British Food Journal Paying t o see a bug on my f ood: how regulat ions and inf ormat ion can hamper radical innovat ions in t he European Union Tiziana de-Magistris Stefano Pascucci Dimitrios Mitsopoulos Article information: To cite this document: Tiziana de-Magistris Stefano Pascucci Dimitrios Mitsopoulos , (2015),"Paying to see a bug on my food: how regulations and information can hamper radical innovations in the European Union", British Food Journal, Vol. 117 Iss 6 pp. Permanent link t o t his document : http://dx.doi.org/10.1108/BFJ-06-2014-0222 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) Downloaded on: 01 May 2015, At : 06: 14 (PT) Ref erences: t his document cont ains ref erences t o 0 ot her document s. To copy t his document : permissions@emeraldinsight . com The f ullt ext of t his document has been downloaded 19 t imes since 2015* Access t o t his document was grant ed t hrough an Emerald subscript ion provided by 189852 [ ] For Authors If you would like t o writ e f or t his, or any ot her Emerald publicat ion, t hen please use our Emerald f or Aut hors service inf ormat ion about how t o choose which publicat ion t o writ e f or and submission guidelines are available f or all. Please visit www. emeraldinsight . com/ aut hors f or more inf ormat ion. About Emerald www.emeraldinsight.com Emerald is a global publisher linking research and pract ice t o t he benef it of societ y. The company manages a port f olio of more t han 290 j ournals and over 2, 350 books and book series volumes, as well as providing an ext ensive range of online product s and addit ional cust omer resources and services. Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive preservation. *Relat ed cont ent and download inf ormat ion correct at t ime of download. 1 2 3 4 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 5 In this paper we present and discuss a paradox which seems to characterise the agri 6 food innovation landscape of the European Union (EU). On one hand the EU is 7 supporting and fostering ground breaking innovations, especially in the direction of 8 exploring novel and sustainable source of food ingredients, often by subsidising 9 research organizations, Small and Medium Enterprises (SMEs) and start up companies 10 (Dries , 2014). On the other hand EU policy makers are investing resources to set 11 up and enforce an effective regulatory environment, in order to promote quality 12 standards, to foster food safety, and ultimately protect consumers from risks derived by 13 novel food products (Frewer 14 conditions to create a safe and profitable environment to foster innovation, and to 15 provide win win solutions to EU stakeholders. However we also highlight that a lack of 16 coordination and alignment of these two streams of regulation can hamper innovation 17 potentials, reducing potential benefits for producers and consumers, and thus creating 18 what can be defined as an example of “regulatory or nonmarket failures1” (Borrás, 19 2006; Wijnands 20 is based on the idea that European consumers, as well as other stakeholders, are very 21 sensitive to information, and strongly “conditioned” by their cultural patterns and issues 22 of neo phobia2 when it comes to novel food (Sylvester 2007; Rollin , 2011). In principle this is guaranteeing the 2011; Brouwer 2013). Such a criticism 2009; Frewer , 2011). 1 Nonmarket (or regulatory) failures can be defined as “outcomes that depart from the efficiency or distributional goals by which also market outcomes are judged to fail. Although the touchstones of success are similar, the ways in which nonmarket solutions fail differ from those in which market outcomes fail”. (Wolf, 1987, p. 115). In line with other studies in this paper we use the term nonmarket failure and regulatory failure as synonyms (see also Fox and Cranfield, 2014) 2 Food related neophobia can be defines as fear of eating new or unfamiliar foods. A neophobic food consumer is less willing to try new things and take risks (Logue, 2004). 1 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 23 The result is a lower degree of openness for radical innovations, compared to other 24 areas in the world (Sylvester 2009; Bunte 2011). 25 In our view, one important example of “regulatory failures” within the EU context, 26 is provided by the case of insect based products. In the original plan of the European 27 Commission, an update of the European Novel Food Regulation (EC) 258/97 (ENFR) 28 should have been established in 2012, in order to clarify the “legal status” (among other 29 products) of insect based food products3 (Belluco 30 be marketed in the EU as whole, for both human and animal consumption, but they 31 cannot be processed and used as ingredients in food products (e.g., as source of proteins 32 or micro nutrients) because they are considered as novel food4. However, the use of 33 insects as whole may not be advantageous to producers since this would mean that 34 whole insects can be visualised on food products. Visualization may create several 35 concerns among consumers, leading them to rejection due to disgust and/or neo phobia. 36 In line with these concerns, our broader research question is focused on determining 37 whether the ENFR is creating a regulatory failure, thus hampering the market potential 38 of insect based food products in the European Union. More specifically we have 39 examined the role of the ENFR on consumers’ acceptance of and willingness to pay 40 (WTP) for insect based food products. Using a choice experiment, we assess the 41 presence and relevance of a regulatory failure through the analysis of consumers’ 42 preferences and WTPs for insect based food products, with different product attributes 43 directly imposed by the ENFR. Namely, we assess the effect of the 44 insects in the product and the use of , 2013). At this stage, insects can such as of and 45 on consumers’ WTP. The results show that consumers prefer and are 46 willing to pay a premium price for insect based products with a nutritional health claim 3 4 At the time we are writing this paper, no such update has been put in place yet. A more detailed description of the state of the art can be found at http://www.food.gov.uk/multimedia/pdfs/ipletter110812.pdf 2 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 47 and logo, but they are not willing to pay (otherwise willing to be compensated) for a 48 product with a visualized insect. These results may indicate the risk of regulatory failure 49 for insect based food products due to the ENFR. 50 The rest of the article is organized as follows: next section discusses regulatory 51 failures in the context of (radical) food innovation pathways. In section 3 we describe 52 the experimental design and the model specification, and section 4 presents the results. 