The Price Elasticity of Demand for
Fresh Dairy Products
M.J.A. Van den Tillaart1
ABSTRACT
This study examines the price elasticity of demand for fresh dairy products. In addition, this study examines the differences between the price elasticity of demand for national brands and store brands of fresh dairy products. Using the Almost Ideal Demand System with retail-scanner data from 2016-2018, 9 different fresh dairy categories are considered. I find that the price elasticity of demand for fresh dairy products is inelastic for basic dairy products and inelastic for specialty dairy product. Moreover, I show that the price elasticity of demand for national brands is smaller than the price elasticity of demand for store brands on average. Furthermore, I found that there exists a relationship between the budget share of national brands and the magnitude of the price elasticity of demand of store brands.
Key words: dairy, price elasticity, price effects, brands, national brands, store brands
Supervisor: prof. dr. M.A. Allers
Course Code: EBM877A20
Organization: Faculty of Economics and Business, University of Groningen
Date: 11-01-2019
1 Martijn van den Tillaart is a Msc Economics student at the University of Groningen
2 1. INTRODUCTION
In 2018, the prices of dairy products rose sharply with 4.1% in the Netherlands, more than four times higher than the rise of other food prices according to the Central Bureau of Statistics (CBS, 2018). The increase in the prices of dairy products was nowhere as high in Europe as in the Netherlands (CBS, 2018). Although the prices of dairy products have increased strongly, the demand for dairy has not decreased: in 2018, the total volume of dairy products consumed has even increased with 0.4% in the Netherlands according to Nielsen (2018). Food prices are main determinants of consumption patterns and rising food prices may have negative effects on health, dietary quality and food security, especially for poor people (Green, Cornelsen, Dangour, Turner, Shankar, Mazzocchi, and Smith, 2013; Andreyeva, Long, and Brownell, 2010). Moreover, Bellemare (2015) argues that rising food prices can even cause social unrest. According to the CBS (2018), the consumption of dairy products takes a great share in the total food consumption in the Netherlands, namely 14%. Therefore, rising prices of dairy products have a significant impact on total food expenditures and affects not only the poor, but the entire society. According to Nielsen (2018), the price rise of dairy products is mainly because of an increase in the price of store brands, which increased with 11% on average in 2018.
On top of that, on January the first 2019, the government of the Netherlands imposed an increase in the Value-Added Tax (VAT) rate from 6% to 9% on food products. This means that the prices of food products increase with 3% more in 2019. Taxation, such as the VAT rate, on food and non-food products have been used worldwide to decrease consumption, e.g. VAT on tobacco or alcohol, and showed to be very effective (Godfrey and Maynard, 1988). Since the VAT rate rise in the Netherlands is imposed on the entire food category, it affects the prices of dairy products and the consumption of dairy products. Therefore, there is a strong relevance for analyzing the price elasticity of demand for dairy categories, especially for fresh dairy products. According to Nielsen data (2018), fresh dairy products take up a volume share of 75% within the dairy category in the Netherlands and prices of fresh dairy products rose stronger than the average of dairy products (CBS, 2018).
3
products or categories may compete amongst each other (Bouamra-Mechemache et al., 2008). Moreover, this study is the first to investigate the relationship of price changes and demand for all fresh dairy categories (i.e. 9 product categories) including a comparison between national brands on the one hand and store brands on the other in the Netherlands. The scope of this study limits itself to the price elasticities of demand within the dairy category. We do not study the price elasticity of demand for dairy products regarding other food categories. Next to that, due to data limitations we ignore the income elasticity of demand.
Previous studies focus on the effects of price changes on the demand for aggregated dairy within other food categories (e.g. beverages, meat, fruit and vegetables). Existing literature presenting the effects of price changes on the demand for disaggregated dairy products is small and there exists ambiguity in the results. Heien and Wesseils (1990) show that the price elasticity of demand for dairy products is inelastic and Bouamra-Mechemache, Réquillart, Soregaroli, and Trévisiol (2008) argue that the price elasticity of demand for fresh dairy products is also inelastic. However, Davis, Dong, Blayney, and Owens (2010) present that numerous fresh dairy products are price elastic. Similarly, in the same year Davis, Blayney, Muhammad, Yen, and Cooper (2010) show price elasticities of demand for fresh dairy products that are elastic. Next to that, the difference in the price elasticity of demand for national brands and store brands is extensively discussed for several categories, however not so extensively for dairy categories. Previous studies (Huang, Jones, and Hahn, 2007; Bijmolt, van Heerde, and Pieters, 2005) show that store brands are less price sensitive compared to national brands, however some studies show the contrary (Bèrges, Hassan, and Monier-Dilhan, 2009; Binkley, Eales, Jekanowski, and Dooley, 2001). So the question remains to what extent price changes affect the demand for fresh dairy products and the demand for brands within fresh dairy products exactly.
Many models were used to estimate the price elasticity of demand for dairy products. Past few decades however, the Almost Ideal Demand System (AIDS) model, developed by Deaton and Muellbauer (1980), is mainly used for demand estimations. Also in this study I use the AIDS model with retail scanner data to obtain the price elasticities of demand (i.e. the own- and cross price elasticities of demand) for fresh dairy products and the comparison between national brands and store brands. After estimating the AIDS coefficients, we compute both the uncompensated own- and cross price elasticities of demand.
4
of demand for national brands and store brands of fresh dairy products. Results show that the price elasticity of demand for fresh dairy products vary amongst the fresh dairy categories. Distinguishing fresh dairy products in basic dairy products, e.g. milk or yogurt, and specialty dairy products, e.g. custard or desserts, is helpful in the understanding of the price elasticity of demand. Basic dairy products are price inelastic and specialty dairy products are price elastic. Furthermore, we present some limitations and recommendations for future research regarding the outcomes of the price elasticity of demand for fresh dairy products.
These outcomes are not only relevant for academics interested in consumer behavior regarding price effects on demand, but also for professionals. For example, the price elasticity of demand for fresh dairy products is relevant for policy makers who try to achieve social benefits by influencing consumption patterns and improving dietary quality. Then, the price elasticity of demand for fresh dairy products contains valuable information for important implications such as tax rate policy on food products. Furthermore, the effects of price changes on the demand for fresh dairy products is important information for professionals active in the dairy industry, both at the farm-level as at the industry-level (i.e. manufacturers and retailers). This information can help in the determination of price strategies, forecasting or profit-maximizing for firms.
5 2. ECONOMIC THEORY
Let us begin by analyzing the economic theory that explains the effects of price changes on the demand for fresh dairy products. The aim of this section is to elaborate on the consumer demand theory that is the foundation for the hypotheses in section (4). Moreover, it provides a basic explanation for the empirical demand system in section (5) to estimate the price elasticity of demand for fresh dairy products including a brand comparison. The price elasticity of fresh dairy products is theoretically explained by basic microeconomic theory, where consumers are striving for utility maximization or cost minimization (Huq and Arshad, 2010). In previous economic literature, this theoretical framework is well defined and thoroughly discussed (Deaton and Muellbauer, 1980). Therefore, this section presents the relevant aspects of the theoretical framework to reflect on consumer behavior regarding price changes on the demand for fresh dairy products. First, we will discuss the law of demand and the relationship with the price elasticity of demand. Secondly, we elaborate more on the price elasticity of food and zoom in on fresh dairy products. Finally, we look into differences between the price elasticity of demand for national brands and store brands.
