Advances in Consumer Research Volume 16, 1989 Pages 454-459
EXPLORING THE EFFECTS OF COUNTRY OF ORIGIN LABELS: AN INFORMATION PROCESSING FRAMEWORK
Carl Obermiller, University of Washington
Eric Spangenberg, University of Washington
Theories of information processing are used to develop a framework for the effects of country of origin labels. The framework identifies three types of effects: cognitive--the traditional hierarchy of effects, affective--an emotional response to country stereotypes that influences attitude directly without intervening belief changes, and normative--a direct effect an behavior
The growth of international trade and the academic debate over the merits of global versus customized marketing have increased awareness of and interest in the effects of country of origin labels on consumer evaluations. Although a large number of studies have been reported, no firm conclusions can be drawn on the pervasiveness or the strength of country of origin (CO) label effects. Lacking a basis in theory, most CO research has been limited by a narrow descriptive focus and marred by methodological flaws. Our objective is the development of a theoretical framework that can account for the existing diverse findings and generate both useful directions for further research and insights into marketing practice.
Most published work exploring CO phenomena has reported descriptive empirical work. These studies primarily detail which countries, product classes, or consumer groups, exhibit or possess which stereotypical effects. That countries connote different product images or stereotypes in consumers' minds has been firmly established (Bannister and Saunders 1978; Eroglu and Machleit 1986; Gazda, Light and Brown 1986; Khanna 1986; Lillis and Narayana 1974; Nagashima 1970; Reierson 1966; Schooler and Wildt 1968; Schooler 1971). For example, Gaedeke (1973) found that attitudes toward a specific product or brand are substantially changed, both favorably and unfavorably when the country of origin of the product/brand is revealed to the consumer. The CO phenomenon is also observed in industrial buyers (White and Cundiff 1978; White 1979) and has been shown to change over time (Nagashima 1977). CO stereotypes have been shown to vary by demographic characteristics: age and gender (Bannister and Saunders 1978; Schooler 1971), race and education (Schooler 1971). Older, better educated, and higher income consumers tend to have more positive attitudes toward foreign products but no one has determined whether the correlation is the result of these consumers' using the CO label more or less than other consumers. In general, attitudes toward products made in less developed countries have been shown to be considerably less favorable than toward products from more developed countries (Gaedeke 1973).
Other research has indicated psychographic traits significantly influence the effect of CO images on domestic versus foreign auto purchase (Andersen and Cunningham 1972). In a similar fashion, Johansson and Nebenzahl (1986) proposed a link between CO stereotyping and social norms in the consumer's country. Douglas and Urban (1977) supported this idea with evidence that women in a given country hold differing social norms than those of women in the same product markets of different nations/cultures. Complicating the roles of psychographic traits and normative beliefs is the finding that stereotypes of a country or culture were not equivalent to stereotypes of products from that country (Etzel and Walker 1974).
While most of the work done on CO biases has been descriptive, a few studies have been explanatory and/or predictive in nature. These studies set out to explain why people have whatever CO stereotypes they exhibit, and how those stereotypes will manifest themselves in different situations. Schooler (1965) indicated that CO product stereotypes result from attitudes toward people and governments of other countries, although he also concludes, counter-intuitively, that CO biases are not developed through travel or related to attitudes toward industry or labor organizations within a country. Erickson, Johansson and Chao (1986) and Johansson, Douglas and Nonaka (1985), attempted to model the effects of CO stereotypes based on information processing principles. Unfortunately, the work done in model development has many of the same limitations of earlier descriptive pieces, e.g. lack of generalizability. Using survey data from real world situations puts clear limits on the findings of these studies.
A few studies have examined the influence of marketing activities on CO effects. Communication and promotion devices were shown to alter CO biases favorably by Reierson (1967). The effect, however, was contingent upon consumers' holding CO biases of low intensity. The moderating effect of price was examined by Schooler and Wildt (1968), who found that the effect of product bias on the selection decision between similar, alternative domestic and foreign goods could be offset by manipulation of the price differential. Thus, a disinclination to purchase a product, resulting from a negative CO bias, should not be equated with an unalterable unwillingness to buy.
