- Email: [email protected]
What is the value of bottled water? Empirical evidence from the Italian retail market
Author’s Accepted Manuscript What is the value of bottled water? empirical evidence from the italian retail market Domenico Carlucci, Bernardo De Gennaro, Luigi Roselli www.elsevier.com/locate/wre
PII: DOI: Reference:
S2212-4284(16)30049-4 http://dx.doi.org/10.1016/j.wre.2016.07.001 WRE77
To appear in: Water Resources and Economics Received date: 3 November 2015 Revised date: 25 May 2016 Accepted date: 13 July 2016 Cite this article as: Domenico Carlucci, Bernardo De Gennaro and Luigi Roselli, What is the value of bottled water? empirical evidence from the italian retail m a r k e t , Water Resources and Economics, http://dx.doi.org/10.1016/j.wre.2016.07.001 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
What is the value of bottled water? Empirical evidence from the Italian retail market Domenico Carlucci Department of Agricultural and Environmental Science, University of Bari Aldo Moro, Bari, Italy, Via Giovanni Amendola 165/A, 70126, Bari, Italia E-mail: [email protected] Bernardo De Gennaro Department of Agricultural and Environmental Science, University of Bari Aldo Moro, Bari, Italy, Via Giovanni Amendola 165/A, 70126, Bari, Italia E-mail: [email protected] Luigi Roselli (corresponding author) Department of Agricultural and Environmental Science, University of Bari Aldo Moro, Bari, Italy, Via Giovanni Amendola 165/A, 70126, Bari, Italia E-mail: [email protected] Abstract Bottled water has become a global business, and Italy is one of the largest producers and consumer countries in the world. However, the Italian bottled water market seems to have reached the maturity stage of its life cycle, and because competitive pressure has strongly increased in this market, producers need to revise their marketing strategies. A variety of products with different features and prices are now available on the market of bottled water, however it is unknown how the retail price of bottled water is affected by its attributes. We thus measured whether, and to what extent, the price of bottled water depends upon its extrinsic (brand, packaging, origin), and intrinsic characteristics (mineral composition). We estimated a hedonic price model using data collected via direct observation of the shelves in various Italian modern retail stores. Our results show that bottled water is highly differentiated and its retail price is mainly affected by extrinsic characteristics. On the basis of these results, several insights are provided for both practitioners and policy makers.
Keywords bottled water, natural mineral water, Italy, hedonic price, competitive advantage
1. Introduction Over the last few decades, the global consumption of bottled water has been increasing consistently even in countries where tap water is largely available and considered to be of an excellent quality (Wilk, 2006). This increase has been mostly linked to the changes in lifestyles and eating habits associated with economic growth (Ferrier, 2001; Wilk, 2006). Several studies have investigated consumer behaviour in order to identify the main reasons why people decide to buy more expensive bottled water rather than tap water. These studies have shown that bottled water consumption is mostly driven by consumers’ dissatisfaction
with the organoleptic quality of tap water (especially taste), followed by health-related concerns (Abrahams, Hubbell and Jordan, 2000; Doria, 2006; Doria, Pidgeon and Hunter, 2009; Van Der Linden, 2015; Viscusi, Huber and Bell, 2013). Although safety concerns and health improvement are sometimes considered as equivalent, the two factors act differently. In contexts where distrust in tap water suppliers exists (e.g. information regarding current or prior problems with tap water), risk perception and safety concerns are more prominent in driving bottled water consumption (Abrahams, Hubbell and Jordan, 2000; Doria, 2006; Doria, Pidgeon and Hunter, 2009; Hu, Morton and Mahler, 2011; Viscusi, Huber and Bell, 2013). In other cases, consumers may prefer bottled water simply because it is considered as healthier, but not necessarily safer, than tap water (Black, 2009; Doria, 2006; Wilk, 2006). The need for convenience is another important reason for choosing bottled water especially for out-of-home consumption (Doria, 2006; He, Jordan and Paudel, 2008; Van Der Linden, 2015; Viscusi, Huber and Bell, 2013). Today, a bottle of water can be bought almost everywhere and, when finished, it is easily disposed of in the trash, thus it is not necessary to carry around a bulky container all day. Italy is one of the highest producers and consumers of bottled water in the world (IBWA, 2015). In Italy, the consumption of bottled water began in the 1970s and, since then, annual per capita consumption has increased considerably from 47 litres in 1980 to a maximum of 192 litres in 2008 (Bevitalia, 2015). However, in the last few years (2005-2013), Italian consumption of bottled water has stagnated considering that annual per capita consumption has been ranging between 187 and 192 litres. This is probably due to the persistent economic crisis that has reduced the purchasing power of Italian households, but also other factors are relevant. Purification devices are increasingly used in bars and restaurants as well as homes to treat tap water by removing chlorine, as well as the bad taste and odours in order to provide safe and palatable drinking water as an alternative to bottled water (Greenlee et al., 2009). In addition, there is a growing consumer sensitivity regarding the environmental impacts of increasing bottled water consumption (Rani et al., 2012; Van Der Linden, 2015). In fact, the process of extraction, processing, packaging and transportation of bottled water involves considerable environmental impacts, including pollution, climate change and depletion of natural resources (Botto et al., 2011; Gleick and Cooley, 2009; Lagioia, Calabrò and Amicarelli, 2012; Nessi, Rigamonti and Grosso, 2012; Niccolucci et al., 2011; Torretta, 2013). Essentially, the Italian market of bottled water seems to have reached the maturity stage of its life cycle (Niccolucci et al., 2011; Rani et al., 2012) and, thus, competitive pressure has strongly increased. Bottled water producers thus need to revise their marketing strategies in order to reinforce their competitive advantages and defend their profits and market shares from competitors. A review of the economic literature concerning the bottled water market reveals that most studies have focused on two mainstreams: the analysis of consumer behaviour in order to investigate the reasons for the increase in bottled water consumption (Abrahams, Hubbell and Jordan, 2000; Black, 2009; Doria, 2006; Doria, Pidgeon and Hunter, 2009; Hu, Morton and Mahler, 2011; Van Der Linden, 2015; Viscusi, Huber and Bell, 2013; Wilk, 2006), and the assessment of the environmental impacts of bottled water consumption ( Botto et al., 2011; Gleick and Cooley, 2009; Lagioia, Calabrò and Amicarelli, 2012; Nessi, Rigamonti and
Grosso, 2012; Niccolucci et al., 2011; Torretta, 2013). To the best of our knowledge, to date, only a few studies have performed analyses in order to support firms in their decision-making problems (He, Jordan and Paudel, 2008; Rani et al., 2012). This study starts with the observation that a variety of products with different features and prices is now available on the Italian market of bottled water despite the simplicity of “water products”, however it is unknown how the retail price of bottled water is affected by its attributes. Our survey aims to fill this gap by measuring whether, and to what extent, the price of bottled water depends upon its extrinsic (brand, packaging, origin), and intrinsic characteristics (mineral composition). We used the hedonic price method (Rosen, 1974) to estimate the implicit prices associated with the main characteristics of bottled water. Estimates of the implicit prices of bottled water attributes can provide useful insights for both practitioners and policy makers. First, being aware of their production costs, firms involved in this market can use implicit prices to devise an optimal mix of attributes and more profitable marketing strategies. Second, the hedonic approach isolates the premium for various attributes that affect the environmental impact of bottled water consumption, such as the use of eco-friendly bottles and the distance between the water source and the point of sale (Botto et al., 2011; Gleick and Cooley, 2009; Gironi and Piemonte, 2010; Lagioia, Calabrò and Amicarelli, 2012; Nessi, Rigamonti and Grosso, 2012; Niccolucci et al., 2011).
2. An overview of the Italian market of bottled water In 2013, the Italian consumption of bottled water was over 12 billion litres, the seventh highest in the world after China (39), the USA (38), Mexico (31), Indonesia (18), Brazil (18), and Thailand (15) (IBWA, 2015). In terms of annual per capita consumption, Italy (190 L) is the first European country and third in the world after Mexico (255L) and Thailand (225L) (IBWA, 2015). The bottled water consumed in Italy is almost totally produced domestically and foreign trade is very limited (Bevitalia, 2015). Contrary to other countries in the world like the USA1, where purified water is largely sold as bottled water (IBWA, 2015), in Italy bottled water is essentially represented by so-called “natural mineral water” (Bevitalia, 2015) as specified in the Legislative Decree n. 176/2011 implementing the European Directive 2009/54/EC. By law, natural mineral water is clearly distinguished from other drinking water by its original purity (absence of chemical treatments) and its favourable health properties. The annual report of Bevitalia (2015) provides an overview of the Italian market of bottled water which is briefly described below. In 2013, the Italian bottled water industry included 143 companies of different sizes, with overall annual gross sales of 2.4 billion euros. The industry is highly concentrated considering that the four largest players (Nestlé Waters, San Benedetto, Norda, and Fonti Vinadio) control more than half of the domestic market. Since many companies adopt a multi-branding strategy, about 270 brands are available on the Italian market of bottled water. Most of these brands are regional or local, and only a few (about a dozen) 1
In the United States, about 44% of all bottled water is sold as “purified water” which is produced by distillation, deionization, reverse osmosis or other processes (Gleick and Cooley, 2009). The three largest brands of purified water sold in the US market are “Dasani” (Coca-Cola Company), “Aquafina” (Pepsi-Cola Company), and “Pure Life” (Nestlé Waters) (Gleick and Cooley, 2009).
are national brands. There are also various store brands that have reached a significant cumulative market share accounting for 9% of total sales. The largest segment of the Italian market of bottled water is represented by still (non-sparkling) water which accounts for 65% of sales volume, while carbonated and naturally carbonated waters account for 19% and 16%, respectively. Still water is ideal for ordinary daily consumption, while sparkling water, given its particular taste, is preferred at mealtimes and is mainly sold in restaurants. Bottled water with a low mineral content (50-500 mg/L) is the best-selling, accounting for 59% of sales volume, while bottled water with a very low mineral content (<50 mg/L) and medium-high mineral content (>500 mg/L) represent minor market segments (14% and 27%, respectively). Water with a very low mineral content is almost totally tasteless although it is recommended for people who have diuretic problems. It is also more suitable for special purposes such as the formulation of baby foods (e.g. milk powder). Conversely, water with a medium-high mineral content has a stronger and diversified taste, depending on the composition of the dissolved salts, and is recommended for people who need to increase their daily intake of minerals (e.g. people who play sport). Bottled water is mainly packaged (83% of sales volume) in plastic bottles (polyethylene terephthalate - PET) with different sizes: larger plastic bottles (1L, 1.5L, 2L2), being preferred for daily water consumption at home, are sold the most (75% of sales volume); smaller plastic bottles (0.25L, 0.33L, 0.50L) are specifically aimed at meeting consumers’ convenience needs (out-of-home consumption) and account for only 8% of sales volume. Plastic bottles are often closed with a practical “push-pull” cap which makes drinking easier, especially for children (this kind of bottle is thus often called a “baby-bottle”). Glass bottles (usually of 1L) are used to a lesser extent (17% of sales volume) and are mostly sold in restaurants, as they are considered to be more elegant. In the last few years, some bottled water producers have also introduced bottles made from eco-friendly materials, represented by recycled, partly recycled or biodegradable plastic (Lagioia, Calabrò and Amicarelli, 2012; Gironi and Piemonte, 2010; Nessi, Rigamonti and Grosso, 2012; Niccolucci et al., 2011). These are also called “bio-bottles” for which, however, market data are not yet available. Finally, concerning where the retail sales of bottled water occur, modern retail stores (hypermarkets, supermarkets, minimarkets and discount stores) play a predominant role, accounting for 71% of sales volume. The remaining sales follow the Ho.Re.Ca. channel (18%) and traditional stores (11%).
