Information Technology Risk and Return Trade- Off: New Evidence from Recent Data Gokcen Arkali, PVAMU with Indranil Bardhan, UT-Dallas Vish Krishnan, UC-San Diego YAEM 2010 July 1, 2010 Research Problem To investigate the impact of IT investments on overall firm risk. 1 Positive Impact of IT on Firm Output Levels Prior literature is rich in terms of productivity and profitability impact of IT but is limited in terms of explaining the IT risk- return relationship. A significant amount of research has claimed that productivity paradox has been disappeared by showing positive impact of IT on firm output levels. (Brynjolfsson and Hitt 1996, Dewan and Min 1997) Dedrick et al. 2003 reviewed more than 50 published articles and concluded that higher levels of IT investments are associated with higher levels of productivity growth. Do higher IT investments impact the overall firm risk? How? 2 Observed Impact of IT Kobelsky et al. (2008) empirically examine the impact of IT investments on earnings volatility and suggest two different explanations of the observed impact of IT: Implementation risk view: IT result in higher earning volatility compared to other capital investments. Information processing view: IT reduces the volatility of earnings. 3 Prior Literature on the Impact of IT on Firm Risk Kobelsky et al. (2008): IT increases the volatility of earnings however the increase is conditional upon firm-level contextual variables such as sales growth, unrelated diversification, and size. Dewan, Shi, and Gurbaxani (2007): Established a link between IT risk and return using a production function frame. IT is the most risky form of investment among all other non- IT capital investments. Firms that have higher IT risk also have higher IT returns compared to the firms that have lower IT risk. 4 IT Risk and Return Trade-off We empirically explore the relationship between IT risk and firm returns. Dewan, Shi, and Gurbaxani (2007) empirically test the relationship between IT risk and firm returns and measure the magnitude of IT risk relative to other investments. We conduct a replication analysis to contribute to the stream of research initiated by Dewan et al. (2007) by a recent and unique data set to see the impact of newer technologies on overall firm risk. 5 Modeling IT Risk i ,t 0 1 IT _ Assetsi ,t 2 K _ Assetsi ,t 3 Sizei ,t 4 Leveragei ,t 5 R & Di ,t 6 Advi ,t 7 IndQi 8 InConci 9,k Yeark i ,t We use pooled regression to estimate the average contribution of IT risk to the overall firm risk. 6 Modeling IT Risk-Industry Segment Level We investigate IT risk at the industry segment level where the industry segments are defined using two-digit NAICS codes. J i ,t 0 1, j IT _ Assetsi ,t IndCodei , j 2 K _ Assetsi ,t 3 Sizei ,t j 1 4 Leveragei ,t 5 R & Di ,t 6 Advi ,t 7 IndQi 8 InConci 9,k Yeark i ,t IndCodei,j : 1 if firm i resides in industry segment j otherwise the value is 0. α1,j : represents the average contribution of IT investment of industry sector j on the overall firm risk. 7 Modeling IT Return We model the impact of IT capital, non-IT capital, and other control variables on a standard Cobb-Douglas production function following Dewan et al. (2007) in a log-linear form as shown below. ln VAi ,t 0 1 ln IT _ Assetsi ,t 2 ln K _ Assetsi ,t 3 ln LaborCosti ,t 4 IndQi ,t 5 IndConci ,t 6,k Yeark i ,t VAi,t denotes the valued-added for firm i in year t. Value-added is essentially the sales (data12) minus the cost of goods sold (data41). LaborCost is obtained from Compustat data item data42. As consistent with Dewan et al. (2007), the marginal product of IT capital is defined as follows. 8 Data Collection IT Data: Obtain firm-level IT spending data from an international research firm through the annual surveys whereas Dewan et al. (2007) obtains firm- level IT stock data from the Computer Intelligence Infocorp (CII) installation database. Our baseline dataset spans a recent time period from 1997 to 2004 permitting us to see the impact of post-Internet era whereas Dewan et al. (2007) dataset covers years 1987 to 1994. Financial and Accounting data are obtained from Standard and Poor’s Compustat database. Daily raw return data are obtained from Center for Research in Security Prices (CRSP) stock price database. Missing data for advertising from year 2002 to 2006 has been supported via the TNS Media Intelligence (TNSMI) database. 9 Descriptive Statistics Variable Mean Std. dev. Minimum Maximum SdStock 0.033 0.050 0.007 1 SdEarnings 0.593 0.336 0.006 1 SdEquity 0.150 0.260 0.001 1 ITCapital 381.82 873.62 0 14599.95 KCapital 12894.01 35292.75 0.001 529040.54 IT_Assets 0.031 0.032 0 0.396 K_Assets 0.559 0.213 0.009 0.960 Size 8.44 1.79 0.155 13.03 Leverage 0.274 0.209 0 2.14 R&D 0.055 0.141 0 3.35 Adv 0.021 0.055 0 1.80 Notes. 