Aggregate statistics on per capita alcohol consumption are important in tracking trends in and for identifying the determinants and consequences of alcohol use. Though wine's share of overall ethanol consumption is lower than other beverages it is known to have varied considerably in strength over time. However, the details of this variation have not been carefully addressed. The amount of wine sold is well measured but the amount of ethanol contained in that wine has relied to date on ad hoc conversion factors accounting only for broad changes in the wine market. The purpose of this study is to develop detailed empirical estimates of the alcohol content and per capita consumption of wine. Differences over time and across states in the shares of key wine types and and changes in the alcohol concentration of brands within each type have resulted in considerable variation in the mean alcohol content of wine sold. Data on the percentage alcohol contained in specific brands and their market share in two states are used to determine the mean alcohol content by type for the years 1984 to 2002, while extrapolated approximations for each type are used for 1962 to 1983. These averages were combined with national and state wine type market shares to estimate the ethanol content of wine by state and year. Average alcohol content for each state and year are then applied to consumption figures to estimate the per capita ethanol consumption from wine. Results showed that the mean alcohol content of wine sold in the U.S. declined consistently from the early 1960s through the mid 1980s due to the reduced share of fortified wine and increased share of wine coolers. After leveling in the late 1980s mean ethanol content has gradually risen from 1992. Differences across states are found to be considerable through the 1980s but have diminished in recent years. Empirically based estimates of per capita ethanol consumption from wine are found to be generally lower than those from ad hoc estimates. The two estimation approaches yielded markedly different cross state and time variations.
