Consumer Demand and Willingness to Pay for Healthy Food ...

Consumer Demand and Willingness toPay for Healthy Food Diversity inGermanyDr. Larissa DrescherDepartment of Rural EconomyUniversity of AlbertaMaking Choices: Consumer and Their Impact onCanada’s Agriculture and FoodNovember, 5th, 2007

Outline1. The impact of food diversity2. Measuring healthy food diversity- Traditional diversity measures- Healthy Food Diversity-Index (HFD-Index)3. Theory- Demand for healthy eating- Consumer Goods Characteristics Model (CGCM)4. Data & estimation5. Selected results- Demand analysis- Willingness to pay (WTP)6. Conclusions & Implications2

1. The impact of food diversityFood supply in the German retailer over timeNumber of food items14000120001000080006000400020000SupermarketConsumer marketDepartment storesAverage1979 1991 1994 2000 2003YearEurohandelsinstitut, several years3

1. The impact of food diversityPositive health aspects of food diversity- Assures supply with as many of 40 nutrients that arenecessary for survival (Royo-Bordonada et al., 2003)- Increases dietary quality (Hatloy et al., 1998)- Reduces probability of excessive contaminant intakefrom high amounts of any single food (Krebs-Smith etal., 1987)- Reduces morbidity: variety in vegetables decreasescancer risk about 36% (Jansen et al., 2004)- Decreases mortality risks: higher variety lowersmortality risk for all-sites cancer about 30% in women(Michels & Wolk, 2002)High diversity of plant foods: healthy food diversity4

1. The impact of food diversityFood guidelines- Advise food choices that promote health and decrease risk of chronicdiseases of all healthy people- Some differ across countries due to differences in food availablity,environment and eating patterns- But: diversity guideline almost always the first oneVariety message in USDA‘s guidelines over timeEdition YearWording1980-1999 Eat a variety of foods.2000 Let the pyramid guide your food choice. Choose a variety of grainsdaily, especially whole grains. Choose a variety of fruits andvegetables daily.2005 Consume a variety of nutrient-dense foods and beverages withinand among the basic food groups while choosing foods that limitthe intake of saturated and trans fats, cholesterol, addedsugars, salt, and alcohol.5

1. The impact of food diversity- 1st of 10 guidelines of the German NutritionSociety (2005): “Enjoy the great variety of foods.Selection from among a great many different foods,appropriate combination and adequate quantities ofhigh-nutrient and low-energy food are characteristics ofa well-balanced diet“- Eating well with Canada‘s Food Guide (2007):„Enjoy a variety of foods from the four food groups.“6

2. Measuring healthy food diversityTraditional diversity measuresDistribution measuresn• Berry-Index BI = 1 − ∑Count Measures• Most frequently used• Number of foods or foodgroups• Disadvantages:– Equal weight forchocolate and apple– Distribution of foodsnot considered2s ii = 1• Number and distribution offoods• Disadvantages:– Equal weight forchocolate and apple– Diversity higher, themore foods are eatenin equal amountsCount measure = 250% 1% 50% 99%Berry – Index = 0,5 0,00997

2. Measuring healthy food diversityHealthy Food Diversity (HFD) - Index⎛ n ⎞ ⎛ n= ⎜ −21 ⎟ ⎜⎜ ∑si*⎟ ⎜∑hf⎝ i = 1 ⎠ ⎝ i = 1i* si⎞ s⎟ i= quantity share⎟⎠ hf i= health factorBerry-Index x health value= diversity (of the food basket)health factors = hf i= describes the „value“ of single foods derived from the German food pyramide (German NutritionSociety)8

2. Measuring healthy food diversityHealthy Food Diversity – Indexqualitative dietary guidelines: pyramide sidesquantitative dietary guidelines: (modified) nutritioncircle 15 health factors9

2. Measuring healthy food diversityShare of foodgroup(quantitativedietary guideline)Plant foods 73%Animal foods 25%Share of food sub group(qualitative guideline quantified)Vegetables/fruits/leaf salads/juices 36%Wholemeal products/paddy 28%Potatoes 20%White meal products/peeled rice 12%Snacks and sweets 4%Fish/low-fat meat/low-fat meat products 36%Low-fat milk, low-fat dairy products 28%Milk, dairy products 20%…Health factor0.73 *0.36 = 0.260.28 = 0.200.20 = 0.150.12 = 0.090.04 = 0.030.25 *0.36 = 0.090.28 = 0.070.20 = 0.05Fats & Oils 2& … …Health value = ∑health factor of food i * quantitative share of food i10

