Размещено на http://www.allbest.ru/

Пермский филиал федерального государственного автономного образовательного учреждения высшего образования

«Национальный исследовательский университет

«Высшая школа экономики»

Факультет экономики, менеджмента и бизнес-информатики

Выпускная квалификационная работа - Бакалаврская работа

по направлению подготовки 38.03.01 Экономика

образовательная программа «Экономика»

«Потребительское отношение к генетически модифицированным пищевым продуктам в России»

(Consumer Attitudes to Genetically Modified Food in Russia)

Рецензент

Д. К. Коутс

Руководитель

М. М. Ёрмирзоев

Пермь 2016 

Abstract

This study explores the Russian market of genetically modified products, namely bread with GM components. It is widely and rapidly developing area of investigations, so these results shed a light on consumer preferences, the value of willingness to buy and the impact of factors, which include attitude and individual characteristics in Russia. The double-bounded dichotomous choice methodology and contingent valuation were used with two model specifications. The obtained results evidence a generally negative attitude toward GMOs and the zero meaning of WTA. There are influence of risk perceptions, attitude regarding use of biotechnologies in food industry, self-reported knowledge and importance of labeling. Unexpectedly, all socio-demographic variables are insignificant. In sum, there are unfavorable conditions of developing GMO production in Russia.

Данная работа исследует Российский рынок генетически модифицированных пищевых продуктов, а именно хлеб с добавлением ГМО. Эта область исследований является широко и быстро развивающейся, так полученные результаты прольют свет на потребительские предпочтения, на значение готовности покупать и на влияние различных факторов, включающих мнение и персональные характеристики индивидов в России. В исследовании применяется методология субъективной оценки и дважды взвешенный дихотомичный выбор, построены две спецификации модели. Полученные результаты свидетельствуют о негативном отношении потребителей к ГМО и нулевом значении показателя WTA. Влияющими факторами стали ожидаемый риск, отношение к использованию биотехнологий в производстве пищевых продуктов, уровень знаний индивида и значимость маркировки. Неожиданным итогом стала незначимость всех социально-демографических переменных. В итоге условия для развития производства ГМО в России являются неблагоприятными. genetic modified food

Introduction

Nowadays, biotechnologies and genetically modified organisms are essential for development of food production. Genetically modified foods are foods received from organisms whose DNA has been modified in an unnatural way, so through the introduction of a gene from a different organism. There are many different advantages and disadvantages of GM foods, and this is a debatable aspect of this technology to what extent they can be useful or harmful for the humanity and the environment (Pinstrup -Andersen and Schiшler, 2001).

Using GM foods can potentially allow producing larger quantity of crop yields, which could solute the problem of starvation in developing countries. They are also considered as more profitable, despite the initial higher costs. Using of biotechnologies reduces the quantity of pesticides and herbicides as well as the number of human resources needed to prosperous growing the crops and high income. The main disadvantage of using GMO`s is that there are continuing debates around possible influence on health. Despite the variety of researches on this subject, it still remains uncertain question whether genetically modified food is safe for people. At the present times, we can observe the diversity of policies regarding producing, distribution and labeling of products, containing genetically modified components. The European Union uses the Precautionary principle exacting a pre-market permission for any GMO to enter the market, moreover a post-market environmental monitoring takes place. The European Food Safety Authority (EFSA) and the member states author a risk assessment.

This assessment must indicate that the food is safe for human and animal health and the environment under its intended circumstances of use (European Commission, 2015). Also The European Union sets out a framework for ensuring the traceability and labeling of GMOs and products produced from these organisms throughout the food chain. Traceability of GMOs allows the monitoring and controlling of information given on labels, the checking of impacts on the environment and the withdrawal of products from the market in happenings where new science data prove the using of GMOs present an environmental or health risk (European Parliament, 2003).There is no any federal legislation in the United States that is specific to genetically modified organisms. As conventional products, GMOs are regulated pursuant to health, safety, and environmental legislation.  The US treatment to regulating GMOs is premised on the hypothesis that regulation should focus on the nature of the products, rather than the process in which they were. Compared to other countries, regulation of GMOs in the US is relatively advantageous to their extension.  GMOs are an economically important constituent element of the biotechnology industry, which now plays a substantial role in the US economy. As we can see, the US is the world's key producer of genetically modified crops.  In 2012, the United States accounted for 69.5 million, over 40% of the total - 170.3 million hectares of biotech crops globally (Guide to U.S. Regulation of Genetically Modified Food and Agricultural Biotechnology Products, 2001).

