.svg)
Transaction Data for Microeconomics Research
Transaction data records what was purchased, at what price, in what quantity, and at which merchant, usually at the grain of a single line on a single receipt or card authorisation. That grain gives microeconomists observed choices rather than recalled ones, at a frequency close enough to see substitution occur within weeks of a price change, a tax change, or a new market opening. Household expenditure surveys ask people to reconstruct spending after the fact, and published scanner indices arrive already summed to a category and a region. Purchase level panels sit much closer to the demand system a researcher is trying to estimate, because they carry the basket the good was bought alongside, the merchants a consumer moved between, and the consumers who stopped buying altogether.
How microeconomists use transaction data
Substitution between competing goods after a market opens
When a legal market opens in some states before others, purchase level records let researchers test directly whether spending on an incumbent product falls, instead of inferring substitution from aggregate state revenue totals that mix price, participation and tax changes together. The usual design is a difference in differences on store level weekly sales, with treatment defined by the legalisation date and the outcome measured on the narrow category expected to lose sales. The binding constraint is coverage: transaction panels are built from the merchants or cardholders a provider can observe, so states and store formats enter unevenly, and a null can reflect thin coverage in the treated states as easily as an absence of substitution.
Payment frictions and the size of a purchase
Staggered adoption of a payment technology across stores in one chain gives an event study that holds the store, the assortment and the local market fixed, so the estimate turns on the change in the transaction itself. The outcome that matters is basket size or visit frequency rather than the count of observed transactions, since the latter moves mechanically when a payment method becomes easier to use. Card and receipt panels see only what clears through the observed channel, so cash purchases drop out entirely and part of any measured increase is migration between payment methods rather than new spending.
Repeat purchase, churn and cross merchant demand
Consumer panels that follow a stable identifier across merchants and months support estimates of repeat purchase, churn and cross merchant substitution that no single firm's own records can produce, because a firm observes only the customers who stayed with it. Researchers typically benchmark the panel against a known aggregate first, then model purchase incidence and spend conditional on incidence separately. These panels are opt in or derived from forwarded e-receipts, so they skew toward younger, online and urban consumers, and levels should be read as relative movements rather than as population spending without reweighting.
Transaction datasets available on Dewey
PDI Technologies
Transaction records at the level of the individual purchase, which microeconomists reach for when a question turns on what was bought at a given price and moment rather than on a monthly category total.
View datasetConsumer Edge
Consumer transaction data covering spending across many merchants, which supports demand and substitution questions that require following the same consumer from one retailer to another.
View datasetMeasurable AI
E-receipt data from emerging markets, useful where the research question concerns household purchasing in countries with little published scanner or expenditure microdata.
View datasetMore published work using transaction data
Poet Larsen, Sriniketh Vijayaraghavan, Uttara M Ananthakrishnan · Working Paper · 2025
Matthew Brown; Jeffrey Ohl · Working Paper · 2026
Coady Wing, Wei-Lun Lo, Maddie Potter, Tarik Yuce, Alberto Ortega, John Cawley, Thuy D. Nguyen & Kosali I. Simon · NBER Working Paper Series · 2026
.svg)