Willingness to pay or intention to pay: The attitude-behavior relationship in contingent valuation

Willingness to pay or intention to pay: The attitude-behavior relationship in contingent valuation

Willingness to Pay or Intention to Pay: The Attitude-Behavior Relationship in Contingent Valuation E. JANE LUZAR KELLI J. COSSE" Louisiana Agricultur...

1MB Sizes 7 Downloads 114 Views

Willingness to Pay or Intention to Pay: The Attitude-Behavior Relationship in Contingent Valuation

E. JANE LUZAR KELLI J. COSSE" Louisiana Agricultural Experiment Station ABSTRACT" Increased awareness of factors influencing the environment suggests it is reasonable to assume that salient environmental attitudes exist and influence behavior. Appropriate elicitation of environmental attitudes tot inclusion in nonmarket valuation models may improve the descriptive and predictive ability of these models, especially in the case of contingent valuation studies eliciting willingness-topay values. This study identifies an appropriate conceptual model of the attitudebehavior relationship that is conceptually consistent with the process of contingent valuation. Using primary data collected from a survey of rural residents, willingnessto-pay to accept changes in individual and state level water quality is estimated with and without attitudinal explanatory factors. In both models, attitudinal variables are significant explanatory factors, enhancing the explanatory and predictive power of the estimations.

Increased awareness of factors influencing environmental quality suggests that it is now appropriate to assume that salient, measurable environmental attitudes exist and influence behavior (Ladd & Bowman, 1996; Arcury, et al., 1987). Environmental attitudes have, for example, been found to be significant explanatory factors in behavioral analyses of environmental participation decisions (Lynne, et al., 1988), consumer response (Allen,. Machleit & Kleine ,1992; Wilson et al., 1990), *Direct all correspondence to: E. Jane Luzar, Department of Agricultural Economics and Agribusiness. Louisiana State University, Baton Rouge, LA 70803-5604. Journal of Socio-Economics, Volume 27, No. 3, pp. 427-444 Copyright © 1998 by JAI Press, Inc. All rights of reproduction in any form reserved. 1SSN: 1053-5357

428

THE JOURNAL OF SOCIO-ECONOMICS Vol. 27/No. 3/1998

environmental management (Heberlein, 1989), and recreation decisions (Manfredo, Yuan & McGuire, 1992). However, in empirical research attitudes are often confused or intermingled with the related concepts of perceptions, beliefs, and values, introducing some question as to what has actually been analyzed (Ajzen & Fishbein, 1980; Ajzen, 1991). Appropriate elicitation of environmental attitudes for inclusion in economic explanatory models may improve the descriptive and predictive ability of these models, especially in the special case of contingent valuation studies eliciting the behavioral intention of willingness-to-pay. The following section reviews the social-psychological foundation of attitudes, emphasizing their relationship to behavior and intentions toward a behavior. A model of the attitude-behavior relationship that is conceptually consistent with many explanatory economic behavioral models, including the process of contingent valuation, is then presented. For purposes of illustration, this conceptual framework is applied in a contingent valuation example which elicits willingnessto-pay for changes in rural water quality. In order to explore the contribution of attitudinal explanatory variables in this setting, the variation in willingness-to-pay for changes in water quality is evaluated with and without attitudinal explanatory factors for individual and state level water quality. A summary and conclusions follow. THE ATTITUDE-BEHAVIOR RELATIONSHIP

Attitudes are abstract psychological constructs, and therefore some disagreement exists among social scientists as to their definition and the relationship between perceptions, attitudes, values, and beliefs. Perceptions refer to an individual's current appraisal of an object as experienced in the immediate situation. Attitudes guide these appraisals of the object. While perceptions are linked to attitudes, attitudes have been the focus of most social science research due to their relationship to behavior (Fazio, Powell, & Williams, 1989). An attitude is the degree of belief that a consequence will occur, multiplied by the evaluation of that consequence. Therefore, if an individual agrees with a certain statement, it suggests that the individual places a high probability on the consequence occurring. Attitudes can, as a result, be defined as the degree of belief, or probability, that some consequence will occur, multiplied by the evaluation of that consequence (Summers, 1970; Dawes, 1972; Henderson, et al., 1987; Ajzen & Fishbein, 1980; Ajzen, 1991). Although early attitudinal research debated whether attitudes influence behavior & therefore whether attitudes could predict behavior, current research is consistent in its assertion that attitudes do influence behavior. The issue of current interest is determining when the attitude-behavior link is most accurate. This body of research suggests that the degree of influence of attitudes on behavior varies depending on factors such as multiple attitudes, prior information, or contemplation (Lynne, et al., 1988; Heberlein, 1989; Manfredo, 1992).