3. Method 3.1 Hedonic price model We used the hedonic price model to analyse the relationship between the price and the main attributes of bottled water. The hedonic price model has been successfully employed to analyse the markets of several food products, including wine (Nerlove, 1995; Oczkowski, 1994; Schamel, 2006; Steiner, 2004; Boatto, Defrancesco and Trestini, 2011; Panzone, 2011), carbonated beverages (Martinez-Garmendia, 2010), fresh meat (Loureiro and McCluskey, 2000; Ward, Lusk and Dutton, 2008), eggs (Satimanon and Weatherspoon, 2
Italian Regulation (Legislative Decree n. 176/2011) prohibits the use of containers holding more than 2 liters for bottling natural mineral water.
2010), apples (Carew, Florkowski and Smith, 2012), yogurt (Carlucci et al., 2013), coffee (Schollenberg, 2012) and olive oil (Carlucci et al., 2014; Karipidis, Tsakiridou and Tabakis, 2005; Muñoz, Moya and Gil, 2015; Roselli, Carlucci and De Gennaro, 2016). The hedonic approach is borrowed from Lancaster’s (1966) theory of demand which states that consumers derive utility directly from the quality attributes embedded in a product rather than from the product itself. In other words, any differentiated product can be considered to be a bundle of quality attributes that are independently valued by consumers at the time of purchase. Rosen (1974) subsequently developed a theoretical model demonstrating that the observed price of a product can be considered as the sum of the prices associated with each of its quality attributes. Although these prices are not explicitly expressed by the market, they can be estimated by employing a regression equation, i.e. the hedonic price model, which expresses the price of a product (directly observable) as a function of its attributes (directly or indirectly observable). According to Rosen’s (1974) formulation, a hedonic price model can be specified as follows: (1) where P is the price of a product, and Z = z1, z2, . . . , zj, . . . , zn is a vector of n objectively measured attributes that completely describe the product quality. After estimating the hedonic price equation, a partial derivative with respect to the attribute j, ∂P(Z)/∂z j, can be interpreted as the implicit or shadow price of the specific attribute j. This theoretical model is based on the assumption that the market is in equilibrium and there is perfect competition. In this situation, consumers maximize utility by choosing the available products under budget constraints, and firms maximize profits given the available technology and factor prices (Rosen, 1974). Consequently, being related to both supply and demand conditions, implicit prices cannot be considered merely as indicators of consumer preferences (Costanigro and McCluskey, 2011; Oczkowski, 1994; Rosen, 1974; Schamel, 2006).
3.2 Data collection Data on the prices and characteristics of bottled water were collected via direct observation of the shelves in various modern retail stores located in the province of Bari. As already mentioned, almost three-quarters of Italian retail sales of bottled water occur in hypermarkets, supermarkets, minimarkets and discount stores (Bevitalia, 2015). We collected data in the first months of 2015 (January and February) from two hypermarkets (Auchan and Coop), two supermarkets (Famila and Sigma), two minimarkets (Despar and Simply) and two discount stores (Lidl and Eurospin) of the main retail chains operating in the province of Bari. Obviously, this is only an explorative analysis and the reliability of the results could be further improved by extending the survey to a larger area where differences may arise in price levels and product
variety. However, the area of data collection seems to be sufficiently representative at a national level3 to test the mechanisms of price formation. In each of the selected stores, we adopted a snapshot-type procedure for data collection: each store was visited just once, when we directly and simultaneously recorded the retail prices and the characteristics of all drinking bottled waters placed on dedicated shelves. Because in some cases the same product was offered in different types of packaging or at different prices, each item was always considered as a separate observation. We only excluded products sold at reduced prices (special offer) when the “normal” prices were not indicated in order to prevent any possible bias in the results due to the inclusion of prices that were only temporarily reduced4. By examining the products and their labelling, the following information was recorded for each product: price, packaging characteristics (type of pack, bottle size, bottle material, type of cap), chemical composition (total salt content, sodium content, presence/type of effervescence), brand and place of bottling.
3.3 Data set Using the criteria described above, we collected a data set containing 374 observations. A preliminary analysis of the data set was carried out by calculating descriptive statistics regarding both the total sample and specific sub-samples grouped according to particular attributes (Table 1).
Table 1. Summary statistics of the sample
374
Min 0.08
Price/litre* Max 3.52
315 59
0.08 0.12
51 133 54 136
N. cases Total sample Type of pack single bottle multipack Bottle size 2.00 L 1.50 L 0.75-1.00 L ≤ 0.50 L Bottle material plastic glass eco-friendly material Type of cap standard push-pull Salt content
Mean 0.50
Std dev 0.45
3.52 3.49
0.48 0.59
0.44 0.51
0.08 0.09 0.25 0.26
0.28 0.66 3.52 3.49
0.17 0.28 0.83 0.70
0.05 0.11 0.72 0.42
332 17 25
0.08 0.89 0.13
1.67 3.52 0.49
0.45 1.74 0.31
0.32 0.94 0.11
347 27
0.08 0.64
3.52 1.67
0.45 1.15
0.42 0.40
3
The province of Bari is located in Apulia, southern Italy. With about 1.27 million inhabitants, the province of Bari represents the fifth most-populous province of Italy, after Rome (4.34 million), Milan (3.19 million), Naples (3.12 million), and Turin (2.29 million) (Comuni Italiani, 2016). The main demographic indicators of the province of Bari are very similar to the national averages. For the province of Bari, the average age of the population is 43, the share of younger people (below 15) is 14%, the share of elders (over 65) is 20%, the average number of people per household is 2.6, and annual per capita income is 9,300 Euro. At a national level, the average age is 44, the share of younger people is 14%, the share of elders is 21%, the average number of people per household is 2.4, and annual per capita income is 12,100 Euro (Comuni Italiani, 2016). 4
The number of products sold at reduced prices (special offer) and thus excluded from the data collection was very limited (one or two products for each of the selected stores).
medium-high mineral (500-1500 mg/L) low mineral (50-500 mg/L) very low mineral (<50 mg/L) Effervescence still naturally carbonated carbonated Sodium content low sodium (<20 mg/L) medium-high sodium (>20mg/L) Store format hypermarket supermarket minimarket discount store Distance <500 km 500-1,000 km >1,000 km Brand San Benedetto Levissima Rocchetta Sant'Anna Vera Ferrarelle Uliveto Sangemini store brands other brands
109 234 31
0.08 0.09 0.17
1.40 3.52 1.25
0.43 0.53 0.48
0.23 0.54 0.30
215 92 67
0.08 0.16 0.09
3.52 1.40 3.52
0.52 0.45 0.50
0.45 0.23 0.66
296 78
0.08 0.09
3.49 3.52
0.49 0.52
0.39 0.63
160 117 75 22
0.08 0.12 0.13 0.09
3.52 1.67 1.67 0.54
0.54 0.51 0.47 0.20
0.57 0.34 0.34 0.10
139 161 74
0.08 0.11 0.16
1.40 1.60 3.52
0.33 0.51 0.78
0.24 0.32 0.76
40 31 25 23 18 17 14 12 9 185
0.22 0.28 0.33 0.25 0.16 0.26 0.35 0.53 0.11 0.08
1.60 1.67 0.86 0.90 0.60 1.40 0.74 1.00 0.38 3.52
0.63 0.70 0.49 0.41 0.30 0.52 0.55 0.74 0.24 0.45
0.45 0.46 0.15 0.17 0.14 0.28 0.14 0.20 0.10 0.54
*Prices are expressed in €
First, a wide variability in price was detected in the overall sample considering that the unit price of bottled water ranged from a minimum of 0.08 Euro/L to a maximum of 3.52 Euro/L, with a mean of 0.50 Euro/L. However, the frequency distribution of unit price was skewed to the right, with low unit prices showing much higher frequencies: the unit price was lower than 1.39 Euro/L for 95% of the observations (Figure 1).
Figure 1. Price frequency distribution of bottled water
0.18
Percentiles 5% 25% 50% 75% 95%
0.16
Relative frequency
0.14
Price (euro/L) 0.13 0.23 0.39 0.59 1.39
Obs 374 Mean 0.50 Std. Dev. 0.45 Variance 0.21
0.12 0.10 0.08 0.06 0.04 0.02 0.00 0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
Price (Euro/L)
The price variability of the data set seems to be primarily related to the different types of packaging. In fact, water is sold as single bottles or multipacks (including four or six bottles), and bottle size varies from a minimum of 0.25L to a maximum of 2L. In addition, bottles are made of plastic (PET), glass, or eco-friendly materials (recycled, partly recycled or biodegradable plastic) and the bottles are closed with standard or push-pull caps. We found that the average price of water sold in smaller bottles (1 L or less), in glass bottles, and in bottles with push-pull caps was much higher than the average price of water sold, respectively, in larger bottles (1.5-2L), in plastic bottles, and in bottles with standard caps. On the other hand, regarding the intrinsic attributes of bottled water (total salt content, sodium content, presence/type of effervescence), only limited differences in the average prices were observed, while distance (calculated between the place of bottling and the point of sale) affected the average price of bottled water notably. Finally, we found that there were large price differences among different types of store (in particular, for discount stores compared to the other stores) and different brands.
3.4 Empirical Model We used a stepwise procedure to specify several hedonic price models which were estimated through the Ordinary Least Squares (OLS) method and checked for statistical validity. Among the valid models, the best (most parsimonious) model was selected through a comparison of Akaike Information Criteria (AIC) and Bayesian Information Criteria (BIC) (Burnham and Anderson, 2003; Snipes and Taylor, 2014). The best model is the following:
(2)
However, we also present and discuss an extended version of the best model including both significant and non-significant predictors which were also tested:
(3) The variables included in the empirical models are listed in Table 2. The unit price of bottled water is the dependent variable (price) which is a continuous variable. Just one explanatory variable, the distance between the place of bottling and the point of sale (distance), is also a continuous variable, while since the other explanatory variables are categorical, they were transformed into one or more dummy variables. We chose the double-log formulation due to its better fit to the data.
Table 2. Variables of the empirical models Variables Dependent variable price Regressors multipack bottle size
bottle material
push-pull salt content
effervescence
low sodium store
distance brand
Type
Description
continuous variable
price per litre (€)
dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy continuous variable dummy dummy dummy dummy dummy dummy dummy dummy
multipack (4 or 6 bottles) = 1; single bottle = 0 small (≤0.50 L) = 1; otherwise = 0 medium (0.75-1.00 L) = 1; otherwise = 0 large (1.50 L) = 1; otherwise = 0 extra-large (2.00 L) = 1; otherwise = 0 (baseline) glass = 1; otherwise = 0 eco-friendly material = 1; otherwise = 0 plastic = 1; otherwise = 0 (baseline) push-pull = 1; standard = 0 medium-high mineral (>500 mg/L) = 1; otherwise = 0 low mineral (50-500 mg/L) = 1; otherwise = 0 (baseline) very low mineral (<50 mg/L) = 1; otherwise = 0 still = 1; otherwise = 0 (baseline) naturally carbonated = 1; otherwise = 0 carbonated = 1; otherwise = 0 low sodium (<20 mg/L) = 1; medium-high sodium (>20mg/L) = 0 hypermarket = 1; otherwise = 0 (baseline) supermarket = 1; otherwise = 0 minimarket = 1; otherwise = 0 discount store = 1; otherwise = 0 distance between the place of bottling and the point of sale (km) San Benedetto = 1; otherwise = 0 Levissima = 1; otherwise = 0 Rocchetta = 1; otherwise = 0 Sant'Anna = 1; otherwise = 0 Vera = 1; otherwise = 0 Ferrarelle = 1; otherwise = 0 Uliveto = 1; otherwise = 0 Sangemini = 1; otherwise = 0
dummy dummy
other brands = 1; otherwise = 0 store brands = 1; otherwise = 0 (baseline)
4. Results The estimation results are reported in Table 3, while the most important performance indicators and statistical tests are summarized in Table 4. The two models show only negligible differences in the values of estimated parameters. In addition, both models show a good overall significance (F-statistic with a P-value much lower than 0.01 in both cases) and a high capability to explain the variability in the data set (adjusted R-squared equal to 0.90 in both cases). Checks for multicollinearity, heteroscedasticity and non-normality of residuals ruled out the existence of such statistical problems for both models.
