This project proposes a Big Data solution to increase the income tax base and boost public revenue. The idea is to leverage consumption data obtained from credit card transactions, ATM withdrawals and online purchases to construct an income proxy, and compare it to reported income from administrative tax records. Consumption data will be obtained through collaboration with the largest credit-card supplier and banking regulator for at least one country. To design an algorithm for mapping consumption measures into income proxies, using non-parametric estimation and statistical machine learning. We propose to randomly notify taxpayers of discrepancies between proxied and reported taxable income to estimate the causal effect of the program on tax payments.