Willingness or Intention to Pay

429

Multiple Attitudes A person may have several attitudes toward an object. Napier and Forster (1982) suggest that one reason it is difficult to determine the role of attitudes in predicting behavior is that people may have many contradictory attitudes. Depending on how these contradictions are reconciled, the discrepancies may not be inconsistent (Millar & Tesser, 1990). For example, an individual may have a strong positive attitude toward a natural resource and its stewardship and also have a positive attitude toward making a profit from that resource. The two attitudes would be consistent if the profit making activities did not destroy the resource so that it was contrary to the stewardship beliefs (Napier & Forster, 1982).

Prior Information The attitude reported at any specific time may be a function of the information presented to a respondent. For instance, a rural person may initially have a neutral attitude toward changes in water quality. However, when given some information about the effects of agricultural pesticides in drinking water, she may develop a different attitude. These differing attitudes may provoke different behaviors. Therefore, knowledge of a person's attitudes does not always guarantee that behavior can be predicted (Ajzen & Fishbein, 1980). When given prior information about an attitude object, the attitude a person forms is a more accurate predictor of behavior than an attitude formed without prior information. For example, while Nowak and Korsching (1983) found that many farmers had a positive attitude toward stewardship, this attitude did not allow conservation behavior to be predicted. In the area where an educational program was in place, farmers were more likely to perceive soil erosion and water quality as a problem and therefore were more likely to adopt BMPs than those farmers in the control area, without an educational program. Prior information may therefore introduce a positive or negative bias when evaluating the attitude-behavior relationship.

Contemplation It has also been established that thought or reflection about an object can increase the correlation between attitudes and behavior or decrease the correlation, an effect known as the Cognitive-Affective Mismatch Hypothesis (Millar & Tesser, 1990). Attitude reports are based on an aspect of the object that is salient at the time of the attitude report (Millar & Tesser, 1990; Fazio, Powll, & Herr, 1983). For instance, a respondent may express a neutral attitude toward rural water quality and thus not be willing to pay for changes in water quality. If asked to explain this neutral attitude, the respondent may feel compelled to offer a "good" reason and in that process, state concern about water quality. Even though the attitude changed from neutral to positive, the behavior (willingness to pay) did not change, suggesting that thought about the attitude decreased its correlation to behavior (Millar & Tesser, 1990).

430

THE JOURNAL OF SOCIO-ECONOMICS Vol. 27/No. 3/1998

In contrast, other studies suggest that thought about an attitude object can increase its ability to predict behavior (Millar & Tesser, 1990). Persons who are self-focused, or introspectively in touch with their values, beliefs and attitudes are likely to have a high correlation between attitudes and behavior. Similarly, when respondents are instructed to think about the attitude, the attitudes' ability to predict actual behavior increases (Millar & Tesser, 1990). Summarizing to this point, it has been suggested that attitudes play an important role in behavioral outcomes through factors including multiple attitudes, prior information, and contemplation. However, it is also important to recognize the association between attitudes and other concepts such as values and beliefs.