Table 3. Estimation results: OLS regression
constant multipack bottle size large medium small bottle material glass eco-friendly push-pull salt content medium-high mineral very low mineral effervescence naturally carbonated carbonated low-sodium store supermarket minimarket discount store ln distance brand San Benedetto Levissima Rocchetta Sant'Anna Vera Ferrarelle Uliveto Sangemini other brands
coefficient -3.61 -0.01
extended model S.E. p-value 0.18 <0.001 *** 0.04 0.80
coefficient -3.60
parsimonious model S.E. p-value 0.17 <0.001 *** not included
0.27 0.69 0.97
0.05 0.07 0.05
<0.001 <0.001 <0.001
*** *** ***
0.28 0.70 0.99
0.04 0.06 0.04
<0.001 <0.001 <0.001
*** *** ***
1.18 -0.03 0.68
0.09 0.05 0.06
<0.001 0.60 <0.001
***
1.18
***
***
0.68
0.08 <0.001 not included 0.06 <0.001
-0.05 -0.13
0.09 0.05
0.53 0.01
***
-0.13
not included 0.05 <0.001
***
0.11 -0.02 -0.21
0.08 0.03 0.05
0.18 0.51 <0.001
***
-0.21
not included not included 0.04 <0.001
***
0.09 0.09 -0.45 0.28
0.03 0.03 0.06 0.03
<0.001 <0.001 <0.001 <0.001
*** *** *** ***
0.09 0.10 -0.46 0.28
0.03 0.03 0.06 0.03
<0.001 <0.001 <0.001 <0.001
*** *** *** ***
0.28 0.44 0.66 0.42 0.16 0.53 0.30 1.02 0.38
0.06 0.06 0.06 0.07 0.07 0.10 0.09 0.08 0.07
<0.001 <0.001 <0.001 <0.001 0.01 <0.001 <0.001 <0.001 <0.001
*** *** *** *** ** *** *** *** ***
0.28 0.44 0.66 0.43 0.18 0.60 0.37 1.09 0.40
0.06 0.06 0.05 0.07 0.06 0.09 0.07 0.06 0.06
<0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001
*** *** *** *** *** *** *** *** ***
Notes: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.
Table 4. Estimation checks Type of check Goodness of Fit
OLS (extended model) Dependent variable = ln price per litre F (25 / 348) = 141.53 P-value (F) <0.0001 R2 = 0.91 Adjusted R2 = 0.90
OLS (parsimonious model) Dependent variable = ln price per litre F (20, 353) = 177.46 P-value (F) <0.0001 R2 = 0.91 Adjusted R2 = 0.90
***
Heteroscedasticity
Multicollinearity
Normality of residuals
Log-likelihood = 48.10 Akaike Information Criteria = -44.19 Bayesian Information Criteria = 57.83
Log-likelihood = 45.24 Akaike Information Criteria = -48.48 Bayesian Information Criteria = 33.93
Breusch-Pagan / Cook-Weisberg test for heteroscedasticity H0: the variance is constant H1: the variance is not constant Chi-square (1) = 0.62 p-value = 0.4307 Variance Inflation Factor (VIF) for regressors: multipack: 1.3; large: 3.5; medium: 2.8; small: 3.4; glass: 1.8; eco-friendly material: 1.4; push-pull: 1.4; medium mineral: 7.0; very low mineral: 4.1; naturally carbonated: 6.8; carbonated: 1.6; low-sodium : 2.4; supermarket: 1.3; minimarket: 1.3; discount: 1.3; ln distance: 2.6; San Benedetto: 5.5; Levissima: 4.5; Rocchetta: 3.7; Sant'Anna: 6.6; Vera: 3.1; Ferrarelle: 3.8; Uliveto: 3.5; Sangemini: 2.8; other brands: 9.2 (VIF greater than 10 indicates a multicollinearity problem) Jarque-Bera Test H0: residuals are distributed normally H1: residuals are not distributed normally Chi-square(2) = 1.947 p-value 0.378
Breusch-Pagan / Cook-Weisberg test for heteroscedasticity H0: the variance is constant H1: the variance is not constant Chi-square (1) = 0.25 p-value = 0.6144 Variance Inflation Factor (VIF) for regressors: large: 2.7; medium: 2.4; small: 2.7; glass: 1.5; push-pull: 1.3; very low mineral: 4.1; lowsodium : 2.1; supermarket: 1.3; minimarket: 1.3; discount: 1.2; ln distance: 2.3; San Benedetto: 5.3; Levissima: 4.3; Rocchetta: 3.6; Sant'Anna: 6.5; Vera: 3.0; Ferrarelle: 3.3; Uliveto: 3.0; Sangemini: 2.4; other brands: 9.0 (VIF greater than 10 indicates a multicollinearity problem)
Jarque-Bera Test H0: residuals are distributed normally H1: residuals are not distributed normally Chi-square(2) = 1.859 p-value 0.395
Because the frequency distribution of the unit price is skewed to the right, with low unit prices showing much higher frequencies, as shown in Figure 1, the sample mean may be sensitive to the outliers in the upper percentiles, and thus OLS regression may yield biased results. In order to test the robustness of the OLS estimates, we also performed a Median Regression which uses the median (50th percentile) as a measure of central tendency rather than the mean5. Estimation results of Median and OLS regressions applied to the most parsimonious model are compared in Table 5 which shows only minimal differences in the magnitude of parameters obtained from the two estimation methods. In any case, these differences do not substantially alter the nature of the phenomena. We thus conclude that the results of the OLS regression can be considered as sufficiently robust.
Table 5. OLS vs. Median regression (parsimonious model)
constant bottle size large medium small bottle material 5
OLS coefficient -3.60*** 0.28*** 0.70*** 0.99***
marginal effect N/A 33% 101% 169%
Median regression coefficient marginal effect -3.35*** N/A 0.25*** 0.57*** 0.89***
28% 76% 142%
Median Regression parameters are less sensitive to outliers than OLS parameters analogous to the median vs mean of the sample (Buchinsky, 1998; Koenker and Bassett, 1978; Koenker and Hallock, 2001). This is because Median Regression parameters are estimated by minimizing the sum of absolute residuals, while OLS parameters are estimated by minimizing the sum of squared residuals.
glass push-pull salt content very low mineral low-sodium store supermarket minimarket discount store ln distance brand San Benedetto Levissima Rocchetta Sant'Anna Vera Ferrarelle Uliveto Sangemini other brands
1.18*** 0.68***
224% 98%
1.22*** 0.70***
240% 101%
-0.13*** -0.21***
-12% -19%
-0.19*** -0.06***
-17% -5%
0.09*** 0.10*** -0.46*** 0.28***
9% 10% -37% N/A
0.08*** 0.12*** -0.49*** 0.23***
8% 13% -39% N/A
0.28*** 0.44*** 0.66*** 0.43*** 0.18*** 0.60*** 0.37*** 1.09*** 0.40***
32% 55% 94% 54% 19% 82% 45% 197% 49%
0.27*** 0.40*** 0.63*** 0.50*** 0.15*** 0.60*** 0.52*** 1.07*** 0.40***
31% 49% 87% 64% 16% 82% 68% 191% 49%
Notes: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.
These results show that the type of packaging has a strong effect on the unit price of bottled water. First, the size of the bottle has a significant effect on price and, specifically, the large, medium and small dummies have positive and increasing coefficients equal to +0.28, +0.70 and +0.99, respectively. Considering the functional form of the equation, the coefficient of dichotomous explanatory variables can be transformed into the percentage change in price due to the presence of a given quality attribute (marginal effect) applying the formula: {exp (coefficient) - 1}. It follows that, assuming that the largest bottles (2L) are the baseline, the unit price increases by +33%, +101% and +169%, respectively, when water is sold in progressively smaller bottles (1.5L, 0.75-1.5L, ≤0.5L). This can be explained considering both the higher costs of packaging for smaller bottles (Gleick and Cooley, 2009), and the higher willingness to pay of those consumers who prefer smaller bottles because they are considered as more practical, especially outside the home (Doria, 2006; He, Jordan and Paudel, 2008; Van Der Linden, 2015; Viscusi, Huber and Bell, 2013). The same considerations can be made about the type of cap. In fact, bottled water with a push-pull cap has a much higher unit price (+98%) related to both the higher cost of packaging and consumer preferences. The kind of material used for the bottle (codified by the glass, plastic and eco-friendly dummies) also has a significant effect on price. Water in glass bottles gains a relevant premium price (+224%), while, unexpectedly, a non-significant effect was found for bottles made from eco-friendly materials (recycled, partly recycled or biodegradable plastic), assuming plastic bottles as the baseline. The premium for water in glass bottles can be explained considering the higher cost of glass compared to plastic, but also the high willingness to pay of those consumers who prefer a more elegant glass bottle rather than an ordinary plastic one. Conversely, the result regarding the bottles made from eco-friendly material means that consumers are not willing to pay any premium for this attribute, even considering its positive impact on the environment (Lagioia, Calabrò and Amicarelli, 2012; Gironi and Piemonte, 2010; Nessi, Rigamonti and Grosso, 2012; Niccolucci et al., 2011).
The type of pack (multipack) is another non-significant variable, meaning that single bottles and multipacks are sold at the same unit price, all other characteristics being equal. This is also an unexpected result because a discount on the unit price is usually given when a larger number of products are purchased. The intrinsic attributes of bottled water only seem to have a slight influence on price. Specifically, water with a very low mineral content had a significant but moderate discount price (-12%), while water with mediumhigh mineral content did not have any premium or discount price compared to water with low mineral content used as the baseline. Water with low sodium content also showed a discount price (-19%) compared to water with a medium-high sodium content, while the presence/type of effervescence did not have any effect on price. These results show that consumers do not appear to be very interested in the mineral composition of bottled water even though they express less preference for water with a very low content of minerals, including sodium. The type of retail store, the distance between the place of bottling and the point of sale, and the brand also affect the price of bottled water considerably. Bottled water sold in discount stores had a relevant discount price of -37% compared to bottled water sold in hypermarkets used as the baseline. On the other hand, bottled water sold in supermarkets and minimarkets showed only a moderate premium price equal to +9% and +10%, respectively. These price differences may be related to the logistics costs and store management which are, obviously, much lower for discount stores and higher for supermarkets and minimarkets compared to hypermarkets. The distance between the place of bottling and the point of sale is also a significant variable with a coefficient equal to +0.28. Taking into account the logarithmic form of the equation, the coefficient of a continuous variable can be directly interpreted in terms of elasticity. Therefore, the positive (but less than one) coefficient of the distance variable means that an increase in the distance leads to a less-thanproportional increase in the unit price of bottled water. Obviously, the positive impact of distance on the price of bottled water may be mainly related to the transportation costs. Finally, the brand affects the unit price of bottled water considerably. Compared to store brands used as the baseline, all major brands (San Benedetto, Levissima, Rocchetta, Sant’Anna, Vera, Ferrarelle, Uliveto, Sangemini) as well as minor brands gain a significant premium price ranging from a minimum of +19% (Vera) to a maximum of +197% (Sangemini).
