can your corner store spot a recession?
Convenience stores and the recession question: what everyday purchases reveal about household strain. Early clues deserve scrutiny. Prediction remains unproven.
An investigative report by Kelly EmrickDHSc, PhD, MBA, BSRT(ARRT)R
- $341.2B inside sales, 2025
- +1.7% sales, −1.7% transactions
- −5.7% fuel revenue, +0.8% gallons
- 0 validated U.S. c-store recession models found
01 · The question behind the everyday purchase
A skipped snack. A smaller basket. An early signal?
A customer may skip a drink, change a lunch purchase, or stop a routine visit. Each decision is minor. Repeated across many stores, these changes could reveal shifts in household finances before monthly economic data are released.
Convenience-store trips are “a leading indicator of consumer spending.”
A consumer-goods representative on the Philadelphia Fed’s Economic and Community Advisory Council, April 16, 2025, reporting recent declines. One council member’s view. Not a Federal Reserve recession forecast and not statistical validation of the indicator.
Scope of this investigation
Source: Federal Reserve Bank of Philadelphia (2025), Economic and Community Advisory Council meeting summary, p. 2.
02 · Three meanings of early
“Early” is three different claims.
1
Early observation
Detecting a change shortly after it occurs.
2
Early publication
Receiving a useful measure before a slower official release.
3
Leading indicator
A change that comes before the economic outcome itself.
Recession dating is retrospective
The Great Recession peak and its official announcement
The NBER assesses broad declines by depth, diffusion, and duration, not a mechanical two-quarter GDP rule. A signal appearing in 2008 could precede the announcement even though it followed the downturn’s beginning. Comparing an alert only with the announcement would exaggerate its predictive lead.
“Earliest” is a comparative claim, not merely the availability of high-frequency data.
Sources: NBER (2008), Business Cycle Dating Committee announcement; NBER (n.d.), dating procedure FAQ.
03 · What the industry numbers actually show
Sales rose. Transactions fell.
2025 inside sales
$341.2B
reported by NACS, up 1.7% from 2024.
Inside sales and transaction counts moved in different directions
View the published figure
Figure 1. Published annual changes for 2025. The series are from separate releases; identical store coverage is not established. A descriptive comparison, not a forecast or matched-store decomposition.
Transaction counts measure purchase events, not unique individuals. The chart does not show that the typical customer spent more, and dividing the two rates does not estimate average ticket growth without matched definitions and coverage.
Revenue can rise without more customers through
2025 was the 23rd consecutive year of inside-sales growth
Year of inside-sales growth Growth during a recession year (2008, 2009, 2020)
A discrepancy that deserves reporting
−2.7%
combined pump and inside transactions, NACS release, April 2026
−3.0%
the same measure, NACS analysis, August 2026
The public material does not reconcile this difference. The figures are not averaged here, and no undocumented revision is assumed.
35,000+
stores in company submissions, including a self-reported subscription database. Scale alone does not establish random sampling, and public annual summaries cannot demonstrate a weekly lead over the business cycle.
May be represented differently
- Large chainsvsSmall independents
- Participating storesvsNon-participants
- Successful survivorsvsClosed locations
Obtain before modeling
Sources: NACS (2026, April 15); Blasinsky (2026, August 31). Recession years marked for reference.
04 · Why a convincing warning can be misleading
Falling fuel revenue did not mean falling gallons.
Fuel revenue and gallons sold, annual change
Revenue is nominal; gallons measure physical volume
View the published figure
Figure 2. Reported annual changes, not estimates from a new model. The chart does not establish recession timing.
Sources: NACS (2025, 2026); Kilian (2009), American Economic Review.
05 · Rival explanations
Some purchases shift. They don’t disappear.
Circana’s public first-quarter 2025 overview describes weaker convenience-store sales and units, reduced foodservice traffic, and shoppers increasing visits to club and e-commerce channels. Commercial descriptive evidence, not a recession test, but a specific rival explanation.
Signal explainer: before you call it a recession warning
Choose something a store might observe. The panel shows what else could explain it and what to check first, drawn from the report.
If: Fuel revenue fell
Could also mean
- Lower pump prices, not lower demand. NACS reported fuel revenue down 5.7% in 2024 while gallons rose 0.8%.
- A supply-driven price decline, which carries different information than a demand-driven one.
- Changes in vehicle use, fuel efficiency, or competing fuel sellers.
Check first
- Gallons per store-open-day and per completed fuel purchase.
- Prices reported separately from quantities.
If: Fewer inside transactions
Could also mean
- Transactions are purchase events, not unique shoppers or visits.
- Payment-processing changes rather than behavior.
- Purchases shifting to club and e-commerce channels instead of disappearing.
Check first
- Purchase events per store-open-day, with consistent treatment of returns and duplicate receipts.
- The same trend in other retail channels.