Attitudes, Values and Beliefs In addition to their relationship with behavior, attitudes are also related to values and beliefs. Like attitudes, values are also abstract constructs, often lacking a unified definition. A definition that is accepted by most researchers is one offered by Mueller, (1986) which states that values are an "enduring belief that a specific mode of conduct or end state of existence is personally or socially preferable to an opposite or converse mode of conduct or end state of existence." Values are more permanent than attitudes. Generally, an individual develops a set of values in the early stages of life, based on family or social ideals or childhood surroundings. Attitudes are generated from this set of values, and a single attitude may be caused by many values. For instance, if a person values health and the environment, he may have a positive attitude toward improved rural water quality. Conversely, a person with a family tradition of farming may have strong values regarding the farm, and thus a neutral attitude toward water quality. Beliefs are defined as "a concept's position on the probability dimension" (Fishbein, 1967). More specifically, beliefs can be separated into beliefs in, the probability that an object exists, and beliefs about, the probability that a relationship involving the object exists (Fishbein, 1967). The controversy that surrounds competing definitions of attitudes and values does not appear to exist for beliefs. There is a reciprocal relationship between beliefs and attitudes. That is, beliefs about objects affect attitudes and beliefs are influenced by attitudes. For example, if an individual believes that a person has good qualities, one tends to like that person. The reciprocal is that an individual is more likely to believe positive information about someone that she likes than someone that the individual does not like (Mueller, 1986). Beliefs are sometimes considered to be expressions of attitudes and are often used as the means for measuring attitudes.

Willingness or Intention to Pay

431

Fishbein and Ajzen formalized the relationship between attitudes, values, and beliefs in the following expression: N

Ao =

~

Biai

(1)

i=1

where A o is an attitude toward an object o, B i is the strength of belief I that the object o is associated with another object, a i is the evaluative aspect (value) of B i, the object with which the attitudinal object is believed to be associated, and N is the number of beliefs about o (Ajzen & Fishbein, 1980; Mueller, 1986; Ajzen, 1991). Ajzen's (1991) Theory of Planed Behavior is the model most often used by economists to link attitudes to the prediction of a behavioral intention (Lynne, et al., 1988; Ajzen & Driver, 1992). As shown in Figure 1, key psychological concepts are combined in this model, resulting in the prediction of a behavioral intention which leads to an actual behavior. First, the relevant behavior must be identified, for example, willingness to pay. If the appropriate behavior is not identified, the relevant attitudes may not be elicited. This model depicts two main elements that are combined to form the prediction of a behavioral intention: an attitude and a subjective norm. An attitude is the product of an individual's evaluation of one' s behavior' s consequence, multiplied by the strength of belief that the behavior will lead to the consequence. The other main component in the prediction of an intention is the subjective norm. The subjective norm is the product of a set of normative beliefs multiplied by the individual's motivation to comply with the normative beliefs (Ajzen & Fishbein, 1980; Ajzen, 1988; Ajzen, 1991). AN EMPIRICAL ILLUSTRATION The following illustration reports the empirical results of an effort to reconsider a model explaining the variation in willingness to pay, in this case for water quality changes, by including appropriately constructed and measured attitudinal constructs, including a subjective norm, behavioral intention, and attitude toward the behavior. Primary data for the modeling activities were collected through a mail survey of rural Louisiana residents. The study area chosen for this survey was the rural, rice producing region of southwest Louisiana. A number of agricultural production regions of Louisiana have experienced water quality problems due to the production requirements for specialized crops, such as rice. Because irrigated rice requires a tremendous amount of water in the production process, it can have a profound impact on the ground and surface water quality in the area. One of the main environmental concems in rice production areas involves runoff water which can contain high levels of suspended solids, causing low dissolved oxygen and "fill-

432

O F SOCIO-ECONOMICS Vol. 27/No. 3/1998

THE JOURNAL

......................................................... ,~ _~

• .......................................................

©

r~

o_

. ~N

.<

"D

¢,.m

ii

u_

i

i_ @

Z

Willingness or Intention to Pay Table

Variable

1.

