AN EMPIRICAL STUDY OF BIG DATA–ENABLED PREDICTIVE ANALYTICS AND THEIR IMPACT ON FINANCIAL FORECASTING AND MARKET DECISION-MAKING

Abstract

This study addresses the problem that enterprises in data intensive capital market operations may invest in big data and predictive analytics, yet decision quality remains uneven because forecasting signals are not always reliable, interpretable, or embedded in governance routines. The purpose was to test a capability to outcome model in which Big Data Capability (BDC) and Predictive Analytics Capability (PAC) improve Financial Forecasting Accuracy (FA), and FA improves Decision-Making Quality (DMQ). Using a quantitative cross-sectional, case-based survey design, data were collected once from professionals in an enterprise case context, yielding 226 responses and 210 usable cases after screening (max item missingness 1.9%; two outliers removed). Key variables were BDC and PAC (predictors), FA (mediator), and DMQ (outcome), measured on five-point Likert scales with internal consistency (α = .86 to .91). Analysis followed data screening, reliability testing, descriptive statistics, Pearson correlations, multiple regression (Model 1: FA; Model 2: DMQ), and mediation via 5,000 bootstrap resamples. Descriptively, respondents reported moderate to high capability and outcomes (BDC M = 3.78, SD = 0.61; PAC M = 3.69, SD = 0.66; FA M = 3.62, SD = 0.58; DMQ M = 3.74, SD = 0.55). All bivariate associations were positive and significant (BDC, FA r = .52; PAC, FA r = .57; FA, DMQ r = .63; p < .001). In regression, BDC (β = 0.28, p < .001) and PAC (β = 0.39, p < .001) jointly explained 44% of the variance in FA (R² = 0.44), and FA was the strongest predictor of DMQ (β = 0.49, p < .001) in a model explaining 52% of DMQ (R² = 0.52), with smaller direct effects from BDC (β = 0.14, p = .013) and PAC (β = 0.17, p = .004). Mediation results showed partial mediation of the PAC to DMQ link through FA (indirect effect = 0.19, 95% CI [0.12, 0.28]; direct effect = 0.11, 95% CI [0.04, 0.18]). These findings imply that improving decision quality requires stronger data governance and analytics discipline, plus integration of forecasts into decision protocols across analytics enabled workflows.