2. Measuring healthy food diversity• HFD-Index = Berry-Index x health value• Properties of HFD– Within health factor groups (eggs and meat)• increases with increasing number of foods eaten• Increases the more equally distributed the foods are– If the distribution between health factor groups (meat andapple) changes in favour of healthy (unhealthy) foodgroups:• HFD increases (decreases)• HFD-Index better reflects a healthy diet than traditionaldiversity measures (Drescher, Thiele, Mensink, 2007;Journal of Nutrition)11

3. TheoryDemand for healthy eatingDemand for Selected studies Demand theory CriticismNutrientsFoodsAbdulai and Aubert(2004)Brown andSchrader (1990)Lancaster‘scharacteristicsapproachTraditionaldemand theoryConsumer don‘tdemand nutrients;confounder problemIncomplele picture ofthe diet; confounderproblemDietary quality(nutrients)Thiele et al. (2004) No reference Consumer don‘tdemand nutrients;confounder problemDietary quality(HealthyEating-Index)Variyam et al.(1998)Lancaster‘scharacteristicsapproachConsumer don‘tdemand nutrients;confounder problem Complete picture of the diet: utilization of the HFD-Index withinLancaster‘s characteristics approach12

3. Theory• Lancaster’s characteristics approach (1966; 1971):– Not goods but rather characteristics provide utility– Linear production technology = amount ofcharacteristics from the consumption of all products• Consumer Goods Characteristics Model (CGCM) extension of Lancaster’s model by Ladd andSuvannunt (1976)• Assumption of the CGCMA1) negative marginal utility of one or more characteristicsprovided that the sum of the marginal utility remainspositiveA2) common, not necessarily linear consumptiontechnologyA3) utility independent of the distribution of characteristicswithin the good• CGCM version suggested by Eastwood et al. (1986)13

3. TheoryConsumer Goods Characteristcs Approach• “Health” is a utility providing good; produced by foodswith the following characteristics:– Healthy food diversity (HFD)– Other nutritional characteristics, e.g. Vitamin C (C j)• Production functions– Healthy food diversity (HFD) (non-linear)• Influenced by human capital (K) (Chern, 2003)• Common:HFD = g( q ,..., q ,K )1 nq i = quantity– Other nutritional characteristics (C j ) (linear)• Education is a proxy variable of nutritionalknowledge (Variyam et al., 1998)• Common:q i = quantityC = fj( q ,..., q ,K )1 n14

3. TheoryConsumer Goods Characteristcs Approach• Representative utility functionU = U C( ( q q , K ),...,C ( q ,..., q , K) , HFD( q q , K)1 1,..., n m 1 n1,...,n• Budget restriction (1 st constraint)n∑i = 1piqi≤ YY= incomep i= price of good iq i= quantitiy of good i• Quantity restriction (2 nd constraint)Q=n∑i = 1q iQ = total quantity (food basket)15

3. TheoryConsumer Goods Characteristcs Approach• Solving maximization problem with LagrangeL = U( C ( q ,..., q , K) ,..., C ( q ,... q , K) , HFD( q ,..., q K))⎛+ λ ⎜1⎜⎝n∑i = 11 1 nm 1 n1 n,qi⎞ ⎛− Q⎟+ λ ⎜ −⎟2 Y⎜⎠ ⎝n∑i = 1p q• First order condition (FOC)ii⎞⎟⎟⎠∂L∂qi=m∑j = 1∂U∂Cj∂Cj ∂U∂HFD* + * + λ1− λ2p∂q∂HFD∂qiii= 0• 2 ways of solving FOC:– Derivation of the demand function for characteristicsHFD = dHFD( p ,Y,K)– Solving FOC for prices (p i) implicit prices for thecharacteristics willingness to pay (WTP)nm∑p ⎜iqi= ∑βjCj + βHFD∑i = 1 j = 1i = 1⎛⎜⎝n∂HFD∂qi⎞* q ⎟i + β Q⎟0⎠16

4. Data & estimation• German Nutrition Survey (1998): 4030 participants– Representative for Germany– 133 different foods• External price data: MADAKOM and ZMP• Demand analysis: multiple regression– Dependent variable: HFD-Index (logit transformation)– Independent variables: socioeconomic, attitude andknowledge variables– Endogeneity problem: two-stage least squares estimation• Willingness to pay: multiple regression– Dependent variable: expenditure per unit (Box-Coxtransformed)– Independent variables:• (Instrumented) characteristics of the food basket (HFD,Vitamin C, Calcium, quantity approximated by calories)• Socioeconomic variables (Nerlove, 1995)17

