Understanding consumption behaviour of individuals and households is crucial for many economic questions, particularly for those with policy implications. For instance, given that consumption is the largest component of aggregate demand in most economies, it is important for policymakers to understand how consumers respond to changes in their economic environment – so called ‘shocks’ – including those caused by new policies. Consumer behaviour is at the heart of many debates about the effectiveness of counter-cyclical policies, in particular fiscal policies; see, for example, the surveys by Auerbach et al. (2010) and Ramey (2011). From a scientific point of view, it is also important to relate consumption patterns observed in the data to basic economic models. For example, how do we reconcile the large swings in expected asset returns we see across asset markets (equity, debt, currencies, commodities, etc.) without observing correspondingly large swings in aggregate risk, typically measured by changes in aggregate consumption (see Campbell and Cochrane 2000)?
While we have made tremendous progress over the years in better understanding consumer behaviour, there still remain many open questions. One such question is whether consumers form habits over time based on past behaviour, how persistent such habits are, and what implications they have for policy and economic welfare. While the use of habit formation in various models has a long tradition (examples include Eichenbaum et al. 1988 in macroeconomics, Becker and Murphy 1989 in health economics, or Sundaresan 1989 in asset pricing), it is difficult to empirically study habits and to separate them from other factors that determine consumer behaviour. However, much progress has recently been made on this front; Bronnenberg et al. (2012), for instance, study how preferences of migrants in the US toward specific brands of consumer packaged goods depend on where consumers have previously lived. Similarly, Atkin (2013) shows that migrants in India keep consuming food varieties that mimic the typical diet of the region in which they were born, not the region of current residence. Finally, Malmendier and Nagel (2011) show that investors’ risk preferences depend more heavily on asset returns that they have experienced in their lifetime compared to other historical returns.
New micro evidence from Russia’s alcohol market
In recent work (Kueng and Yakovlev 2014), we study the persistence of consumption habits toward alcohol among Russian males, and we estimate when such habits are formed. Russia provides a unique setting for studying the persistence of habit formation not available in most other countries.
- First, the Russian Longitudinal Monitoring Survey (RLMS) is the consumption panel with the longest panel dimension we know.
- Second, the RLMS has a health module that asks each household member in privacy about his alcohol consumption.
Hence, these data provide individual consumption information of high quality (as opposed to household-level expenditures that previous studies had to rely on; e.g., Dynan 2000) – a feature that turns out to be very important for answering the questions at hand.
Third, alcohol consumption is widely believed to be the main contributing factor to the high male mortality rate in Russia, where male life expectancy is 15 years lower than female life expectancy. Understanding what determines alcohol preferences is thus tremendously important.
Finally, Russia’s history offers two large shocks that can be used to study habit formation.
As Figure 1 shows, Russia’s beer sales expanded rapidly after the collapse of the Soviet Union, both in levels but also relative to vodka sales – the other main alcoholic beverage consumed by Russians.
Figure 2, which shows the share of alcohol consumed by males in each year of the RLMS for different birth cohorts (measured in grams of ethanol), provides strong evidence that:
- These changes in the alcohol markets strongly affected the consumption habits of Russian males;
- These habits are persistent, and
- They are formed around the time when consumers start to consume alcohol – usually around age 18.
Males born in the 1960s or earlier who spent their early adulthood in the Soviet Union still prefer vodka today, whereas younger generations overwhelmingly prefer beer. These differences are also quantitatively large: vodka constitutes on average 60% of total alcohol intake for males born in the 1960s or earlier, but only 48% for those born in the 1970s who spent their early adulthood in the transition period after the collapse of the Soviet Union, and 32% and 19% for those born in the 1980s and 1990s, respectively, who spent their early adulthood in the post-Soviet era. In contrast, the share of beer in total alcohol intake for these age groups constitutes 20%, 36%, 56%, and 68%, respectively.
More formal statistical analysis reveals, for example, that males who turned 18 in 2002 exhibit, on average, a 12% higher share of beer consumption compared with males who are only two years older. This is presumably due to the fact that these younger consumers had access to a significantly larger beer market when they started to consume alcohol than males that are only slightly older.
At this point, it is important to note that our results are not driven by a heavy drinking or alcoholism. First, we measure consumption as shares, not levels, in order to make the results robust to extreme observations. Second, in all our formal analysis we control for the total level of alcohol intake. Finally, our results are robust to excluding the top quartile of total alcohol consumers.
