On the Reliability of Software Piracy Statistics

Despite tremendous debate and policy interest in software piracy, statistics compiled by the Business Software Alliance (BSA) have generally been accepted at face value by policy makers and scholars. However, the accuracy of BSA statistics has not been independently verified. Based on a review of the BSA methodology and empirical analysis, I conclude the following. A change in the BSA consultant and methodology around 2002-03 had systematic effects on published piracy rates. First, the trend rate of decrease of piracy rates fell from 2.0% points per year to 1.1% points per year. Second, piracy rates apparently became more sensitive to changes in income. Third, piracy rates depended on projections of software usage based on per capita incomes in the respective countries, and were subject to a bias that was greater for higher-income countries