53 In the final sections we discuss conclusions and main implications of our study. 54 55 56 Fostering ground breaking, radical innovation at both business and consumer level 57 is one of the main priorities for the EU (Dries , 2014). Particularly, the 58 development and exploitation of new source of sustainable food ingredients is one of 59 the new frontiers in the European Union when it comes to radical innovation (Hoek 60 , 2013). For example, there is increasing awareness in the scientific as well as in the 61 business community to reduce meat consumption, and find ways to develop large scale 62 oriented meat alternative products. Meat production and consumption are considered 63 one of the most relevant sources of environmental degradation as well as health costs 64 related to diabetes, cardiovascular disease and obesity, mainly due to over consumption 65 (Oonincx and de Boer 2012; Hoek et al., 2013). From a more regulatory and policy 66 making point of view, we can observe the presence of two distinct but related regulatory 67 frameworks influencing the pathways of radical food innovation within the EU (figure 68 1). 69 70 >Figure 1. Regulatory frameworks in the EU radical innovation landscape 71 3 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 72 On one hand the EU is more and more supporting Research and Development 73 (R&D) policy measures to stimulate and foster ground breaking innovations, especially 74 by subsidizing SMEs and start ups. As a consequence a growing number of food 75 scientists and business actors have started to cooperate, particularly in the domain of 76 alternative and sustainable sources of food ingredients. On the other hand the EU food 77 regulations are more and more keen in creating a safe environment for EU consumers, 78 therefore setting up and enforcing measures of food safety, such as mandatory labelling 79 about quality and health claims (Frewer 80 regulations is dedicated to control the introduction of novel food products and 81 ingredients in the EU (ENFR). One of the main principles of the ENFR is based on the 82 “absence of evidence of consumption within the EU” of the specific food product or 83 ingredient before 1997 as a role to classify a product as novel (Belluco 84 Therefore, even when a food product or ingredient is widely consumed outside the EU, 85 without evidence of treat to food safety, it is still considered novel. In this case any 86 product or ingredient needs to go through a “dedicated” procedure, in order to get the 87 permission to be marketed within the EU. From our point of view, the presence of this 88 principle is not supported by a scientific rationale, while in many case it constitutes the 89 premise to create tensions and misalignment between regulations and ultimately create 90 the room for regulatory failures (Hobbs et al., 2014) (see figure 1). More specifically, 91 in our perspective, the ENFR is creating the room for a regulatory failure in the domain 92 of insect based food products and ingredients. , 2011). Moreover a distinct body of , 2013). 93 From a business perspective, insect based products can represent a profitable 94 business due to the potential of the meat substitutes market. Insect based products can 95 also represent a sustainable source of food for humans with relatively high nutritional 96 value. Past research suggests that insect based food products can be “win win” products 4 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 97 for the environment and businesses (Oonincx 2010; Oonincx and de Boer, 2012). 98 However, European consumers consider insect based food products as a radical 99 innovation and consider entomophagy, the technical term to define insect consumption, 100 as a major challenge (Pascucci and de Magistris, 2013; Tan , 2015). Insect based 101 food products are still generally unknown to European consumers, and information can 102 have a relevant impact on the adoption of different entomophagy consumption patterns 103 (Tan , 2015). 104 This literature also emphasises how these products are potentially challenging the 105 traditional European food culture. Within the EU, entomophagy is mainly framed as 106 related to niche and ethnic products, sometimes as alternative to meat or within health 107 seeking diets (Ten 108 associate insects to food (Verkerk 109 reinforced by the fact that in many European social contexts, entomophagy is a cultural 110 taboo (Pascucci and de Magistris, 2013; Ten 111 slow process given that westernized societies are still reluctant to use insects, despite 112 being a good source of animal protein (Yen, 2009). Thus, the main attitude towards 113 insects as (part of) food products in European societies is related to either fear or disgust 114 (Verkerk , 2015). Still the great majority of European consumers do not 2007; Pascucci and de Magistris, 2013). This is , 2015). Changing such a taboo is a 2007; Pascucci and de Magistris, 2013), or curiosity (Yen, 2009). 115 116 117 118 119 In order to preliminary test the hypothesis that ENFR is creating a regulatory failure, 120 we looked at how existing regulations can shape different insect based food designs. 121 More precisely we decided to conduct a study on consumers’ preferences and WTP for 122 different designs of an insect based product. In order to limit the problem of unrealistic 5 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 123 or hypothetical choices for consumers, we decided to implement the experiment in an 124 EU country where insect based food products are “less unfamiliar” to the large audience 125 than anywhere else. From this point of view the Netherlands is an ideal setting because 126 the uncertainties and tensions create by ENFR on insect based products didn’t 127 discourage the national government, business actors, and researchers and to keep 128 investing and exploring the potentials of this innovation pathway (Veldkamp 129 2012). In the Netherlands insect based foods or ingredients have been widely promoted 130 or exposed to the public, and several restaurants or groceries have tried to market them 131 (Veldkamp et al., 2012; Pascucci and de Magistris, 2013). 