G-RAM framework for software risk assessment and mitigation strategies in organisations

• Published in: Journal of enterprise information management 2018-03, Vol.31 (2), p.276-299
• Main Authors: Biswas, Baidyanath, Mukhopadhyay, Arunabha
• Format: Article
• Language: English
• Subjects: Autoregressive processes; Business; Clustering; Computer viruses; Correlation analysis; Cybersecurity; Econometrics; Empirical analysis; Exploitation; Growth models; Information systems; Information technology; Literature reviews; Model accuracy; Network security; Optimization; Parameter estimation; Risk analysis; Risk assessment; Security management; Software; Software reliability; Statistical distributions; Time dependence; Variance; Volatility
• ISSN: 1741-0398; 1758-7409
• DOI: 10.1108/JEIM-05-2017-0069

Purpose Malicious attackers frequently breach information systems by exploiting disclosed software vulnerabilities. Knowledge of these vulnerabilities over time is essential to decide the use of software products by organisations. The purpose of this paper is to propose a novel G-RAM framework for business organisations to assess and mitigate risks arising out of software vulnerabilities. Design/methodology/approach The G-RAM risk assessment module uses GARCH to model vulnerability growth. Using 16-year data across 1999-2016 from the National Vulnerability Database, the authors estimate the model parameters and validate the prediction accuracy. Next, the G-RAM risk mitigation module designs optimal software portfolio using Markowitz’s mean-variance optimisation for a given IT budget and preference. Findings Based on an empirical analysis, this study establishes that vulnerability follows a non-linear, time-dependent, heteroskedastic growth pattern. Further, efficient software combinations are proposed that optimise correlated risk. The study also reports the empirical evidence of a shift in efficient frontier of software configurations with time. Research limitations/implications Existing assumption of independent and identically distributed residuals after vulnerability function fitting is incorrect. This study applies GARCH technique to measure volatility clustering and mean reversal. The risk (or volatility) represented by the instantaneous variance is dependent on the immediately previous one, as well as on the unconditional variance of the entire vulnerability growth process. Practical implications The volatility-based estimation of vulnerability growth is a risk assessment mechanism. Next, the portfolio analysis acts as a risk mitigation activity. Results from this study can decide patch management cycle needed for each software – individual or group patching. G-RAM also ranks them into a 2×2 risk-return matrix to ensure that the correlated risk is diversified. Finally the paper helps the business firms to decide what to purchase and what to avoid. Originality/value Contrary to the existing techniques which either analyse with statistical distributions or linear econometric methods, this study establishes that vulnerability growth follows a non-linear, time-dependent, heteroskedastic pattern. The paper also links software risk assessment to IT governance and strategic business objectives. To the authors’ knowledge, this is the first study in IT secur