A Comparison of Two Methods for Testing the Utility Maximization Hypothesis When Quantity Data Are Measured With Error

Abstract

The Generalized Axiom of Revealed Preference (GARP) can be violated because of random measurement errors in the observed quantity data. We study two tests proposed by Varian (1985) and de Peretti (2004), which test GARP within an explicit stochastic framework. Both tests compute adjusted quantity data that are compliant with GARP. We compare and contrast the two tests in theoretical terms and in an empirical application. The empirical application is based on testing a large group of monetary assets for the United States over multiple sample periods spanning 1960-1992. We found that both tests provided reasonable results and were largely consistent with each other

Belgium relief fund, post war food shortages and the “True” cost of living

“…These procedures reject utility maximization under rationing for a single violation of the modified Afriat inequalities. To allow for a margin of error in consumer optimization, Afriat (1967) and Varian (1985) suggest using an efficiency index, "e", to determine if violations of GARP are from sub-optimal choices, also discussed in Gross (1995) and Jones and de Peretti (2005). Fleissig and Whitney (2015) develop an LP Procedure using an efficiency index to test for rationing under sub-optimal choices where the researcher selects the value for "e".…”

A Simple Method to Account for Measurement Errors in Revealed Preference Tests

“…I do so by applying them to a data set that has previously been used in Jones and de Peretti (2005) to compare and contrast Varian's (1985) error procedure with an alternative procedure proposed by de Peretti (2005). Both of these procedures are based on computing perturbed quantity data that satisfy GARP.…”

“…The data consists of nominal per-capita asset stocks and real user cost prices for the assets in the monetary aggregate L (Liquid Assets) (See Jones and de Peretti, 2005). The data span monthly observations from 1960 to 1992, but because of inconsistencies in the data, Jones and de Peretti (2005) split the data into 8 di¤erent (non-overlapping) sub sets, which they called S1-S8.…”

“…The data span monthly observations from 1960 to 1992, but because of inconsistencies in the data, Jones and de Peretti (2005) split the data into 8 di¤erent (non-overlapping) sub sets, which they called S1-S8. Table 1 presents the results and some summary statistics of S1-S8 7 .…”

“…I demonstrate the practical usefulness of the models and methods by applying them to monetary data from Jones and de Peretti (2005). In this application, I investigate the computational properties of the new algorithm.…”