Introducing a High Definition Retail Trade Index

Juan Murillo News&References

BBVA Research is a reference in macroeconomic analysis and economic research in regions where BBVA is present (e.g. Spain, USA, Turkey, South America) and at a geopolitical level. We encourage you to subscribe to their reports and follow them @bbvaresearch. As part of their observations of consumption in Spain, they exploit National Statistics Institute (INE) figures, and rely on several  conjunctural indicators such as the Retail Trade Index. Traditionally, that indicator is based on the monthly survey of 12.500 merchants and takes a full month to get published with figures aggregated at a Autonomous Community level.

In response, together with BBVA Research, we engaged in a joint project to drastically improve the geographical and temporal granularity of the analysis of consumption in the domestic market. For the first time, we joined our complementary abilities and aligned our objectives to leverage Big Data generated by consumers, governments and companies with a macroeconomic perspective.




Our approach converged into three disciplines with great synergies and data as a common material to manipulate:

  • Data Science to create knowledge from large volume of the electronic transactions at a high spatial and temporal resolution.
  • Traditional statistics with sampling techniques to remove biases by heterogeneous geographical representation of the BBVA customers.
  • Econometrics to develop high frequency structural decomposition  models for time series.


“Our results reveal that we are able to anticipate INE’s monthly Retail Trade Index with a more detailed geographical pulse”


The solution we developed looks into the aggregated information of anonymized transactions carried out with debit or credit card by BBVA customers in Spanish point of sale terminals (PoS). That way, the spatial scope of the analysis covers the activity of more than one million merchants providing a unique descriptive capacity at a municipal level, or even at the neighborhood level. Additionally, using this information source and taking into account the date and time of the purchase we get a timely pulse of sales in the retail sector that allows us to anticipate the publication of the official index. The results show that, when comparing the monthly growth rates of the indexes published by the INE and those constructed from our data, we can prove the existence of a stable and significant  short-term relationship with a correlation that exceeds 95% at both national and regional levels.



This first outcome of our collaboration with BBVA Research indicates that Big Data analytics have a high potential to enrich the study on economic agents decision-making. Currently our approach is being applied in Spain and Mexico, as we are considering other applications in other regions.

A more detailed description of our approach is available (in Spanish) in BBVA Research’s most recent publication on Consumption in Spain for the second semester of 2016 in the section dedicated to Big Data y consumo: el índice BBVA de comercio al por menor.