Slacks (inefficiency) in deterministic DEA models: Does economic theory of slacks predict long-run performance?

Standard deterministic data envelopment analysis (DEA) treats input and output slacks as pure wastes, whereas organizational theory and chance-constrained DEA view them as potential buffers against uncertainty. Using bootstrapped regression model, this study tests whether short-run static slacks predict long-run dynamic growth efficiency (GE), drawing on a panel of 50 Indian commercial banks (2005–2024) and four slack aggregation methods. The predictive power of the model depends entirely on the evaluation horizon. Over a one-year window, slacks fail to predict GE meaningfully, producing negligible model fit and making annual assessments unreliable. Over two- or three-year horizons, positive and significant relationships emerge consistently across all slack aggregation methods, with coefficients and model fit increasing monotonically with horizon length. Introducing controls for ownership and macroeconomic phases more than doubles the estimated coefficients and markedly improves model fit, revealing that omitted-variable bias conceals slack’s true contribution. Foreign banks achieve higher predicted average GE but exhibit negative slack-ownership interactions, implying diminishing returns from slack reduction. Slack-period interactions show that post-asset quality review regulations dampen efficiency gains, while the global financial crisis era lowers GE. By bridging static short-run slack analysis with dynamic long-run GE, we offer a parsimonious strategy that reconciles deterministic DEA with uncertainty: observable static slacks can proxy for strategic reserves, avoiding the demanding joint probability distributions of chance-constrained stochastic DEA. We conclude that static short-run slack forecasts dynamic long-run GE only over horizons of at least two years, repositioning slack from a symptom of inefficiency to a strategic asset.