5. Conclusions The results of this study show that bottled water is, surprisingly, highly differentiated. However its retail price is mainly affected by extrinsic characteristics, in particular brand and packaging (bottle size, bottle material and type of cap). Other extrinsic factors, i.e. the type of store and the distance between the water source and the point of sale, also have a relevant effect on price. Conversely, the intrinsic characteristics affect the price of bottled water only moderately.
Based on these results, it is possible to draw some useful insights for both practitioners and policy makers. Practical implications can be grouped according to three main dimensions: commercial, environmental and regulatory. From a commercial point of view, bottled-water producers need to revise their marketing strategies to tackle a market that seems to have reached the maturity stage of its life cycle. As is known, the two main marketing strategies for a competitive advantage are based on “lower cost” and “differentiation” (Porter, 1985), sometimes combined within the same firm through different brands. In order to extend the lower cost strategy, firms should concentrate on high levels of efficiency and cost reductions. Thus, the type of packaging is crucial because it greatly affects the price of bottled water and, at the same time, the relative production costs. Firms should adopt less-expensive types of packaging represented by larger-sized plastic bottles with standard caps. Another important option to reduce the cost of packaging, although not considered in our study, is to act on bottle weight by adopting thinner plastic (Lagioia, Calabrò and Amicarelli, 2012). Moreover, firms implementing a lower cost strategy should opt for growth based on the penetration of local markets surrounding water sources, rather than expanding the market. In fact, expanding the market would increase transportation costs which would have a strong impact on the final price of bottled water. Finally, discount stores and hypermarkets seem more suitable channels for the distribution of low-priced bottled water, taking into account their more efficient logistics. A product differentiation strategy can be strengthened primarily through the brand. As shown by the results of this study, brand equity plays a key role in differentiating bottled water, considering its strong effect on price. However, a product differentiation strategy merely based on the brand is not sustainable in the longterm especially in mature markets characterized by strong competitive pressure. Our analysis shows that leading brands such as “Vera” and “San Benedetto” are already adopting very aggressive pricing strategies. Considering also the high level of product substitutability that characterizes the market of bottled water, the reinforcement of brand equity should be supported by a real differentiation of products based on both extrinsic and intrinsic attributes specifically aimed at meeting consumers’ needs and preferences. Therefore, firms should diversify their products, firstly, in terms of packaging by offering the same product in bottles made from different materials (glass and plastic), sizes and with different types of cap. Despite our results showing that the mineral composition only has a moderate effect on the price of bottled water, firms interested in implementing a product differentiation strategy should not neglect this aspect. The mineral composition of bottled water varies notably among different brands, however consumers do not seem to perceive these differences. One reason could be that, despite the mineral composition being indicated on the label, consumers have difficulty in interpreting and comparing this information. We therefore argue that firms should emphasize the connection between the brand and the specific intrinsic properties of water. For example, some brands (Ferrarelle and Lete) emphasize the natural effervescence and its effect on taste, while others (Sangemini, Lilia) stress the high calcium or magnesium content and the importance of these minerals for health and wellbeing (Black, 2009). Generic claims about the positive health effects of natural mineral water (e.g. “helps digestion”, “promotes diuresis”), inserted on almost all the labels of bottled water, have no
real value unless they explain the specific intrinsic characteristics that generate these health effects and whether these characteristics are common or not to other products. Considering the expensive advertising campaigns for communicating the specific properties of bottled water to consumers, firms should expand their market and implement a diffuse distribution network. Since market development involves higher transportation costs, logistics plays a key role and should be carefully optimized. National brands should consider implementing a multi-sourcing strategy with a short chain distribution which could be useful to reduce the transportation costs (Lagioia, Calabrò and Amicarelli, 2012). As regards environmental issues, the main result of our analysis is that the market of bottled water does not seem to exhibit any effective signs of self-regulation towards a reduction in its environmental impacts. We found that consumers are willing to pay higher prices for bottled water packaged in small-sized bottles and with special caps (push-pull) in order to satisfy their need for convenience despite this having a more negative impact on the environment (Gleick and Cooley, 2009). In addition, consumers are willing to pay higher prices for bottled water which is transported over long distances (often greater than 1,000 kilometres) with high energy consumption and air pollution (Botto et al., 2011; Gleick and Cooley, 2009; Gironi and Piemonte, 2010; Lagioia, Calabrò and Amicarelli, 2012; Nessi, Rigamonti and Grosso, 2012; Niccolucci et al., 2011). Conversely, consumers are not willing to pay any premium for water packaged in bottles made from eco-friendly materials (recycled, partially recycled or biodegradable plastic) (Lagioia, Calabrò and Amicarelli, 2012; Gironi and Piemonte, 2010; Nessi, Rigamonti and Grosso, 2012; Niccolucci et al., 2011). Considering that the environmental impact of bottled water consumption is a public concern, it also seems interesting to provide some regulatory insights. Since the bottled water market does not seem capable of selfregulation, policy makers should implement measures to encourage producers and consumers towards a more responsible behaviour in terms of environmental sustainability. First, a taxation scheme based on the type of packaging used for bottling natural mineral water could be introduced in order to discourage the excessive use of small bottles and, at the same time, sustain the use of bottles made from eco-friendly materials. It is worth noting that Italian regulation prohibits the use of containers holding more than 2 litres which seems to be a counterproductive constraint. In fact, using bigger containers (e.g. 3L, 5L, 10L), already used for other food products such as wine, would be an effective tool for reducing both the price of bottled water and the environmental impact of its consumption. Another important issue concerns the current taxation system of water abstraction which varies depending on the region, and is mainly calculated as a constant tariff per unit volume of water abstracted (Legambiente, 2014). Inevitably, such regulation promotes an increase in bottled water flows from the regions where taxation is lower to regions where the taxation is higher and, consequently, an increase in environmental impacts of bottled water consumption. Therefore, also considering that Italy is one of the European countries where bottled water is sold at the lowest price (Bevitalia, 2015), increasing and levelling the taxes among the regions could encourage the purchase of cheaper bottled water from local springs with positive effects on the environment.
Acknowledgements
The authors would like to thank the two anonymous reviewers for their useful comments, needless to say that any shortcomings are the responsibilities of the authors alone.
References Abrahams, N. A., Hubbell, B. J., & Jordan, J. L. (2000). Joint production and averting expenditure measures of willingness to pay: do water expenditures really measure avoidance costs?. American Journal of Agricultural Economics, 82(2), 427-437. Bevitalia (2015). Acque Minerali, Bibite e Succhi - Annuari del Bere, Beverfood.com Edizioni. Black, R. (2009). Acqua minerale di Sangemini: the Italian mineral water industry finds a place at the table. Journal of Modern Italian Studies, 14(2), 184-198. Boatto, V., Defrancesco, E., & Trestini, S. (2011). The price premium for wine quality signals: does retailers information provision matter? British Food Journal, 4-5, 669-679. Botto, S., Niccolucci, V., Rugani, B., Nicolardi, V., Bastianoni, S., & Gaggi, C. (2011). Towards lower carbon footprint patterns of consumption: The case of drinking water in Italy. Environmental Science & Policy, 14(4), 388-395. Buchinsky, M. (1998). Recent advances in quantile regression models: a practical guideline for empirical research. Journal of human resources, 88-126. Burnham, K. P., & Anderson, D. R. (2003). Model selection and multimodel inference: a practical information-theoretic approach. Springer Science & Business Media. Carew, R., Florkowski, W. J., & Smith, E. J. (2012). Hedonic analysis of apple attributes in metropolitan markets of Western Canada. Agribusiness, 3, 293–309. Carlucci, D., De Gennaro, B., Roselli, L., & Seccia, A. (2014). E-commerce retail of extra virgin olive oil: an hedonic analysis of Italian SMEs supply. British Food Journal, 10, 1600-1617. Carlucci, D., Stasi, A., Nardone, G., & Seccia, A. (2013). Explaining Price Variability in the Italian Yogurt Market: A Hedonic Analysis. Agribusiness, 2, 194–206. Comuni Italiani (2015). (accessed in April 2016) Costanigro, M., & McCluskey, J. (2011). Hedonic price analysis in food markets, in Lusk, J., & Shogren, J. F. (Eds.). (2011). The Oxford handbook of the economics of food consumption and policy. Oxford University Press. Doria, M. F. (2006). Bottled water versus tap water: understanding consumers' preferences. Journal of water and health, 4(2), 271-276. Doria, M. F., Pidgeon, N., & Hunter, P. R. (2009). Perceptions of drinking water quality and risk and its effect on behaviour: A cross-national study. Science of the Total Environment, 407(21), 5455-5464. Ferrier, C. (2001). Bottled water: understanding a social phenomenon. AMBIO: A Journal of the Human Environment, 30(2), 118-119. Gleick, P. H., & Cooley, H. S. (2009). Energy implications of bottled water. Environmental Research Letters, 4(1), 014009.
Gironi, F., & Piemonte, V. (2010). Life cycle assessment of polylactic acid and polyethylene terephthalate bottles for drinking water. Environmental Progress & Sustainable Energy, 30 (3), 459–468. Greenlee, L. F., Lawler, D. F., Freeman, B. D., Marrot, B., & Moulin, P. (2009). Reverse osmosis desalination: water sources, technology, and today's challenges. Water research, 43(9), 2317-2348. He, S., Jordan, J., & Paudel, K. (2008). Economic evaluation of bottled water consumption as an averting means: evidence from a hedonic price analysis. Applied Economics Letters, 15, 337–342. International Bottled Water Association – IBWA (2015).
Rani, B., Maheshwari, R., Garg ,A., & Prasad, M. (2012). Bottled Water – A Global Market Overview. Bulletin of Environment, Pharmacology and Life Sciences, 1, 1-4. Roselli, L., Carlucci, D., & De Gennaro, B. (2016). What Is the Value of Extrinsic Olive Oil Cues in Emerging Markets? Empirical Evidence from the US E‐Commerce Retail Market. Agribusiness. Rosen, S. (1974). Hedonic prices and implicit markets: Product differentiation in pure competition. Journal of Political Economy, 82, pp. 34–55. Satimanon, T., & Weatherspoon, D. D. (2010). Hedonic analysis of sustainable food products. International Food and Agribusiness Management Review, 4, 57–74. Schamel, G. (2006). Geography versus brands in a global wine market. Agribusiness, 3, 363–374. Schollenberg, L. (2012). Estimating the hedonic price for Fair Trade coffee in Sweden. British Food Journal, 2-3, 428-446. Snipes, M., & Taylor, D. C. (2014). Model selection and Akaike Information Criteria: An example from wine ratings and prices. Wine Economics and Policy, 3(1), 3-9. Steiner, B. E. (2004). Australian wines in the British wine market: A hedonic price analysis. Agribusiness, 3, 287–307. Torretta, V. (2013). Environmental and economic aspects of water kiosks: Case study of a medium-sized Italian town. Waste Management, 33, 1057–1063. Van Der Linden, S. (2015). Exploring Beliefs About Bottled Water and Intentions to Reduce Consumption The Dual-Effect of Social Norm Activation and Persuasive Information. Environment and Behavior, 47(5), 526-550. Viscusi, W. K., Huber, J. C., & Bell, J. (2013). The private rationality of bottled water drinking. Contemporary Economic Policy, 14-1. Wilk, R. (2006). Bottled water: The pure commodity in the age of branding. Journal of Consumer Culture, 6(3), 303-325. Ward, C. E., Lusk, J. L., Dutton, J. M. (2008). Implicit value of retail beef product attributes. Journal of Agricultural and Resource Economics, 3, 364–381.