The Russian policy regarding distribution of food containing genetically modified products is characterized as stricter. Despite the fact, that the Russian government authorized the import of products containing GMOs in 1996 (The Federal Law of Russian Federation, 1996), later it had been chosen the policy of GMOs restriction, taking the example of the European Union. In September of 2013 was officially published the decision of state registration of GMOs, which implies the control of using genetically modified organisms on Russian territory (Russian Federation Government Resolution, 2013).

Nowadays, it can be seen the continuing negative position toward GMO and biotechnologies in Russia. In April of 2015 deputes of Russian Federation unanimously adopted a bill on a total ban on the use of GMOs in agriculture and animal husbandry on first reading. The main quotations are following: «Do not use for seeding (planting) plant seeds produced using genetic engineering methods, including the inability to reproduce or transfer of hereditary genetic material, with the exception of their seeding (planting) with the expertise and research work» and «Prohibited the cultivation and breeding of plants and animals whose genetic program has been altered using genetic engineering techniques, with the exception of cultivation and breeding during the expertise and research work.» What regards to imported products containing GMOs, nothing about it say with confidence: «The Government of the Russian Federation have the right to prohibit the import into the Russian Federation of genetically modified organisms intended for release into the environment, and (or) products derived from the use of such organisms or containing such organisms.» This bill should be reviewed on the second reading by the Council of the Federation in March of 2016. Moreover, in the September of 2015 the State Duma introduced a bill banning palm oil, genetically modified organisms and a number of food additives. The reason for this proposal is high mortality rates from cancer and cardiovascular diseases, poor quality of food products consumed in Russia.

Public attitude toward GMOs in Russia is also leaning toward disapproval. Analytical Center "Levada-center" conducted a survey on a representative nationwide sample of urban and rural population of 1,600 people aged 18 years and older in 130 settlements of 45 regions of the country (Analytical Center of Yuri Levada, 2011). The results are: since 2003 to 2011 the percentage of knowledgeable people has increased from 30% to 75%; since 2003 to 2011 the percentage of generally positive respondent`s view about GMO distribution has decreased from 12% to 4%, and conversely the percentage of generally negative respondent`s view has increased from 41% to 81%. Thereby, we can see growing awareness of Russians about GMOs and negative dynamic of public attitude toward distribution of food with genetically modified internals (see Tab. 1, Tab. 2).

Table 1. The Distribution of Answers from «Levada-center»Statistical Survey* (1)

*(In a percentage of the total number of the respondents) 

Table 2. The Distribution of Answers from «Levada-center»Statistical Survey* (2)

*(In percentage of the total number of the respondents)

The introduction of genetically modified crops into world markets has created a new niche and possible consumers. The business leading companies, producers and scientists involved in biotech manufacturing are interested in better understanding of consumer attitudes and behavior toward GM products. Moreover, there is a significant feature of GMO food in contradistinction to other food products: the lack of public awareness about influence on health and possible consequences of GMO`s. A similar situation on Russian market of GM products leads us to investigate consumer attitudes to genetically modified organisms in food. Based on people preferences concerning GMO we could recommend the winning strategy for the producers of selling modified products in Russia.

The results of this research may be useful and practically valuable for both consumers and producers. First of all they provide producers by winning strategy of introduction of genetically modified products on Russian market. The preferable amount of discount and personal characteristics of target audience are the most useful information for the producers. As a result, the demand of consumers, who are willing to purchase GMO`s products according to their attitudes and preferences, will be satisfied. Moreover, entering of product made with genetically modified components on the lower price comparing to similar products, will make other producers to reduce their prices because of competition.

Namely, the goal of the study is to estimate willingness to accept GM-food with discounts compared to non-GM food on consumer market in Russia. Willingness to accept (WTA) characterizes people`s ability to purchase particular good, putting up with something negative. There are some reasons: using GM product reduces production costs, hence price of the product should be lower; genetically modifications of the food are perceived negatively by the consumers in contrast with natural products. To assess the value of WTA the contingent valuation method and dichotomous choice are used, it assumes choice of particular discount on GM bread or its absence in a hypothetical market.

In order to reach main goal of the study we need the following: to examine and analyze theoretical and practical researches; to suggest a number of testable hypotheses about the possible determinants of consumer behavior on hypothetical market; to collect and analyze the data required for examining hypotheses; to build an econometric model describing the consumer choice; to interpret the results. To get the data 300 Perm citizens will be chosen as respondents for the survey, and will be asked about hypothetical consumption of GM bread. Survey data will be collected through in-person interviews near grocery stores.