433

Variable Descriptions: Water

Hypothesized Sign

WTP1 WTP2 TREATED FARMER INCOME KIDS AGE AGESQ GENDER EDUC1 EDUC2 OIL AGRIC

(+ ) (+ ) (+) (-} (+) (+) (-) (+) (+)

PRIVATE ATTITUDE

(+-) (+)

SUBNORM

(+)

(+) (- )

Willingness Quality

t o Pay f o r C h a n g e s in

Variable Description Willingness to pay for changes in individual water quality Willingness to pay for changes in state level water quality 1 if the respondent remediates home water; 0 otherwise 1 if the respondent lives on a farm; 0 otherwise Respondent's income 1 if children under 12 present in the household; 0 otherwise Respondent's age Respondent's age, squared 1 if the respondent is female; 0 otherwise 1 if more than high school; I) otherwise 1 if grade school; 0 otherwise 1 if oil is believed to be a source of water pollution; 0 otherwise 1 if agriculture is believed to be a ~our~ e of water pollution; (I otherwise 1 if private well is primary drinking water source; 0 otherwise 1 if respondent's attitude toward willingness to pay for changes in water quality is at least good; 0 otherwise 1 if subjective norm is likely; 0 otherwise

ing-in". Other concerns include pesticide and organic matter contamination, and saltwater intrusion. Groundwater for the irrigation of rice and drinking water in the study area comes from the Chicot Aquifer, a sole source aquifer. As a rice monoculture production area heavily reliant on groundwater for agricultural irrigation, this region is characterized by competing demands for water among agricultural, residential, and industrial users, especially oil production. Water quality was considered to be a salient environmental consideration for this empirical example because it consistently ranks first and strongly in surveys of state level environmental concerns. In order to evaluate the contribution of socio-psychological factors in this model, two versions of the model were specified. A model without attitudinal factors was first specified. The second specification included this base model plus three socio-psychological variables represented in the Fishbein-Ajzen Attitude Model presented earlier. Table 1 presents descriptions and hypothesized signs of variables used in the analysis. In order to evaluate attitude accessibility in this model, two specifications of willingness to pay were used: willingness to pay for changes in an individual's own water quality and willingness to pay for changes in state-wide water quality. It was hypothesized that issues of state level water quality would be less focused and accessible for individuals. As a result, it was expected that the attitudinal modeling would contribute more to the model explaining variation in willingness to pay for the individual's water quality.

434

THE JOURNAL OF SOCIO-ECONOMICS Vol. 27/No. 3/1998

Survey Procedures The mail survey used to collect data for this study was designed and implemented according to the Dillman Total Design Method (TDM) (Dillman, 1978; 1991) which has been shown to result in improved response rates and response quality. A sample of 1,938 rural residents owning a water well in the study area was acquired through the Louisiana Department of Transportation and Development, the National Agricultural Statistical Service, and the Louisiana Department of Agriculture. Using TDM procedures of multiple mailings and follow up letters, the overall survey response rate was 34%, reflecting a return of 664 usable questionnaires. In the survey, individuals were asked to provide general background information about the study area and their opinions about their own water quality. In addition, individuals were asked about their main source of water and experiences with water quality. If the resident had a private water well, questions were asked about the location of the well and if it had been tested. The survey asked individuals about their level of education, income, employment status, age, gender, and race. In order to evaluate the economic model, it was necessary to elicit individuals' willingness to pay values for changes in water quality. Because it was assumed that individuals were not familiar with valuing environmental goods such as water quality, an important step in the construction of this section of the survey was to present the respondents with a hypothetical market for changes in water quality. This market compared water quality to many other publicly provided goods, such as police and fire protection, highways, and education, reminding respondents that all contribute to the provision of those goods. This familiarized respondents with how to value a public good such as water quality, allowing meaningful willingness to pay values to be revealed. The open-ended question format was used for the willingness to pay questions in which respondents were asked to disclose the dollar amount they would be willing to pay annually for a change in water quality. Although the single noniterative bid valuation process typically results in more zero value bids than alternative methods, it avoids other well documented contingent valuation biases (Mitchell & Carson, 1989). In order to test the hypothesis that individuals would be willing to pay more as the changes moved from a general to a personal level, the questions began with valuing water quality on a state-wide level and progressed to valuing water in the respondents' homes.