2.1 Law of demand and the price elasticity of demand
According to Perloff (2012), one of the most important empirical findings in economics is the law of demand. This law explains that demand for a product increases when the price of a product decreases, holding constant taste, household income, prices of substitutes or complements and other factors that could influence consumption. Since the law of demand tells us something about the response to quantity demanded when the price of a product changes, it is strongly related to the price sensitivity, or the price elasticity of demand. The price elasticity of demand is the percentage change in the quantity demanded for a product in response to a percentage change in the price of a product. In mathematical form
(1) 𝜖𝑖𝑗 = ∆𝑄𝑖 𝑄𝑖 ∆𝑝𝑗 𝑝𝑗 = ∆𝑄∆𝑝𝑖 𝑗 𝑝𝑗 𝑄𝑖
where the left-hand side of (1), 𝜖𝑖𝑗, indicates the price elasticity of demand, ∆𝑄𝑄𝑖
𝑖 is the
percentage change in quantity demanded of product i and ∆𝑝𝑝𝑗
𝑗 is the percentage change in the
6
price elasticity of demand in the form of a negative sign, which indicates that when a price of a product increases, the quantity demanded decreases. When the relative change of quantity demanded is smaller than the relative change in price, the price elasticity is said to be inelastic, i.e. the price elasticity is between 0 and 1 in magnitude. If the magnitude of the price elasticity is bigger than 1, the price elasticity of a product is elastic.
Generally, the price elasticity of demand depends on the availability of any substitutes. When there are relatively more substitutes available, the magnitude of the price elasticity of demand will increase. Something crucial to understand the mechanism behind this we have to define the own-price and cross-price elasticity of demand, since it defines in which rate the quantity demanded of a product will change relative to price fluctuations of, respectively, that same product or another product. According to Pindyck and Rubinfeld (2013), the own-price elasticity of demand can be defined as the percentage change in the quantity demanded for a product in response to a percentage change in the price of that same product. In mathematical form we can write the own-price elasticity as in (1) for i = j. Economic theory tells us that the own-price elasticity of a product is negative (Pindyck and Rubinfeld, 2013). Very rarely, the demand for a product increases when the price of that product increases. When this exception is the case for a product, we call it a giffen good. Since the demand increases when the price increases for a giffen good, the own-price elasticity must be positive.
The cross-price elasticity can be defined as a percentage change in the quantity demanded for a product i in response to a percentage change in the price of another product j for i≠j. Next to that, the cross-price elasticity provides an explanation of the economic relationship between two products, i.e. whether these products are substitutes or complements. A positive cross-price elasticity indicates that the products considered are substitutes and a negative cross-price elasticity indicates that the products considered are complements. When there is no effect of a price change on the quantity demanded, the products are considered independent (Pindyck and Rubinfeld, 2013).
7
effect is taken into account, products are called net substitutes or net complements (Huq and Arshad, 2010). We will come back to this in section (5).
2.2 The price elasticity of demand for food and fresh dairy products
Generally, the price elasticity of demand for food is inelastic, since it is a basic necessity (Barkley, 2016). According to Eaton and Lipsey (1989), when prices of basic necessities increase, e.g. food or electricity, the quantity demanded decreases relatively little. Therefore, basic necessities, such as fresh dairy products and especially milk, are considered to be price inelastic. Maslow’s theory (1970), where the hierarchy of product needs are discussed, is very relevant for this topic. Maslow, Frager, Fadiman, McReynolds, and Cox (1970) suggest that a distinction between luxury and necessity goods can be expressed in the form of demand elasticities. One of these distinctions in the demand elasticity is the income elasticity of demand, where luxury goods have an income elasticity that is larger than 1 in magnitude and basic necessities, also called normal or inferior goods, have an income elasticity that is smaller than 1 in magnitude. Moreover, interesting for this research is the distinction in the price elasticity of demand. Maslow’s theory explains that when prices of luxury goods increase, such as tickets for a concert or a holiday trip, the quantity demanded decreases relatively more, i.e. being price elastic. The contrary holds for basic necessities, i.e. being price inelastic.
Food is a basic necessity that is not substitutable, everyone needs to eat food. Although food in general might be price inelastic, the magnitude of the price elasticity varies amongst food categories. We provide several reasons for the difference in the price elasticity of demand for food categories. One of the reasons is that various food categories are substitutes amongst each other, e.g. rice and potatoes or chicken and beef, which increases the magnitude of the price elasticity. Another reason is that amongst various food categories, products are more competitively distinctive in substitutes and quality, which increases the likelihood of being relatively more elastic (Cotterill et al., 2000; Bèrges et al., 2009).
8
are relatively less distinctive in quality but they do have a wide range of substitutes, which increases the magnitude of the own- and cross price elasticity. In specialty fresh dairy categories, e.g. desserts or custard, products are relatively more distinctive in quality, which decreases the magnitude of the own- and cross-price elasticity. However, as discussed before these products are relatively more luxury products, which increases the price elasticity (Eaton and Lipsey, 1989). Moreover, also in the quark and custard categories substitutes are widely available. Another increasing factor of the price elasticities for fresh dairy products is the fresh characteristic. Since consumers buy fresh dairy products weekly, they are able to switch easily to other products when prices increase relatively more compared to substitutes (Bunte, Van Galen, Kuiper and Bakker, 2007).
All in all, the own-price elasticities of demand for aggregated fresh dairy products is supposed to be price inelastic, i.e. possessing a price elasticity of demand between 0 and -1. However, the price elasticity of disaggregated categories within the fresh dairy category is ambiguous and dependent on many variables, which could turn the price elasticity from inelastic to elastic. One important variable is the characteristic of their basic needs (Maslow et al., 1970). For a full understanding of the change in quantity demanded due to price changes, we review empirical evidence on the price elasticity of demand for fresh dairy products in section (2).
2.3 Price elasticity of demand for national brands and store brands
9
brands may have a lower price elasticity than national brands and continue with why store brands may have a higher price elasticity than national brands.
2.3.1 Low price elasticity of demand for store brands
We argue several reasons why the price elasticity of demand for store brands might be lower than the price elasticity of demand for national brands. Chardon and Dumartin (1998) discuss that in some food categories the intrinsic value perceived of national brand products and store brand products is equal, which decreases the price elasticity of demand for store brands relatively to national brands. Another reason is that store brands are mainly active in categories where the price elasticity of demand is inelastic. Evidence that supports this argument exists in a study by Raju, Sethuraman and Dhar (1995). Raju et al. (1995) found that high store brand share is correlated with an inelastic price elasticity of demand for national brands. Furthermore, consumers’ lack of price knowledge, e.g. due to non-transparency of retailer’s price changes, could also affect the price elasticity of demand. If there are relatively stronger price changes of store brands compared to national brands and consumers lack knowledge about this information, a possibility could be that consumers might not notice these movements and still buy the product of the store brand. Sequentially, consumers tend to trust the price characteristic of store brands and buy these products even with relatively higher price rises (Inman, MacAlister, Hoyer, 1990). Therefore, the price elasticity of demand could be lower for store brands than for national brands.
2.3.2 High price elasticity of demand for store brands
10
and Chintagunta, 2006) . If a relatively great share of store brand buyers are price-sensitive households, we might expect that the price elasticity of demand is greater for store brands than for national brands.
Regarding economic theory, it is difficult to say if national brands have a higher price elasticity of demand versus store brands or contrary. Although we have discussed some valid arguments for both situations, in section (3) we discuss empirical evidence on the comparison between the price elasticity of demand for national brands and store brands to get a more comprehensive understanding of the relationship between national brands and store brands.