Theoretical background

This section is devoted to description of the theoretical framework of the research. Also, we will consider the main results of the similar studies. In recent papers have been made numerous attempts to understand and investigate public attitudes toward new food containing GMO`s in different countries all over the world. Willingness to pay describes consumer`s behavior on hypothetical market and represents people`s ability and readiness to buy some product at premium price, because of its usefulness or other features. Mathematically, WTP is defined as the particular sum of individual`s income which should be taken to keep his utility as constant amount. Also there is willingness to accept, which is about consumer choice of product with negative or uncertain impact. WTA is defined as the amount of money that individual needs to keep his utility constant. The contingent valuation with dichotomous choice questions is the most appropriate and well-known method to asses these values. There are some differences among researchers toward applicable econometric estimation and modeling (Economic and Social Development Department).

One of the first studies was devoted to of UK consumer attitudes to genetically modified organisms in food, representing by willingness to pay to avoid these products. Michael Burton and Dan Rigby (2001) used choice modeling approach to estimate consumer preferences. Significant differences are found between attitudes to GM food in which plants are modified by the introduction of genes from other plants and those in which plants are modified by the introduction of genes from animals and plants.

It is exceedingly important to emphasize the contribution of Michael Hanemann`s, John Loomis`s and Barbara Kanninen`s (1991) approach proved statistical efficiency of double-bounded dichotomous choice contingent valuation. The authors noted that this methodology successfully value nonmarketed products. Californians data regarding their willingness to pay for wetlands were used to compare single- and double-bounded dichotomous choice models and their outcomes. The main idea is that each respondent of a survey is presented with two bids.

The amount of the second bid is contingent regarding the response to the first bid. If the individual answer «yes» to the first bid, the following bid is some amount bigger than the first bid; if the individual responds "no" to the first bid, the second bid is some amount less than the first bid. This approach allows determining more precisely interval of true willingness to pay. The basic conclusion of the paperwork is that adding a follow-up bid to a conventional, dichotomous choice CV survey essentially improved the statistical information provided by the data.

In 2002 Quan Li, Kynda R. Curtis, Jill J. McCluskey, and Thomas I. Wahl focused on the research of GM rise market and GM soybean oil in China. This investigation used data collected from interviewing of 599 consumers. As in previous research, authors used the standard double-bounded logit model (Hanemann et al, 1991). Their hypothesis supported by the main ideas, that consumer choice depends on safety of products, such as the use of genetically modified components. In this study the contingent valuation methodology was implied to elicit consumer`s willingness to pay. It was concluded that Chinese consumers have generally positive attitudes toward genetically modified (GM) foods, especially for GM foods with product-enhancing attributes, despite the fact that the majority of surveyed individuals reported that they had little or no knowledge of biotechnology. The willingness to pay (WTP) of Chinese for GM rice and GM soybean oil was positively influenced by respondents` positive opinion toward GM foods for both products and by higher levels of self-reported knowledge for soybean oil. Nevertheless, toward GM rice, WTP was negatively influenced by the individual`s age - the older consumers were less likely to choose GM rice. These unexpected results may have place due to the fact, that China is the fourth largest producer of biotech crops in the world and has been a major supporter of biotech research since the 1980s.

Directly related to my research are the investigations from the similar study conducted in Japan. The authors Kristine M. Grimsrud and Jill J. McCluskey (2003) have interviewed 400 citizens of Matsumoto City in June of 2001 with regards to consumption of GM products. Contingent valuation methodology with double-bounded dichotomous questions was used and logit model was applied to estimate WTP. This case showed the lack of gender in determining consumer response to genetically modified food. Simultaneously, such factors as self-reported risk perceptions, concern about food safety, self-reported knowledge about biotechnology, education and income are in positive relation with the amount of discount. Interestingly, about eighty percent of consumers would not choose GM noodles even with a discount set at one of following levels: 5%, 10%, 25%, 40% and 50%. The main findings of this work indicate Japanese consumers` attitude opposes successful for entering GM products to the Japanese market.

Kristine M. Grimsrud and Jill J. McCluskey (2004) investigated factors that stimulated Norwegian consumers to buy GM-products and estimated their willingness to purchase GM-food, namely bread. The authors used the contingent valuation method through dichotomous choice. Actually, the received econometric model represented a special case of the double-bounded logit model. The collected data contains 400 answers of in-person interviews. According to K. Grimsrud and J. McCluskey, it was identified that willingness to buy GM-food was higher for people, who are favorable about use of biotechnology in food. Also females and older consumers are more skeptical about GMO that men and youth. On average, Norwegian consumers are willing to purchase GM bread with 49, 9% discount, which is close to the maximum offered discount. Nevertheless, the Norwegian market may be more open to GM products in future, because of acceptance of youth.