Most CO studies to date provide us with little generalizable knowledge. Results are typically product, country, or sample specific. Moreover, much of the research has been flawed by demand effects. Early experimental studies artificially heightened the salience of CO labels. Some provided little information other than the CO label, others used within-subjects designs that sensitized subjects to the manipulation of the country of origin (Bilkey and Nes 1982). Survey approaches that attempted to assess the role of CO labels in evaluations of real products suffered from gross self-report measures of the 'importance" of the CO label in overall evaluations.
THE NEED FOR THEORY
Central to the problem with CO research is a common weakness in business research--the focus on results from "realistic" applications leads to specious attempts to generalize from research at the level of operational variables rather than theoretical constructs. Lacking a generalizable theoretical framework, one can make little sense of the collection of positive, negative, and null effects in CO research. Nor can one understand the role of other managerial actions that may mediate CO effects. Although we know that some CO effects are positive and some negative, we do not know why. "Made in Germany" -(relative to "made in Italy") for instance, enhances evaluation of a car but diminishes evaluation of dress shoes. And, although descriptive research may uncover such instances, it does not tell firms how to take advantage of the positive or minimize the negative. Vast differences across purchase situations have prevented marketers from developing useful hypotheses that relate concrete tactics, such as CO --labels, to outputs, such as attitude or purchase. Rather, we must focus on the meaning of these tactics within the framework of consumer information processing. We hope to take some first steps toward the development and testing of a generalizable framework for CO effects.
Considerable research indicates that CO labels influence overall product evaluations or beliefs about product quality (e.g. Nagashima 1970; Reierson 1967). This effect may result from two processes: In the first, the CO label has no direct effect on evaluation. Rather, from it, consumers infer the level of some other attribute(s) that determine(s) a global attitude or assessment of quality. We suspect this cognitive process to be the most frequent role of CO labels in product evaluation. Erickson, Johansson, and Chao (1984) demonstrated this process for automobiles; CO labels affected quality evaluations through their effects on beliefs about such attributes as workmanship, durability, and reliability.
Less likely would be CO label's triggering an affective process, resulting in an emotional response that bypasses the purely cognitive inferential evaluation. Consumers may, in fact, infer positive beliefs on key attributes yet respond negatively to the country of origin. This emotional response will occur only when the CO is, itself, directly tied to an evaluative response. Americans of Arab descent, for example, may evaluate an Israeli-made precision instrument high on craftsmanship yet have a strong negative reaction overall. It is this affective process that has been cited as the country of origin stereotypes effect (Bannister and Sanders 1978; Reierson 1966).
A third process in which CO labels may be involved would reveal an influence on preferences without a change in overall evaluation or product attitude. The effect of the CO label would intervene between evaluation and behavioral intention. The extended Fishbein model of behavior (Fishbein and Ajzen 1975) allows for anticipated circumstances to moderate the effect of attitude on choice and cites normative pressure as an important type of anticipated circumstance. Thus, one may evaluate Nissan automobiles favorably and have no emotional response to Japan as a country of origin yet comply with a "Buy American" norm that operates in one's family or peer group.
The figure models the three processes that involve CO effects. It identifies situational and individual difference variables that mediate CO effects and output variables that indicate change.
The cognitive effects of CO labels result from an inference to some unknown attribute value. The constraints on any inference process are identified as individual difference variables: Consumers will use CO labels to infer other attributes if they perceive the two attributes to be associated, if they are confident in their knowledge of the country of origin, and if there are no better indicators of the target attribute value.