If: Smaller baskets or single packs
Could also mean
- A smaller purchase may reflect limited funds at the moment of purchase.
- But a larger package may reflect economizing through a lower unit price, consistent with Nevo and Wong (2019).
- A rule that reads every smaller basket as a warning ignores both mechanisms.
Check first
- Standardized physical quantities within stable categories, adjusted for package size.
- Comparable prices, not raw item counts.
If: One store’s sales dropped
Could also mean
- Road construction, weather, or stockouts.
- Changes in operating hours, nearby competition, or commuting patterns.
- Promotional activity, shifting health preferences, or regulation.
Check first
- Breadth: does the change appear across stores, regions, and categories?
- Persistence over successive weeks, after alternatives are examined.
If: Annual inside dollars rose
Could also mean
- Higher prices, product mix, larger purchases, or a changing store population.
- In 2025, inside sales rose 1.7% while inside transactions fell 1.7%.
- Inside sales grew for 23 straight years, through two recessions.
Check first
- Physical quantities and visits in a stable-store panel.
- A separate account of store openings and closures.
Sources: Circana (2025); NACS (2025, 2026); Nevo and Wong (2019). Store-level causes are potential confounders to investigate, not established explanations for the national figures.
06 · What the academic evidence supports
Strong related research. No direct proof.
Electronic payments can improve current estimates of economic activity, and households change where and how they shop during downturns. These findings establish plausible mechanisms. They do not demonstrate the forecasting performance of a convenience-store index.
| Study and setting | Finding relevant to this investigation | Limit on interpretation |
|---|---|---|
| Galbraith and Tkacz (2018)Canadian payments | Payments data reduced errors in GDP and retail-sales nowcasts. | Broad payment activity and current-period estimates, not proof of a U.S. convenience-store recession lead. |
| Chetty et al. (2024)U.S. private-sector data | Granular data revealed spending and employment responses to the pandemic in near real time. | Observation after a shock does not establish prediction before the shock. |
| Lewis et al. (2022)U.S. weekly activity | A composite weekly index tracked rapid changes using several economic series. | Supports combining information; the index measures current activity. |
| Nevo and Wong (2019)Great Recession shopping | Households increased economizing, including buying on sale and shopping at discount stores. | Store spending may change through shopping choices rather than an equivalent decline in consumption. |
| Griffith et al. (2016)UK household food purchases | Households cut real food spending while maintaining calories and nutritional quality. | A UK food result cannot be directly generalized to U.S. convenience-store forecasting. |
Table 1. Research relevant to the proposed signal.
Measure stress first. Test recession second.
View the published figure
- H1Adjusted changes in visits and quantities improve estimates of current household consumption.
- H2These measures improve recession-onset forecasts beyond established economic information, at horizons such as three, six, or twelve months (proposed, not demonstrated lead times).
- H3The improvements stay robust to changes in retailer coverage, geography, category mix, and economic episode.
Figure 3. Author-developed conceptual model. Arrows show hypothesized relationships and a proposed analytical sequence, not measured causal effects, a validated index, or an inevitable progression toward recession.
07 · What a weekly monitor should measure
Six measures. Six safeguards.
Start with a transparent set of component series. Set composite weights and alert thresholds only after validation. Aggregate daily records weekly while preserving the dates the information became available.
| Measure | Operational definition | Interpretation safeguard |
|---|---|---|
| Inside transactions | Purchase events per store-open-day, with consistent treatment of returns and duplicate receipts. | Transactions are not unique shoppers or visits. Separate payment-processing changes from behavior. |
| Physical quantities | Comparable units or volume within stable product categories, adjusted for package size. | Do not add unlike items into an unexplained total. Retain assortment and stockout flags. |
| Basket composition | Quantities, standardized prices, and shares of selected categories within matched stores. | Promotions, product launches, and price changes can alter the mix without a macroeconomic shift. |
| Fuel purchases | Gallons per store-open-day and per completed fuel purchase, reported separately from prices. | Vehicle use, efficiency, and competing fuel sellers can affect gallons. |
| Fuel and inside attachment | Share of eligible fuel visits with a linked inside purchase, only where reliable linkage exists. | Unmatched pump and till totals cannot establish conversion. Use privacy-preserving aggregate reporting. |
| Breadth and persistence | Distribution of adjusted changes across stores, regions, and categories over successive weeks. | A widespread signal is more persuasive than one chain’s result; persistence rules require validation. |
Coverage and records the panel needs
Table 2. Proposed measures and required safeguards (author proposal). Report geographic patterns in aggregate; never infer an individual shopper’s income or distress from a purchase.
08 · How to test the recession claim
Five gates before anyone says “predict.”
Validation checklist: mark each gate for a monitor you are evaluating
- 1
Timing
Reconstruct weekly information sets using release dates, processing delays, and data vintages.