Based on the use of contingent valuation methodology, Ricardo I. Diaz Carcamo investigated the consumer`s acceptance of genetically modified food products in Chile. The data used in this work was collected through in-person survey conducted in Santiago, in June of 2005, including 600 interviews. The explored product was second-generation genetically modified apples. As in similar studies, author applied double-bounded logit model, including dichotomous choice questions. The results of this paper indicate customers of Chile were engaged in food safety, had knowledge about biotechnologies, and felt positive about using of biotechnology in food production.

Nevertheless, the majority of them associated a level of risk to consuming biotech products. Family size and strict attitudes about food safety had significantly negative influence on WTP for GM apples. Being employed full-day had a positive effect on WTP. Finally, the amount of the bid offered to the respondent had a positive effect on WTP, which was unexpectedly to author. Most respondents would not choose GM apples even with a high discount. So, there is likely a consumer segment that views GM food as risky and are not responsive to changing in price. Chilean consumers need to be enlightened toward the safety of GM foods and told about the benefits associated with using of biotechnologies in production.

The double-bounded dichotomous choice question approach has gained few followers so far. For instance, Simon Chege Kimenju and Hugo De Groote revealed attitudes and willingness to pay for GM maize meal in Nairobi, Kenya. Researchers used double-bounded logit model to estimate WTP. The main results from 604 respondents in total are: 38 percent of respondents are aware of GM crops, most consumers believe in the positive impact of biotechnologies, but are interested in environmental and health risks. The major part of individuals (68%) would buy GM maize meal at the same price as their favorite maize meal brand. The mean WTP for GM maize meal is 13.7% premium over average current maize meal prices. Considering socio-economic factors, only income and education of individual significantly affect WTP. Subjective elements are seemed to be the main determinants of WTP. Health risk perception and ethical and equity concerns have negative influence on WTP, while trust in government to ensure food quality positively influence on consumers WTP.

The similar study was conducted in the USA by Naoya Kaneko and Wen S. Chern. Their paper published results from a national telephone survey on GMO. The investigated products were vegetable oil, cornflakes and salmon. Authors used contingent valuation methodology in which participants of a survey choose between GM and non-GM products with an option of indifference. The binomial and multinomial logit models were involved to estimation of willingness to pay to avoid the GM alternatives. The mean WTP to avoid were 20, 9% of the base price for GM vegetable oil, 14, 8% for GM cornflakes, 28, 4% for GM fed-salmon and 29, 7% for GM salmon. The inclusion of indifference option has increased the sample size. Moreover, it was found that people were less likely to accept GM products if they rated it as risky to human health. Once more conclusion represents the insignificance of consumer`s demographic and social characteristics.

Food markets in developing countries are characterized as experiencing hence it is attractive to investigate. Later study by Alok Anand, Ron C. Mittelhammer, and Jill J. McCluskey (2007) investigates the Indian market for GM wheat. The data collection in this study was through in-person interviewing of 600 consumers in New Delhi and Patna, India, in June of 2004. Authors utilized dichotomous-choice contingent valuation methodology to estimate willingness to pay and determine factors that affect consumer decision. Moreover they investigated the extent to which the provision of information about GM wheat has any significant effect on consumers' willingness to pay and the degree to which the effect varies based on the information treatment. Following the results of this study, Indian respondents were willing to pay a small premium for GM wheat when no information was provided to them. Also, in situation, when positive “producer friendly” information was provided to the respondents, their willingness to pay for the GM wheat increased, but only by a small amount. The results were dramatically different when negative information about potential health effects was provided. There was a strong negative consumer reaction and average respondent was not willing to buy the GM wheat at any price. The generally positive response in the case of India supports the findings of other studies in countries with emerging economies. If GM crop production is to expand in the future, then it is natural to look for alternative markets such as India with its massive population and fast growing economy.

Alejandro Lopez-Feldman in his paperwork «Introduction to contingent valuation using Stata» (2012) considers econometric models, applied to contingent valuation with dichotomous choice. The way to proceed with the estimation of double-bounded model, used in this paper, is to construct a likelihood function to directly obtain estimates for в and д, using maximum likelihood estimation. The considerable contribution of Alejandro Lopez-Feldman is the creation of specialized command «doubleb» in Stata program, allowing direct estimation of в and д using maximum likelihood.

Besides, there are different approaches of estimation willingness to pay or willingness to accept. For example, Timothy C. Haab and Kenneth E. McConnell in their book «Valuing environmental and natural resources, the econometric of non-market valuation» (2002) investigate models used in carrying out studies with contingent valuation and follow-up questions (double-bounded dichotomous choice) approach. Authors demonstrated that formulation of econometric model for double-bounded data is referred to as the bivariate discrete choice model. It said that the bivariate probit model is a common parametric model of two-response surveys.