5. Selected resultsDemand analysisHFD-Index(transformed)Independent variables: Parameter t-valueConstant 0.0320 0.13Household income in DM/1000 0.010* 1.83Age (logarithm) 0.380***Low education level -0.112**High education level 0.059**3.79-2.742.46Supplement taker 0.064** 3.12Vegetarian 0.290*** 6.57Sportive activity 0.100*** 4.48Smoker -0.129*** -5.07R 2 0.175***,**,* significance level 1%, 5%, 10%18

5. Selected resultsWillingness to pay (WTP)Dependent variable: food expenditure per unitParameter Chi 2 -valueHFD’ 164.638*** 1016.25Caloric density 19.442*** 921.51Vitamin C- density 37.043*** 136.11Calcium – density -10.729*** 200.56Net equivalence income(DM/1000)-0.162*** 24.24Age -0.205*** 163.80Age 2 0.002*** 131.96Low education 0.218* 2.91High education -0.839*** 85.55Supplement taker -0.226** 10.87Smoker 1.080*** 168.78Vegetarian -2.488*** 263.58***,**,*significancelevel1%, 5%, 10%19

5. Selected resultsWillingness to payVariable Implicit prices ElasticitiesHFD’ 0.066 DM/HFD 1.721Caloric density 0.008 DM/kcal 2.397Vitamin C-density 0.015 DM/mg 0.298Calcium-density -0.004 DM/mg -0.46620

6. Conclusions & ImplicationsLinking the results:• Higher income:- Demand for HFD ↑, food expenditures per unit ↓- No significant interaction effect between HFD and income onfood expenditure per unit High-income consumers are more efficient in selecting a goodquality with less costly foods• Higher education- Demand for HFD ↑, food expenditures per unit ↓- If low-educated consumers demand HFD of higher quality:food expenditures per unit would be higher Low-educated consumers are less efficient in food demand• Positive attitudes towards health– Vegetarians, supplement takers and nonsmokers have ahigher demand for HFD but lower food expenditure per unit A positive attitude towards health enables to obtain HFD withless costly foods21

6. Conclusions & ImplicationsImplications for German nutrition and health policy• Dietary information actions geared at consumers– With lowest education level– Who already reveal undesired health behavior (e.g.smoking)– Younger consumers• Public health strategies– To increase awareness of HFD and to establishincentives that sustain consumer’s self responsibility forhealth– To support changes in dietary behavior in favor of morenutrient- but less energy-dense diets• Communicate how to obtain healthier diets with less costlyfood items22

THANK YOU FOR YOUR ATTENTION!23

3. TheoryConsumer Goods Characteristcs Approach• Derivation of demand functions– Solving for optimal quantities of the characteristics byusing the implicit function theorem– Demand function for Healthy Food Diversity:HFD=Cdj =HFDdj( p ,Y,K)– Demand function for other nutritional characteristics( p ,Y,K)• Derivation of implicit pricespi⎛ m1 ⎜ ∂U=λ ⎜∑2 ∂C⎝ j = 1 j*∂C∂qij+∂U∂HFD*∂HFD∂qi⎞+ λ ⎟1⎟⎠24

3. TheoryConsumer Goods Characteristcs ApproachDerivation of implicit prices• Assumptions:– λ 1 = marginal utility of the total quantity of foodsconsumed (assumed to be constant)– λ 2 = marginal utility of income (Ladd and Suvannunt,1976; Eastwood et al., 1986)• With respect to the budget constraint, Y is replaced byexpenditure (E)• Hedonic price equationpi=m∑j = 1∂C∂qij∂E∂Cj∂HFD+∂qi∂E∂HFD∂E+∂QMarginal rate of substitution of consumer‘s Implicit price expenditure for total for quantity each HFD of theothe implicit rnutritional price for characteristics one additional unit marignal HFD implicit price for an additionalunit of characteristic j25

3. TheoryConsumer Goods Characteristcs Approach• Empirically estimable hedonic price equationpi=m∑j = 1β cjji+ βHFD∂HFD∂qi+β 0• Prices of homogenous food products are similaramong consumers and do not vary substantially multiplication by q inm∑p ⎜iqi= ∑βjCj + βHFD∑i = 1 j = 1i = 1⎛⎜⎝n∂HFD∂qi⎞* q ⎟i + β Q⎟0⎠• Assumption of constant implicit prices not consistentfor a unique maximum (Morse and Eastwood, 1989) variable implicit prices: Box-Cox transformation26

6. Summary & Conclusions• Demand for HFD positively influenced by– Income– Education– Positive attitudes towards health• WTP for healthy food diversity– Highly significant WTP for HFD (0.66DM/HFD)– WTP for caloric density is higher than for HFD– Participants with• high income• high education• positive health attitudes have lower food expenditures per unit compared totheir counterparts27