Weakens the claim: the warning appears only in revised data or after recession onset.
- 2
Incremental value
Compare a benchmark model with an otherwise comparable model that adds convenience variables.
Weakens the claim: convenience measures add nothing beyond information already available.
- 3
Chronological generalization
Train on earlier periods; test on later untouched periods, leaving entire downturns out when feasible.
Weakens the claim: performance depends on random splitting or one unusual episode.
- 4
Reliability
Evaluate calibration, Brier score or log loss, sensitivity, precision, and false alarms.
Weakens the claim: frequent warnings, poor calibration, or benefits that vanish under reasonable specifications.
- 5
Robustness
Check regions, store types, coverage changes, category exclusions, and the pandemic separately.
Weakens the claim: one chain, one category, sample turnover, or pandemic closures drive the result.
0 of 5 gates passed
Mark each gate. Until all five pass, treat the signal as information about household pressure, not as a recession predictor.
Benchmarks to beat, at the same forecast horizon and information cutoff
Predefine a limited set of variables. Report lead time against the economic peak or onset convention, with its range across episodes, and report false alerts during expansions separately. Choose the alert threshold from a stated tolerance for misses and false alarms, never afterward to flatter the chart. Follow any favorable history with a prospective test under frozen rules.
Table 3. Proposed validation gates (author proposal).
09 · A practical research sequence
Advance on evidence, not on a schedule.
- 1
Feasibility audit
Available history, store coverage, physical units, and real-time availability.
- 2
Transparent component series
Build the series and document every correction.
- 3
Historical comparison
Test against benchmark models and report both successful and unsuccessful specifications.
- 4
Prospective evaluation
Run under fixed rules, recording every alert and its subsequent outcome.
Useful now
Operators can use unusual purchase patterns to investigate local demand, staffing, stock availability, and customer preferences. Economists can weigh them alongside broader measures.
Not yet supported
A national recession declaration or a fixed lead-time claim would exceed the available evidence.
Author recommendations. The reporting agenda (written data requests to NACS and providers, operator interviews, independent forecasters) is proposed; no interviews or data agreements were completed for this report.
Early clues deserve scrutiny. Prediction remains unproven.
The case for studying convenience stores is credible: the sector records frequent, adjustable purchasing decisions, and related research shows transaction data can improve economic measurement. The gaps are timing, generalizability, and false alarms. This review found descriptive sector evidence and relevant mechanisms, but no validated U.S. convenience-store model that provides superior advance warning of recession onset. Whether these stores detect the earliest signs of a national recession remains an empirical question.
An investigative report by Kelly EmrickDHSc, PhD, MBA, BSRT(ARRT)R
References
- Blasinsky, C. (2026, August 31). State of the c-store industry: Profit under pressure. NACS. convenience.org
- Chetty, R., Friedman, J. N., Stepner, M., & Opportunity Insights Team. (2024). The economic impacts of COVID-19. Quarterly Journal of Economics, 139(2), 829 to 889. doi:10.1093/qje/qjad048
- Circana. (2025, May 22). U.S. c-store landscape: Q1 2025. circana.com
- Federal Reserve Bank of Philadelphia. (2025, April 16). Economic and Community Advisory Council [Meeting summary]. philadelphiafed.org
- Galbraith, J. W., & Tkacz, G. (2018). Nowcasting with payments system data. International Journal of Forecasting, 34(2), 366 to 376. doi:10.1016/j.ijforecast.2016.10.002
- Griffith, R., O’Connell, M., & Smith, K. (2016). Shopping around: How households adjusted food spending over the Great Recession. Economica, 83(330), 247 to 280. doi:10.1111/ecca.12166
- Kilian, L. (2009). Not all oil price shocks are alike. American Economic Review, 99(3), 1053 to 1069. doi:10.1257/aer.99.3.1053
- Lewis, D. J., Mertens, K., Stock, J. H., & Trivedi, M. (2022). Measuring real activity using a weekly economic index. Journal of Applied Econometrics, 37(4), 667 to 687. doi:10.1002/jae.2873
- National Association of Convenience Stores. (2025, April 9). C-store foodservice delivered exceptional growth in 2024. convenience.org
- National Association of Convenience Stores. (2026, April 15). U.S. convenience in-store sales top $340 billion. convenience.org
- National Bureau of Economic Research. (2008, December 1). Business Cycle Dating Committee announcement. nber.org
- National Bureau of Economic Research. (n.d.). Business cycle dating procedure: Frequently asked questions. nber.org
- Nevo, A., & Wong, A. (2019). The elasticity of substitution between time and market goods: Evidence from the Great Recession. International Economic Review, 60(1), 25 to 51. doi:10.1111/iere.12343
© Kelly Emrick, DHSc, PhD, MBA, BSRT(ARRT)R. Adapted from the investigative report “Convenience Stores and the Recession Question.”