Some investigators applied such model specification to get results from contingent valuation double-bounded dichotomous choice approach. So, Ayooth Yooyen and Nisachon Leerattanakorn studied consumer preferences and willingness to pay for organic pork in Thailand. Authors used a sample of 400 Thais consumers to determine the willingness of individuals to pay a premium price for this product. In order to receive the mean value of WTP and to obtain information about the key factors that influence the WTP a bivariate probit model was applied. The main results indicated that factors which better affect WTP are lifestyle variables and knowledge about organic foods rather than the common socioeconomic factors. The mean premium price or WTP for organic pork is approximately 34.30 Bath per kg. Concerning the market potential study showed that the characteristic of organic pork which corresponds with consumer preferences are composed of modernized and environmental packaging with special product details.

There is once more study based on using of bivariate probit model to estimation of willingness to pay. Alem Mezgebo, Worku Tessema and Zebene Asfaw (2013) have investigated the economic cost of irrigation water in Wondo Genet zone by detecting households' WTP using contingent valuation method in the form of double bounded questions. There was a sample of 154 responses collected through in-person interviewing. The important conclusion is that bivariate probit and probit models were successfully applied to measure the mean WTP and factors affecting this value for irrigation water, respectively.

In spite of this fact that it is rapidly developing area of the study, there are many gaps in practice and many countries are not investigated in regards to consumer acceptance of GMO`s. The following paper will shed a light on consumer preferences of Russians, examining WTP, factors of acceptance and consumer behavior toward GMO`s.

Research question

The main research question of this study is to estimate consumers' willingness to accept genetically modified (GM) food products in Russia. This value could be received through conducting a survey including preference questions related with price on particular GM product. Also, consumer attitude toward genetically modified products seems to be substantially affected by individual`s awareness and propensity. The next factor that is supposed to be significant is suggested discount. Hence, we need to examine which factors and in what extent influence consumer preferences of GMO.

The choice of investigated genetically modified food product is the one of the primary task. Leaning on previous studies this product should be available, often consumed and has a possibility to be genetically modified.

The most appropriate method of data collection is a contingent valuation because there is an assumption of hypothetically of investigated GM product, which is not available within the real market in Perm. Moreover the double-bounded dichotomous choice questions format or «interval-data model» was selected for more accurate determination of willingness to accept (Hanemann et al. 1991, Arrow et al. 1993). One of the most important conditions of data collection is conducting a survey through in-person interviews near grocery stores (Grimsrud et al, 2003).

The next task is to compose questionnaires which involve possible determinants of consumer attitude and dichotomous choice questions. These determinants should be ranged by personal characteristics of the respondents, individual preferences and life conditions. Based on previous investigations we need to select possible factors influencing Russian`s willingness to accept and to include appropriate questions in survey structure.

Availability of data set allows to construct econometrical models and to estimate the amount of willingness to pay and the influence of specified variables. The precise analysis of related literature and papers revealed the most applicable model specifications appropriate for contingent valuation interval-data methodology. Referring the investigations by Quan Li, Kynda R. Curtis, Jill J. McCluskey, and Thomas I. Wahl (2002), Kristine M. Grimsrud and Jill J. McCluskey (2003), Alok Anand, Ron C. Mittelhammer, and Jill J. McCluskey (2007) and the core work written by Michael Hanemann`s, John Loomis`s and Barbara Kanninen`s (1991) the double-bounded logit model and likelihood function maximization are chosen to achieve the purpose of the following research. Moreover it is helpful to pay attention to such specification as bivariate probit model which appropriateness was proved by Timothy C. Haab and Kenneth E. McConnell (2002) and Ayooth Yooyen and Nisachon Leerattanakorn (2012). Both these specifications have common features and fit the assessing of consumer preferences in consistent choice conditions, where the second response depends from the answer on the initial question.

The practice of previous researchers allows to suggest the following dependence between factors and to construct hypothesis. It is expected that the higher level of consumer knowledge regarding GMO and more preferable position toward use of biotechnologies have positive impact on choice decision as in results from Chinese, Norway and Japanese investigations. Also, a strict attitude toward possible risk will lead to higher meaning of necessary discount. The higher meaning of individual`s family income has negative influence on purchasing decision toward GMO`s. In turn, lower income and uncertain risk perceptions may lead people to purchase genetically modified products with particular discount which correspond with results from paper written by Kristine M. Grimsrud and Jill J. McCluskey (2003). The next suggestion is that women and elders tend to be more skeptical about GMO`s as consumers from China and Norway. Based on results from Kristine M. Grimsrud and Jill J. McCluskey (2003) and Ricardo I. Diaz Carcamo (2005) the respondents who prefer safety of food products to lower price need higher discount or choose not to purchase genetically modified food.