132 During December 2011 March 2012 we have randomly recruited participants in 133 different locations across three Dutch cities (Wageningen, Utrecht and Den Bosch). We 134 used a stratified random sampling procedure (by age, gender), we then target 135 respondents who were the primary food buyers of households. As reported in Appendix 136 1, we prepared 5 different versions of the questionnaire, each containing different 137 treatments in terms of information about insect based products. Some 30/40 138 interviewees have been recruited per version, thus leading to a complete sample of 153 139 consumers5. The questionnaire was facilitated by use of tablets, and it contained 140 questions on socio demographic characteristics (i.e. gender, family size and 141 composition, age, education level, income). Respondents were asked if they had 142 previously tried insect based products. Table 1 presents the main features of the sample. 143 About 63.4% of respondents declared to have never tried an insect based product in the 144 past. However all participants were found familiar with the concept of using insects as 145 food or ingredient. 146 5 In a previous study we present the preliminary results related to a restricted sample, in which the number of baseline consumers were reduced since version E of the questionnaire was still not complete. For further details about the pilot study please refer to Pascucci and de Magistris (2013). 6 147 > Table 1. Definition and Means of Demographic Variables (%) 148 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 149 150 Based on the ENFR procedures insect based products cannot be purchased on the 151 market if insects have been processed, even just smashed or separated in parts (Belluco 152 , 2013). Given this scenario, our empirical strategy has been to identify a number 153 of attributes directly linked to the ENFR framework, such as the presence of insect as 154 whole, and other attributes related to the general EU food regulation, such as 155 and . To identify the “effect” of the ENFR regulation we identify 156 as source of potential barrier for consumers’ WTP insect based products. 157 The main rationale is since ENFR is “imposing” to food producers the use of insects as 158 a whole, the only flexibility in the design of insect based foods is to either let the insects 159 being visualised or not. A positive contribution to WTP will mean that visualization of 160 insect is not hampering the potential capacity of this product to be appreciated and 161 valued by consumers. In this case the ENFR is not creating substantial regulatory 162 failure, since it is not misaligned with the main objectives and investments made via the 163 R&D polices. On contrary, a negative impact on WTP due to visualization will indicate 164 issues of regulatory failure. To fully assess the role of visualization, we used two 165 alternative product designs of equal amount or weight, one in which the insect is clearly 166 visible, and an alternative design where the insect is not visible (see table 2). To assess 167 WTP we designed a hypothetical choice experiment with consumers. Arguably, the 168 choice experiment (CE) approach is now the most widely used stated preference method 169 in valuing products or attributes. Some of the reasons for CE’s popularity include its 170 flexibility to take into account several attributes which can be estimated simultaneously 171 and its consistency with random utility theory and Lancaster’s consumer theory 7 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 172 (Lancaster, 1966). Individual CE questions are also framed in a manner that closely 173 resembles consumer shopping situations (Lusk and Schroeder, 2004). 174 The type of insect based product we used in our investigation is a delicatessen 175 similar to sushi food but insect based (see table 2). This looks like the insect based 176 products that can be purchased at some Dutch restaurants and that have been 177 “advertised” through the national media. Therefore, they are more potentially familiar to 178 the Dutch consumers. In the choice setting, we presented a box containing 4 sushi 179 pieces, each at four different price levels (1.50, 2.50, 3.50 and 4.50 euro). The first price 180 level represents the base price which resembles the average price for a sushi box in a 181 Dutch retail shop, while level two, three and four can be associated to premium prices.. 182 The attributes referring to the general EU food regulation are related to information 183 and health claims, and namely the use of logo, resembling voluntary quality 184 information, and health claims, resembling mandatory quality information. The logo is 185 named “ 186 use of food logos has been indicated as a relevant attribute for conditioning consumer 187 choice and willingness to pay, for example increasing the quality perception of the 188 product (Golan 189 clear reference to insects. To present nutritional health claims we introduced a 190 information related to the content of Omega 3. This kind of information is relevant in 191 the consumer framing of the product (Grunert and Willis, 2007; Anders and Mőser, 192 2010). 193 ”, representing a stylized butterfly chrysalides. In many studies the 2001; Gao and Schroeder, 2009). Therefore we used a logo with a The choice set design follows Street and Burgess (2007), and we used an orthogonal 194 main effect plan (OMEP) to develop the profiles in the first option (Street 2005). 