Food Expenditures at Home and Away from Home in the United States--A Switching Regression Analysis

Abstract

The results of this study showed that the decision to eat away from home was positively affected by income while the amount spent away from home was only affected by income, again positively, at higher levels of household income. Food expenditures at home were similarly affected by income only at higher levels of household income provided the household also ate away from home. If the household did not eat away from home, food expenditures at home were positively affected by income at all levels, the effect decreasing in magnitude with the level of income. A SIGNIFICANT economic trend in recent )&Vdecades is the declining share of consumer expenditures on food. Expenditures for food fell steadily from 20.0% of disposable personal income in 1960 to 16.1% in the third quarter of 1983 (USDA). This drop occurred as a result of a decline in the share of consumer expenditures on food purchased for consumption at home. The share of consumer expenditures on meals purchased at restaurants, cafeterias, and fast food chains remained constant at about 4% of disposable personal income. Hence the consumption of purchased meals away from home has become more important relative to food consumed at home. A number of studies on food consumption at home and away from home have been made. Houthakker and Taylor (1970) obtained results that indicate away from home food expenditures were more responsive to total consumption expenditures than at home expenditures based on 1960-61 household budget data. Time-series data were also employed by Houthakker and Taylor to investigate the dynamics of food consumption. Prochaska and Schrimper (1973) found that along with income the opportunity cost of time is an important factor in determining away from home food consumption. Fishe et al. (1979), employing simple and heteroscedastic Tobit models, found that income and household specific circumstances are important determinants of restaurant expenditures. Lamm (1982) estimated a translog demand system for food consumed at home, purchased meals, and non-food items. His results indicated that consumers have been eating away from home more often primarily as a result of increases in their incomes. Recently, Huang and Raunikar (1984) estimated Engel equations for at home and away from home food expenditures using ordinary least squares and Tobit regressions, respectively. In their study Huang and Raunikar found that, in general, at home food consumption is less income responsive than away from home food consumption. The major objective of Huang and Raunikar's study was to investigate the importance to household food expenditures of consigned income for committed expenditures. This study further investigates food consumption at home and away from home employing a switching regression model. The same basic data analyzed by Huang and Rauniker (HR) are also analyzed here. However, since the HR study focused on consigned income, households not reporting complete information with regard to committed expenditures, e.g., rent or mortgage, or after-tax income, were not included in their study. In this study, the emphasis is placed on the household's choice of whether to eat out and the factors affecting consequent away from home and at home food expenditures. The sample includes all households that reported before-tax income and is significantly larger than the sample used in the HR study.