PII: DOI: Reference:
S2212-4284(16)30049-4 http://dx.doi.org/10.1016/j.wre.2016.07.001 WRE77
To appear in: Water Resources and Economics Received date: 3 November 2015 Revised date: 25 May 2016 Accepted date: 13 July 2016 Cite this article as: Domenico Carlucci, Bernardo De Gennaro and Luigi Roselli, What is the value of bottled water? empirical evidence from the italian retail m a r k e t , Water Resources and Economics, http://dx.doi.org/10.1016/j.wre.2016.07.001 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
What is the value of bottled water? Empirical evidence from the Italian retail market Domenico Carlucci Department of Agricultural and Environmental Science, University of Bari Aldo Moro, Bari, Italy, Via Giovanni Amendola 165/A, 70126, Bari, Italia E-mail: [email protected] Bernardo De Gennaro Department of Agricultural and Environmental Science, University of Bari Aldo Moro, Bari, Italy, Via Giovanni Amendola 165/A, 70126, Bari, Italia E-mail: [email protected] Luigi Roselli (corresponding author) Department of Agricultural and Environmental Science, University of Bari Aldo Moro, Bari, Italy, Via Giovanni Amendola 165/A, 70126, Bari, Italia E-mail: [email protected] Abstract Bottled water has become a global business, and Italy is one of the largest producers and consumer countries in the world. However, the Italian bottled water market seems to have reached the maturity stage of its life cycle, and because competitive pressure has strongly increased in this market, producers need to revise their marketing strategies. A variety of products with different features and prices are now available on the market of bottled water, however it is unknown how the retail price of bottled water is affected by its attributes. We thus measured whether, and to what extent, the price of bottled water depends upon its extrinsic (brand, packaging, origin), and intrinsic characteristics (mineral composition). We estimated a hedonic price model using data collected via direct observation of the shelves in various Italian modern retail stores. Our results show that bottled water is highly differentiated and its retail price is mainly affected by extrinsic characteristics. On the basis of these results, several insights are provided for both practitioners and policy makers.
Keywords bottled water, natural mineral water, Italy, hedonic price, competitive advantage
1. Introduction Over the last few decades, the global consumption of bottled water has been increasing consistently even in countries where tap water is largely available and considered to be of an excellent quality (Wilk, 2006). This increase has been mostly linked to the changes in lifestyles and eating habits associated with economic growth (Ferrier, 2001; Wilk, 2006). Several studies have investigated consumer behaviour in order to identify the main reasons why people decide to buy more expensive bottled water rather than tap water. These studies have shown that bottled water consumption is mostly driven by consumers’ dissatisfaction
with the organoleptic quality of tap water (especially taste), followed by health-related concerns (Abrahams, Hubbell and Jordan, 2000; Doria, 2006; Doria, Pidgeon and Hunter, 2009; Van Der Linden, 2015; Viscusi, Huber and Bell, 2013). Although safety concerns and health improvement are sometimes considered as equivalent, the two factors act differently. In contexts where distrust in tap water suppliers exists (e.g. information regarding current or prior problems with tap water), risk perception and safety concerns are more prominent in driving bottled water consumption (Abrahams, Hubbell and Jordan, 2000; Doria, 2006; Doria, Pidgeon and Hunter, 2009; Hu, Morton and Mahler, 2011; Viscusi, Huber and Bell, 2013). In other cases, consumers may prefer bottled water simply because it is considered as healthier, but not necessarily safer, than tap water (Black, 2009; Doria, 2006; Wilk, 2006). The need for convenience is another important reason for choosing bottled water especially for out-of-home consumption (Doria, 2006; He, Jordan and Paudel, 2008; Van Der Linden, 2015; Viscusi, Huber and Bell, 2013). Today, a bottle of water can be bought almost everywhere and, when finished, it is easily disposed of in the trash, thus it is not necessary to carry around a bulky container all day. Italy is one of the highest producers and consumers of bottled water in the world (IBWA, 2015). In Italy, the consumption of bottled water began in the 1970s and, since then, annual per capita consumption has increased considerably from 47 litres in 1980 to a maximum of 192 litres in 2008 (Bevitalia, 2015). However, in the last few years (2005-2013), Italian consumption of bottled water has stagnated considering that annual per capita consumption has been ranging between 187 and 192 litres. This is probably due to the persistent economic crisis that has reduced the purchasing power of Italian households, but also other factors are relevant. Purification devices are increasingly used in bars and restaurants as well as homes to treat tap water by removing chlorine, as well as the bad taste and odours in order to provide safe and palatable drinking water as an alternative to bottled water (Greenlee et al., 2009). In addition, there is a growing consumer sensitivity regarding the environmental impacts of increasing bottled water consumption (Rani et al., 2012; Van Der Linden, 2015). In fact, the process of extraction, processing, packaging and transportation of bottled water involves considerable environmental impacts, including pollution, climate change and depletion of natural resources (Botto et al., 2011; Gleick and Cooley, 2009; Lagioia, Calabrò and Amicarelli, 2012; Nessi, Rigamonti and Grosso, 2012; Niccolucci et al., 2011; Torretta, 2013). Essentially, the Italian market of bottled water seems to have reached the maturity stage of its life cycle (Niccolucci et al., 2011; Rani et al., 2012) and, thus, competitive pressure has strongly increased. Bottled water producers thus need to revise their marketing strategies in order to reinforce their competitive advantages and defend their profits and market shares from competitors. A review of the economic literature concerning the bottled water market reveals that most studies have focused on two mainstreams: the analysis of consumer behaviour in order to investigate the reasons for the increase in bottled water consumption (Abrahams, Hubbell and Jordan, 2000; Black, 2009; Doria, 2006; Doria, Pidgeon and Hunter, 2009; Hu, Morton and Mahler, 2011; Van Der Linden, 2015; Viscusi, Huber and Bell, 2013; Wilk, 2006), and the assessment of the environmental impacts of bottled water consumption ( Botto et al., 2011; Gleick and Cooley, 2009; Lagioia, Calabrò and Amicarelli, 2012; Nessi, Rigamonti and
Grosso, 2012; Niccolucci et al., 2011; Torretta, 2013). To the best of our knowledge, to date, only a few studies have performed analyses in order to support firms in their decision-making problems (He, Jordan and Paudel, 2008; Rani et al., 2012). This study starts with the observation that a variety of products with different features and prices is now available on the Italian market of bottled water despite the simplicity of “water products”, however it is unknown how the retail price of bottled water is affected by its attributes. Our survey aims to fill this gap by measuring whether, and to what extent, the price of bottled water depends upon its extrinsic (brand, packaging, origin), and intrinsic characteristics (mineral composition). We used the hedonic price method (Rosen, 1974) to estimate the implicit prices associated with the main characteristics of bottled water. Estimates of the implicit prices of bottled water attributes can provide useful insights for both practitioners and policy makers. First, being aware of their production costs, firms involved in this market can use implicit prices to devise an optimal mix of attributes and more profitable marketing strategies. Second, the hedonic approach isolates the premium for various attributes that affect the environmental impact of bottled water consumption, such as the use of eco-friendly bottles and the distance between the water source and the point of sale (Botto et al., 2011; Gleick and Cooley, 2009; Gironi and Piemonte, 2010; Lagioia, Calabrò and Amicarelli, 2012; Nessi, Rigamonti and Grosso, 2012; Niccolucci et al., 2011).
2. An overview of the Italian market of bottled water In 2013, the Italian consumption of bottled water was over 12 billion litres, the seventh highest in the world after China (39), the USA (38), Mexico (31), Indonesia (18), Brazil (18), and Thailand (15) (IBWA, 2015). In terms of annual per capita consumption, Italy (190 L) is the first European country and third in the world after Mexico (255L) and Thailand (225L) (IBWA, 2015). The bottled water consumed in Italy is almost totally produced domestically and foreign trade is very limited (Bevitalia, 2015). Contrary to other countries in the world like the USA1, where purified water is largely sold as bottled water (IBWA, 2015), in Italy bottled water is essentially represented by so-called “natural mineral water” (Bevitalia, 2015) as specified in the Legislative Decree n. 176/2011 implementing the European Directive 2009/54/EC. By law, natural mineral water is clearly distinguished from other drinking water by its original purity (absence of chemical treatments) and its favourable health properties. The annual report of Bevitalia (2015) provides an overview of the Italian market of bottled water which is briefly described below. In 2013, the Italian bottled water industry included 143 companies of different sizes, with overall annual gross sales of 2.4 billion euros. The industry is highly concentrated considering that the four largest players (Nestlé Waters, San Benedetto, Norda, and Fonti Vinadio) control more than half of the domestic market. Since many companies adopt a multi-branding strategy, about 270 brands are available on the Italian market of bottled water. Most of these brands are regional or local, and only a few (about a dozen) 1
In the United States, about 44% of all bottled water is sold as “purified water” which is produced by distillation, deionization, reverse osmosis or other processes (Gleick and Cooley, 2009). The three largest brands of purified water sold in the US market are “Dasani” (Coca-Cola Company), “Aquafina” (Pepsi-Cola Company), and “Pure Life” (Nestlé Waters) (Gleick and Cooley, 2009).
are national brands. There are also various store brands that have reached a significant cumulative market share accounting for 9% of total sales. The largest segment of the Italian market of bottled water is represented by still (non-sparkling) water which accounts for 65% of sales volume, while carbonated and naturally carbonated waters account for 19% and 16%, respectively. Still water is ideal for ordinary daily consumption, while sparkling water, given its particular taste, is preferred at mealtimes and is mainly sold in restaurants. Bottled water with a low mineral content (50-500 mg/L) is the best-selling, accounting for 59% of sales volume, while bottled water with a very low mineral content (<50 mg/L) and medium-high mineral content (>500 mg/L) represent minor market segments (14% and 27%, respectively). Water with a very low mineral content is almost totally tasteless although it is recommended for people who have diuretic problems. It is also more suitable for special purposes such as the formulation of baby foods (e.g. milk powder). Conversely, water with a medium-high mineral content has a stronger and diversified taste, depending on the composition of the dissolved salts, and is recommended for people who need to increase their daily intake of minerals (e.g. people who play sport). Bottled water is mainly packaged (83% of sales volume) in plastic bottles (polyethylene terephthalate - PET) with different sizes: larger plastic bottles (1L, 1.5L, 2L2), being preferred for daily water consumption at home, are sold the most (75% of sales volume); smaller plastic bottles (0.25L, 0.33L, 0.50L) are specifically aimed at meeting consumers’ convenience needs (out-of-home consumption) and account for only 8% of sales volume. Plastic bottles are often closed with a practical “push-pull” cap which makes drinking easier, especially for children (this kind of bottle is thus often called a “baby-bottle”). Glass bottles (usually of 1L) are used to a lesser extent (17% of sales volume) and are mostly sold in restaurants, as they are considered to be more elegant. In the last few years, some bottled water producers have also introduced bottles made from eco-friendly materials, represented by recycled, partly recycled or biodegradable plastic (Lagioia, Calabrò and Amicarelli, 2012; Gironi and Piemonte, 2010; Nessi, Rigamonti and Grosso, 2012; Niccolucci et al., 2011). These are also called “bio-bottles” for which, however, market data are not yet available. Finally, concerning where the retail sales of bottled water occur, modern retail stores (hypermarkets, supermarkets, minimarkets and discount stores) play a predominant role, accounting for 71% of sales volume. The remaining sales follow the Ho.Re.Ca. channel (18%) and traditional stores (11%).
