Augmenting productivity analysis with data mining: An application on IT business value Simon K. Poon * , Joseph G. Davis, Byounggu Choi School of Information Technologies, University of Sydney, NSW 2006, Australia Abstract In this paper we use a large firm-level dataset to extend previous studies by augmenting the endogenous growth accounting framework with a data mining technique to analyze the complex relationships between the use of IT and organizational practices. There is emerging evidence of recent emphasis on organizational factors and a greater shift towards ‘‘IT complementarities” in which value addition is linked to combining complementary organizational practices with IT investments. Our findings indicate that the set of interrelated orga- nizational practices that complement positively to IT use is different from the set of practices hindering IT use. The presence of clustering among organizational practices clearly implies that some combinations of practices make it difficult to precisely empirical examine. We have found that our technique was able to show some organizational factors may have different pathways to affect organizational per- formance and such organizational practices have often been overlooked but can play a weak yet non-trivial role in production and orga- nizational processes. Ó 2008 Elsevier Ltd. All rights reserved. Keywords: Association rules mining; Complementarities; Information technology business value; Productivity paradox 1. Introduction and motivation In the mid-1980s, Robert Solow 1 and others argued that there was insufficient evidence to link the massive investments in information technology (IT) to productiv- ity growth. This phenomenon which came to be known as the productivity paradox has since emerged as an important research topic for economists, information sys- tems researchers, and management theorists. Since then, a range of models and methods has emerged (Brynjolfsson & Hitt, 1996; Lehr & Lichtenberg, 1998). However, there are still considerable variations in the findings on produc- tivity impacts attributable to IT investments. Stiroh (2004) analyzed the extensive econometric literature on the economic impact of IT and found that much of the observed variation in estimates of the impact attributable to IT is due to differences in model specification and ana- lytical technique. Earlier disappointing results suggested that while IT was a critical technology in organizations, it was modeled inad- equately (Brynjolfsson & Yang, 1996). Recent studies have suggested that important organizational practices were often omitted in the productivity analyses, and the time- consuming nature of many organizational changes would take a much longer time to observe (Brynjolfsson & Hitt, 2003). These studies have highlighted that more nuance analytical techniques are needed to enhance our under- standing of the complementarities of the technology-orga- nizational phenomenon. 0957-4174/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.eswa.2007.12.028 * Corresponding author. Tel.: +61 2 9351 4920; fax: +61 2 9351 3838. E-mail address: spoon@it.usyd.edu.au (S.K. Poon). 1 Solow (1987) argued that, ‘‘You can see the computer age everywhere but in the productivity statistics”. www.elsevier.com/locate/eswa Available online at www.sciencedirect.com Expert Systems with Applications 36 (2009) 2213–2224 Expert Systems with Applications