The four situational variables mediate the inference process by affecting one or more of the three constraints. Product category heterogeneity refers to the familiarity of the product category to the consumer and the perceived variation of brands on the target attribute. As with other proxy variables, such as price and brand name, one can hypothesize that when consumers are unfamiliar with the product class they will rely more on perceived associations between CO and target attributes. Johansson, Douglas, and Nonaka (1985), however, found a positive interaction between product familiarity and CO influence. Similarly, Johansson and Nebenzahl (1986) found a significant positive correlation between "Knowledge about product class" and "Importance of country-of-origin." We would argue that high familiarity does not preclude inferences from CO. The interaction is a complicated one, depending on situational factors. When product heterogeneity is high, CO may be a valid indicator of quality or other target attributes, and knowledgeable consumers will use it. Note that product heterogeneity refers to brand variation across countries of origin.
Country brand heterogeneity refers to the consumer's knowledge of the various brands from a given country in a single product category or small set of related categories and the perceived variation within that country's brands. Whereas product category heterogeneity refers to variations across countries, country brand heterogeneity refers to variations within a country. Lack of familiarity with a country's product offerings would prevent any perceived association between CO and other attributes. On the other hand, high familiarity with a country's offerings will permit such an association only when the country's brands have little variance. Thus, "California" wine does not connote a specific quality level to those who know that California vintages fall across a wide range. Consequently, confidence in the implication of the CO label should vary inversely with variation in the country's offerings. Variation within a country's offerings limits the informativeness of the CO label, as will any factor that reduces the clarity of country of origin. A specific threat to clarity is multiple countries of origin, represented by hybrid products, e.g. a car that might have parts manufactured in several countries and assembled in still another. Thus, clarity of the CO label relates to confidence in its informativeness.
A THEORETICAL FRAMEWORK FOR COUNTRY OF ORIGIN EFFECTS
The final mediator of the cognitive inference process is the presence of other information. Consumers will rely on CO labels as referents only if no better indicators are equally available. Whether some other information is an indicator of a target attribute is a function of the same set of issues that have been discussed with respect to CO labels as indicators. Research has demonstrated, however, that "intrinsic" cues, which are informative of composition or functional performance, are better indicators of quality than such "extrinsic" cues as CO label (Olson and Jacoby 1972; Wheatley and Goldman 1981).
The remaining two processes, affective and normative influences, bypass cognitive processing and are subject to less mediation from external factors. Both are mediated by the clarity (and salience) of the CO label, which acts as a trigger. An affective response requires an evaluative response to the country stimulus, a stereotype, that overrides any attribute-based evaluation. Overall product evaluations may be colored by other product related cues. Research on stereotypes, schematic processing, and halo effects indicates that such affective responses influence attitude directly, without affecting attribute beliefs. An implication for marketers with respect to affective processes regards focus of promotional effort. Thus, promotional effort may be aimed at positive aspects of the country of origin rather than the specific product. Rather than trying to change product beliefs, marketers can attempt to change negative affective responses to their products. If the country stereotype cannot be changed the alternative is to reduce the clarity of the CO label by deemphasizing or obscuring it.
Normative processing of CO labels occurs when a country-relevant norm exists. Attitude-behavior theorists have proposed a combination of normative pressure and motivation to comply with such norms as a mediator of attitude's effect on behavior. When strong norms exist regarding country-specific purchases, CO may affect purchase intention without changing attribute beliefs or attitudes. Some consumers, for example, will "Buy American" due to normative pressure irrespective of their belief that other countries produce superior automobiles. Marketers utilizing the "Buy American" theme are trying to establish normative processes, subrogating consumer attitudes and attribute beliefs in advertising campaigns. Countries that face normative obstacles have three strategic options: They may attempt to change the social norm, often a difficult task. They may attempt to change individuals' motivation to comply with the norm, which may often be done by appealing to appropriate reference groups for support of noncompliance. A final option is, again, to obscure the relationship between product and CO.
