Increasing Block Rate Tariffs as Faulty Transmitters of Marginal Willingness to Pay

Abstract

Econometric demand estimation is conducted under the assumption that market prices provide accurate signals of consumers' marginal willingness to pay for associated observed quantities demanded. But, increasing block rate tariffs are faulty transmitters of marginal willingness to pay (MWTP). The problem arises because consumers may not be able to adjust quantity demanded so as to equate MWTP with price. Under such tariffs, price increases discretely at the block boundaries, and at the rate of consumption associated with a block boundary a consumer's MWTP may exceed the price in the preceding block but be less than the price in the next block. Thus, such a consumer would consume at the quantity associated with the boundary, with observed MWTP (i.e., the price in the preceding block) less than true MWTP. The observed marginal price is an erroneous measure of MWTP for those consumers falling into this price gap, and these measurement errors are neither independent of the MWTP nor do they have a zero mean. As a consequence, the estimated parameters of demand functions for goods sold under increasing block rate tariffs are likely to be biased, hypotheses tests concerning the parameters are likely to be invalid, and estimated price and income elasticities in error. The faulty transmitter problem may also explain, in part, the failure to obtain hypothesized results in most demand studies using a Taylor/Nordin demand specification.1 In the next section the faulty transmitter problem is described. A demand model is specified for a simple case, and the biases in the estimated parameters, resulting from the errors in the observed marginal prices, are examined for predictions as to direction. Then follows a description of a simulation conducted, using tariffs and aggregate data for 19 Wisconsin communities, to examine the direction and magnitude of these biases in an empirical residential water demand study. The biases are examined for demand functions estimated using both micro data and aggregate, community level, data. Finally, some conjecture on methods of circumventing the faulty transmitter problem is offered.