3. Method 3.1 Hedonic price model We used the hedonic price model to analyse the relationship between the price and the main attributes of bottled water. The hedonic price model has been successfully employed to analyse the markets of several food products, including wine (Nerlove, 1995; Oczkowski, 1994; Schamel, 2006; Steiner, 2004; Boatto, Defrancesco and Trestini, 2011; Panzone, 2011), carbonated beverages (Martinez-Garmendia, 2010), fresh meat (Loureiro and McCluskey, 2000; Ward, Lusk and Dutton, 2008), eggs (Satimanon and Weatherspoon, 2
Italian Regulation (Legislative Decree n. 176/2011) prohibits the use of containers holding more than 2 liters for bottling natural mineral water.
2010), apples (Carew, Florkowski and Smith, 2012), yogurt (Carlucci et al., 2013), coffee (Schollenberg, 2012) and olive oil (Carlucci et al., 2014; Karipidis, Tsakiridou and Tabakis, 2005; Muñoz, Moya and Gil, 2015; Roselli, Carlucci and De Gennaro, 2016). The hedonic approach is borrowed from Lancaster’s (1966) theory of demand which states that consumers derive utility directly from the quality attributes embedded in a product rather than from the product itself. In other words, any differentiated product can be considered to be a bundle of quality attributes that are independently valued by consumers at the time of purchase. Rosen (1974) subsequently developed a theoretical model demonstrating that the observed price of a product can be considered as the sum of the prices associated with each of its quality attributes. Although these prices are not explicitly expressed by the market, they can be estimated by employing a regression equation, i.e. the hedonic price model, which expresses the price of a product (directly observable) as a function of its attributes (directly or indirectly observable). According to Rosen’s (1974) formulation, a hedonic price model can be specified as follows: (1) where P is the price of a product, and Z = z1, z2, . . . , zj, . . . , zn is a vector of n objectively measured attributes that completely describe the product quality. After estimating the hedonic price equation, a partial derivative with respect to the attribute j, ∂P(Z)/∂z j, can be interpreted as the implicit or shadow price of the specific attribute j. This theoretical model is based on the assumption that the market is in equilibrium and there is perfect competition. In this situation, consumers maximize utility by choosing the available products under budget constraints, and firms maximize profits given the available technology and factor prices (Rosen, 1974). Consequently, being related to both supply and demand conditions, implicit prices cannot be considered merely as indicators of consumer preferences (Costanigro and McCluskey, 2011; Oczkowski, 1994; Rosen, 1974; Schamel, 2006).
3.2 Data collection Data on the prices and characteristics of bottled water were collected via direct observation of the shelves in various modern retail stores located in the province of Bari. As already mentioned, almost three-quarters of Italian retail sales of bottled water occur in hypermarkets, supermarkets, minimarkets and discount stores (Bevitalia, 2015). We collected data in the first months of 2015 (January and February) from two hypermarkets (Auchan and Coop), two supermarkets (Famila and Sigma), two minimarkets (Despar and Simply) and two discount stores (Lidl and Eurospin) of the main retail chains operating in the province of Bari. Obviously, this is only an explorative analysis and the reliability of the results could be further improved by extending the survey to a larger area where differences may arise in price levels and product
variety. However, the area of data collection seems to be sufficiently representative at a national level3 to test the mechanisms of price formation. In each of the selected stores, we adopted a snapshot-type procedure for data collection: each store was visited just once, when we directly and simultaneously recorded the retail prices and the characteristics of all drinking bottled waters placed on dedicated shelves. Because in some cases the same product was offered in different types of packaging or at different prices, each item was always considered as a separate observation. We only excluded products sold at reduced prices (special offer) when the “normal” prices were not indicated in order to prevent any possible bias in the results due to the inclusion of prices that were only temporarily reduced4. By examining the products and their labelling, the following information was recorded for each product: price, packaging characteristics (type of pack, bottle size, bottle material, type of cap), chemical composition (total salt content, sodium content, presence/type of effervescence), brand and place of bottling.
3.3 Data set Using the criteria described above, we collected a data set containing 374 observations. A preliminary analysis of the data set was carried out by calculating descriptive statistics regarding both the total sample and specific sub-samples grouped according to particular attributes (Table 1).
Table 1. Summary statistics of the sample
374
Min 0.08
Price/litre* Max 3.52
315 59
0.08 0.12
51 133 54 136
N. cases Total sample Type of pack single bottle multipack Bottle size 2.00 L 1.50 L 0.75-1.00 L ≤ 0.50 L Bottle material plastic glass eco-friendly material Type of cap standard push-pull Salt content
Mean 0.50
Std dev 0.45
3.52 3.49
0.48 0.59
0.44 0.51
0.08 0.09 0.25 0.26
0.28 0.66 3.52 3.49
0.17 0.28 0.83 0.70
0.05 0.11 0.72 0.42
332 17 25
0.08 0.89 0.13
1.67 3.52 0.49
0.45 1.74 0.31
0.32 0.94 0.11
347 27
0.08 0.64
3.52 1.67
0.45 1.15
0.42 0.40
3
The province of Bari is located in Apulia, southern Italy. With about 1.27 million inhabitants, the province of Bari represents the fifth most-populous province of Italy, after Rome (4.34 million), Milan (3.19 million), Naples (3.12 million), and Turin (2.29 million) (Comuni Italiani, 2016). The main demographic indicators of the province of Bari are very similar to the national averages. For the province of Bari, the average age of the population is 43, the share of younger people (below 15) is 14%, the share of elders (over 65) is 20%, the average number of people per household is 2.6, and annual per capita income is 9,300 Euro. At a national level, the average age is 44, the share of younger people is 14%, the share of elders is 21%, the average number of people per household is 2.4, and annual per capita income is 12,100 Euro (Comuni Italiani, 2016). 4
The number of products sold at reduced prices (special offer) and thus excluded from the data collection was very limited (one or two products for each of the selected stores).
medium-high mineral (500-1500 mg/L) low mineral (50-500 mg/L) very low mineral (<50 mg/L) Effervescence still naturally carbonated carbonated Sodium content low sodium (<20 mg/L) medium-high sodium (>20mg/L) Store format hypermarket supermarket minimarket discount store Distance <500 km 500-1,000 km >1,000 km Brand San Benedetto Levissima Rocchetta Sant'Anna Vera Ferrarelle Uliveto Sangemini store brands other brands
109 234 31
0.08 0.09 0.17
1.40 3.52 1.25
0.43 0.53 0.48
0.23 0.54 0.30
215 92 67
0.08 0.16 0.09
3.52 1.40 3.52
0.52 0.45 0.50
0.45 0.23 0.66
296 78
0.08 0.09
3.49 3.52
0.49 0.52
0.39 0.63
160 117 75 22
0.08 0.12 0.13 0.09
3.52 1.67 1.67 0.54
0.54 0.51 0.47 0.20
0.57 0.34 0.34 0.10
139 161 74
0.08 0.11 0.16
1.40 1.60 3.52
0.33 0.51 0.78
0.24 0.32 0.76
40 31 25 23 18 17 14 12 9 185
0.22 0.28 0.33 0.25 0.16 0.26 0.35 0.53 0.11 0.08
1.60 1.67 0.86 0.90 0.60 1.40 0.74 1.00 0.38 3.52
0.63 0.70 0.49 0.41 0.30 0.52 0.55 0.74 0.24 0.45
0.45 0.46 0.15 0.17 0.14 0.28 0.14 0.20 0.10 0.54
*Prices are expressed in €
First, a wide variability in price was detected in the overall sample considering that the unit price of bottled water ranged from a minimum of 0.08 Euro/L to a maximum of 3.52 Euro/L, with a mean of 0.50 Euro/L. However, the frequency distribution of unit price was skewed to the right, with low unit prices showing much higher frequencies: the unit price was lower than 1.39 Euro/L for 95% of the observations (Figure 1).
Figure 1. Price frequency distribution of bottled water
0.18
Percentiles 5% 25% 50% 75% 95%
0.16
Relative frequency
0.14
Price (euro/L) 0.13 0.23 0.39 0.59 1.39
Obs 374 Mean 0.50 Std. Dev. 0.45 Variance 0.21
0.12 0.10 0.08 0.06 0.04 0.02 0.00 0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
Price (Euro/L)
The price variability of the data set seems to be primarily related to the different types of packaging. In fact, water is sold as single bottles or multipacks (including four or six bottles), and bottle size varies from a minimum of 0.25L to a maximum of 2L. In addition, bottles are made of plastic (PET), glass, or eco-friendly materials (recycled, partly recycled or biodegradable plastic) and the bottles are closed with standard or push-pull caps. We found that the average price of water sold in smaller bottles (1 L or less), in glass bottles, and in bottles with push-pull caps was much higher than the average price of water sold, respectively, in larger bottles (1.5-2L), in plastic bottles, and in bottles with standard caps. On the other hand, regarding the intrinsic attributes of bottled water (total salt content, sodium content, presence/type of effervescence), only limited differences in the average prices were observed, while distance (calculated between the place of bottling and the point of sale) affected the average price of bottled water notably. Finally, we found that there were large price differences among different types of store (in particular, for discount stores compared to the other stores) and different brands.
3.4 Empirical Model We used a stepwise procedure to specify several hedonic price models which were estimated through the Ordinary Least Squares (OLS) method and checked for statistical validity. Among the valid models, the best (most parsimonious) model was selected through a comparison of Akaike Information Criteria (AIC) and Bayesian Information Criteria (BIC) (Burnham and Anderson, 2003; Snipes and Taylor, 2014). The best model is the following:
(2)
However, we also present and discuss an extended version of the best model including both significant and non-significant predictors which were also tested:
(3) The variables included in the empirical models are listed in Table 2. The unit price of bottled water is the dependent variable (price) which is a continuous variable. Just one explanatory variable, the distance between the place of bottling and the point of sale (distance), is also a continuous variable, while since the other explanatory variables are categorical, they were transformed into one or more dummy variables. We chose the double-log formulation due to its better fit to the data.
