This replication package contains the Stata do-files used to produce all results and figures reported in the paper. The analysis draws on the Italian Household Budget Survey (HBS) conducted by ISTAT, covering the years 2014–2022.

STATA, 18

The underlying microdata are not included in this package, as access is restricted to authorised researchers via ISTAT. Information on how to apply for access is available at https://www.istat.it/informazioni-sulla-rilevazione/spese/

See more information here (in Italian): https://www.istat.it/informazioni-sulla-rilevazione/spese/

The package comprises the following do-files:

Dependent_variables_FINAL.do — construction of binary store-visit indicators and aggregate purchase measures across seven venue types Control_variables_FINAL.do — preparation of sociodemographic and socioeconomic control variables Descriptives_FINAL.do — summary statistics for all key variables Descriptives_over_time_FINAL.do and Summary_table_descriptives_over_time_FINAL.do — time-trend descriptives and tabulation (2014–2022) Histograms_FINAL.do — distribution plots of venue-specific purchase counts Countfit_for_nbreg_FINAL.do — overdispersion diagnostics to justify negative binomial estimation Model_1__total__FINAL.do — OLS regression of total store voracity on household characteristics, with marginal means by income decile and education Model_2__HHI__FINAL.do — zero-one-inflated beta (ZOIB) regression of the Herfindahl–Hirschman Index of shopping concentration Model_3__logit_and_nbreg__FINAL.do — two-part models (logit + negative binomial) for each of the seven store types, with combined margin plots Poisson_margins_FINAL.do — alternative Poisson-based margins for robustness Model_4_Margins_per_food_item_Q5-Q1_FINAL.do — Q5–Q1 margin differences by food product and store type Heatmap_Google_Colab_Script.txt — link to the Python/Google Colab script used to produce the food-item × store-type heatmap

All do-files are self-documented with inline annotation. File paths in the scripts refer to the authors' local environment and must be updated to match the replicator's directory structure.