Essays on Energy, Technical Change and Competition

This thesis reports research on energy in growth and distribution and on sample selection in industrial organization in three essays.;The first essay presents a new method for sample selection from one-dimensional noisy data, by conditioning the selection on the data structure. Gibbs sampling from a Bayesian mixture model of data cross sections, each observation is assigned an indicator latent variable and sorted into signal or noise component. Application to US firms' profit rates from COMPUSTAT from 1962 to 2012 yields an asymmetric Laplace distributed signal, confirming earlier estimates. The model retains 97% of the raw data, compared with as little as 20%, in a previous application, and 93% for a symmetric Laplace distribution. A change point analysis demonstrates a Bayesian filtering method based on co-variates.;The second essay analyzes the extent to which productivity growth can be decoupled from fossil fuel combustion. A hypothesis derived from ecological economics of a high positive correlation between average national labor productivity and the fossil energy-labor ratio is contrasted with hypotheses extant in the literature. Constructing a dataset from 1950 to 2012, previous regression results by Rada and Taylor (2008) are replicated, extended and explained in their economic context. The elasticity of the fossil energy-labor ratio with respect to labor productivity grows to unity after 2000, explained well by the ecological but not other hypotheses. Decoupling via technology hinges critically on switching to renewables, not saving energy.;The third essay measures weighted international per capita energy use inequality corrected for net embodied energy exports (consumption approach) and interprets the energy distribution from a Jaynesian maximum entropy perspective, extending and critiquing Lawrence et al. (2013). Application of standard tools from economic inequality analysis to Eora MRIO embodied energy data from 1990 to 2011 shows that consumption approach inequality has been higher throughout, with net embodied energy transfers to rich regions. Reinterpretation of the exponential distribution as arising from Jaynes-Shannon informational entropy elides problematic assumptions about energy trade and shows the exponential distribution to be a good first approximation to the data. Residual unexplained information suggests the need for additional economic theorizing about energy inequality.