What to Eat when Having a Millennial Over for Dinner

That's the title of a new working paper co-authored with Ph.D. student Kelsey Conley.  There is a lot of talk about how millennial's food preference may differ from previous generations, but much less is available in terms of hard evidence.  Here's what we write as the challenge with a lot of the previous research in this area (this criticism is also be true of previous blog posts I've written on the subject, such as this one and this one): 

A key downside of the previous research in this area is that today’s millennials are typically compared to today’s older generations. This sort of analysis presents a confound because older people are likely to differ from younger people at any point in time. That is, an “age effect” is confounded with a “millennial effect.” The more difficult question is whether today’s young people are different than younger people decades ago. To address this issue, cohort analyses are often conducted (Pitta et al., 2012). However, cohort comparisons are also confounded by a myriad of factors that change over time. For example, falling prices or rising incomes may lead young people today to make different purchases than young people in previous decades. To sort out this conundrum, this paper uses a difference-in-difference estimator to identify the causal effect of millennials. Using high quality government survey data [the Bureau of Labor Statistic’s Consumer Expenditure Survey], we compare the difference in food spending between the young and old in 1980 to the difference in young and old in 2015. We refer to the resulting estimate as the “millennial effect” that provides insights into whether millennials truly have different food preferences than other generations. To our knowledge, no previous research has used difference-in-difference methods to determine the millennial effect in a theoretically consistent demand framework.

Some summary statistics and preliminary analysis:

To preview our subsequent analysis, a crude difference-in-difference effect can be calculated. In 1980, the young spent $25.48-$38.98=-$13.5/week less than the old on meat. In 2015, however, the young (millennials) only spent $14.28-$18.43 = -$4.15/week on meat than the old. Thus, the millennial effect (or difference-in-difference) for meat is -$4.15+13.43 = $9.28/week. So, even though spending on meat has declined among both the old and the young from 1980 to 2015, it has declined even more among the old than the young. Thus, this crude estimate suggests young people are spending $9.28/week more in 2015 than they otherwise would have because they are millennials.

Our main findings are likely to be somewhat unexpected.  We find that the "millennial effect" is positive on food expenditure shares for three meat categories (beef, pork, and poultry), eggs, cereal, and fresh fruit.  A statistically significant negative ‘millennial’ effect is found for non-alcoholic beverages and food away from home.  This doesn't mean millennials are spending less of their food budget eating out (or spending more of their food budget on meat) than young people from the 1980s, only that they're spending less of their food budget eating out (and spending more of their budget on meat) compared to older folks today than in the past.