Table 2. Variables of the empirical models Variables Dependent variable price Regressors multipack bottle size
bottle material
push-pull salt content
effervescence
low sodium store
distance brand
Type
Description
continuous variable
price per litre (€)
dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy dummy continuous variable dummy dummy dummy dummy dummy dummy dummy dummy
multipack (4 or 6 bottles) = 1; single bottle = 0 small (≤0.50 L) = 1; otherwise = 0 medium (0.75-1.00 L) = 1; otherwise = 0 large (1.50 L) = 1; otherwise = 0 extra-large (2.00 L) = 1; otherwise = 0 (baseline) glass = 1; otherwise = 0 eco-friendly material = 1; otherwise = 0 plastic = 1; otherwise = 0 (baseline) push-pull = 1; standard = 0 medium-high mineral (>500 mg/L) = 1; otherwise = 0 low mineral (50-500 mg/L) = 1; otherwise = 0 (baseline) very low mineral (<50 mg/L) = 1; otherwise = 0 still = 1; otherwise = 0 (baseline) naturally carbonated = 1; otherwise = 0 carbonated = 1; otherwise = 0 low sodium (<20 mg/L) = 1; medium-high sodium (>20mg/L) = 0 hypermarket = 1; otherwise = 0 (baseline) supermarket = 1; otherwise = 0 minimarket = 1; otherwise = 0 discount store = 1; otherwise = 0 distance between the place of bottling and the point of sale (km) San Benedetto = 1; otherwise = 0 Levissima = 1; otherwise = 0 Rocchetta = 1; otherwise = 0 Sant'Anna = 1; otherwise = 0 Vera = 1; otherwise = 0 Ferrarelle = 1; otherwise = 0 Uliveto = 1; otherwise = 0 Sangemini = 1; otherwise = 0
dummy dummy
other brands = 1; otherwise = 0 store brands = 1; otherwise = 0 (baseline)
4. Results The estimation results are reported in Table 3, while the most important performance indicators and statistical tests are summarized in Table 4. The two models show only negligible differences in the values of estimated parameters. In addition, both models show a good overall significance (F-statistic with a P-value much lower than 0.01 in both cases) and a high capability to explain the variability in the data set (adjusted R-squared equal to 0.90 in both cases). Checks for multicollinearity, heteroscedasticity and non-normality of residuals ruled out the existence of such statistical problems for both models.
Table 3. Estimation results: OLS regression
constant multipack bottle size large medium small bottle material glass eco-friendly push-pull salt content medium-high mineral very low mineral effervescence naturally carbonated carbonated low-sodium store supermarket minimarket discount store ln distance brand San Benedetto Levissima Rocchetta Sant'Anna Vera Ferrarelle Uliveto Sangemini other brands
coefficient -3.61 -0.01
extended model S.E. p-value 0.18 <0.001 *** 0.04 0.80
coefficient -3.60
parsimonious model S.E. p-value 0.17 <0.001 *** not included
0.27 0.69 0.97
0.05 0.07 0.05
<0.001 <0.001 <0.001
*** *** ***
0.28 0.70 0.99
0.04 0.06 0.04
<0.001 <0.001 <0.001
*** *** ***
1.18 -0.03 0.68
0.09 0.05 0.06
<0.001 0.60 <0.001
***
1.18
***
***
0.68
0.08 <0.001 not included 0.06 <0.001
-0.05 -0.13
0.09 0.05
0.53 0.01
***
-0.13
not included 0.05 <0.001
***
0.11 -0.02 -0.21
0.08 0.03 0.05
0.18 0.51 <0.001
***
-0.21
not included not included 0.04 <0.001
***
0.09 0.09 -0.45 0.28
0.03 0.03 0.06 0.03
<0.001 <0.001 <0.001 <0.001
*** *** *** ***
0.09 0.10 -0.46 0.28
0.03 0.03 0.06 0.03
<0.001 <0.001 <0.001 <0.001
*** *** *** ***
0.28 0.44 0.66 0.42 0.16 0.53 0.30 1.02 0.38
0.06 0.06 0.06 0.07 0.07 0.10 0.09 0.08 0.07
<0.001 <0.001 <0.001 <0.001 0.01 <0.001 <0.001 <0.001 <0.001
*** *** *** *** ** *** *** *** ***
0.28 0.44 0.66 0.43 0.18 0.60 0.37 1.09 0.40
0.06 0.06 0.05 0.07 0.06 0.09 0.07 0.06 0.06
<0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001
*** *** *** *** *** *** *** *** ***
Notes: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.
Table 4. Estimation checks Type of check Goodness of Fit
OLS (extended model) Dependent variable = ln price per litre F (25 / 348) = 141.53 P-value (F) <0.0001 R2 = 0.91 Adjusted R2 = 0.90
OLS (parsimonious model) Dependent variable = ln price per litre F (20, 353) = 177.46 P-value (F) <0.0001 R2 = 0.91 Adjusted R2 = 0.90
***
Heteroscedasticity
Multicollinearity
Normality of residuals
Log-likelihood = 48.10 Akaike Information Criteria = -44.19 Bayesian Information Criteria = 57.83
Log-likelihood = 45.24 Akaike Information Criteria = -48.48 Bayesian Information Criteria = 33.93
Breusch-Pagan / Cook-Weisberg test for heteroscedasticity H0: the variance is constant H1: the variance is not constant Chi-square (1) = 0.62 p-value = 0.4307 Variance Inflation Factor (VIF) for regressors: multipack: 1.3; large: 3.5; medium: 2.8; small: 3.4; glass: 1.8; eco-friendly material: 1.4; push-pull: 1.4; medium mineral: 7.0; very low mineral: 4.1; naturally carbonated: 6.8; carbonated: 1.6; low-sodium : 2.4; supermarket: 1.3; minimarket: 1.3; discount: 1.3; ln distance: 2.6; San Benedetto: 5.5; Levissima: 4.5; Rocchetta: 3.7; Sant'Anna: 6.6; Vera: 3.1; Ferrarelle: 3.8; Uliveto: 3.5; Sangemini: 2.8; other brands: 9.2 (VIF greater than 10 indicates a multicollinearity problem) Jarque-Bera Test H0: residuals are distributed normally H1: residuals are not distributed normally Chi-square(2) = 1.947 p-value 0.378
Breusch-Pagan / Cook-Weisberg test for heteroscedasticity H0: the variance is constant H1: the variance is not constant Chi-square (1) = 0.25 p-value = 0.6144 Variance Inflation Factor (VIF) for regressors: large: 2.7; medium: 2.4; small: 2.7; glass: 1.5; push-pull: 1.3; very low mineral: 4.1; lowsodium : 2.1; supermarket: 1.3; minimarket: 1.3; discount: 1.2; ln distance: 2.3; San Benedetto: 5.3; Levissima: 4.3; Rocchetta: 3.6; Sant'Anna: 6.5; Vera: 3.0; Ferrarelle: 3.3; Uliveto: 3.0; Sangemini: 2.4; other brands: 9.0 (VIF greater than 10 indicates a multicollinearity problem)
Jarque-Bera Test H0: residuals are distributed normally H1: residuals are not distributed normally Chi-square(2) = 1.859 p-value 0.395
Because the frequency distribution of the unit price is skewed to the right, with low unit prices showing much higher frequencies, as shown in Figure 1, the sample mean may be sensitive to the outliers in the upper percentiles, and thus OLS regression may yield biased results. In order to test the robustness of the OLS estimates, we also performed a Median Regression which uses the median (50th percentile) as a measure of central tendency rather than the mean5. Estimation results of Median and OLS regressions applied to the most parsimonious model are compared in Table 5 which shows only minimal differences in the magnitude of parameters obtained from the two estimation methods. In any case, these differences do not substantially alter the nature of the phenomena. We thus conclude that the results of the OLS regression can be considered as sufficiently robust.
Table 5. OLS vs. Median regression (parsimonious model)
constant bottle size large medium small bottle material 5
OLS coefficient -3.60*** 0.28*** 0.70*** 0.99***
marginal effect N/A 33% 101% 169%
Median regression coefficient marginal effect -3.35*** N/A 0.25*** 0.57*** 0.89***
28% 76% 142%
Median Regression parameters are less sensitive to outliers than OLS parameters analogous to the median vs mean of the sample (Buchinsky, 1998; Koenker and Bassett, 1978; Koenker and Hallock, 2001). This is because Median Regression parameters are estimated by minimizing the sum of absolute residuals, while OLS parameters are estimated by minimizing the sum of squared residuals.
glass push-pull salt content very low mineral low-sodium store supermarket minimarket discount store ln distance brand San Benedetto Levissima Rocchetta Sant'Anna Vera Ferrarelle Uliveto Sangemini other brands
1.18*** 0.68***
224% 98%
1.22*** 0.70***
240% 101%
-0.13*** -0.21***
-12% -19%
-0.19*** -0.06***
-17% -5%
0.09*** 0.10*** -0.46*** 0.28***
9% 10% -37% N/A
0.08*** 0.12*** -0.49*** 0.23***
8% 13% -39% N/A
0.28*** 0.44*** 0.66*** 0.43*** 0.18*** 0.60*** 0.37*** 1.09*** 0.40***
32% 55% 94% 54% 19% 82% 45% 197% 49%
0.27*** 0.40*** 0.63*** 0.50*** 0.15*** 0.60*** 0.52*** 1.07*** 0.40***
31% 49% 87% 64% 16% 82% 68% 191% 49%
Notes: ***, **, and * denote significance at the 1%, 5%, and 10% levels, respectively.
These results show that the type of packaging has a strong effect on the unit price of bottled water. First, the size of the bottle has a significant effect on price and, specifically, the large, medium and small dummies have positive and increasing coefficients equal to +0.28, +0.70 and +0.99, respectively. Considering the functional form of the equation, the coefficient of dichotomous explanatory variables can be transformed into the percentage change in price due to the presence of a given quality attribute (marginal effect) applying the formula: {exp (coefficient) - 1}. It follows that, assuming that the largest bottles (2L) are the baseline, the unit price increases by +33%, +101% and +169%, respectively, when water is sold in progressively smaller bottles (1.5L, 0.75-1.5L, ≤0.5L). This can be explained considering both the higher costs of packaging for smaller bottles (Gleick and Cooley, 2009), and the higher willingness to pay of those consumers who prefer smaller bottles because they are considered as more practical, especially outside the home (Doria, 2006; He, Jordan and Paudel, 2008; Van Der Linden, 2015; Viscusi, Huber and Bell, 2013). The same considerations can be made about the type of cap. In fact, bottled water with a push-pull cap has a much higher unit price (+98%) related to both the higher cost of packaging and consumer preferences. The kind of material used for the bottle (codified by the glass, plastic and eco-friendly dummies) also has a significant effect on price. Water in glass bottles gains a relevant premium price (+224%), while, unexpectedly, a non-significant effect was found for bottles made from eco-friendly materials (recycled, partly recycled or biodegradable plastic), assuming plastic bottles as the baseline. The premium for water in glass bottles can be explained considering the higher cost of glass compared to plastic, but also the high willingness to pay of those consumers who prefer a more elegant glass bottle rather than an ordinary plastic one. Conversely, the result regarding the bottles made from eco-friendly material means that consumers are not willing to pay any premium for this attribute, even considering its positive impact on the environment (Lagioia, Calabrò and Amicarelli, 2012; Gironi and Piemonte, 2010; Nessi, Rigamonti and Grosso, 2012; Niccolucci et al., 2011).
The type of pack (multipack) is another non-significant variable, meaning that single bottles and multipacks are sold at the same unit price, all other characteristics being equal. This is also an unexpected result because a discount on the unit price is usually given when a larger number of products are purchased. The intrinsic attributes of bottled water only seem to have a slight influence on price. Specifically, water with a very low mineral content had a significant but moderate discount price (-12%), while water with mediumhigh mineral content did not have any premium or discount price compared to water with low mineral content used as the baseline. Water with low sodium content also showed a discount price (-19%) compared to water with a medium-high sodium content, while the presence/type of effervescence did not have any effect on price. These results show that consumers do not appear to be very interested in the mineral composition of bottled water even though they express less preference for water with a very low content of minerals, including sodium. The type of retail store, the distance between the place of bottling and the point of sale, and the brand also affect the price of bottled water considerably. Bottled water sold in discount stores had a relevant discount price of -37% compared to bottled water sold in hypermarkets used as the baseline. On the other hand, bottled water sold in supermarkets and minimarkets showed only a moderate premium price equal to +9% and +10%, respectively. These price differences may be related to the logistics costs and store management which are, obviously, much lower for discount stores and higher for supermarkets and minimarkets compared to hypermarkets. The distance between the place of bottling and the point of sale is also a significant variable with a coefficient equal to +0.28. Taking into account the logarithmic form of the equation, the coefficient of a continuous variable can be directly interpreted in terms of elasticity. Therefore, the positive (but less than one) coefficient of the distance variable means that an increase in the distance leads to a less-thanproportional increase in the unit price of bottled water. Obviously, the positive impact of distance on the price of bottled water may be mainly related to the transportation costs. Finally, the brand affects the unit price of bottled water considerably. Compared to store brands used as the baseline, all major brands (San Benedetto, Levissima, Rocchetta, Sant’Anna, Vera, Ferrarelle, Uliveto, Sangemini) as well as minor brands gain a significant premium price ranging from a minimum of +19% (Vera) to a maximum of +197% (Sangemini).