195 We then added one of the generators suggested by Street and Burgess (2007) to obtain 196 the profiles in the second option. The orthogonal main effect plan was calculated using 8 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 197 the SPSS orthoplan, which generated 8 profiles. We used these 8 profiles to obtain the 198 products for the second option using one of the generators derived from the suggested 199 difference vector (1 1 1 1) by Street and Burgess (2007) for 4 attributes with 4, 2, 2 and 200 2 levels, respectively, and the two options. This design is 97.2% D efficient compared 201 to the D optimal. Each respondent was asked to make choices in the 8 choice sets and 202 they had to choose between two products with different attributes and prices. They were 203 also given a no buy option just in case they choose not to pick either of the two 204 products. Finally, in order to reduce potential hypothetical bias, the respondents were 205 asked to read a cheap talk script (see Appendix A1 and A2) similar to one used by Silva 206 et al., (2009), because results from several studies imply that cheap talk is potentially 207 effective in reducing hypothetical bias in stated preference studies (Carlsson 208 2005; Champ 209 Murphy 210 presented to respondents are shown in table 2. 2009; Cummings and Taylor, 1999; List, 2001; Lusk, 2003; 2005; Silva 2009). The information on the attributes and the label 211 212 >Table 2. Attributes and Levels Used in the Choice Design. 213 214 215 Our theoretical model is based on the Lancastrian consumer theory of utility 216 maximization (Lancaster, 1966), with consumers’ preferences for food attributes 217 modelled within a random utility framework (McFadden, 1974). The choice questions 218 were analysed using the random utility framework. Thus, the final specification of the 219 utility function is assumed to depend on the attributes and attribute levels considered in 220 the choice questions. Hence, the utility that individual 221 choice situation is: 222 # = α + β1 $% & + β 2! "# + β3 + β4 obtains from alternative +ε at (1) 9 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 223 where is the number of respondents, denotes each of the three options available in 224 the choice set (alternative A, alternative B and no buy option) and 225 choice sets (eight). The constant α represents the alternative specific constant coded as a 226 dummy variable that takes the value 1 for the no buy option and 0 otherwise. It is 227 expected that the constant α is negative and significant, indicating that consumers obtain 228 lower utility from the no buy option than from the two other product options or 229 designed alternatives (A and B). 230 is the number of The price ($% &) variable is the price of a box for 4 sushi pieces, while the rest of 231 the attributes enter the model as an effect code ( , and ! "# ). In 232 particular, the 233 and 1 if the logo was not present, whereas 234 corresponding claim was present in the product and 1 if corresponding claim was not 235 present. Finally, the visualization of the insect attribute (! "# 236 of +1 if the insect is visualized in option , and 1 if the insect is used in form of 237 ingredient in option in choice situation . variable was set equal to 1 if the logo was present in the product attribute was set equal to 1 if the ) is coded with a value 238 Louviere and colleagues (2002) argued that “scale heterogeneity” is a major source of 239 taste heterogeneity in choice models. Moreover, they also stated that the Random 240 parameter model (RPL) is seriously mis specified since it ignores the scale 241 heterogeneity. Their argument led Fiebig 242 multinomial logit (G MNL) model that nests random parameter model (RPL) and 243 multinomial model and takes into consideration the scale heterogeneity (S MNL). (2010) to develop the generalized 244 Hence, we estimate equation (1), using the Generalized Multinomial Mixed Logit 245 (GMML), and following Layton and Brown (2000) and Revelt and Train (1998), we did 246 not allow the price coefficient to vary in the population. Moreover, The G MNL model 247 is defined as follows in equation (2): 10 248 # = α + [σ β1 + γ 1η + (1− γ 1)σ η ]$% & + [σ β2 + γ 2η + (1− γ 2 )σ η ]! "# + 249 (2) [σ β3 + γ 2η + (1− γ 3 )σ η ] + [σ β4 + γ 4η + (1− γ 4 )σ η ] +ε 250 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 251 where: σ = exp( τ2/2 + τ ), ~ N[0,1].; γ is a parameter between 0 and 1 that 252 indicates how the variance of residual taste heterogeneity varies with scale in a model; σ 253 is the scale of error term which captures scale heterogeneity assuming that σ is 254 heterogeneous in the population, and η captures the residual taste heterogeneity. ε is the 255 he standard i.i.d Gumbel distributed unobservable. 256 The parameter γ is only present in G MNL model. If γ→1, the standard 257 deviation of ηn is independent of the scaling of β . Therefore, the equation (2) is called 258 G MNL I model: 259 # = α + [σ β1 +η ]$% & + [σ β2 +η ]! "# + [σ β3 +η ] + 260 + [σ β4 +η 261 262 ] (3) +ε On the other hand, when γ→0 the standard deviation of ηn is proportional to σn, the equation (2) is called the G MNL II model: 263 # = α + [σ (β1 +η )]$% & + [σ (β2 +η )]! "# 264 + [σ β3 +η )] + (4) + [σ β4 +η )] +ε 265 266 Finally, because σn represents the scale of the idiosyncratic error, it should be 267 positive. Thus, σn is assumed to be log normal with mean 1 and standard deviation τ. 268 Hence, in accordance to this G MNL model, τ is the key parameter that captures the 11 269 scale heterogeneity. When τ and the variance covariance matrix of ηn are 0, we 270 approach MNL (Fiebig 271 contrast, when τ>0 and the diagonal elements of variance covariance matrix of ηn 272 approach 0, G MNL approaches S MNL. The estimations of the G MNL models with 273 corresponding restrictions are made in NLOGIT 5.0. 2010), whereas when τ→0, G MNL approaches RPL. In 274 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 275 276 The estimation results are shown in table 3. The first column presents the results for 277 a simple multinomial model (MNL model 1), whereas the second column contains the 278 results of G MNL that nests a scale multinomial model (S MNL model 2) assuming 279 that τ>0 and the diagonal elements of variance covariance matrix of ηn approach 0. The 280 next two columns report the results from random parameter model (RPL model 3) and 281 the Generalized multinomial mixed logit (G MNL model 4) with Cholesky matrix, 282 respectively. 