With regards to description of the situation on Russian genetically modified products market, we can assume that the majority of respondents will have negative expectations toward possible health risks and will be against of using biotechnologies in food productions. These could be explained by Russian Government policy of GMO restriction, which affect people opinions and perceptions. Such interrelation between government policy and people attitude takes place in investigations conducted in the USA, Japan and countries-members of the European Union.

Methodology and procedure

This part of the study explains the methods using within the study. To assess the value of WTA in a hypothetical market the contingent valuation method and double-bounded dichotomous choice are used generally. This methodology is based on a dichotomous choice approach in which individual WTA can be estimated using responses to market-type questions (Kanninen, 1993). Because of absence of available data that can be used in the following research, information is obtained from a consumer survey, through in-person interviews, including dichotomous choice questions. Location for the interviewing will be near grocery stores to collect opinions of direct consumers. It will be the first part of the research.

Following Kristine M. Grimsrud and Jill J. McCluskey (2003), ordinary consumers are asked about their willingness to accept bread made from genetically modified wheat. This product is the most appropriate for analysis since bread is commonly and daily used product among all segments of the population in Russia. Questionnaires for research are consisting of three parts: the first part includes questions related with consumer attitude toward genetically modifications and biotechnology using; the second is about dichotomous-choice questions, that is necessary for willingness to pay estimation; and the third contains personal characteristics of the respondent. As we can see the first and the third part of the survey provide determinants of consumer behavior, while the second part represents consumer choice.

More detailed consideration of factors influencing choosing of GM-food, questionnaires for research should comprise the following: questions about attitudes concerning the environment and economic growth, safety preferences, knowledge about GMO and biotechnologies, possible risks related with consuming GM-food, GM labeling, and also preferences of domestic over against imported food products. Also, personal information of the respondents should comprise: age and male of the respondent, the level of education, household size - the number of family members, children, status concerning shopping, employment status and household income. A similar format was used by many researchers such as: Quan Li, Kynda R. Curtis, Jill J. McCluskey, and Thomas I. Wahl (2002), Kristine M. Grimsrud and Jill J. McCluskey (2003), Ricardo I. Diaz Carcamo(2005), Simon Chege Kimenju and Hugo De Groote (2005) and Alok Anand, Ron C. Mittelhammer, and Jill J. McCluskey (2007). The appropriateness of previously mentioned attitude and social variables was proved by their papers and precise results that fully describe consumer behavior.

It is assumed that there are some insignificant difficulties in the following study. The first problem is the possible biasness of a sample by some parameters for example: age, male and income. To avoid this and to make a sample more homogenous it is necessary to interview respondents of different age and male. Also it is helpful to conduct a survey in grocery stores belonging to different chains of stores with the aim of income differentiating. The second problem that we need to avoid is false personal information, hence it is important to inform respondents about an anonymous profile. There are some limitations of the study relating with the interviewing. Respondent`s age should be equal to18 or higher, so only adults can take part in a survey. Also, it is preferable to interview every third consumer and to avoid respondents from the same household.

The survey is conducted during a several months in Perm, the capital city of the subject of Russian Federation - Perm Krai. This city is the one of the most populated in Ural region and represents the center of economic activity. The survey data is collected through in-person interviews of 300 consumers. We had chosen three stores for interviewing, on the assumption of differentiation of chain stores. These stores belong to Semia, Vivat and Magnit companies, aiming on diverse target audience to a certain degree. All chosen stores are placed near the city center. Data collection from consumers occurs at the same time: about from 5 to 7 pm. Respondents are selected randomly with the criterion that the interviewer was to ask every third customer who came in the survey area. There are no gifts or bonuses.

The double-bounded dichotomous choice assumes that every respondent is asked firstly if he is willing to pay the same price for the GM-bread as the non-GM bread. If the answer was “no”, then he is suggested to make the choice between discounted (at the particular level) GM bread and not discounted non-GM bread. The discount is set at one of the following levels: 5%, 15%, 25%, 40% and 50%. All levels of discounts should be used the same amount of time. That is sixty of 300 surveys have a 5% discount for GM bread, another sixty surveys have a 15% discount and so on. Respondents have to fill out questionnaires randomly, so the discounts will be offered randomly too. The answer “yes” at the initial question assumes no following questions (Hanemann et al, 1991).

