Digital Infrastructure Integration and Capital-Market Resilience in a Frontier Exchange: A Longitudinal Panel and Coefficient-Decomposition Analysis of the Nairobi Securities Exchange

Digital infrastructure has become the backbone of commerce across Sub-Saharan Africa, yet evidence on whether firm-level integration protects listed companies' capital-market outcomes during macroeconomic stress remains scarce, dominated by studies of Chinese firms.This study examines whether sectoral digital infrastructure exposure is associated with market-based resilience (lower equity price volatility and stronger returns, not organizational resilience) among the 53 companies continuously traded on the Nairobi Securities Exchange over the turbulent 2021-2025 cycle, and tests the financial channel involved.Daily transactional records are aggregated into a balanced firm-year panel and analysed with three estimators: random effects (REs), twoway fixed effects (FEs), and correlated REs (CREs, Mundlak), all with firm-clustered standard errors.A Hausman test rejects REs for the timevarying regressors (χ²(3) = 41.11,p < 0.001).Because the exposure tier is a coarse, sector-based proxy correlated with firm size and liquidity, five additional identification checks extend the robustness battery: an unbalanced panel reintroducing previously excluded, thinly traded counters (including a bank omitted from the original classification, which is assigned Tier 2 but contributes no usable observation); a lagged, predetermined liquidity specification; a quarterly-frequency replication; a continuous market-wide shock index; and a traded-value size/turnover proxy, given the absence of market-capitalization data.Market liquidity, not digital exposure directly, carries the association.A coefficient-decomposition analysis, distinguished from causal mediation, identifies the channel: between-firm market liquidity (firm-mean log volume) is strongly and negatively associated with volatility (-0.510, p < 0.01), and a Gelbach (2016) decomposition within a single model family attributes essentially all of the small RE-CRE coefficient gap to this liquidity term (-0.120 of -0.099 total).The association survives every extension: an unbalanced 58-firm sample (-0.448), quarterly aggregation (-0.384), 1-year-lagged liquidity (-0.433), a continuous shock index (-0.482),and an independent traded-value proxy (-0.380).Descriptively, digitally integrated banking and telecommunications firms trade with materially lower volatility (2.02% versus 3.01-3.42%)and higher mean returns (14.73%) than traditional-sector counterparts, but this raw tier-level pattern does not stand as a direct digital-exposure effect: the linear digital-exposure index is not itself a significant predictor of volatility, any resilience signal is concentrated in the highexposure tier, and it does not survive controlling for between-firm liquidity in the preferred CRE model (direct effect 0.097, n.s.).Digital exposure does not steepen the liquidity-return relationship; a minimum-detectable-effect calculation shows the design can reliably detect interaction effects only above roughly 4.5-5.3 at 80% power, several times the observed estimates, a bounded, power-limited null.Market liquidity and scale are therefore the proximate, measurable channel linking digital infrastructure to capital-market resilience here, though the design is associational, single-market, and limited by a coarse sector-level exposure measure.