IMPLICATIONS FOR EMPIRICAL WORK
CO effects research is a classic case of research "falling in the crack" between theory-based research and effects-application research (see Calder, Phillips and Tybout 1981). Research interest has been motivated almost exclusively by interest in applied effects, with the result that CO labels have been considered almost exclusively at the level of operational variables. Yet, researchers have attempted to draw general conclusions from their results. If nothing else, our framework indicates that CO labels have different meanings depending on a number of mediating and moderating factors. In large part, the hodgepodge of past findings has resulted from a failure to consider consumers' subjective interpretations of CO labels.
In order to generalize the effects of CO labels, researchers must understand them at the theoretical level, which implies a focus on the construct level of the variable, rather than the operational level. It is insufficient to observe differences in consumer responses to products with labels "Made in Japan," "Made in Great Britain," or no label; we need to investigate the basic information processing content of such labels.-The role of a theoretical framework is to permit one to translate from the level of operations to the level of theoretical constructs. The researcher who wishes to test hypotheses derived from such a framework has two general options: correlational designs that measure all the specified constructs or experimental designs that manipulate and control them.
We advocate the experimental approach for two reasons. First, correlational studies rely on the validity of their measures to elucidate effects that result from naturally-occurring (read "small") variations on the factors in the model. To date, we have no valid measures of the factors identified in our framework. Any measures of perceived association, confidence, availability of other information, and prior beliefs in variation of products across or within countries would be gross self-report scales. We have more confidence in our ability to manipulate levels of these factors than in our skill at identifying naturally-occurring differences.
Second, the CO label is a complex phenomenon. We do not presume to have captured all that complexity in our framework. Even if one assumed perfect measures of the factors that have been specified, a correlational study might be misleading due to underspecification of the model. A failure to identify causal variables in an experimental design can also threaten internal validity, but an experiment allows the researcher to use very strong manipulations, which may override the effects of uncontrolled variables.
The advantages of experimental designs in determining causal relationships are elementary; the problems with using experiments to address the CO label phenomenon may be less obvious. In general, the concerns are of two types: (1) assuring that experimental manipulations are construct valid and (2) assuring that the experimental situation does not become a causal factor by introducing expectations, increased sensitivity, or other biases. Both concerns are relevant to investigations of the roles of the factors we have identified. For example, how does one manipulate consumers' perceived associations between a CO label and overall product quality, or the valence of a country stereotype, or a country-relevant norm? These factors are the result of accumulated learning. If a subject comes to an experiment with a strong perception that country of origin indicates overall quality of the product in question, it is unlikely that any experimental manipulation will change that belief. It is necessary, therefore, to avoid situations where the relevant consumer beliefs are intractable. As a result, in order to understand important "real life" effects, the experimenter may purposefully investigate CO effects for products and countries that are unimportant or even fictitious.
The approach we suggest is a scenario paradigm. Subjects are given information designed to create desired levels of specific prior beliefs. The subjects subsequently evaluate products with CO labels. Since both the specific brands and the countries are either little known or fictitious, one expects that subjects will not have well-formed prior beliefs and will, therefore, form beliefs based on the scenario information provided in the experiment. This paradigm has been used successfully in prior research on price as a cue to quality. Obermiller and Wheatley (1984) used several scenarios to create beliefs of either high or low quality variance across brands. In their study, manipulation checks indicated that scenarios succeeded in creating the intended prior beliefs, and quality ratings supported the hypothesized link between prior beliefs and the price-perceived quality relationship. We expect that the scenario paradigm may be applied similarly to the study of prior beliefs that are proposed to affect the relationship between CO labels and perceived quality.
We conducted a study using the scenario paradigm (See Obermiller and Spangenberg 1988). Due to space limitations we cannot present a complete report of the empirical study. Hence, we will briefly summarize the results of that study and discuss the implications of both the results and the overall framework.
The study was designed to test one small part of the framework -- the hypothesized mediation of CO effects by product heterogeneity and country heterogeneity. The results of the experiment provided mixed support for the model's hypothesized mediation of CO effects. By and large, evaluations of one of the product categories were influenced by CO labels as hypothesized; evaluations of a second category were not.
