QUANTITATIVE EVALUATION OF INNOVATION POLICY: LIMITS OF THE USE OF PATENTS AND EVIDENCE FROM BRAZILIAN INNOVATION LEGISLATION

Abstract

The Innovation Law was enacted 20 years ago as a national policy to support innovation and industrial development in Brazil. Because of this, it is important to verify the effects produced by this regulation. This article investigates the effectiveness of Law No. 10,973/2004 using the number of patent applications in Brazil, and discusses the limits of a quantitative evaluation based only on this indicator. The central question of the study is whether the Innovation Law was able to generate statistically significant changes in patent filings over time. Although patent data are widely used as an indicator of innovative activity in policy evaluation, the study examines whether this measure is sensitive enough to capture the effects of a single legal instrument inside a complex and dynamic innovation system. The methodological approach followed four steps. First, we used the annual historical series of patent filings (1997–2021), obtained from the Intellectual Property Database (BADEPI) provided by the National Institute of Industrial Property. Second, we tested the series for stationarity to confirm the use of time-series models. Third, we estimated AutoRegressive Integrated Moving Average (ARIMA) and ARIMAX models with intervention and control variables. The goal was to test different intervention structures and include macroeconomic and institutional controls. Finally, we used Welch’s t-test to compare the mean differences between the pre- and post-law periods. This methodological framework offers a replicable strategy to evaluate public policy effects using time-series evidence. Although the post-2004 period shows a higher average number of patent filings (t = −7.30; p < 0.001), the properly specified time-series models ARIMA (0,2,1) and ARIMAX with different interventions — did not identify statistically significant effects attributable to the law. The quantitative results were complemented by an examination of the legal and institutional framework related to innovation in Brazil. The findings indicate that the evolution of patent filings reflects the trend of the series and structural factors not captured by the models, showing the limits of using this variable alone as a proxy for innovation and for the effectiveness of Law No. 10,973/2004. The study did not find statistically significant evidence that the law was able to accelerate national innovation. This suggests that its effects may occur through institutional, collaborative, or technological mechanisms that are not reflected in patent filings. These results reinforce the need for broader and multidimensional approaches to evaluate innovation policies.