283 284 >Table 3. Model estimates and mean WTPs 285 286 Since we expect that some attributes may be inter dependent, we assumed the 287 correlation structure of β to follow a multivariate normal distribution (normal with 288 vector mean X and variance covariance matrix Y). If at least some of the estimates for 289 elements of the Cholesky matrix C (where C’C= Y) show statistical significance, then 290 the data are supportive of dependence across tastes (Scarpa and Del Giudice, 2004). To 291 determine which of the four models are preferable, we looked at the log likelihood, 292 information criteria AIC and, the R2. Firstly, we can notice that S MNL model provided 293 a great improvement in fit by adding only one parameter, ( 1225.61 in MNL vs. 12 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 294 994.99 in S MNL ), it leads to substantial improvements in information criteria AIC, 295 AIC3 and R2 . Moreover, the parameter τ is statistically significant, implying that 296 allowing for such heterogeneity leads to a further substantial improvement in fit. 297 However, even if S MNL model provides a great improvement in fit compared to the 298 simple MNL, the improvements achieved by RPL model was considerably greater 299 because the corresponding likelihood, improved from 994.99 in MNL to 857.49 in 300 RPL. In the same line information criteria AIC improved from 2261.22 in MNL to 301 1742.98 in RPL, AIC3 improved from 2266.22 in MNL to 1742.98 in RPL and R2 was 302 greater in RPL (0.358). Finally, we can notice that G MNL with Cholesky matrix 303 provided a better fit than either the S MNL or RPL model, because by adding two 304 parameters, it achieved a log likelihood improvement of 137 points over S MNL and 305 22.47 points over RPL. To illustrate, the inclusion of scale heterogeneity and variance 306 of residual taste heterogeneity led to an improvement not only in the likelihood over the 307 RPL model (i.e., from 857.49 in RPL to 828.52 in GMN L), but also in the 308 information criteria AIC (i.e. from 1742.98 in RPL to 1682.98 in GMN L) AIC3 ( i.e. 309 from 876.09 in RPL to 849.71 in GMN L and, R2(i.e. from 0.358 in RPL to 0.379 in 310 GMN L). In particular, in the GMNL model, while the parameter τ is statistically 311 significant, implying the existence of such heterogeneity in scale, the estimate γ is not 312 statistically significant, implying that the data are closer to the G MNL II models and 313 that the variance of residuals taste heterogeneity increases with scale (see equation 2). 314 Therefore, we selected the latter G MNL model with Cholesky matrix for further 315 analysis. 316 Firstly, the standard deviations for the random variables are statistically significant 317 indicating that consumers’ preferences for these attributes of the product are 318 heterogeneous. Likewise, all standard deviation parameter estimates of the diagonal 13 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 319 elements in the Cholesky decomposition matrix are statistically significant, except for 320 the VISUAL and LOGO. 321 As expected, the alternative specific constant is negative and significant, indicating 322 that consumers gained a lower utility from the no buy option than from one of the two 323 product alternatives. As expected, the price variable (PRICE) is negative and 324 statistically significant in accordance with economic theory. The logo label (LOGO) 325 variable is positive and statistically significant at the 1% significance level indicating 326 that consumers had higher utility for a box with a logo than a box without a logo. 327 Moreover, the variable corresponding to the nutritional claim related to the content of 328 Omega 3 (CLAIM) in the product is also positive and statistically significant. This 329 result indicates that consumers gained higher utility from a package of 4 sushi pieces 330 with a nutritional claim than without a nutritional claim. In contrast, as we expected, 331 consumers gained lower utility when the visualization of the insect was present 332 (VISUAL). This result confirms our hypothesis that when consumers visualise the 333 insect this could dampen the market acceptance and value creation potentiality of insect 334 based food products. 335 Finally, we also calculated the WTP values. Because the non monetary attributes are 336 effect coded with two levels, the mean WTP values for individual attributes are 337 calculated by taking the ratio of the mean parameter estimated for the non monetary 338 attributes to the mean price parameter multiplied by minus two. Table 4 reports the 339 mean and the t values of WTPs for different insect based products as well as the WTPs 340 among the respondents who declared to have tried them (38.6%) and those who have 341 never tried them before (64.4%). 342 343 >Table 4 WTPs for different insect based products 14 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 344 345 Results generally indicate that consumers were willing to pay a premium price of 346 2.92€ for a box of 4 sushi insect based products when the logo is shown; and they were 347 willing to pay 2.18€ for a box of 4 sushi insect based products when they knew that the 348 product contained omega 3. In contrast, consumers were willing to pay 11.79€ less (thus 349 they were willing to be compensated) for the products with visualization of the insect. 350 Moreover, the findings show that the WTPs for those consumers who had already 351 consumed an insect based product are significantly higher than for those who stated that 352 they have never tried an insect based product. Interestingly, the latter group showed the 353 highest WTPs for nutritional claim attribute. 354 355 356 The ENFR has been set up by public authorities to mitigate the risks for the 357 European society that could surface from unregulated food markets (Borrás, 2006; 358 Rollin 359 enforcing safeguard mechanisms, and reducing information asymmetries among 360 producers and consumers are just a few examples of what the ENFR provided. 361 However, as described in figure 1, when new food products are so radically innovative, 362 such as the case of meat substitutes and insect based food products, a consequent lack 363 of alignment of ENFR with other EU food regulatory and R&D measures can create 364 room for a “regulatory failure”. 