3. LITERATURE REVIVEW
In this section, we review previous studies on the price elasticity of food, fresh dairy products and brands. We use these insights in combination with the economic theory in section (2) as a foundation for the hypotheses stated in section (4). Generally, previous studies on the demand for food categories or specifically the dairy category show that the price elasticity for both is inelastic. When we divide dairy into product categories, we see that various dairy categories are indeed price inelastic, but for some dairy categories it holds that the price elasticity of demand is price elastic. Although most studies agree on the price elasticity of demand for dairy products, the magnitude of the results differ from one study to another. The same holds for studies regarding the differences between national brands and store brands. The variation amongst studies is hard to declare, since in many studies various approaches, e.g. different models and types of data, are used to analyze the demand for dairy products. Furthermore, demand is heavily dependent on supply and consumption trends, which could affect the results obtained. Therefore, we try to provide a brief summary of previous studies on the price elasticity of demand for (fresh) dairy products and discuss the most relevant results. A comprehensive synthesis of previous studies regarding dairy products is presented in Table 1. This section is organized as follows: In section 2.1 and 2.2 we analyze the price elasticity of demand for, respectively, food and fresh dairy products and in section 2.3 we analyze the price elasticity of demand for national brands and store brands.
3.1 The price elasticity of demand for food
11
and Galametsos, 1970), the price and income elasticities of demand for various consumer goods, such as food, clothing and fuel, have been estimated using a linear expenditure demand system in combination with consumer expenditure data. These studies used data from, respectively, the United Kingdom and several OECD countries. Generally, both studies show that for food products demand is responding inelastic to price and income movements, which is in line with economic theory.
In previous decades, the most commonly used model in the analysis of demand for food products is the Almost Ideal Demand System (AIDS), which we use in this research and discuss in section (5). Deaton and Muellbauer (1980) introduced this model as an alternative to the linear expenditure demand system, the Rotterdam model and the translog model, which are all demand estimation models. Deaton and Muellbauer (1980) present the demand elasticities of 8 non-durable consumer product groups by using annual British data (1954-1974) of consumer expenditure. Results show that the demand elasticities are in line with economic theory and previous studies. In here, the price elasticities of demand for necessities, such as food or housing, are inelastic and for more luxury goods, such as transport and communication, the price elasticities of demand appear to be elastic, ranging from -0.01 for food to -1.21 for transport and communication. This is also in line with Maslow’s theory (1970).
12
Another method used to estimate the price elasticity of demand is a meta-analysis. Andreyeva et al. (2010) reviewed an amount of 160 U.S. food demand studies for main food categories to assess mean price elasticities of demand. They concluded that the price elasticity for food is inelastic, ranging from -0.27 for eggs to -0.79 for soft drinks. More interestingly, they also found a price elasticity for dairy and milk which is smaller than the price elasticity for meat or fruit. They report that the mean price elasticity of the reviewed studies is -0.65 for dairy and -0.59 for milk. This means that the price elasticity for dairy is inelastic and that these products can be observed as basic necessities.
In another perspective, Ecker and Qaim (2008) found that dairy products are not always assumed to be basic necessities. Ecker and Qaim (2008) found that the own-price elasticity for dairy products & milk is elastic (-1.27) in Malawi. Although this seems very high, they also report a high own-price elasticity for other food products, such as eggs and meat. Ecker and Qaim (2008) discuss that in a poor country, such as Malawi, animal-source foods have a high own-price elasticity compared to starchy food or fruits and vegetables, which are cheaper and more easily accessible. Interesting to see is that we observe a contrast between poor and developed countries. For poor countries such as Malawi, animal-source food is relatively more price elastic than for developed countries, such as the United States. Andreyeva et al. (2010) report same results with respect to food in general, since they find that an increase in food prices results in a relatively greater decrease of consumption in poor countries than in developed countries: in high and low income countries the price elasticity of food is, respectively, -0.43 and -0.61 (Andreyeva et al., 2010).
Considering previous studies, we see that the price elasticity of food and dairy products is inelastic in general. However, we also see that the price elasticities of demand differ amongst disaggregated categories and amongst countries. Therefore, we take a closer look at the disaggregation of dairy products in the following paragraph.
3.2 The price elasticity of demand for fresh dairy products
13
First of all, Heien and Wesseils (1988) studied the demand elasticities for 12 different food categories, amongst 4 dairy categories, using the Almost Ideal Demand System as mentioned before. For this study, they used household survey data (1977-1978) from the United States, which contains data on prices and expenditures, to estimate the price and expenditure elasticity of demand. Furthermore, they used household income data to estimate the relation between total income and total food consumption. With a focus on the dairy categories, Heien and Wesseils (1988) show that the own-price elasticity of demand for the dairy categories milk (-0.62), cheese (-0.52) and butter (-0.73) are mostly price inelastic, except for the own-price elasticity of demand for cottage cheese with an absolute value of -1.1. Interesting to see is that the cross-price elasticities of demand for the dairy categories also show to be inelastic, but mostly negative. This indicates that the income effect is greater than the substitution effect so that dairy categories could be complements (Heien and Wesseils, 1988).
Secondly, Bouamra-Mechemache et al.(2008) estimated the price and income elasticity of demand for dairy products by using consumer panel and scanning data for two European countries, France and Italy, in the period of 1952-2003. Also in this study, the AIDS model of Deaton and Muellbauer (1980) is used to provide own- and cross-price elasticities and income elasticities of demand for various dairy categories. Bouamra-Mechemache et al. (2008)
used a multi-stage budgeting process, which they define as a process where consumers
14
is positive but insignificant. The income elasticity of demand is significantly positive and inelastic for all dairy categories.
Since our interest is mainly focused on the price elasticities of demand for fresh dairy products, we now review earlier studies specifically focused on the price elasticities of demand for fresh dairy categories, e.g. milk, buttermilk and yogurt. Earlier studies had the tendency to analyze aggregated dairy categories, however over time segment differentiation became more important so that more data became available on various fresh dairy products to analyze (Davis, Blayney, Cooper and Yen, 2009).
In one of the earlier studies where milk got differentiated from aggregate dairy was by Boehm and Babb (1975). In this study, consumer panel data from the United States was used to identify eight beverage milk groups to estimate the demand elasticities by using cross-sectional and time-series models. Boehm and Babb (1975) mention the distinction between a long-term and short-term price elasticity and assign household consumption to long-term effects and per capita consumption to short-term effects. Own-price elasticities of demand ranged from -1.7 for whole milk to -0.83 for 1-percent fat milk in the long run and from -0.14 for total fluid milk to -1.78 for buttermilk in the short run. For buttermilk and total fluid milk, the own-price elasticity of demand was respectively -1.52 and -1.63 in the long run. This means that the own-price elasticity of demand for buttermilk and total fluid milk is elastic and that they are no necessities in the United States.
Gould, Cox and Perali (1990) studied the demand elasticities for several food categories, with a focus on whole and low-fat milk by using the AIDS model including time-series data over the period of 1955-1985. Here they find that low-fat milk has a significantly greater own-price elasticity of demand in magnitude than whole milk, being respectively -0.437 and -0.324. Also, the positive cross-price elasticities of demand show that low-fat milk and whole milk are strong substitutes. Furthermore, they also examine the income elasticity of demand for both milk products, being 0.658 for whole milk and 0.062 for low-fat milk. This means that the demand for whole milk reacts stronger to an increase in income than the demand for low-fat milk.
15
In a well-known and more recent study of Davis, Dong, Blayney and Owens (2010) the household demand for dairy in the United States has been analyzed. A censored demand system including Nielsen home-scan data and demographic and socioeconomic variables is used to derive the demand elasticities for 12 dairy products and margarine in 2007. Here they found that the own-price elasticities of demand for dairy categories is mostly elastic, ranging from -1.81 for butter to -0.77 for bulk ice cream. Furthermore, interesting to see is that whole milk has an compensated own-price elasticity of -1.65 and drinkable yogurt -1.72. Moreover, Davis et al. (2010) also found that fluid milk products, i.e. whole milk, reduced-fat milk and canned milk, are strong substitutes. In contrast to Gould et al. (1990) they find that the own-price elasticity of demand for reduced-fat milk is smaller than for whole milk, which means that they found that reduced-fat milk is less price sensitive than whole milk.