Despite the mixed empirical results, the study has provided valuable understanding. The model provides a framework for understanding some of the mixed results in past research and from it one can generate further testable hypotheses. Thus, it appears to be a useful framework for the effects of country of origin cues on consumer evaluations and preferences.
Care was taken in the design of the research to control the constructs under investigation by fictionally creating prior beliefs rather than merely representing these constructs with existing brands. This artificiality is necessary if one is to separate the effects of prior beliefs regarding the brand and country variances from any other beliefs. There are two risks to this paradigm of fiction we have used. The first is that subjects will refuse to believe it. The second is that its very artificiality will create demand effects. Our manipulation checks argue against the former and our mixed results argue against the latter. An explanation based on simple demand effects cannot account for one product supporting the hypotheses while the other product failed to do so.
We expect the differences across products resulted from one's being agricultural (coffee beans) and the other's being manufactured (camping stoves). Subjects may have relied on available stereotypes or schema for judging manufactured goods -- evaluating camping stoves according to beliefs about country specific appliances for example. The extent to which consumers make inferences from one product category to another is an area in need of research.
The role of schematic or stereotype processing in CO effects is but one question for future research. Although the model does not help us understand when consumers will regard an unfamiliar product as representative of some larger class or what that class will be (another consumer might make inferences about lawn mowers from prior beliefs about farm equipment), it could describe the processing of CO labels for that larger class as well as for a specific product.
The model also implies individual and contextual effects that could be examined in future research. Greater expertise and familiarity have been suggested to reduce reliance on the CO cue, reasoning that consumers who are more qualified will use other, better indicators of quality. The model predicts a more complex relationship: Greater expertise would increase confidence in the information value of the CO labels, but the perceived association between CO label and product quality may be high as well as low. Thus, an expert may feel confident that a "made in" label is a good indicator of quality; whereas, a novice may only weakly suspect so.
The model might be used to investigate the effects of "hybrids"--products with multiple countries of origin. A hybrid CO label might reduce the clarity of the cue and potentially create conflicting associations. With the increasing globalization of the marketplace, marketers will be interested in the effects both of alternative and combined CO labels.
A final question suggested by the model is the mediating effect of other indicators of quality. If consumers are able and predisposed to use it, they are predicted to rely on such information, when available, rather than making inferences from CO labels. The model does not give any insight into what other indicators would supply better information about quality, but future research might begin with physical composition information, price, non-commercial endorsements, packaging, and store image.
As a concluding comment, we wish to broaden the context of our work. The model has been presented as a framework for CO effects, but it should be seen as applicable to cue utilization in general. The effects of price, brand name, package design, and other peripheral or extrinsic cues to quality are still poorly understood despite considerable research. We feel that our model provides an integrative framework, from which we can move beyond studies of isolated main effects to the more complex interactions and mediated relationships that characterize consumer evaluations. We hope that other researchers in the area will regard country of origin effects as an instance of the larger field of the role of cue utilization in consumer information processing.
Andersen, W. and W.H. Cunningham (1972), "Gauging Foreign Product Promotion," Journal of Advertising, 12 (1), 29-34.
Bannister, J.P. and J.A. Saunders (1978), "UK Consumers' Attitudes Towards Imports: The Measurement of National Stereotype Image,"European Journal of Marketing, Vol. 12, 562-70.
Calder, Bobby J., Lynn W. Phillips and Alice M. Tybout (1981), "Designing Research for Application+" Journal of Consumer Research, 8 (September), 197-207.
Bilkey, Warren J. and Erik Nes (1982), "Country-of-Origin Effects on Product Evaluations," Journal of International Business Studies, (Spring/Summer), 89-99.
Douglas, Susan P. and Christine D. Urban (1977), "Life-Style Analysis to Profile Women in International Markets," Journal of Marketing, (July), 46-54.