Hence we can observe three possible outcomes and three groups of individuals, based on their view toward GMO. The first group includes consumers who are ready to buy GM products without any discounts at the same price as the non-GM product implying a “yes”, call it favorable view. The second is about individuals who need some discount to buy such products: the respondent is not willing to purchase the GM product neither at the same price but is willing to buy the GM food at the random discount offered, so a “no” followed by a “yes”. And the third group involves people with negative attitude, who are not ready to buy GMO neither at the same price nor at a discount relative to the non-GM product, so “no” to both bids.

The building of econometric model will be the next stage of the work. It is known that if the dependent variable in the regression model is continuous the investigation can be carried out using linear regression model. Also, if the dependent variable is binary the analysis could be carried out using linear probability, logit or probit models (Pindyck and Rubinfeld, 1981). However, the using of linear probability model may formed estimates less than zero or higher than one as a result, which contradicts the basic principles of probability (Gujarati 2004).

The first model that is appropriate to investigate the results of our study can be considered as a special case of the double-bounded logit model (Hanemann et al, 1991). In the model, the initial bid B0 equals zero and represents no price difference between GM and the non-GM food product. The second bid - BD - is a random percentage discount on the GM-food product offered at relative to the corresponding, non-GM food product. This bid is only suggested to those individuals who answer that they would not purchase GM food products at equal prices.

The sequence of questions distinguishes the span in which the respondent's true WTA discounts for GM products relative to non-GM food products lie. The discount for the GM product can be zero, B0, or it can be located in one of the intervals (B0, BD],? BD,???,?where BD is the discount bid offered. Because of the relation of the second bid, BD, and the individual`s response to the initial preference decision, there are bounds to be placed on the respondent's unobservable true WTA for GM food. The lower bound on the WTA discounts for GM food was established as no discount on GM-food compared to non-GM food. There is an assumption, that the genetic modification did not add any positive value to the product for the consumer, which based on the "Levada-center" survey, confirming negative attitude to GM food of the Russian customers in 2011 (81%).

The randomly appointed discount bids were BD = {0.05, 0.15, 0.25, 0.4, and 0.5}, or pronounced as percentages: 5%, 15%, 25%, 40%, and 50% of discount compared to the price of the normal product. Let BD ? 0 denote the discount bid on the GM food product relative to the non-GM food conventional product. The following discrete outcomes of the bidding process are observable:

D = (1)

Respondents who require no discount fall into the first group, because they are willing to buy the GM bread at zero discounts. People falling into the second group need a discount that is positive but less than or equal to B0 for GM bread compared to the non-GM bread. They will choose the GM bread, if the price of the GM bread will be discounted relative to the corresponding non-GM bread by an amount less than or equal to the offered discount bid. Finally, respondents who require the largest discount to purchase GM bread fall into the third group. These customers are not willing to buy the GM-product at the discount offered. The WTA function for individual i is:

(2)

Where Bi is the ultimate discount bid offered to individual i, Zi is a column vector of observable characteristics of the individual, еi is a random error term. Unknown parameters to be estimated are б, с, and л. Linearity in Z and е is assumed. Moreover, the distribution of the error term is assumed as following: е ~ G (0, 2), where G (0, 2) denotes a cumulative distribution function with mean zero and variance 2. Under these hypotheses, the choice probabilities individual i can be described as The clause of linearity on z is a streamline hypothesis broadly used in RUM (random utility models). :

(3) 

Thus, the log-likelihood function becomes:

(4)

Where IK is a function, pointing the event K, and Di = j indicates that the j th alternative derived. In the empirical implementation of the model, we assume G (·) it is the standard logistic distribution function with mean zero and standard deviation у =р / 3.

The final econometric equation will represents the dependence of consumer willingness to accept genetically modified food (WTA) from the offered discount and a column vector of observable characteristics of the individual. Hence we will able to evaluate the importance of factors that affect Russian consumers' choice of whether to purchase GM-foods and, specifically, how much of a relative discount Russian consumers will need to pay for GM bread.

Some authors - Cameron and Quiggin (1994), Aprahamian, Chanel, and Luchini (2007) - noted that the answer at the second response may be influenced by the initial answer, where this first bid takes the role of an anchor. It was asserted that when the individual responses at two questions, then he may reconsiders his willingness to pay and the distribution of WTP to the follow-up (contingent) question may differs from the initial question. The bivariate probit model allows the estimation of varied distributions of WTP across the first and second questions whereas the interval data model assumes the same distribution of WTP. In the frame of bivariate probit model, we model two equations of dichotomous choice responses as single-bounded simultaneously. Hence, the bivariate probit model is the second appropriate specification to evaluate consistent mean values in a doubled bounded model.