Furthermore, Maynard and Liu (1999) found relatively high elasticity estimates for dairy products by using weekly-scanner data. Capps and Nayga (1990) argue that short-term data could lead to more elastic estimates because of storage activity, also stockpiling behavior. However, storage activity for fresh dairy products is hard because it is a perishable product, so they discuss that this argument is not applicable. Therefore, weekly-scanner data seems more compatible to fully reflect consumers’ responses to any price changes of necessity products (Maynard and Liu, 1999).
17 3.3 The price elasticity of demand for national brands and store brands
As discussed before, we argued that store brands are often dominated by national brands concerning perceived quality and that national brands and store brands are vertically differentiated: with equal prices consumers prefer national brands over store brands. Experimental evidence by Richardson et al. (1994) confirms these arguments. Although store brands are often chosen over national brands because of their price advantage, some studies place this argument in perspective by showing that it deviates from product to product. Livesey and Lennon (1978) show that consumers stick to orange juice store brands even when the price gap between store brands and national brands increases. Baltas (1997) argues that the perceived quality for orange juice store brands can be equivalent to national brands. This suggests that the preference for national brands over store brands is related to the perceived purchasing risk (Batra and Sinha, 2000). Therefore, the perceived purchasing risk has a significant role in the choice between national brands and store brands at category level. Erdem, Zhao and Valenzuele (2004) discuss that this partly explains the variation in the penetration rate or budget share of store brands in some categories: high for basic products, such as milk and yogurt, but low for specialty products where consumers trust and taste is crucial, such as infant nutrition or custard. Eventually this influences consumers price sensitivity of national brands and store brands and the price sensitivity decreases when the quality of products increases. As discussed before, Raju et al. (1995) found that store brands are mainly active in price inelastic categories and that a high store brand budget share is correlated with a low price elasticity of demand for that category. Next to that, Cotterill et al. (2000) discuss that the own- and cross price elasticities of demand for national brands and store brands are dependent on budget share of a store brand product. For instance, the price elasticity of demand for national brands would be greater than store brands in categories where store brands have a relatively higher budget share.
18
national brands (-1.51) have a lower own-price elasticity than store brands (-3.39) over the period of 1989-1990. Moreover, in a study by Bergès et al. (2009) the own-price elasticity for pasta, biscuits and jam is relatively low for national brands compared to store brands. In Table 2, previous studies about the differences between the own-price elasticity of demand for national brands and store brands are presented. Four out of six studies show an own-price elasticity of demand that is smaller for store brands than for national brands.
Table 2.
Synthesis of own-price elasticities of demand for national brands and store brands
Author(s) Country Sample and Period Method Own-price elasticity of demand*
NBR SBR SBR-NBR Huang et al. (2009) United
States
Retail scanner data, 2000-2002
AIDS/LA-AIDS
-2.20 -1.90 0.30
Cotterill et al. (2000) United States
Retail scanner data, 1991-1992
LA-AIDS -1.07 -0.98 0.09
Bergès et al. (2009) France Household panel data, 2004-2005
AIDS -1.04 -1.22 -0.18
Binkley et al. (2001) United States
Retail scanner data, 1989-1990
AIDS -1.51 -3.39 -1.88
Fong et al. (2011) United States
Retail scanner data, 2003-2009
Log-log -0.77 -0.22 0.55
Bijmolt et al. (2005) Mixed 86 studies, 1961-2004 Meta-analysis
-2.67 -2.59 0.08
# of studies with smaller price elasticity 2 4
4. HYPOTHESES
In this section we formulate the hypotheses on the price elasticity of demand for fresh dairy products and the difference between the price elasticity of demand for national brands and store brands. Based on the economic theory and previous studies discussed in section (2) and (3), we devise several hypotheses. If no economic theory or previous study material is available, we formulate our hypotheses based on ad hoc rationale. We estimate the price elasticity of demand for fresh dairy products including sub groups. Thereafter, we estimate the difference between the price elasticity of demand for national brands and store brands regarding fresh dairy products.
19
products and formulate a hypothesis for each subgroup, which is stated in hypothesis 2 and 3. The first hypothesis states that the price elasticity of demand for fresh dairy products in The Netherlands is inelastic.
Hypothesis 1
In the Netherlands, the own-price elasticity of demand for fresh dairy products is inelastic. H0 : 0 < 𝜂𝑖𝑗 < −1 𝑓𝑜𝑟 𝑖 = 𝑗
vs.
H1 : 𝜂𝑖𝑗 < −1 𝑓𝑜𝑟 𝑖 = 𝑗
Secondly, we can disaggregate fresh dairy products into relatively more necessity products, from here also called basic dairy products, and in relatively more luxury products, also called specialty dairy products. Following economic theory and previous studies, we see that the own-price elasticity estimates for milk, cream and yogurt are inelastic on average. Therefore, we suspect that these products are so called basic dairy products, which we expect to be price inelastic. Therefore, we hypothesize:
Hypothesis 2
In the Netherlands, the own-price elasticity of demand for basic dairy products is inelastic. H0 : 0 < 𝜂𝑖𝑗 < −1 𝑓𝑜𝑟 𝑖 = 𝑗 = 𝑚𝑖𝑙𝑘, 𝑐𝑟𝑒𝑎𝑚 𝑜𝑟 𝑦𝑜𝑔ℎ𝑢𝑟𝑡
vs.
H1 : 𝜂𝑖𝑗 < −1 𝑓𝑜𝑟 𝑖 = 𝑗 = 𝑚𝑖𝑙𝑘, 𝑐𝑟𝑒𝑎𝑚 𝑜𝑟 𝑦𝑜𝑔ℎ𝑢𝑟𝑡
Thirdly, following economic theory and previous studies, we see that the relatively more luxury goods, also specialty dairy products, are price elastic. Therefore, in the same context, we hypothesize on the foundation of empirical evidence and ad hoc rationale that the specialty products are price elastic.
Hypothesis 3
In The Netherlands, the own-price elasticity of demand for specialty dairy products is elastic.
H0:0 < 𝜂𝑖𝑗 < −1 𝑓𝑜𝑟 𝑖 = 𝑗 = 𝑏𝑢𝑡𝑡𝑒𝑟𝑚𝑖𝑙𝑘, 𝑞𝑢𝑎𝑟𝑘, 𝑝𝑜𝑟𝑟𝑖𝑑𝑔𝑒, 𝑑𝑒𝑠𝑠𝑒𝑟𝑡𝑠, 𝑎𝑑𝑑𝑒𝑑 𝑣𝑎𝑙𝑢𝑒 𝑑𝑟𝑖𝑛𝑘𝑠 𝑎𝑛𝑑 𝑐𝑢𝑠𝑡𝑎𝑟𝑑
vs.
20
Fourthly, if we consider the difference between the price elasticity of demand for fresh dairy products of national brands and store brands, we see that economic theory suggests that national brands have an advantage in perceived quality over store brands. In some cases, empirical evidence confirms these arguments. However, in section (2) we also discuss that the gap of quality between national brands and store brands is decreasing, which affects the price elasticity of demand for both national brands and store brands. One of the most important arguments is from Raju et al., who found that the price elasticity of demand is correlated with the budget share of brands. Furthermore, we have seen that store brands are mainly present in price insensitive categories where the distinction amongst products in quality is very small and where store brands take a significant budget share. This is also one of the main reasons why store brands have a smaller own-price elasticity than national brands. Moreover, the empirical evidence we discussed in section…. also suggest that the own-price elasticity of demand for store brands is smaller in most studies than the own-price elasticity of national brands. Therefore, the fourth hypothesis is as follows
Hypothesis 4
The price elasticity of demand for store brands is smaller or equal to the price elasticity of demand for national brands if the budget share of store brands is greater than the budget share of national brands.
