This article investigates the impact of ARM initial rate (teaser) discounts on mortgage choice and housing demand. Because discounted ARM loans may reduce expected user costs, theoretical models predict a positive impact on housing expenditures. To test the hypothesis, a simultaneous model of housing expenditures conditioned upon mortgage instrument choice is estimated using a national sample of transactions for the 1986 to 1988 period. The results indicate that overall housing demand would have been reduced by approximately 13 percent during the period in the absence of ARM loans.