5. Conclusions The results of this study show that bottled water is, surprisingly, highly differentiated. However its retail price is mainly affected by extrinsic characteristics, in particular brand and packaging (bottle size, bottle material and type of cap). Other extrinsic factors, i.e. the type of store and the distance between the water source and the point of sale, also have a relevant effect on price. Conversely, the intrinsic characteristics affect the price of bottled water only moderately.
Based on these results, it is possible to draw some useful insights for both practitioners and policy makers. Practical implications can be grouped according to three main dimensions: commercial, environmental and regulatory. From a commercial point of view, bottled-water producers need to revise their marketing strategies to tackle a market that seems to have reached the maturity stage of its life cycle. As is known, the two main marketing strategies for a competitive advantage are based on “lower cost” and “differentiation” (Porter, 1985), sometimes combined within the same firm through different brands. In order to extend the lower cost strategy, firms should concentrate on high levels of efficiency and cost reductions. Thus, the type of packaging is crucial because it greatly affects the price of bottled water and, at the same time, the relative production costs. Firms should adopt less-expensive types of packaging represented by larger-sized plastic bottles with standard caps. Another important option to reduce the cost of packaging, although not considered in our study, is to act on bottle weight by adopting thinner plastic (Lagioia, Calabrò and Amicarelli, 2012). Moreover, firms implementing a lower cost strategy should opt for growth based on the penetration of local markets surrounding water sources, rather than expanding the market. In fact, expanding the market would increase transportation costs which would have a strong impact on the final price of bottled water. Finally, discount stores and hypermarkets seem more suitable channels for the distribution of low-priced bottled water, taking into account their more efficient logistics. A product differentiation strategy can be strengthened primarily through the brand. As shown by the results of this study, brand equity plays a key role in differentiating bottled water, considering its strong effect on price. However, a product differentiation strategy merely based on the brand is not sustainable in the longterm especially in mature markets characterized by strong competitive pressure. Our analysis shows that leading brands such as “Vera” and “San Benedetto” are already adopting very aggressive pricing strategies. Considering also the high level of product substitutability that characterizes the market of bottled water, the reinforcement of brand equity should be supported by a real differentiation of products based on both extrinsic and intrinsic attributes specifically aimed at meeting consumers’ needs and preferences. Therefore, firms should diversify their products, firstly, in terms of packaging by offering the same product in bottles made from different materials (glass and plastic), sizes and with different types of cap. Despite our results showing that the mineral composition only has a moderate effect on the price of bottled water, firms interested in implementing a product differentiation strategy should not neglect this aspect. The mineral composition of bottled water varies notably among different brands, however consumers do not seem to perceive these differences. One reason could be that, despite the mineral composition being indicated on the label, consumers have difficulty in interpreting and comparing this information. We therefore argue that firms should emphasize the connection between the brand and the specific intrinsic properties of water. For example, some brands (Ferrarelle and Lete) emphasize the natural effervescence and its effect on taste, while others (Sangemini, Lilia) stress the high calcium or magnesium content and the importance of these minerals for health and wellbeing (Black, 2009). Generic claims about the positive health effects of natural mineral water (e.g. “helps digestion”, “promotes diuresis”), inserted on almost all the labels of bottled water, have no
real value unless they explain the specific intrinsic characteristics that generate these health effects and whether these characteristics are common or not to other products. Considering the expensive advertising campaigns for communicating the specific properties of bottled water to consumers, firms should expand their market and implement a diffuse distribution network. Since market development involves higher transportation costs, logistics plays a key role and should be carefully optimized. National brands should consider implementing a multi-sourcing strategy with a short chain distribution which could be useful to reduce the transportation costs (Lagioia, Calabrò and Amicarelli, 2012). As regards environmental issues, the main result of our analysis is that the market of bottled water does not seem to exhibit any effective signs of self-regulation towards a reduction in its environmental impacts. We found that consumers are willing to pay higher prices for bottled water packaged in small-sized bottles and with special caps (push-pull) in order to satisfy their need for convenience despite this having a more negative impact on the environment (Gleick and Cooley, 2009). In addition, consumers are willing to pay higher prices for bottled water which is transported over long distances (often greater than 1,000 kilometres) with high energy consumption and air pollution (Botto et al., 2011; Gleick and Cooley, 2009; Gironi and Piemonte, 2010; Lagioia, Calabrò and Amicarelli, 2012; Nessi, Rigamonti and Grosso, 2012; Niccolucci et al., 2011). Conversely, consumers are not willing to pay any premium for water packaged in bottles made from eco-friendly materials (recycled, partially recycled or biodegradable plastic) (Lagioia, Calabrò and Amicarelli, 2012; Gironi and Piemonte, 2010; Nessi, Rigamonti and Grosso, 2012; Niccolucci et al., 2011). Considering that the environmental impact of bottled water consumption is a public concern, it also seems interesting to provide some regulatory insights. Since the bottled water market does not seem capable of selfregulation, policy makers should implement measures to encourage producers and consumers towards a more responsible behaviour in terms of environmental sustainability. First, a taxation scheme based on the type of packaging used for bottling natural mineral water could be introduced in order to discourage the excessive use of small bottles and, at the same time, sustain the use of bottles made from eco-friendly materials. It is worth noting that Italian regulation prohibits the use of containers holding more than 2 litres which seems to be a counterproductive constraint. In fact, using bigger containers (e.g. 3L, 5L, 10L), already used for other food products such as wine, would be an effective tool for reducing both the price of bottled water and the environmental impact of its consumption. Another important issue concerns the current taxation system of water abstraction which varies depending on the region, and is mainly calculated as a constant tariff per unit volume of water abstracted (Legambiente, 2014). Inevitably, such regulation promotes an increase in bottled water flows from the regions where taxation is lower to regions where the taxation is higher and, consequently, an increase in environmental impacts of bottled water consumption. Therefore, also considering that Italy is one of the European countries where bottled water is sold at the lowest price (Bevitalia, 2015), increasing and levelling the taxes among the regions could encourage the purchase of cheaper bottled water from local springs with positive effects on the environment.
Acknowledgements
The authors would like to thank the two anonymous reviewers for their useful comments, needless to say that any shortcomings are the responsibilities of the authors alone.
References Abrahams, N. A., Hubbell, B. J., & Jordan, J. L. (2000). Joint production and averting expenditure measures of willingness to pay: do water expenditures really measure avoidance costs?. American Journal of Agricultural Economics, 82(2), 427-437. Bevitalia (2015). Acque Minerali, Bibite e Succhi - Annuari del Bere, Beverfood.com Edizioni. Black, R. (2009). Acqua minerale di Sangemini: the Italian mineral water industry finds a place at the table. Journal of Modern Italian Studies, 14(2), 184-198. Boatto, V., Defrancesco, E., & Trestini, S. (2011). The price premium for wine quality signals: does retailers information provision matter? British Food Journal, 4-5, 669-679. Botto, S., Niccolucci, V., Rugani, B., Nicolardi, V., Bastianoni, S., & Gaggi, C. (2011). Towards lower carbon footprint patterns of consumption: The case of drinking water in Italy. Environmental Science & Policy, 14(4), 388-395. Buchinsky, M. (1998). Recent advances in quantile regression models: a practical guideline for empirical research. Journal of human resources, 88-126. Burnham, K. P., & Anderson, D. R. (2003). Model selection and multimodel inference: a practical information-theoretic approach. Springer Science & Business Media. Carew, R., Florkowski, W. J., & Smith, E. J. (2012). Hedonic analysis of apple attributes in metropolitan markets of Western Canada. Agribusiness, 3, 293–309. Carlucci, D., De Gennaro, B., Roselli, L., & Seccia, A. (2014). E-commerce retail of extra virgin olive oil: an hedonic analysis of Italian SMEs supply. British Food Journal, 10, 1600-1617. Carlucci, D., Stasi, A., Nardone, G., & Seccia, A. (2013). Explaining Price Variability in the Italian Yogurt Market: A Hedonic Analysis. Agribusiness, 2, 194–206. Comuni Italiani (2015).
Gironi, F., & Piemonte, V. (2010). Life cycle assessment of polylactic acid and polyethylene terephthalate bottles for drinking water. Environmental Progress & Sustainable Energy, 30 (3), 459–468. Greenlee, L. F., Lawler, D. F., Freeman, B. D., Marrot, B., & Moulin, P. (2009). Reverse osmosis desalination: water sources, technology, and today's challenges. Water research, 43(9), 2317-2348. He, S., Jordan, J., & Paudel, K. (2008). Economic evaluation of bottled water consumption as an averting means: evidence from a hedonic price analysis. Applied Economics Letters, 15, 337–342. International Bottled Water Association – IBWA (2015).
Rani, B., Maheshwari, R., Garg ,A., & Prasad, M. (2012). Bottled Water – A Global Market Overview. Bulletin of Environment, Pharmacology and Life Sciences, 1, 1-4. Roselli, L., Carlucci, D., & De Gennaro, B. (2016). What Is the Value of Extrinsic Olive Oil Cues in Emerging Markets? Empirical Evidence from the US E‐Commerce Retail Market. Agribusiness. Rosen, S. (1974). Hedonic prices and implicit markets: Product differentiation in pure competition. Journal of Political Economy, 82, pp. 34–55. Satimanon, T., & Weatherspoon, D. D. (2010). Hedonic analysis of sustainable food products. International Food and Agribusiness Management Review, 4, 57–74. Schamel, G. (2006). Geography versus brands in a global wine market. Agribusiness, 3, 363–374. Schollenberg, L. (2012). Estimating the hedonic price for Fair Trade coffee in Sweden. British Food Journal, 2-3, 428-446. Snipes, M., & Taylor, D. C. (2014). Model selection and Akaike Information Criteria: An example from wine ratings and prices. Wine Economics and Policy, 3(1), 3-9. Steiner, B. E. (2004). Australian wines in the British wine market: A hedonic price analysis. Agribusiness, 3, 287–307. Torretta, V. (2013). Environmental and economic aspects of water kiosks: Case study of a medium-sized Italian town. Waste Management, 33, 1057–1063. Van Der Linden, S. (2015). Exploring Beliefs About Bottled Water and Intentions to Reduce Consumption The Dual-Effect of Social Norm Activation and Persuasive Information. Environment and Behavior, 47(5), 526-550. Viscusi, W. K., Huber, J. C., & Bell, J. (2013). The private rationality of bottled water drinking. Contemporary Economic Policy, 14-1. Wilk, R. (2006). Bottled water: The pure commodity in the age of branding. Journal of Consumer Culture, 6(3), 303-325. Ward, C. E., Lusk, J. L., Dutton, J. M. (2008). Implicit value of retail beef product attributes. Journal of Agricultural and Resource Economics, 3, 364–381.