2012; Belluco , 2013). As said, setting safety and quality standards, 365 In this study, we have tried to assess whether ENFR is indeed creating such a failure 366 in the specific case of insect based products. Although our study should be still 367 considered as exploratory, we believe that some general discussion points can be drawn 368 from our findings. Specifically, we assessed the effect of ENFR on consumers’ WTP 15 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 369 through the use of a quality label, nutritional claim (presence of Omega 3), and 370 visualization of insects on food products. This last attribute has been identified to have 371 the potential to signal the presence of regulatory failure because it could make 372 consumers demand a compensation (i.e., willingness to accept instead of WTP) to try 373 the product. When this happens, then the product can no longer be considered a market 374 potential, since there is no reason why producers would develop and market the product 375 if they cannot make a profit from selling it. Hence, the regulatory failure component 376 arises because if the ENFR would extend or more quickly define the procedures for 377 using processed insects as food or ingredients, then producers could more easily 378 overcome the problem of visualization in their food designs. This assertion is supported 379 by our findings since visualization of insects in the products can indeed hamper 380 consumers’ WTP for the products. From a business perspective it is interesting to see 381 that we found that insect based products with logos and nutritional claims, but without 382 visualization, have higher WTP values. In this sense they could have potentials for more 383 success in the market. 384 These results seem to imply that while the ENFR is keeping insect based products in 385 a niche for a limited number of consumers, just seeking a “new food experience”, the 386 market potential for insect based products could lead producers to explore a broader set 387 of consumer segments, for example more interested in the functional properties of the 388 products. 389 390 ! 391 This study highlights the effect of ENFR on consumers’ WTP for radical food 392 innovations, such as insect based products in the European Union. Indeed, we found 393 that the sensitiveness of consumers to visualization of insects in the products can 16 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 394 dampen the demand for insect based products, thus indicating the need to re define the 395 standards for EU insect food innovators. For instance, since food products with 396 processed insects, used as ingredients, are not allowed by ENFR, food producers are 397 forced to sell these products with a high level of visualization. As showed by our study, 398 this is undermining the possibility of producers to increase the value added of these 399 products, while facing the risk of consumer’s rejection. This result suggests that the 400 ENFR is indeed creating a “regulatory failure” in the context of food innovations in the 401 European Union. 402 From a business point of view, and in line with prior literature and results from 403 Pascucci and de Magistris (2013) preliminary study, we confirm that opportunities to 404 create profitable business models associated to ground breaking and radical food 405 innovations is heavily dependent on how the different policies and regulations are set 406 and aligned (Belluco et al., 2013; Hobbes 407 design derived by ENFR seem to have a relevant impact on the type of marketing and 408 innovation strategy to implement. , 2014). Particularly constrains on food 409 In conclusion our results imply that if the ENFR will continue to impose the 410 “visualization” requirement, the (financial) effort to support new businesses in this 411 domain may be likely to be unsuccessful. Therefore from a policy making perspective, 412 this study highlights the urgency of having a clear plan of actions from the EU 413 Commission on this issue, an particularly on the misalignment of ENFR principles and 414 R&D innovation measures. For example the principle of “historical evidence” of 415 consumption can be revised, for instance in cases in which a long history of 416 consumption without risks for consumers can be observed in context in which food 417 regulations are as rigorous as in the EU. In this sense it is really questionable whether 418 insect based food should be considered as a novel food and whether they really 17 419 constitute a risk to consumers. To the best of our knowledge there is no scientific 420 evidence highlighting that insects as a whole are safer than processed insects. 421 422 423 Anders, S. and Mőser, A. (2010), “Consumer Choice and Health: The Importance of 424 Health Attributes for Retail Meat Demand in Canada”, Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 425 & ' Vol.58, No. 2, pp. 249–271. 426 Belluco, S., C. Losasso, M. Maggioletti, C. C. Alonzi, M. G. Paoletti and A. Ricci 427 (2013), “Edible Insects in a Food Safety and Nutritional Perspective: A Critical 428 Review”, 429 296 313. 430 Borrás, S. (2006), “Legitimate governance of risk at the EU level? The case of 431 genetically modified organisms”, + 432 73 No. 1, pp. 61 75. 433 Brouwer, F., Fox, G. and Jongeneel, R. (2013), + 434 ( , % ( ) ) * " * " * ( ( Vol. 12, No. 3, pp. " Vol. Cab International, pp. 278. 435 Bunte, F., van Galen, M., de Winter, M., Dobson, P., Bergès Sennou, F., Monier 436 Dilhan, S., Juhász, A., Moro, D., Sckokai, P., Soregaroli, C., van der Meulen, B. and 437 Szajkowska, A. (2011), + 438 & 439 Luxembourg: Publications Office of the European Union, 2011. ISBN 978 92 79 440 19149 7, pp. 201. 441 Carlsson, F., Frykblom, P. and Lagerkvist, C.J. (2005), “Using cheap talk as a test of 442 validity in choice experiments”, & ( ( (( ( & ( ( & ( , Vol. 89, No. 2, pp.147 152. 18 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 443 Champ, P.A., Moore, R. and Bishop, R.C. (2009), “A comparison of approaches to 444 mitigate hypothetical bias”, 445 2, pp.166 180. 446 Cummings, R.G. and Taylor, L.O. (1999), “Unbiased value estimates for environmental 447 goods: a cheap talk design for the contingent valuation method”, + 448 & 449 Dries, L., Pascucci, S., Török, Á. and Tóth, J. (2014), “Keeping your secrets public? 