Attitude and Subjective Norm The attitude and subjective norm are the two nontraditional components of the economic models presented here. Since both are determinants of a behavioral intention, the immediate determinant of actual behavior, they are hypothesized to improve the explanation of variance in willingness to pay values. Once the behavior of interest (willingness to pay) was identified, respondents were asked about

Willingness or Intention to Pay

435

beliefs associated with the behavior and the intention to act on the behavior. Also included was a general Likert scale attitude question and an overall subjective norm question. However, to arrive at an attitude, its two components, the outcome evaluations and the behavioral beliefs, must be measured and then multiplied together. The same is true for the subjective norm. Its two elements, the normative beliefs and motivations to comply, must be identified and multiplied together. The evaluation of the consequences of one's belief is a component of the attitude. The outcome evaluations were multiplied by the "strength of belief' questions to form the attitude. Individuals were asked about their own evaluation of Louisiana's environmental conditions. Questions moved from general state-wide environmental quality to specific issues involving water quality in the respondents' homes. For this question set, respondents were given five choices to answer: Very Poor, Poor, Undecided, Good, or Excellent. Numerical values were assigned to each answer choice in order to make the multiplication procedure possible. A value of two was assigned to Excellent, one to Good, continuing on to a negative two assigned to Very Poor. Each question in this set was multiplied by its complimentary "outcome evaluation" question to form an attitude. These questions were designed to elicit the strength of the beliefs by allowing respondents to indicate the likelihood that performing a behavior would result in a certain outcome. For example, respondents were asked about the likelihood that their willingness to pay for water quality changes would result in improvements in Louisiana's water quality. The answer choices included: Very Likely, Somewhat Likely, Undecided, Somewhat Unlikely, Very Unlikely. The numerical values were assigned differently to these questions. Because they represent the strength of beliefs, the values attached to each answer choice ranged from zero, representing no strength of belief, to four representing a very strong strength. Very Likely was assigned a value of four and the remaining values declined to zero, which was assigned to Very Unlikely. The subjective norm, which is an important part of predicting actual behavior is formed by multiplying normative beliefs by the respondents' motivation to comply. Normative beliefs involve the respondent' s view of how others may view that respondent's willingness to pay for changes in water quality. Normative beliefs were elicited through responses to a statement such as: "The likelihood that most members of my family think I should be willing to pay for changes in water quality." The answer choices included: Very Likely, Somewhat Likely, Undecided, Somewhat Unlikely, Very Unlikely. Numerical values attached to these answer choices were consistent with those of the outcome evaluation section. The motivation to comply refers to how likely the respondent is to behave in a manner consistent with the opinion of others. Each of the questions complimented a normative belief question. The compliment to the example given above is: "The likelihood that I do what most members of my family think I should." The answer choices for this set of questions included: Very Likely, Somewhat Likely, Unde-

436

THE JOURNAL OF SOCIO-ECONOMICS Vol. 27/No. 3/1998

cided, Somewhat Unlikely, Very Unlikely. The numerical assignments to these answer choices also paralleled that of the behavioral beliefs.

Other Hypotheses In addition to the attitude variables described above, it was hypothesized that socio-economic characteristics of the respondents would explain the variation in willingness to pay due to their documented association with environmental concern and activity. Age was hypothesized to influence WTP because it signifies different points in an individual's economic life cycle. To account for a hypothesized nonlinearity in this variable, age squared was included in the specification. The presence of young children in the household was hypothesized to positively influence WTP due to their increased vulnerability to poor environmental quality. Gender, specified here as female, was hypothesized to be positively associated with WTP, as was a higher education level. Lower levels of educational attainment and lower income levels were hypothesized to have a negative influence on WTP. The role of agriculture as a potential source of pollution suggested introducing an indicator variable for employment status as farmer. The relationship between a farm occupation and the WTP value was hypothesized to be positive. As major commercial consumers as well as individual consumers of water, farmers were hypothesized to be willing to pay higher amounts for improved quality. Additionally, farmers may be more informed of the adverse affects of contaminated water, thus increasing their willingness to pay values. Perceived sources of water quality problems (agriculture and oil production) were included in the model and hypothesized to be positive influences on WTP due to their regional employment importance. A 0-1 indicator variable was also included for individuals who treated their water, regularly purchased bottled water, used a filter, or somehow attempted to remediate poor water quality. There was no a priori hypothesis formed for the relationship between WTP and having a private well. If respondents felt that having a private well made their water supply more vulnerable to pollution, they may have expressed a higher WTP value. Alternatively, the relationship could be negative if respondents were confident that water quality problems did not exist, or that they could protect their own well without contributing to "public" protection. SURVEY RESULTS Survey responses indicate that about half of the survey sample used a private well and half used a rural public water system as the main source of drinking and cooking water. Approximately 75% of the residents using private wells classified their wells as being located in a safe and protected area, while a significant portion (18%) classified the location as being near a source of pollution. Only 8% stated that the well was not in use. Of the private wells in use, 62% had not been tested for contaminants. When asked what action would be taken if their drinking water

