Econometric analysis of the real exchange rate, energy consumption and instrumental variables

Abstract

The present thesis consists of four chapters: the introduction (Chapter 1), and three econometrics-based research papers (Chapters 2 through 4). The introduction sets out the research questions explored in the subsequent chapters and previews the results. Chapter 2 uses Monte Carlo studies to investigate a pre-test bias problem potentially associated with using Stock and Yogo’s (2005) critical values to determine whether the instrumental variables in the model are strong or weak. A solution is proposed based on Angrist and Krueger’s (1995) Split-Sample IV estimator. Chapter 3 tests the Balassa-Samuelson hypothesis using a new dataset, and with a focus on oil-producing countries. A relationship between oil rents, which represent productivity measure in the tradable sector of oil exporters, and the real exchange rates are examined using unique data constructed from Wood Mackenzie’s Global Economic Model. The results provide evidence in favour of the Balassa-Samuelson effect in most countries in the sample, apart from OPEC countries. Chapter 4 investigates the long-run relationship between GDP per capita, energy consumption and energy prices in a set of 28 OECD countries using newer econometric techniques than have been prevalent in the prior literature. The results of the analysis suggest that the long-run bi-directional relationship exists, but is likely to be heterogeneous across countries. Also, the bi-directionality is not symmetric: energy consumption was found to be more strongly affected by economic growth than vice versa. In a narrative sense, the three research papers reflect the progression of my interests over the past several years: from a work focused on econometric theory (‘how should we do IV pre-testing?’), to a work of applied econometrics testing a piece of economic theory (‘does the Balassa-Samuelson hypothesis account for real exchange rate movements in oil exporters?’), and finally to a work of applied econometrics which is oriented more towards answering policy-relevant questions (‘if we cut our energy usage to meet climate targets, will we choke off economic growth?’).