Erickson, Gary M., Johny K. Johansson and Paul Chao (1984), "Image Variables in Multi-Attribute Product Evaluations," Journal of Consumer Research, 11 (September), 694-9.
Eroglu, Sevgin A. and Karen A. Machleit (1986), "Country of Origin as a Product Quality Cue: An Improved Methodological Perspective," working paper.
Etzel, Michael J. and Bruce J. Walker (1974), "Advertising Strategy for Foreign Products," Journal of Advertising Research, (June), 41-4.
Fishbein, M. and I. Ajzen (1975), Belief, Attitude, Intention and Behavior, An Introduction to Theory and Research. Reading, MA: Addison-Wesley.
Gaedeke, Ralph (1973), "Consumer Attitudes Towards Products 'Made In' Developing Countries," Journal of Re ailing, 49 (Summer), 13-24.
Gazda, Gregory M., C. David Light and Jacqueline J. Brown (1986), "U.S. Consumer Attitudes Toward Foreign Automobiles," Presented at the Academy of International Business Annual Conference, (November).
Gazda, Gregory M., Susan P. Douglas and Ikujiro Nonaka (1985), "Assessing the Impact of Country-of-Origin on Product Evaluations: A New Methodological Perspective," Journal of Marketing Research, 22 (November), 388-96.
Gazda, Gregory M. and Israel D. Nebenzahl (1986), "Multinational Expansion: Effect on Brand Evaluations," Journal of International Business Studies, (Fall).
Khanna, Sri Ram (1986), "Asian Companies and the Country Stereotype Paradox: An Empirical Study," Columbia Journal of World Business, (Summer), 29-38.
Lillis, Charles M. and Chem L. Narayana (1974), "Analysis of 'Made In' Product Images - An Exploratory Study," Journal of International Business Studies, (Spring), 119-27.
Nagashima, Akira (1970), "A Comparison of U.S. and Japanese Attitudes Toward Foreign Products, Journal of Marketing, 34 (January), 68-74.
Obermiller, Carl and John J. Wheatley (1984), "Belie& in Quality Differences and Brand Choice," in Advances in Consumer Research, vol. 12, eds. Elizabeth C. Hirschman and Morris B. Holbrook, Association for Consumer Research, 75-8.
Obermiller, Carl and Eric Spangenberg (1988), "An Information Processing Framework for Predicting the Effects of Country of Origin Labels," University of Washington working paper.
Olson, Jerry C. and Jacob Jacoby (1972), "Cue Utilization in the Quality Perception Process," In Proceedings, Third Annual Conference of the Association for Consumer- Research, ed. M Venkatesan, 167-179.
Reierson, Curtis (1966), "Are Foreign Products Seen as National Stereotypes?" Journal of Retailing, (Fall), 3340.
Reierson, Curtis (1967). "Attitude Changes Toward Foreign Products," Journal of Marketing Research, (November), 385-7.
Schooler, Robert D. (1965), "Product Bias in Central American Common Market," Journal of Marketing Research, 2 (November), 394-7.
Schooler, Robert D. (1971), "Bias Phenomena Attendant to the Marketing of Foreign Goods in the U.S.," Journal of International Business Studies, (Spring), 71-80.
Schooler, Robert D. and Albert R. Wildt (1968), "Elasticity of Product Bias," Journal of Marketing Research, 5 (February), 78-81.
Wheatley, John J. and A. Goldman (1981), 'The Effect of Physical Quality and Price Cues on Consumers' Product Quality Perceptions: Implications for Retailers," Journal of Retailing, Summer, 100-124.
White, Phillip D. (1979), "Attitudes of U.S. Purchasing Managers Toward Industrial Products Manufactured in Selected Western European Nations," Journal of International Business Studies, 10 (Spring/Summer), 81-90.
White, Phillip D. and Cundiff (1978), "Assessing the Quality of Industrial Products," Journal of Marketing, (January), 80-6.