Figure 1 also reveals the second large change in the alcohol market. From 1986 to about 1991, Mikhail Gorbachev’s anti-alcohol campaign heavily restricted the sale of official alcohol markets. While this policy sharply reduced the official production of alcohol, it simultaneously also lead to a dramatic increase in the illegal production of homemade vodka called samogon as shown in Figure 3. Crucial for our study of habit formation is the fact that the increase in home-produced samogon was much more prevalent in rural areas than in the densely populated urban areas for reasons we discuss in more detail in the paper. Therefore, rural males had more readily access to vodka during the anti-alcohol campaign. In our analysis of current consumption shares in the RLMS, we find that rural males who turned 18 during the campaign still consume a 5 percentage points larger vodka share today than rural males that turned 18 either before or after the campaign. Moreover, the difference in consumption shares between rural and urban males that turned 18 during the campaign is 7 percentage points larger (in absolute value) than the same difference among males that turned 18 either before or after the campaign. Hence, these two large shocks show that consumption habits can be very persistent, and that the habits for different types of alcohol are formed when people first start consuming alcohol.
The effect of alcohol habits on Russian male life expectancy
We study the implications of our results for the evolution of life expectancy of working age males, taking into account the persistent habits we uncovered. It is important to note that of the 40% alcohol-related deaths each year, only about 7% are due to alcohol poisoning. Over 30% are due to external causes related to alcohol intoxication, including vehicular and other accidents and homicides, and hence are also unrelated to long-run consequences of alcohol consumption, such as liver cirrhosis. To proceed with the analysis, we estimate a hazard model of death as a function of the share of alcohol consumed, controlling for many individual characteristics, in particular the level of total alcohol intake. The estimates show that consuming the same level of alcohol but doing so by using more vodka rather than beer is associated with a significantly higher mortality rate, both statistically and economically.
Using our results, we estimate that male mortality in Russia will decrease by one quarter within 20 years even under the status quo, that is, under the current set of policies and current levels of relative prices of alcoholic beverages. This will happen simply because new generations will be more accustomed to beer and will replace older generations with strong preferences for vodka. Since much of the gap in male life expectancy is due to occasional binge drinking (even holding fixed the average level of alcohol intake), which in turn is more likely to occur for males who prefer vodka, this shift in consumption habits toward beer has strong effects on life expectancy. Hence, this reduction in the male mortality rate will be the result of changes that occurred several decades ago.
Atkin, D (2013), “Trade, Tastes, and Nutrition in India”, American Economic Review, 103 (5): 1629–63.
Auerbach, A J, W G Gale, and B H Harris (2010), “Activist Fiscal Policy”, Journal of Economic Perspectives, 24(4): 141-64.
Becker, G S and K M Murphy (1988), “A theory of rational addiction”, Journal of Political Economy, pp. 675–700.
Bronnenberg, B J, J P H Dubé, and M Gentzkow (2012), “The Evolution of Brand Preferences: Evidence from Consumer Migration”, American Economic Review, 102 (6): 2472–2508.
Campbell, J Y and J H Cochrane (2000), “Explaining the Poor Performance of Consumption-based Asset Pricing Models”, Journal of Finance, 55: 2863–2878.
Dynan, K E (2000), “Habit formation in consumer preferences: Evidence from panel data”, American Economic Review, pp. 391–406.
Eichenbaum, M S, L P Hansen, and K J Singleton (1988), “A Time Series Analysis of Representative Agent Models of Consumption and Leisure Choice Under Uncertainty”, Quarterly Journal of Economics, 103 (1): 51–78.
Kueng, L and E Yakovlev (2014), “How Persistent Are Consumption Habits? Micro-Evidence from Russia's Alcohol Market”, NBER Working Paper No. 20298.
Malmendier, U and S Nagel (2011), “Depression Babies: Do Macroeconomic Experiences Affect Risk Taking?” Quarterly Journal of Economics, 126 (1): 373–416.
Ramey, V A (2011), “Can Government Purchases Stimulate the Economy?” Journal of Economic Literature, 49(3): 673-85.
Sundaresan, S M (1989), “Intertemporally dependent preferences and the volatility of consumption and wealth”, Review of Financial Studies, 2 (1): 73–89.