Digitalization, Innovation and Economic Growth in the United States: Analysis by Classical and Quantile Regression (QR And QQR)

Abstract

Introduction: Digitalization and innovation have become major drivers of economic growth, but their effects may vary considerably across different economic conditions. While conventional analyses often focus on average effects, they may overlook important heterogeneities and asymmetric relationships, particularly across low and high levels of economic growth. The United States provides a particularly relevant context for this analysis due to its highly digitized economy and its leading position in innovation and digital technologies.

Objectives: This study examines the impact of digitalization and innovation on US economic growth by comparing classical quantile regression (QR) and quantile-on-quantile regression (QQR). The objective is to identify whether the effects of digitalization and innovation differ across the distribution of economic growth and to determine whether these effects vary according to different economic regimes.

Methods: The analysis employs annual US data on economic growth, digitalization, and innovation. Digitalization is measured through the spread of broadband infrastructure, while innovation is proxied by research and development (R&D) expenditure. Quantile regression is first applied to examine the effects of these variables across different conditional GDP growth quantiles. Quantile-on-quantile regression is then employed to capture two-dimensional and conditional heterogeneity, allowing the effects of digitalization and innovation to be examined across different combinations of their own distributions and the GDP growth distribution.

Results: The results indicate that digitalization has a limited, and in some cases negative, effect at certain GDP growth quantiles, suggesting the presence of diminishing marginal returns in an already highly digitized economy. In contrast, innovation has a stronger positive effect on economic growth, particularly during periods of rapid expansion, highlighting its procyclical nature. The comparison between QR and QQR shows that conventional quantile regression tends to smooth the relationships and may conceal important heterogeneities. The QQR approach reveals a more complex pattern, characterized by negative effects at certain intermediate quantiles and substantially stronger positive effects at high quantiles.

Conclusions: The findings provide a more nuanced understanding of the relationship between technology and economic growth in the United States. They highlight the importance of accounting for heterogeneous and asymmetric effects when assessing the economic contribution of digitalization and innovation. The results are consistent with the coexistence of diminishing returns to digitalization in a mature digital economy and the increasing importance of innovation for sustained economic expansion. They also suggest that policies supporting digitalization and innovation should take into account the prevailing growth regime in order to enhance their effectiveness.