UBC Theses and Dissertations
Purchasing power parity methods of making international comparisons Hill, Robert J.
The objective of this dissertation is to improve our understanding of the various Purchasing Power Parity (PPP) methods that have been advocated in the literature on international comparisons. The first of three essays builds on the pioneering work of Van Yzeren(1987) to rationalize the literature, by constructing a taxonomy of PPP methods. In particular, the taxonomy reinterprets PPP methods in a graph theoretic context. This reinterpretation yields many useful insights. The second essay was motivated by the realization that virtually all PPP methods have the same underlying graph theoretic structure. This essay develops a new PPP method which allows the data to choose the underlying structure by using Kruskal’s “Minimum Spanning Tree” Graph Theory algorithm to chain PPPs across countries rather than imposing the structure ex ante. The Minimum Spanning Tree (MST) method may potentially dramatically simplify the procedure for constructing PPPs. The MST method also has important implications for time series comparisons. The essay concludes with an empirical comparison using 1990 OECD data between the MST method and the three most widely used PPP methods. The third essay focuses specifically on the Average Price class of PPP methods identified in the taxonomy. Average Price methods have the very desirable property of generating quantity indices that literally add up over different levels of aggregation when measured in value terms. However, it is widely claimed that Average Price methods overestimate the output shares of any outlier countries in a comparison. This is the so-called Gerschenkron effect. In spite of its significant implications, evidence for the Gerschenkron effect remains largely anecdotal. This essay explains the reasoning behind the Gerschenkron effect. As part of this explanation it is necessary to give a precise interpretation to the hitherto vague notion of an “outlier” country. Also frameworks are developed for empirically verifying and measuring the Gerschenkron effect, which are then applied to 1990 OECD data, with some surprising results.
Item Citations and Data