Willingness or Intention to Pay

437

became unsafely polluted, the majority (46%) of residents indicated purchasing bottled water, followed by purchasing water filters (31%). However, 68% of the respondents were not currently paying for water filters or any other methods of improving water quality, suggesting that residents were not acting on concerns about water quality problems at the time of the survey. Of the respondents who had paid that year to improve water quality, the average amount paid was $237. When asked about water quality problems in the future, 74% of the respondents indicated that they did not expect problems in the next five years. The average length of residence was 27 years and the mean number of people residing in each household was three. Over a third of the respondents in the sample (37%) completed high school, while 18% had some college education, 16% completed college, and the small remainder either did not complete high school or had an advanced college degree. Responses about employment status indicated that 37% of the respondents were self-employed, 36% were retired and only 3% were unemployed. Of those employed and retired, the median income was between $20,000 and $29,999, with the majority (21%) falling into the $10,000 to $19,999 category. The majority of respondents were white (96%) and 73% of the respondents were male. The average age of respondent's was 57 years. Forty-four percent of all respondents identified themselves as farmers and on average each had 30 years of experience. The majority (59%) of the sample of southwest Louisiana farmers produced rice as the main commodity, with the second most common agricultural enterprise being livestock (24%). The average number of acres that each farmer had in production was 179 acres of leased land and 239 acres of owned land. The portion of farmers or spouses who held an offfarm job was 41%. In general, respondents indicated that Louisiana's environmental quality, including water, was good. Additionally, respondents indicated that recent efforts to improve environmental and water quality in Louisiana, in their respective parishes and in their own homes, were good. However, respondents were undecided about efforts to use government programs to improve water quality, efforts to collect fines from polluters, and efforts to develop programs to prevent water pollution. Almost a third of the respondents were undecided about the likelihood that they would be willing to pay for improvements in water quality in southwest Louisiana. However, a large percentage (24%) were at least somewhat likely to be willing to pay for the changes. Although 33% of the respondents were undecided about their attitude toward willingness to pay for changes in water quality, a large portion (30%) described the attitude as being good. Twenty-six percent of the respondents believed that their individual willingness to pay for changes in water quality would be somewhat likely to have an impact on actual changes in the water quality in Louisiana, in their respective parishes, or their personal supply. This result suggests that respondents have a fairly strong strength of belief about the outcome of their willingness to pay behavior.

438

THE JOURNAL OF SOCIO-ECONOMICS Vol. 27/No. 3/1998

Responses from the normative belief questions indicated that many (43%) believe that it is very likely that environmentalists, government officials, and water quality experts think respondents should be willing to pay for water quality. However, 25% of the respondents believe that family and friends would be somewhat likely to think residents should be willing to pay for water quality changes. Approximately 36% of the respondents were only somewhat likely to do what family and friends thought they should, indicating the motivation to comply. The majority (27%) of respondents were undecided with regard to how environmentalists and government officials thought respondents should behave. However, 30% of the residents indicated that they would be somewhat likely to do what water quality experts thought they should. Responses from the willingness to pay and willingness to accept questions revealed the typically large disparity between the two values. The open-ended, single noniterative bid question format was used to elicit these values, allowing individuals to reveal true values. The minimum value that respondents were willing to pay to improve water quality and willing to accept as compensation to allow water quality to decrease was zero. The maximum value respondents were willing to pay to improve water quality was $1,070, whereas the maximum value they were willing to accept as compensation to tolerate a decrease in water quality was $1,000,000. When asked to indicate why they had responded with a WTP value of zero dollars, most indicated that they felt constrained by income constraint. Only a small percentage (less than 7%) could be identified as protest bids. Rresidents were willing to pay more to improve overall environmental quality ($33) in Louisiana than to improve water quality ($29) in the state. However, individuals indicated a significantly higher willingness to pay value ($77) when asked about improving their own water quality. Respondents were willing to pay an average of $55 to maintain their current level of personal water quality, still higher than indicated for the state' s water quality. These responses suggest that individuals are willing to pay higher amounts for changes on a personal level rather than for Louisiana in general. The complementary WTA questions showed that respondents required a much larger amount ($3,124) to allow their current level of water quality to decrease than to maintain the current level of water quality ($46). One interpretation of this difference in magnitude is that individuals view a certain level of water quality as a right which should not have to be relinquished. These results are consistent with previous research which resulted in willingness to accept values much larger than willingness to pay values (Randall, 1987; Mitchell & Carson, 1989, Bishop & Heberlein, 1990).

