Error Analysis and Comparison of Two Algorithms Measuring Compensated Income

We analyze and compare the errors of two numerical approaches for measuring compensated income. We prove that Vartia’s algorithm and Breslaw and Smith’s algorithm both converge quadratically; when the price change within each partition step is small, the error of Vartia’s algorithm is approximately half that of Breslaw and Smith’s algorithm. A theorem and the appropriate simulations with three different demand systems correct the existing error analysis of the two algorithms, and demonstrate that Vartia’s algorithm is more accurate than Breslaw and Smith’s. Copyright Springer Science+Business Media New York 2013