Rex Yuxing Du

How Economic Contractions and Expansions Affect Expenditure Patterns

In this study, we attempt to understand how household budget allocations across various expenditure categories change when the economy is in recession or expansion. The common assumption is that a household’s tastes would not change as a function of economic conditions and therefore any adjustments in expenditure patterns during economic contractions/expansions would simply be due to changes in the consumption budget. Standard economic models translate these budgetary effects into lateral movements along a set of fixed Engel curves, which relate category expenditure shares to total expenditures. We propose and test a conceptual framework based on the notion of relative consumption, which prescribes that, for any given total consumption budget, expenditure shares for positional goods/services will decrease during a recession, while shares for nonpositional goods/services will increase (i.e., shifting the entire Engel curve upward or downward, depending on the nature of the expenditure category and the economic conditions).

Where Did All That Money Go? Understanding How Consumers Allocate Their Consumption Budget

All types of consumer expenditures ultimately vie for the same pool of limited resources—the consumers discretionary income. Consequently, consumers’ spending in a particular industry can be better understood in relation to their expenditures in others. Although marketers may believe that they are operating in distinct and unrelated industries, it is important to understand how consumers, with a given budget, make trade-offs between meeting different consumption needs. For example, how much would escalating gas prices affect consumer spending on food and apparel? Which industries would gain most in terms of extra consumer spending as a result of a tax rebate? Answers to these questions are also important from a public policy standpoint because they provide insights into how consumer welfare would be affected as consumers reallocate their consumption budget in response to environmental changes. This study proposes a structural demand model to approximate the household budget allocation decision, in which consumers are assumed to allocate a given budget across a full spectrum of consumption categories to maximize an underlying utility function. The authors illustrate the model using Consumer Expenditure Survey data from the United States, covering 31 consumption categories over 22 years. The calibrated model makes it possible to draw direct inferences about the trade-offs individual households make when they face budget constraints and how their relative preferences for different consumption categories vary across life stages and income levels. The study also demonstrates how the proposed model can be used in policy simulations to quantify the potential impacts on consumption patterns due to shifts in prices or discretionary income.

Measuring Contagion in the Diffusion of Consumer Packaged Goods

This study measures the degree of contagion or interpersonal influence in the diffusion of new consumer packaged goods (CPGs). The authors demonstrate that when an individual-level trial hazard model is properly specified to account for potential sources of biases, substantial contagion effects may be detected in the diffusion of many CPGs. Using longitudinal panel data on individual-level trial and repeat purchases of 67 newly introduced CPGs, they show that standard diffusion models fail to detect contagion. However, after extending the model to allow for spatial and temporal heterogeneity in contagion and controlling for various cross-sectional and temporal confounds, they find statistically significant contagion effects in 33 to 40 of the 67 sample products. The empirical evidence of contagion in the diffusion of many CPGs has important implications because most new product trial models for CPGs have assumed a priori that there is no contagion in the diffusion of these products. Moreover, the individual-level simultaneous analysis of the diffusion of 67 newly introduced CPGs provides useful insights into the unobservable network of influences among consumers. Such analysis allows a vendor to identify the most influential early adopters among its customers, who could help diffuse a new product more effectively in the market.

Decomposing the Impact of Advertising: Augmenting Sales with Online Search Data

Unlike sales data, data on intermediate stages of the purchase funnel (e.g., how many consumers have searched for information about a product before purchase) are much more difficult to acquire. Consequently, most advertising response models have focused directly on sales and ignored other purchase funnel activities. The authors demonstrate, in the context of the U.S. automotive market, how consumer online search volume data from Google Trends can be combined with sales data to decompose advertisings overall impact into two underlying components: its impacts on (1) generating consumer interest in prepurchase information search and (2) converting that interest into sales. The authors show that this decompositional approach, implemented through a novel state-space model that simultaneously examines sales and search volumes, offers important advantages over a benchmark model that considers sales data alone. First, the approach improves goodness-of-fit, both in and out of sample. Second, it improves diagnosticity by distinguishing advertising effectiveness in interest generation from its effectiveness in interest conversion. Third, the authors find that overall advertising elasticity can be biased if researchers consider only sales data.

Advertising and brand attitudes: Evidence from 575 brands over five years

Little is known about how different types of advertising affect brand attitudes. We investigate the relationships between three brand attitude variables (perceived quality, perceived value and recent satisfaction) and three types of advertising (national traditional, local traditional and digital). The data represent ten million brand attitude surveys and

Improving the Performance of Tracking Studies

264 billion spent on ads by 575 regular advertisers over a five-year period, approximately 37% of all ad spend measured between 2008 and 2012. Inclusion of brand/quarter fixed effects and industry/week fixed effects brings parameter estimates closer to expectations without major reductions in estimation precision. The findings indicate that (i) national traditional ads increase perceived quality, perceived value, and recent satisfaction; (ii) local traditional ads increase perceived quality and perceived value; (iii) digital ads increase perceived value; and (iv) competitor ad effects are generally negative.

Size and Share of Customer Wallet

Tracking studies are prevalent in the social sciences. These studies are predominantly implemented via repeated cross-sectional surveys of independent, non-overlapping samples, which are much less costly than recruiting and maintaining a longitudinal panel that track the same sample of respondents over time. In the existing literature, data from repeated cross-sectional surveys are analyzed either independently for each time period, or longitudinally by focusing on the dynamics of the aggregate measures (e.g., sample averages). In this study, we propose a multivariate latent state-space model that can be applied directly to the individual-level data from each of the cross-sectional surveys over time, taking full advantage of three patterns embedded in the data: a) inter-temporal dependence within the population means of each survey variable, b) temporal co-movements across the population means of different survey variables and c) cross-sectional co-variation across individual responses within each sample. We illustrate our proposed model on two applications. In the first application, we have access to all the individual-level purchase data from one large population of grocery shoppers over a span of 36 months. This provides us with a testing ground for benchmarking our proposed model against existing approaches in a Monte Carlo experiment, in order to determine which model performs best in inferring population trends using data sampled through repeated cross-sections. In the second application, we apply the proposed model to repeated cross-sectional surveys that track customer perceptions and satisfaction for an automotive dealer, a situation that is often encountered by marketing researchers.