Green (2007) thus posited the bivariate probit model as a natural extension of the binary choice model in which two solutions are taken jointly: 

(5)

(6)

Where y1 = the first vector of price offered; y2 = the second vector of price offered; x1 and x2 = the first and second bids offered to the consumer respectively; в1 and в2 = coefficients of the first and second bids offered, need to be found.

The best and widely used approach to obtain mean and median willingness to pay and its confidence intervals consists in simulating the confidence intervals using the Krinsky and Robb procedure (Haab and McConnell, 2002). We can to compute the mean WTP using the following expression:

(7)

Where, б is a coefficient for the constant term, and в is a coefficient for offered bids to the respondents.

The STATA package of statistical data processing will be used for all computations.

Results

In this study 300 respondents were conducted with a face-to-face interview. All of the questionnaires gathered were considered valid for the data analysis procedure. In total, 300 consumers were surveyed. The majority of respondents are the primary food shoppers of the household (68%) and female (66%). The respondents' average age is 37 years, which is close to the average age of 38, 7 years for the population of Perm in 2015 (Permstat, 2015). Forty percent of all respondents have children under the age of 18 years living in their household; it can be related with relatively high average age of respondents. The average household size is close to 3 persons. Most of respondents are living in the city center (62%), what can be a biasness of a sample, because of spatial restriction of a survey - all chosen grocery stores are located near the city center. The 60% of people from our sample included themselves in the wageworker's group. The most frequent range of household income as a whole family is between 30000 Russian Ruble (U.S. $423, 73) and 50000 Russian Ruble (U.S. $706, 2) for the 2016 fiscal year. This range is slightly lower than the average Perm income: in 2015, the average income for one person was 31000 RUB, accordingly, considering household with two adults, average income of household is about 62000 RUB. The most frequent level of education includes master's degree. The dataset contains 255 usable observations. Summary statistics and variable descriptions are presented in Table 3.

Table 3. Summary Statistics for the Demographic Variables

The survey asked for information relating to the respondents' attitudes concerning saving jobs or environment, their self-reported knowledge and opinion about biotechnology (see Table 4). Information about environmental attitudes was obtained by presenting a trade-off situation between saving jobs and environmental quality. We find that 39% of the customers assign more importance to saving the environment than jobs security. Furthermore, Perm consumers prefer to choose healthier food rather than cheaper products. Most people consider that GMO consumption implies high risk for health. The most popular position toward using biotechnologies in food production among respondents was indifference. The majority of customers believe they know something about GMO products and biotechnologies, only 9, 6% establishes themselves as well-versed.

Table 4. Summary Statistics for Consumer Information and Perception Variables

Of the 300 respondents, only 12% of the sample would buy bread with GM wheat flour when offered no discount compared to conventional bread. Further,

8, 3% of the sample replied that they would be willing to purchase the GM products if they were cheaper than the conventional products. For more specific statistics on the distribution of responses over the various discounts, see Table 5.

Table5. Range and Distribution of Response Rates to the Randomly Assigned Discount Offers

After data analysis and description we constructed two econometric specifications. Firstly, there is likelihood function maximization, using command Doubleb in STATA package of statistical data processing. This method supposes joint estimation of responses on the first and on the second answers of individuals. The parameter estimates are reported in Table 7.

Table 7. The Likelihood Function Parameter Estimates (Dobleb command)

* - p<0, 05; ** - p<0, 01; *** - p<0,001,

We assessed the willingness to pay using double-bounded dichotomous continent valuation approach and maximum likelihood method. Consequently, we can interpret only significance and signs of the variables. Willingness to accept is a dependent variable, in other words the probability of buying bread with genetically modified components. Also, we can say that the lower willingness to accept, the higher discount required by the individual.

The received results evidence that the willingness to accept compensation for the consumption of GMOs negatively depends on risk perceptions. That is, the more the individual perceives genetically modified food as risky, the less probability that he will buy GMO and the bigger price reduction he needs. Conversely, if the individual regards GMOs as safe food products then there is higher probability of buying and the individual will require lower discount for such product. Further, the variable use of biotechnologies is also significant. Respondents, who consider the use of biotechnologies is normal, are willing to buy GM products with lower discount and with higher probability. So, the more a consumer rejects the use of biotechnologies in food production, the higher discount he needs to choose these products. The next significant variable is the degree of self-reported knowledge of individual concerning GMO and biotechnologies. It was found that the more knowledge individual has the more probability of making a choice in favor of GMO takes place. Moreover, the variable labeling is also significant. People who find that labeling of genetically modified production is very important tend to be more skeptical about it. Hence, the more consumer values labeling, the lower probability of buying GMO commodity exists.