Consumer demand for frozen seafood product categories in the United States

ABSTRACT

Frozen seafood marketing in grocery stores in the United States (U.S.) has undergone substantial transformation as a result of the introduction of value-added and convenience products into the category. However, it is not yet clear whether consumers perceive these value-added products to be substitutes for the traditional unbreaded products. We model the demand for frozen seafood in the United States using the linear approximate almost ideal demand system (LA-AIDS) employing market-level monthly retail scanner panel data. Our emphasis is on the demand relationships between the three aggregate frozen seafood categories, namely, breaded products, entrées and unbreaded products, and on the demand relationships for these categories when disaggregated as finfish and shellfish. We use fixed effects on the spatial and temporal variation in demand and incorporated demographic shifter variables. Our results show that unbreaded products, as compared to value-added categories of breaded seafood and seafood entrées, would gain market share if expenditure on frozen seafood increases. We also find that unbreaded products are stronger substitutes for value-added products than vice versa. We explore similar relationships between frozen shellfish and finfish products. Unbreaded shellfish can be expected to gain market share if expenditure on frozen seafood were to increase.

KEYWORDS:

• Almost ideal demand system
• consumer demand
• panel data
• product differentiation
• scanner data
• seafood
• United States

Acknowledgment

The authors thank anonymous referees for their helpful suggestions.

Notes

This classification is done by the data vending company, and is based on the physical layout of a typical supermarket. More information is in the Data and Econometric Procedure section.

Unbreaded products in the dataset represent all frozen/refrigerated/fresh seafood sold in Universal Product Code (UPC)-coded packs. The data do not allow disaggregation of this category into fresh, refrigerated and frozen products.

Asche, Bjørndal, and Gordon (2007) review econometric model specifications and elasticity estimates from seafood demand literature.

The ACN data/product categorization reflects the “department” or physical layout of the typical supermarket (Hawkes & Piotrowski, 2003); for example, vegetables are categorized into three groupings: frozen, canned and perishable, just as they are physically located in different aisles of a supermarket.

As Chern and Rickertsen (2003) note, the use of the LA-AIDS specification may be necessary when many demographic variables are included in the model, as in our case. In addition, since we are dealing with closely related products within the frozen seafood category, collinearity in their prices is expected. When prices are collinear, simplicity of the LA-AIDS specification can be exploited (Deaton & Muellbauer, 1980).

We imposed the adding-up, homogeneity, and symmetry constraints on parameter estimates. Conventionally, parameter estimates of dropped equations are obtained residually using restrictions. We used an alternative approach in which each system (DD1 and DD2) is estimated twice, each time dropping different equation. Given that the estimates are invariant to dropped equations, we get both parameters and standard errors for all equations this way.

Fixed-effects models allow for correlation between group-specific effects and the other covariates in the model. They can also reduce heterogeneity bias (Hsiao, 1986). And also, since we expect substantial differences to be present within the groups, fixed-effects models can capture these within-group variations (Townsend, Buckley, Harada, & Scott, 2013). The approach is also expected to absorb the impacts of demographic variables not explicitly accounted.

Hypothesis that expenditure elasticities and own-price elasticities equaled unity was rejected in all cases.

Hypothesis that expenditure elasticities and own-price elasticities equaled unity was rejected in all cases.