Theoretical Contributions to Nonparametric Estimation of Production Frontier Functions for Nigerian Manufacturing Firms

Technical efficiency measurement in Nigerian manufacturing firms requires production frontier estimation methods capable of handling the high-dimensional, heterogeneous, and potentially non-standard data structures arising from mixed formal-informal production environments where the Cobb-Douglas functional form assumption may be strongly violated. This dissertation makes original theoretical contributions to nonparametric and semiparametric production frontier estimation relevant to the Nigerian manufacturing context. The first contribution develops a novel multiplicative kernel estimator for the stochastic frontier that is proved consistent and asymptotically normal under weaker distribarity assumptions than existing estimators, specifically relaxing the requirement of symmetric composed error distribution. The second contribution extends the data envelopment analysis efficiency estimator to incorporate statistical inference using a bootstrap procedure adapted to the clustered sampling structure of Nigerian manufacturing surveys. A third contribution develops a theoretical test for the hypothesis of constant returns to scale at the industry level using a ratio of nonparametric frontier efficiency estimates, with the asymptotic null distribution derived analytically. The theoretical framework is applied to panel data from the World Bank Enterprise Surveys for Nigerian manufacturing firms in 2014, 2017, and 2023, spanning food processing, textiles, and metalwork sectors across Lagos, Kano, and Anambra States. Results reveal that the returns-to-scale test rejects constant returns for all three sectors, with increasing returns dominant in food processing. Keywords: production frontier, nonparametric estimation, technical efficiency, Nigerian manufacturing, stochastic frontier.