450 Open versus closed innovation processes in the Hungarian wine sector”, 451 * 452 Fiebig, D.G., Keane, M.P., Louviere, J. and Wasi, N. (2010), “The generalized 453 multinomial logit model: Accounting for scale and coefficient heterogeneity”, % -% . % ) Vol. 38, No. ), Vol. 89, No. 3, pp. 649 665. % 454 & " ), Vol. 17, No. 1, pp. 147 162. , Vol. 29, No. 3, pp. 393 421. 455 Frewer, L.J., Bergmann, K., Brennan, M., Lion, R., Meertens, R., Rowe, G., Siegrist, 456 M. and Vereijken, C. (2011), Consumer response to novel agri‐food technologies: 457 Implications for predicting consumer acceptance of emerging food technologies, + 458 * " -+ , Vol. 22 No. 8, pp. 442–456. 459 Fox, G., and Cranfield, J. (2014), “The Past, Present, and Future of Government 460 Regulation of Agriculture and Food: Is There a New Economics of Food?”, 461 ' 462 pp. 435 440. 463 Gao, Z. and Schroeder, T.C. (2009), “Consumer responses to new food quality 464 information: are some consumers more sensitive than others?”, 465 Vol. 40 No. 3, pp. 339 346. 466 Golan, E., Kuchler, F., Mitchell, L., Green, C. and Jessup, A. (2001), “Economics of 467 food labeling”, ' & /% $ 0 , Vol. 62, No. 4, & , , Vol. 24, pp.117 184. 19 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 468 Greene, W.H. and Hensher, D.A. (2010), “Does scale heterogeneity across individuals 469 matter? An empirical assessment of alternative logit models”, working paper ITLS 470 WP 10 15. 471 http://sydney.edu.au/business/__data/assets/pdf_file/0003/75180/itls wp 10 15.pdf 472 Grunert, K.G. and Wills, J.M. (2007), “A review of European research on consumer 473 response to nutrition information on food labels?”, ' 474 pp. 385–399. 475 Hoek, A.C., Elzerman, J.E., Hageman, R., Kok, F.J., Luning, P.A and de Graaf, C. 476 (2013), “Are meat substitutes liked better over time? A repeated in home use test with 477 meat substitutes or meat in meals”, * 478 Hobbs, J. E., Malla, S., and Sogah, E. K. (2014), “Regulatory Frameworks for 479 Functional 480 & 481 Lancaster, K.J. (1966), “A new approach to consumer theory”, + 482 & 483 Layton, D.F. and Brown, G. (2000), “Heterogeneous preferences regarding global 484 climate change”, % 485 List, J.A. (2001), “Do explicit warnings eliminate the hypothetical bias in elicitation 486 procedures? Evidence from field auctions for sportscards”, + 487 % 488 Logue, A.W. (2004), + 489 Routledge. p. 90. 490 Louviere, J., Street, D., Carson, R., Ainslie, A., DeShazo, J.R., Cameron, T. and 491 Marley, T. (2002), “Dissecting the random component of utility”, 492 Vol. 13, No. 3, pp. 177 193. Available Food and 2 at: $ $ , Vol. 15, , Vol. 28, pp. 253–263. Supplements”, /% 1 ' 0 , Vol. 62, No. 4, pp. 569 594. ' $ , Vol. 74, No. 2, pp. 132 157. ) & " , Vol. 82, No. 4, pp. 616 624. & ), Vol. 91, No. 5, pp. 1498 1507. $ & 3 . , New York: Brunner . , 20 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 493 Lusk, J.L. (2003), “Effects of cheap talk on consumer willingness to pay for golden 494 rice”, 495 McFadden, D. (1974), “The measurement of urban travel demand”, ' 496 & 497 Murphy, J.J., Stevens, T. and Weatherhead, D. (2005), “Is cheap talk effective at 498 eliminating hypothetical bias in a provision point mechanism?”, & 499 % 500 Novel Foods Regulation (EC) 258/97. European Commission. Health and Consumers 501 General Directorate (DG). 502 Oonincx, D.G. and de Boer, I.J. (2012), “Environmental impact of the production of 503 mealworms as a protein source for humans–a life cycle assessment”, $ " 504 No. 12, pp. e51145. 505 Oonincx, D.G., van Itterbeeck, J., Heetkamp, M.J., van den Brand, H., van Loon, J.J. 506 and van Huis, A. (2010), “An exploration on greenhouse gas and ammonia production 507 by insect species suitable for animal or human consumption”, $ 508 12, pp. e14445. 509 Pascucci, S. and de Magistris, T. (2013), “Information bias condemning radical food 510 innovators? The case of insect based products in the Netherlands”, ' 513 , Vol. 85 No. 4, pp. 840 856. $ Vol. 3, No. 4, pp. 303 328. & , Vol. 30, No. 3, pp. 327 343. 511 512 & % " , Vol. 5. No. * ), Vol. 16, No. 3, pp. 1 16. Ramos‐Elorduy, J. (1997), “Insects: A sustainable source of food?”, & 4 , Vol. 7, * , Vol. 36, No. 2 4, pp. 247 276. 514 Revelt, D. and Train, K. (1998), “Mixed logit with repeated choices: households' 515 choices of appliance efficiency level”, % 516 4, pp. 647 657. ) & " , Vol. 80, No. 21 517 Rollin, F., Kennedy, J. and Wills, J. (2011), “Consumers and new food technologies”, 518 + 519 Scarpa, R. and Del Giudice, T. (2004), “Market segmentation via mixed logit: extra 520 virgin olive oil in urban Italy”, ' * Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 521 " -+ , Vol. 22, pp. 99 111. * , Vol. 2, No. 7. 522 Silva, A., Nayga, Jr.R.M., Campbell, B.L. and Park, J.L. (2011), “Revisiting Cheap 523 Talk with New Evidence from a Field Experiment”, ' 524 % 525 Street D., L. Burgess and J.J. Louviere. 2005. Quick and easy sets: Constructing optimal 526 and nearly optimal stated choice experiment 527 & . , Vol., 36, No. 2, pp. 280. ' % 22: 459 470. 528 Street, D. and Burgess, L. (2007), + ( " 529 &5( 530 Sylvester, D.J., Abbott, K.W. and Marchant, G.E. (2009), “Not again! Public 531 perception, regulation, and nanotechnology” % 532 pp.165 185. 533 Tan, H. S. G., Fischer, A. R., Tinchan, P., Stieger, M., Steenbekkers, L. P. A., and van 534 Trijp, H.C. (2015), “Insects as food: Exploring cultural exposure and individual 535 experience as determinants of acceptance”, * 536 doi:10.1016/j.foodqual.2015.01.013 537 Veldkamp T., van Duinkerken, G., van Huis, A., Lakemond, C.M.M., Ottevanger, E. 538 Bosch, G. and van Boekel, M.A.J.S. (2012), “ 539 ( 540 Research Report 638. ISSN 1570 – 8616. New Jersey: John Wiley & Sons Inc. ( 6 - , Vol. 3, No. 2, 2 $ , ”, Wageningen, Wageningen UR Livestock 22 541 Verkerk M.C., Tramper, J., van Trijp, J.C.M. and Martens, D.E. (2007), “Insect cells for 542 human food”, 7 543 Wijnands, J.H.M., van der Meulen, B.M.J. and Poppe, K.J. (2007), “ Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 544 & ( , Vol. 25, pp. 198–202. * ( ”, Report European 545 Commission – Enterprise and Industry Reference no. ENTR/05/75. pp. 328. 546 Wolf, C. (1987), “Market and non market failures: comparison and assessment”, 547 ' 548 Yen, A.L. (2009) “Edible insects: Traditional knowledge or western phobia?”, 549 & $ $ % , Vol. 7, No. 01, pp. 43 70. , Vol. 39, pp. 289–298. 