H0 : 𝑆𝐵𝑅𝜂𝑖𝑗 ≤ 𝑁𝐵𝑅𝜂𝑖𝑗 , 𝑖𝑓 𝑤𝑠𝑏 > 𝑤𝑛𝑏
vs.
H1 : 𝑆𝐵𝑅𝜂𝑖𝑗 > 𝑁𝐵𝑅𝜂𝑖𝑗 , 𝑖𝑓 𝑤𝑠𝑏 > 𝑤𝑛𝑏
This also leads us to the fifth hypothesis, which is stated as follows
Hypothesis 5
The own-price elasticity of demand for national brands is smaller or equal to the price elasticity of demand for store brands if the budget share of national brands is greater than the budget share of store brands.
H0 : 𝑁𝐵𝑅𝜂𝑖𝑗 ≤ 𝑆𝐵𝑅𝜂𝑖𝑗 𝑖𝑓 𝑤𝑛𝑏> 𝑤𝑠𝑏
vs.
21
The estimation and the discussion of the stated hypotheses are evaluated in the following sections. The hypotheses lead to insights that are not only relevant for marketeers, but also for economists interested in consumer behavior regarding price changes of products. For example, by providing estimates on the price elasticity of demand for fresh dairy products, economists might see what effect a proposed price increase would have on the demand for their products. Furthermore, for retailers it is interesting to see what consequences a price increase in a certain category would have on the market share of a brand or product. Governments might be interested in the price elasticity of demand for food products to contemplate tax decreases to encourage people to buy healthy food, such as yogurt or milk.
5. METHODOLOGY
The aim of this section is to provide the empirical model that we use to estimate the price elasticity of demand for fresh dairy products. Next to that, the empirical model is used for the comparison between the price elasticity of demand for fresh dairy products of national brands and store brands. As we have seen in the literature review, the choice of the empirical model and data used to analyze the price elasticity of demand strongly affects the results. In previous decades, one of the most commonly applied empirical models in demand analyses is the Almost Ideal Demand System (AIDS) developed by Deaton and Muellbauer (1980). In this study we use the same AIDS model to estimate the price elasticities of demand by using weekly-scanner data. Deaton and Muellbauer (1980) discuss that the AIDS model can be obtained from the cost function that represents the minimum expenditure for a utility-maximizing consumer
(2) ln 𝑋 (𝑢, 𝑝𝑖𝑡) = 𝛼0+ ∑𝑖=1𝑛 𝛼𝑖ln (𝑝𝑖)+12∑ ∑ 𝛾𝑖 𝑗 𝑖𝑗ln(𝑝𝑖𝑡) ln(𝑝𝑗𝑡)+ 𝑢𝛽0∏𝑛𝑖=1𝑝𝑖𝑡𝛽𝑖
where in (2) 𝑋 in is the minimum level of expenditure to obtain utility level 𝑢 given prices 𝑝𝑖𝑡 (𝑖 = 1, . . . , 𝑛, 𝑤ℎ𝑒𝑟𝑒 𝑛 = 9), 𝑖 𝑎𝑛𝑑 𝑗 index the products and 𝑡 indexes time, and 𝛼𝑖,
𝛽𝑖, 𝑎𝑛𝑑 𝛾𝑖𝑗 are parameters. From equation (2), we can derive the demand equation via the Shephard’s lemma in budget share form.
(3) 𝑤𝑖𝑡 = 𝛼𝑖+ ∑ 𝛾𝑖𝑗ln(𝑝𝑗𝑡) + 𝛽𝑖ln (𝑋𝑃𝑡
𝑡)
𝑛
22
where the left-hand side of equation (3), 𝑤𝑖𝑡, is the budget share of product 𝑖, 𝑝𝑗 is the nominal
price of product 𝑗, with 𝑗 = (1, … … , 𝑛) , 𝛼𝑖 is the constant in the share equation of product,
, 𝛽𝑖 is the total expenditure coefficient, 𝛾𝑖𝑗 is the price coefficient of product 𝑗 in the share
equation of product 𝑖, 𝑡 indexes time, 𝜖𝑖𝑡 is the error term, 𝑋 denotes total expenditures and
𝑃𝑡 is the Paasche price index, often referred to as the ‘corrected’ Stone price index (Moschini,
1995), and used by Asche and Wessels (1997) and Maynard (2000). 𝑃𝑡 can be formulated in
mathematical form as
(4) 𝑃𝑡 = ∑ 𝑤𝑗𝑡ln (𝑝𝑝𝑗𝑡
0𝑡)
𝑛 𝑗
where the left-hand side of equation (4), 𝑤𝑗𝑡, denotes the budget share of product 𝑗, 𝑝𝑗𝑡 is the
nominal price of product 𝑗 and 𝑝0𝑡 is the mean nominal price of product 𝑗 during the study
period. The (a) adding up, (b) homogeneity and (c) symmetry constraints are imposed on the demand system in (2) to be consistent with economic theory.
(a) ∑𝑖𝛼𝑖 = 1 ∑ 𝛾𝑖 𝑖𝑗 = 0 ∑ 𝛽𝑖 𝑖 = 0 (b) ∑𝑗𝛾𝑖𝑗 = 0
(c) 𝛾𝑖𝑗 = 𝛾𝑗𝑖, 𝑓𝑜𝑟 𝑎𝑙𝑙 𝑖, 𝑗 𝑤ℎ𝑒𝑛 𝑖 ≠ 𝑗
when these restrictions hold, the equation in (3) represents a demand system where the
budget shares sum up to unity (∑ 𝑤𝑖 = 1). In section (7) we test if the restrictions hold for
the demand system. If these restrictions do not hold, we impose the restrictions of (a), (b) and (c) on the demand system (3) (Deaton and Muellbauer, 1980). First of all, by dropping the last equation from the model and calculating the parameters for this equation we impose the adding up restriction (a). Secondly, according to economic theory rational consumers do not experience money illusion. Therefore, by changing prices 𝑝𝑗𝑡 into relative prices 𝑝𝑝𝑗𝑡
𝑛𝑡 we
23
in equation (5), we see that the price coefficient, 𝛾𝑖𝑗, of product 𝑗 in the share equation of product 𝑖 is now the relative price coefficient of product 𝑗 with respect to the price of product n, which equation we have dropped. To model the price elasticities of demand for fresh dairy products, 𝑋𝑡 is described as the total expenditures in the category of fresh dairy products.
Furthermore, the demand system is estimated using the method of seemingly unrelated regressions. This model will be applied to a budgeting allocation where consumers only will choose between substitutability within the category, which means that we have weak separability of demand (Deaton and Muellbauer, 1980).
We calculate the price elasticities of demand using the AIDS coefficients and the product expenditure shares. Following Blanciforti and Green (1983), we can calculate the Marshallian, or uncompensated, and the Hicksian, or compensated, own- and cross-price elasticities of demand after estimating the AIDS model coefficients. The Marshallian and Hicksian price elasticity of demand are both derived from respectively the Marshallian demand curve, where the budget constraint is held constant and utility is maximized, and the Hicksian demand curve, where utility is held constant and expenditures are minimized (Varian, 2010). The Marshallian, or uncompensated, price elasticity of demand is defined as the price elasticity of demand including both the substitution and income effect. The Hicksian, or compensated, price elasticity of demand is defined as the price elasticity of demand that ignores the income effect. First of all, we have
(6) 𝜂𝑖𝑗 = 𝛾𝑤𝑖𝑗
𝑖 − 𝛽𝑖(
𝑤𝑗
𝑤𝑖) − 𝛿𝑖𝑗
where the left-hand side of equation (6) is the Marshallian, or uncompensated, price elasticity of demand. Secondly, we have
(7) 𝜂𝑖𝑗 = 𝛾𝑤𝑖𝑗
𝑖 + 𝑤𝑗− 𝛿𝑖𝑗
where the left-hand side of equation (7) is the Hicksian, or compensated, price elasticity of demand. For both formulas holds, 𝛿𝑖𝑗 = 0 if i≠ j and 𝛿𝑖𝑗 = 1 if i=j. As mentioned earlier,
24
effect, we have estimated the uncompensated price elasticity of demand and obtain gross substitutes or complements by looking at the cross-price elasticities of demand.