1. All variables including IT capital, non-IT capital, Assets, Sales, Market value of equity, Debt, and Advertising are measured in millions of dollars. 2. The dependent variables are winsorized by setting the values that are greater than one to be equal to one; we provide the winsorized statistics in the above table. 10 Table 4.2. Descriptive Statistics of Model Variables Industry Partitioning to Estimate IT Risk 2-digit Industry Segment Ind. Code N (firms) NAICS 22 Utilities Ind1 41 31,32,33 Manufacturing Ind2 245 42 Wholesale Trade Ind3 42 44,45 Retail Trade Ind4 27 48,49 Transportation and Ind5 18 Warehousing 51 Information Ind6 40 52 Finance and Insurance Ind7 78 54 Professional, Scientific, Ind8 24 and Technical Services 56 Administrative and Ind9 11 Support and Waste Management and Remediation Services 62 Health Care and Social Ind10 9 Assistance 72 Accommodation and Ind11 13 Food Services Table 4.3. Industry Partitioning 11 Empirical Analyses We present the empirical results for the two underlying empirical models: Estimating the contribution of IT investments on overall firm risk Estimating the relationship between IT risk and return using the production function framework. Then, we perform various robustness checks for the production function analyses. 12 Empirical Results: Estimating IT Risk Replication Original Study (DSG, 2007) Panel A: Dependent variable is earnings volatility (SdEarnings) Variable 2-digit NAICS Industry-structure 2-digit SIC Industry-structure IT_Assets -0.052*** -0.079*** 0.619*** 0.567*** K_Assets 0.046*** 0.020 0.016*** 0.008** Size -0.326*** -0.292*** -0.0008** -0.004*** Leverage 0.095*** 0.059*** 0.028*** 0.009** R&D 0.035** 0.040** 0.182*** 0.182*** Adv -0.086*** -0.081*** 0.048** 0.012 IndQ -0.009 0.0003 IndConc -0.088*** -0.006 Adj. R2 0.15 0.10 0.539 0.546 Panel B: Dependent variable is stock-returns volatility (SdStock) IT_Assets -0.032* -0.025 0.259*** 0.125*** K_Assets -0.009 0.012 0.006*** -0.002** Size -0.349*** -0.364*** -0.0005*** -0.003*** Leverage 0.013 0.011 0.018*** 0.007*** R&D 0.124*** 0.122*** 0.022*** 0.030*** Adv -0.009 -0.016 0.017*** 0.005 IndQ 0.003 -0.0003 IndConc 0.061*** 0.0008*** Adj. R2 0.1748 0.15 0.859 0.885 13 Empirical Results: Estimating IT Risk Replication Panel C: Dependent variable is return on equity volatility (SdEquity) IT_Assets 0.054*** 0.081*** K_Assets -0.033** -0.036*** Size -0.468*** -0.447*** Leverage 0.123*** 0.099*** R&D 0.115*** 0.139*** Adv -0.063*** -0.042*** IndQ 0.083*** IndConc 0.022 Adj. R2 0.25 0.21 F Value 51.06 81.45 14 Empirical Results: Estimating IT Risk Table 4.6. Low versus High IT Risk Industry Segments 15 Empirical Results: Estimating the Return on IT Replication Original Study (DSG, 2007) Panel A: IT risk based on earnings volatility Low IT-risk High IT-risk Low IT-risk High IT-risk Variable subsample subsample subsample subsample Ln (ITCapital) 0.286*** 0.604*** 0.017 0.098*** Ln (KCapital) 0.460*** 0.271*** 0.238*** 0.257*** Ln (LaborCost) 0.258*** 0.049 0.685*** 0.607*** IndQ 0.057 -0.151*** IndConc 0.017* 0.185*** IT marg. prod. 1.880 1.832 1.21 4.08 2 Adj. R 0.71 0.76 0.944 0.869 Panel B: IT risk based on stock-returns volatility Ln (ITCapital) 0.441*** 0.020* 0.094*** Ln (KCapital) 0.283*** 0.277*** 0.197*** Ln (LaborCost) 0.235*** 0.632*** 0.666*** IndQ 0.038*** IndConc 0.018** IT marg. prod. 2.915 1.33 3.36 2 Adj. R 0.75 0.926 0.860 VAavg IT m arg inal product 1 ITavg 16 Robustness Checks Table 4.8. Robustness Checks for Production Function Analyses 17 Key Findings Our results on the impact of IT on firm risk: Impact of IT on the overall firm risk is negative or not significant when the risk is measured by volatility of earnings or stock-returns as opposed to the findings of Dewan et al. (2007) for the same risk measures. Partial support for Dewan et al. (2007) on the positive and significant impact of IT on firm risk as measured by the volatility of ROE. The reducing impact of IT risk reveals the “information processing view of IT” as opposed to the “implementation risk view of IT” supported by Dewan et al. (2007). IT acts as a stabilizer on earnings by better coordinating the information. Our results support the high return on IT across the high IT risk industry segments similar to the findings of DSG (2007). 18 Questions?
Pages to are hidden for
"R_D_ Information Technology_ and Firm Performance"Please download to view full document