550 551 " 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 Tiziana de Magistris is Researcher at Centro de Investigación y Tecnología Agroalimentaria de Aragón, Unidad de Economía Agraria, Zaragoza, Spain. Her research interests are related to consumer behavior, experimental and behavioral economics, food and agribusiness. She has published in several international peer review journals and book chapters. She has been involved in national as well as international research projects. Stefano Pascucci is Assistant Professor in Economics and Organization of Agribusiness at Management Studies Group, Wageningen University, Wageningen, The Netherlands. His research interests are related to the economics and organization of agribusiness, with a specific focus on innovation and sustainability. He has published in several international peer reviewed journals and book chapters, and participated in several national and international funded projects on innovation in the agri food sector, organization of short supply chains and sustainable development. Dimitrios Mitsopoulos has a Master of Science in Organic Agriculture. He is a consultant and freelance specialized in sustainable innovation, management and marketing in the food and agribusiness sector in the Netherlands and European Union. 571 572 573 574 23 575 # $% Definition and Means of Demographic Variables (%) Variable definition Name (type) Number of participants Gender: Male FEMALE (dummy Female 1=female; 0 otherwise) Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) Age (years) Education of respondent: High School 155 49.9 50.1 Between 18 35 years 47.71 Between 35 54 years 30.07 Between 55 64 years 15.03 More than 64 years 7.19 HIGHSCHOOL (dummy 1=high school; 58.17 0 otherwise) Income High income HINCOME (dummy variable =1 if more 28.8 than 2,500 €; 0 otherwise) Have you ever tried some insect based product? Tried (dummy variable =1 if “yes”, =0 38.56 otherwise) 576 24 577 578 # &% Attributes and Levels Used in the Choice Design. (Logo and visualization images created by author) Attributes Levels Acronym PRICE 1.50, 2.50, 3.50 and 4.50 (€ per package) ($% &) Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) No logo LOGO Logo ( ) NUTRITIONAL CLAIM No claim ( ) (! "# ) Omega 3 VISUALIZATION OF INSECT No visualization Visualization (Source: Images created by author) 25 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) '% Model estimates and mean WTPs Model 1 MNL ! $ 86 9 0.197 (5.06) * ** 0.290 (6.95) * ** ! "# 0.652 ( 14.77) * ** $% & 0.447 ( 9.32) * ** 4 67#: 1.721 ( 10.84) * ** " ( # ! "# 3 . ! "# 7 ) Model 2 Model 3 S MNL RPL $ 86 9 $ 86 0.653 (4.05) * ** 0.576 (6.73) * ** 1.928 (14.31) * ** 0.783(6.52) * ** 4.92 ( 19.97) * ** 2.64( 7.87) * ** * ** 3.21 ( 12.83) 0.988( 10.84) * ** * ** 11.17 ( 11.38) 2.96( 11.83) * ** 579 Model 4 GML 9 $ 86 9 1.129(4.47) * ** 1.507(4.25) * ** 6.052( 4.82) * ** 1.031( 15.55) * ** 2.990( 14.51) * ** 0.563(6.04) * ** 1.021(8.14) * ** 3.025(8.14) * ** 0.832(2.93) * ** 2.201(4.61) * ** 7.375(4.46) * ** 0.563(6.04) * ** 0.144(1.13) 0.390(2.09) * * 0.832(2.93) * ** 1.304(2.94) * ** 5.53(4.74) * ** 1.011( 7.71) * ** 0.583(3.08) * ** 2.943( 7.70) * ** 1.772( 3.72) * ** 0.046( 0.14) 4.878(4.04) * ** 1.150(6.56) * ** 0.008(0.02) 3672 1682.98 849.71 828.52 24% 5 5 , ! "# , ! "# , Τ Γ N AIC AIC3 Log likelihood % of no buy option 1.85 (25.48) * ** 3672 2261.22 2266.22 1125.61 24% 3672 2239.3 1854 994.99 24% 3672 1742.98 876.09 857.49 24% 26 580 (% WTPs for different attributes of insect based products Overall Never tried Tried 2.92 € (4.13) *** 2.34 € (1.76)* 3.02 € (3.93) *** 5.01 € (1.55) 4.11 € (3.25) *** 2.18 € (4.36) *** *** ! "# 11.72 € ( 4.63) 18.74 € ( 1.47) 15.26 € ( 2.40) *** # 581 582 583 584 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) # )% Percentage of positive WTPs for different insect based products across socio demographic characteristics * 1 ) Logo Claim Visual 2.65€ 3.29€ 1.92€ 2.59€ 6.13€ 5.86€ 3.18€ 4.39€ 2.09€ 3.13€ 11.3€ 2.64€ 4.97€ 13.7€ 10.22 € 6.44€ 19.02€ 11.59€ ,-$%),. 1.50€ 2.50. 2.50€ 4.50€ >4.50€ )(* 5% ',* 1% )/* 0% &(* 10% )$* 5% 3% &2* )(* 7% &1* 5% (/* 7% 7% &1* )0* 8% 8% &$* ,-$%), ,-$%),. 1.50€ 2.50. 2.50€ 4.50€ ! ,-$%),. 1.50€ 2.50. 2.50€ 4.50€ >4.50€ (1* 3% &0* 7% ((* 5% ,* '$* (&* 5% 1% &2* '/* 5% &)* $1* (2* 0% 7% '$* '1* 6% 7% '(* /* 1% 1% 4% $)* 1% 1% 1% $)* 0% 0% 1% )* 1* 1% 4% 2* 0% 1% 1% )* 0% 0% 7% * +# 585 586 587 588 589 590 591 27 Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) 592 3 $. Regulatory frameworks in the EU radical innovation landscape 593 28 594 " 595 Scheme of the different treatments used in the field survey. $ Downloaded by Wageningen UR Library At 06:14 01 May 2015 (PT) Info treatment (1) Alternative meat 596 597 598 599 600 601 Info treatment (2) Alternative and environmental Cheap talk Version A and E No No No (baseline)* Version B No No Yes Version C Yes No Yes Version D No Yes Yes * version E introduced a social treatment in the design. They both work as baseline compared to cheap talk and the other information treatments, " & # !# 602 Studies show that people tend to act differently when they face hypothetical decisions. 603 In other words, they say one thing and do something different. For example, some 604 people state a price they would pay for an item, but they will not pay the price for the 605 item even when they see this product in a grocery store. 606 There can be several reasons for this different behavior. It might be that it is too difficult 607 to measure the impact of a purchase in the household budget. Another possibility is that 608 it might be difficult to visualize themselves getting the product from a grocery store 609 shelf and paying for it. Do you understand what I am talking about? 610 We want you to behave in the same way that you would if you really had to pay for the 611 product and take it home. Please take into account how much you really want the 612 product, as opposed to other alternatives that you like or any other constraints that might 613 make you change your behavior, such as taste or your grocery budget. Please try to 614 really put yourself in a realistic situation 615 29