6. DATA
In this study we rely on weekly retail scanner data of sales provided by Nielsen from the Dutch grocery market. The data set consists of weekly sales of fresh dairy products divided into 9 categories covering the period from week 1, 2016 until week 44, 2018. The sales of the 9 fresh dairy categories are divided into: buttermilk, quark, milk, porridge, cream, desserts, added value drinks, custard and yogurt. The total revenue of all products in a certain category, adds up to the total revenue of that category, namely fresh dairy. Although most categories are obvious in their definition, some categories might need some further explanation to get a comprehensive understanding of the category. Desserts are sweet specialty dairy products that are often used after dinner or as a treat. Added value drinks (Drinks AV) are dairy drink products which are processed with some added value, e.g. drinkable yogurt, chocolate milk. Cream is a high fat dairy product from milk before homogenization which can be used on top of desserts or in cooking.
25
statistics for the price elasticity of demand for fresh dairy products between national and store brands, also at the same retailer. Thirdly, Table 5 and 6 present the descriptive statistics for the robust estimations regarding the first two estimations with an extended data set by adding all grocery retailers in the Dutch market excluding discount retailers, e.g. Aldi and Lidl.
In Table 3, we see that revenues, or expenditures, reported per category range from €96,000 for porridge to €3,296,000 for yogurt. Furthermore, prices per category range from 0.90€/L for milk to 3.68€/L for desserts, leading to a mean price for fresh dairy of 1.45€/L. Budget shares range from 0.8% for porridge to 29.0% for yogurt. In Table 5, similar descriptions are presented including all retailers.
Comparing Table 3 with Table 5 we see that the budget shares for fresh dairy categories are not equal when we add more retailers, ranging from 1.0% for porridge to 27.5% for yogurt for all retailers. Revenues range from €281,000 for porridge to €7,429,000 for yogurt. The mean price for fresh dairy is 1.40€/L, ranging from 0.88€/L for milk to 3.70€/L for cream. Interesting to see is that the price for fresh dairy at one of the biggest retailers is greater than at all retailers.
Table 3.
Descriptive Statistics of weekly Dutch Retail Sales of Fresh Dairy Products (wk01, 2016 - wk44, 2018), One Retailer
Segments Revenue (€/wk)* Price (€/L) Shares
Mean Standard Deviation
Minimum Maximum Mean Standard Deviation
Minimum Maximum Budget Share Fresh dairy 11365 801 9214 13900 1.45 0.09 1.28 1.71 100% Buttermilk 364 43 270 532 0.92 0.06 0.77 1.00 3.2% Quark 1337 296 896 2268 2.45 0.13 1.93 2.77 11.8% Milk 2795 254 2069 3420 0.90 0.07 0.75 0.98 24.6% Porrdige 96 26 67 216 1.70 0.26 0.98 2.17 0.8% Cream 557 170 366 2055 3.67 0.31 3.17 4.24 4.9% Desserts 608 260 299 1823 3.68 0.71 1.85 6.31 5.4% Drinks, AV 1570 179 1108 2067 1.75 0.19 1.31 2.17 13.8% Custard 741 119 529 1092 1.22 0.10 0.97 1.40 6.5% Yogurt 3296 305 2244 3993 1.78 0.09 1.58 1.99 29.0%
*Note: Revenue reflects values per week divided by 1000
26
national brands of fresh dairy products is 2.01€/L, which is strongly higher than the mean price for store brands of fresh dairy products, which is 1.12€/L.
In Table 4, we present the data for the robust check regarding the estimation of the comparison between national brands and store brands. We see that the budget share of national brands is higher compared to one retailer, namely 58%. Store brands have a budget share of 42%. Next to that, the mean price for national brands is relatively lower than before, 1.78€/L and 1.08€/L for store brands.
Table 4.
Descriptive Statistics of Weekly Dutch Retail Sales Fresh Dairy Products (wk01, 2016 - wk44, 2018), Brands - One Retailer
Segments National Brands Store Brands
Budget Share** Mean Revenue* (€/week) Standard Deviation* (€/wk) Mean Price (€/L) Standard Deviation (€/L Budget Share** Mean Revenue* (€/week) Standard Deviation* (€/wk) Mean Price (€/L) Standard Deviation (€/L) Fresh dairy 52% 5859 465 2.01 0.15 48% 5506 541 1.12 0.09 Buttermilk 40% 144 21 1.07 0.08 60% 220 30 0.84 0.06 Quark 53% 702 295 3.36 0.45 47% 634 63 2.40 0.10 Milk 17% 476 40 1.21 0.07 83% 2319 229 0.85 0.08 Porrdige 100% 96 27 1.70 0.26 0% 0 1 0.28 0.75 Cream 16% 90 29 5.11 0.50 84% 467 152 3.70 0.18 Desserts 73% 442 222 3.83 0.79 27% 166 132 3.74 1.74 Drinks, AV 91% 1434 189 1.84 0.24 9% 136 55 2.27 1.75 Custard 66% 486 128 1.44 0.17 34% 255 52 1.09 0.12 Yogurt 60% 1989 251 2.47 0.20 40% 1307 116 1.39 0.10
Note:* Revenue reflects values per week divided by 1000, **budget share represents percentage of total budget
27
drop the category porridge in the estimation of the comparison between national brands and store brands.
Table 5.
Robust Descriptive Statistics of Weekly Dutch Retail Sales Fresh Dairy Products (wk01, 2016 - wk44, 2018), All retailers
Segments Revenue (€/wk)* Price (€/L) Shares
Mean Standard Deviation
Minimum Maximum Mean Standard Deviation
Minimum Maximum Budget Share Fresh dairy 27,014 1,692 21,500 32,400 1.40 0.07 1.27 1.62 100% Buttermilk 980 126 716 1,452 0.91 0.05 0.82 0.98 3.6% Quark 2,968 392 2,000 4,146 2.46 0.08 2.20 2.68 11.0% Milk 6,596 519 5,058 7,760 0.88 0.06 0.77 0.95 24.4% Porrdige 281 45 202 409 1.66 0.17 1.17 2.01 1.0% Cream 1,327 384 880 4,566 3.70 0.32 3.27 4.19 4.9% Desserts 1,621 412 979 3,956 3.20 0.38 2.16 4.87 6.0% Drinks, AV 3,648 236 2,669 4,178 1.62 0.09 1.38 1.85 13.5% Custard 2,164 219 1,629 2,713 1.20 0.06 1.07 1.32 8.0% Yogurt 7,429 557 5,088 8,490 1.67 0.06 1.56 1.81 27.5%
Note: *Revenue reflects values per week divided by 1000
Table 6.
