Search

Shift‐share is a popular technique used by policymakers and researchers alike to decompose the change in a variable into within and between effects, intensive and extensive margins, or other comparable effects. One reason for the popularity of shift‐share is its relative simplicity compared to econometric techniques. However, often overlooked is that a shift‐share is actually an index number problem that uses differences instead of ratios. Techniques developed in index number theory accentuate the fact that the traditional shift‐share is biased. This paper proposes using the Bennet index to achieve unbiased measurements in shift‐share decompositions. In addition to solving the bias problem, the Bennet index removes the need for a residual and may be even simpler to calculate. While this paper is primarily theoretical, it also explores the differences between the traditional shift‐share and the Bennet index—both chained and fixed base—to measure the competitive and composition effects of Canada's export performance in the United States since 1990. [ABSTRACT FROM AUTHOR]