Robust Descriptive Statistics of Weekly Dutch Retail Sales Fresh Dairy Products (wk01, 2016 - wk44, 2018), Brands - All retailers
Segments National Brands Store Brands
Budget Share** Mean Revenue* (€/week) Standard Deviation* (€/wk) Mean Price (€/L) Standard Deviation Budget Share** Mean Revenue* (€/week) Standard Deviation* (€/wk) Mean Price (€/L) Standard Deviation (€/L) Fresh dairy 58% 15636 863 1.79 0.07 42% 11379 1088 1.08 0.09 Buttermilk 48% 466 62 1.03 0.05 52% 513 69 0.82 0.06 Quark 58% 1729 342 2.95 0.21 42% 1238 110 2.33 0.08 Milk 28% 1880 99 1.10 0.04 72% 4717 439 0.82 0.07 Porrdige 100% 281 45 1.66 0.17 0% 1 1 4.13 3.00 Cream 29% 386 97 3.81 0.17 71% 941 299 3.89 0.25 Desserts 77% 1255 265 3.03 0.32 23% 367 248 5.20 1.80 Drinks, AV 92% 3340 227 1.70 0.11 8% 308 65 2.40 1.70 Custard 69% 1502 195 1.37 0.08 31% 662 75 1.09 0.10 Yogurt 65% 4798 374 2.17 0.08 35% 2631 239 1.27 0.06
28 7. RESULTS
In this section, we discuss the estimation results using model (5) discussed in section (5). The estimation results will include the coefficients of the estimated variables in the demand system. With the aid of the coefficients, we can compute the values of uncompensated price elasticities of demand following Blaniciforti and Green (1983). First, we estimate the model using weekly scanner-data including fresh dairy products from one of the biggest grocery retailers in The Netherlands from week 1, 2016, to week 44, 2018. After that, we present the uncompensated price elasticities of demand for fresh dairy products. Next, we present a comparison between the uncompensated price elasticities of demand for national brands and store brands of fresh dairy products. Second, we do a robust check by estimating the same model with data from the same period but including the full grocery retailer market, excluding discount retailers. Third, we present some limitations with respect to available data to present the price elasticities of demand for fresh dairy products. Before we run the regression, we test if the restrictions imposed by economic theory hold. The estimated coefficients of the first estimation using the model without restrictions are presented in Table A1(Appendix). Table A1 shows that the homogeneity restriction and symmetry restrictions do not hold. Therefore, we impose these restrictions as discussed in section (5). Moreover, when autocorrelation is detected using the Durbin-Watson test, it is corrected by the Prais-Winsten procedure.
7.1 Fresh dairy products
The coefficient estimates of the AIDS model for fresh dairy products including the restrictions are presented in Table A2 (Appendix). Although the estimates are not particularly meaningful, they provide valuable information about the significance of the estimated variables and indirect information about the price elasticities of demand. In Table 7 we see
that for most fresh dairy products the own-price coefficients, 𝛾𝑖𝑗 𝑤𝑖𝑡ℎ 𝑖 = 𝑗, are significant at
the 1% significance level, except for buttermilk. Furthermore, we see that the expenditure coefficient (𝛽𝑖) for buttermilk, milk and custard is statistically significant at the 1% significance level and for desserts at the 10% significance level. Next to that, we see that the
cross-price coefficients, 𝛾𝑖𝑗 𝑤𝑖𝑡ℎ 𝑖 ≠ 𝑗, are mostly insignificant, where 11 out of 36
29
Focusing on the uncompensated price elasticities presented in Table 7, the own-price elasticities are in line with economic theory and the law of demand, i.e. the price elasticity of demand for fresh dairy products is negative. A new valuable insight is that for some products the own-price elasticity is determined, which have not been reported yet (i.e. for custard, desserts, porridge and quark). Amongst the fresh dairy products, only milk 0.34), cream (-0.04) and yogurt (-0.73) show to be price inelastic, suggesting that these fresh dairy products are necessities (Maslow et al., 1970). Moreover, this is in line with Raju et al. (1995), who states that categories with a high store brand share are mainly price inelastic.
The other own-price elasticities of demand for fresh dairy products, when statistically significant, are price elastic and high in magnitude compared to previous studies, ranging from -1.69 for added value drinks to -2.67 for porridge. This means that these dairy products are luxury goods, or specialty products, and therefore relatively more price elastic. When the prices of the specialty products increase, people tend to strongly react to the price change and the demand decreases relatively more than the price increases. We can reject the null-hypothesis of null-hypothesis 1 saying that all fresh dairy products are inelastic. Interestingly, and again in line with Raju et al. (1995), is that the own-price elasticities are mainly high in magnitude for fresh dairy products with a relatively great national brand share, i.e. for added value drinks, porridge, desserts, quark and custard holds that the national brand share is greater than 50%. In addition, Ally, Meng, Chakroborty, Seaton, Holmes, Angus, Guo, Hill, Brennan and Meier (2014) discuss that national brands’ promotional activity is greater than store brands’ promotional activity, which increases effects of price changes on demand. Therefore, categories with relatively high national brand shares (i.e. greater than 50%) show to be price elastic, with only yogurt as an exception. And categories with relatively low national brand shares (i.e. smaller than 50%), show to be price inelastic, except for buttermilk. Buttermilk, although insignificant, is almost perfectly elastic by having a price elasticity of -1.01. We cannot reject hypothesis 2 and we cannot reject hypothesis 3, which is in line with previous empirical results and the ad hoc rationale for the fresh dairy products whose price elasticity of demand have never been reported yet.
30
for custard are positive for fresh dairy products implying that custard is a substitute for other fresh dairy products, i.e. statistically significant for quark, milk, added value drinks and desserts. Next to that, the cross-price elasticity for desserts and quark (0.39 and 0.21) imply that the products are significant substitutes.
All in all, we see that the price elasticity of demand shows that milk, cream and yogurt are basic dairy products. The price elasticity of demand of the other fresh dairy products show that these products are relatively more luxury products, also specialty dairy products. Therefore, we cannot reject the null-hypothesis of hypothesis 1 for milk, cream and yogurt. However, we can reject the null-hypothesis of hypothesis 1 for all other fresh dairy products (i.e. buttermilk, added value drinks, quark, custard and porridge). Furthermore, the results are in line with previous studies and the ad hoc rationale, so we cannot reject the null-hypothesis of hypothesis 2 and hypothesis 3.
Table 7:
Uncompensated Own- and Cross Price Elasticities of Demand for Fresh Dairy Products, One Retailer
Buttermilk Quark Milk Porridge Cream Desserts Drinks AV Custard Yogurt Buttermilk -1.01 0.33 0.20 -0.04 0.06 -0.01 0.28 0.23 0.05 Quark 0.05 -1.82*** -0.08 -0.02 -0.03 0.21*** 0.50*** 0.20*** -0.22 Milk 0.00 0.01 -0.34*** -0.01 -0.15*** 0.02 -0.04*** -0.05** -0.23* Porridge -0.18 -0.22 -0.23 -2.67*** 2.21*** -0.05 0.06 0.08 0.28 Cream 0.01 -0.05 -0.81*** 0.38*** -0.04*** 0.40*** -0.14 0.08 -0.83 Desserts -0.07 0.39*** -0.16 -0.02 0.33*** -1.98*** -0.11 0.09*** -0.29 Drinks AV 0.03 0.47*** -0.08*** 0.00 -0.04 0.01 -1.69*** 0.15*** 0.30 Custard 0.06 0.30*** -0.41** 0.00 0.03 0.08*** 0.20*** -2.13 0.20 Yogurt -0.03 -0.06 -0.25* 0.01 -0.14 -0.01 0.12 0.09** -0.73***
Note: Triple, double and single asterisks denote respectively that parameters are significant at the 1%, 5% and 10% level. Coefficient standard errors in parentheses. Own-price elasticities are bold
7.2 National brands and store brands
31
The uncompensated price elasticities of demand are presented in Table 8. Table 8 shows that the own-price elasticities between national brands and store brands are elastic and mostly in line with previous studies. As in the study of Huang et al. (2004) we see that the own-price elasticities of national brands and store brands are generally close to each other. Furthermore, Table 8 shows that the own-price elasticities of demand for national brands is on average smaller than for store brands. We also see this in other price elasticity studies about competition in national brands and store brands (Binkley et al., 2001; Bergès et al., 2009). However, the own-price elasticity of demand for basic dairy products is greater for national brands than for store brands. Firstly, national brands of basic dairy products, e.g. milk and cream, have an own-price elasticity that is close to or greater than the own-price elasticity of store brands. Especially for milk, we see that national brands (-1.98) have a strong own-price elasticity of demand compared to store brands (-1.21). This means that when the price of national branded milk increases, people tend to greatly decrease their consumption of national branded milk.
32 7.3 Robust results for fresh dairy products
Firstly, the estimated AIDS coefficients of the price elasticity of demand for fresh dairy products are reported in Table A3 (Appendix). As in paragraph (7.1), the estimates present valuable information about the significance of the price coefficients, which are used to compute the uncompensated price elasticities of demand. If we take a look at Table A3, we see immediately, again, that most own-price elasticities of demand are significant at the 1% significance level, except for buttermilk. Next to that, 17 out of 36 cross-price elasticities of demand are significant at the 10% significance level, 14 out of 36 at the 5% significance level and 11 out of 36 at the 1% significance level. From the coefficients in Table A3, we compute the uncompensated price elasticities of demand presented in Table 9.
In Table 9, the uncompensated own-price elasticities of demand for most products are following the law of demand. Comparing the own-price elasticities of demand with earlier results, we see that the magnitude of the own-price elasticities of demand are very similar with only small variations, ranging from a delta in magnitude of 0.68 for porridge to a similar price elasticity for buttermilk. However, the own-price elasticity of demand for cream is positive and inelastic in this estimation (0.636). As discussed before, economic theory
Table 8.
Uncompensated Own- and Cross Price Elasticities of Demand for Fresh Dairy Products – National Brands and Store Brands, One Retailer
Brand NBR SBR Buttermilk*** NBR -1.72 0.82 SBR 0.47 -1.54 Quark*** NBR -1.76 1.38 SBR 0.78 -2.42 Milk*** NBR -1.98 1.02 SBR 0.20 -1.21 Cream*** NBR -1.83 0.87 SBR 0.16 -1.17 Desserts*** NBR -1.13 0.36 SBR 0.32 -1.92 Drinks AV*** NBR -1.14 0.01 SBR 1.46 -1.12 Custard*** NBR -1.62 0.71 SBR 1.15 -2.31 Yogurt*** NBR -1.13 0.11 SBR 0.19 -1.17 Average Own-price -1.54 -1.61 Cross-price 0.66 0.59
33
suggests that products with a positive own-price elasticity of demand are giffen goods, which are rarely observed in real life. General economic explanations for this rarely observed product is that the increase in the price of the giffen good must decrease the purchasing power of people, so that they have less income to buy other products and buy more of the giffen good, or in this case an inferior good. This means that when people observe an increase in the price of cream, they experience a decrease in their purchasing power. Therefore people tend to buy more of cream and less other products.
The cross-price elasticities of demand are very similar to the earlier results with some few exceptions. Interesting to see is that most products have a negative cross-price elasticity of demand with milk, except for buttermilk and quark. Therefore, most products are complements of milk. This means that when the price of milk increases, not only the demand for milk decreases, but also the demand for the complements. An explanation of this fact could be that milk is used to produce other dairy products, including fresh dairy products. When the price of milk increases it affects the prices of other fresh dairy products as well, since milk is used to produce the other dairy products. Therefore, by an increase in the price of milk, the price of the other dairy products increase leading to a decrease in demand. However, this does not explain the substitution relationship with buttermilk and quark. Therefore, the cross-price elasticities of milk is out of scope of this study.
Table 9.
Robust Uncompensated Own- and Cross Price Elasticities of Demand for Fresh Dairy Products, All Retailers
Buttermilk Quark Milk Porridge Cream Desserts Drinks AV Custard Yogurt
Buttermilk -1.010 0.271 0.549 0.154 -0.619*** -0.134** 0.564* 0.758*** -0.334 Quark 0.041 -1.726*** 0.028 0.002 0.015 0.040 0.543*** 0.262** -0.360 Milk 0.047 0.055 -0.270*** -0.018 -0.090*** -0.007 -0.157*** -0.096*** -0.235*** Porridge 0.494 0.034 -0.470 -2.293*** 1.821*** -0.280 -0.284 -0.711* 0.675 Cream -0.512*** 0.027 -0.556*** 0.386*** 0.636*** 0.809*** -0.737*** 0.110 -1.370 Desserts -0.154** 0.003 -0.279 -0.057 0.631*** -1.290 -0.309** 0.011 -0.352*** Drinks AV 0.114* 0.476*** -0.297*** -0.020 -0.248*** -0.079** -1.424*** 0.130* 0.520 Custard 0.284*** 0.326** -0.459*** -0.097* 0.055 0.029 0.135* -2.118*** 0.393 Yogurt -0.089 -0.131 -0.274*** 0.025 -0.235 -0.032*** 0.227 0.147 -0.680***
34 7.4 Robust results for national brands and store brands
The robust estimates of the AIDS coefficients for national brands and store brands are presented in table A5. Here we check if the uncompensated price elasticities of demand for national brands and store brands are also in line with the earlier estimation. 16 own-price and cross-price elasticities of demand were estimated in the model. In Table A5 we see that 12-own price coefficients and 12 cross-price coefficients are significant at the 1% significance level. In the results before, we have seen that all own- and cross price elasticities of demand were significant at the 1% significance level.
Zooming in on the own-price elasticities of demand presented in Table 10, we see that the average for national brands is smaller than in the results before. As in the first analysis, we see that the own-price elasticity of demand is smaller for national brands compared to store brands. Interesting to see is that the own-price elasticity of demand of national brands for milk, cream and desserts is now inelastic. Same goes for store brands, except for desserts. This means that these products are relatively less price sensitive compared to the results before. An explanation of this could be that with the use of more retailers in the data set, promotional effects cancel out in the greater data set. Therefore, the reaction to prices is smaller. Although the own-price elasticities of demand have decreased on average compared to the first estimation, the cross-price elasticities of demand have increased on average. Next to that, the cross-price elasticity of demand for national brands is now larger than the cross-price elasticity for store brands. This means that when the price of national brands increase, the quantity demanded for store brands increase relatively stronger than the other way around.
35
further research needs to be done to obtain stronger results. Therefore, the results regarding hypothesis 4 and 5 are not robust.
8. LIMITATIONS/FUTURE RECOMMENDATIONS
So far, we have estimated the price elasticities of demand for fresh dairy products including a comparison between the price elasticity of demand for national brands and store brands. As most studies, our review has some limitations we have to discuss regarding the obtained results. This study is subject to some caveats regarding to data limitations, data aggregation and omitted variables. Firstly, we start with discussing what data limitations we have, secondly we discuss the problem of aggregation and thirdly we discuss some omitted variables that could influence the computed price elasticities of demand.
8.1 Data limitations
One of the main limitations in this study is data limitation. Retail scanner data provides us with a great data set including information about prices, volume and revenue. In many previous studies, household size or composition information are included in the demand system, specifying facts about income, demographics, composition or any other household
Table 10.
Robust Uncompensated Own- and Cross Price Elasticities of Demand for Fresh Dairy Products – National Brands and Store Brands,
All retailers Brand NBR SBR Buttermilk*** NBR -1.59 0.65 SBR 0.53 -1.59 Quark*** NBR -1.75 0.93 SBR 1.02 -2.27 Milk*** NBR -0.40 -0.57 SBR -0.24 -0.77 Cream NBR -0.81 -0.16 SBR -0.08 -0.93 Desserts NBR -0.94 0.05 SBR -0.20 -1.19 Drinks, AV NBR -1.05 0.00 SBR 0.53 -1.02 Custard*** NBR -1.44 0.46 SBR 0.99 -2.03 Yogurt* NBR -1.06 0.08 SBR 0.12 -1.15 Average Own-price -1.13 -1.37 Cross-price 0.18 0.34