Empirical Results In order to evaluate the contribution of socio-psychological variables to the explanatory power of economic behavioral models, attitudes and the subjective norm were included in economic models reflecting willingness to pay. Although

439

Willingness or Intention to Pay

Table 2. Empirical Results: Willingness to Pay for Changes in Individual Water Quality Variable Intercept TREATED FARMER OIL AGRIC PRIVATE KIDS INCOME AGE AGESQ FEMALE EDUC1 EDUC2 ATTITUDE SUBNORM

Parameter Estimates1 Model 12 Model 23 3.485 -1.509* (0.501) 0.030 (0.111 ) 0.373 (0.137) 0.562 (0.186) 0.480* (0.113) 0.065* (0.039) 0.516" (0.127) -0.026* (0.016) 0.011 * (0.005) 0.084 (0.149) -0.305* (0.161) 0.317* (0.139)

3,010 -1.669* (0.422) 0.289 (0.108) (). ]42 (0.132) 0. ~;88 (0.181 ) 0.470* (0.109) 0.074* (0042) 0431 * (0.124) -0.024* (0.011 ) 0015* (0003) 0.120 (0.144) -0.219" (0132) 0.272* (0 136) 0456* (0 114) 0443* (0.119)

Notes: 1Standarderror in parentheses 2 Model without attitude variablesAdjusted R2 Model 1 - . I ~Modelwith attitude variables * = Significantat 90% confidence level Adiusted R 2 Model 2 - 22

five different willingness-to-pay values were elicited in the survey, only two values were utilized in this empirical analysis, reflecting changes in an individual's personal water quality and state level changes in water quality. Consistent with previous research, these values reflect a relatively large standard deviation, a factor typically contributing to the low explanatory power of the cross-sectional models (Coursey, Hovis & Schulze, 1986). The economic model of willingness to pay was estimated with and without attitudinal variables, for individual as well as state level water quality. Both models were tested for collinearity using condition indexes, condition numbers, and variance proportions. Based on criteria specified by Belsley, Kuh, and Welsch (1980), multicollinearity was not detected as a source of bias by any of these diagnostics. Empirical results of the model are presented in Tables 2 and 3. The adjusted R 2 is

440

THE J O U R N A L OF S O C I O - E C O N O M I C S

Vol. 2 7 / N o . 3 / 1 9 9 8

Table 3. Empirical Results: Willingness to Pay for Changes in State Level Water Quality Parameter Estimates 1 Variable Intercept TREATED FARMER OIL AGRIC PRIVATE KIDS INCOME AGE AGESQ FEMALE EDUC1 EDUC2 ATTITUDE SUBNORM

Model 13 5.074 -1.709* (0.630) 0.078 (0.078) 0.560 (0.504) 1.869 (0.685) 0.706* (0.418) 0.043 (0.149) 1.308* (0.470) -0.216* (0.060) 0.020* (0.005) 0.634

Model 2 3 1.887 -1.369* (0.589) 0.157 (0.385) 0.464 (0.471) 0.905 (0.645) 0.661 * (0.390) 0.074 (0.139) 0.891 * (0.440) -0.177* (0.057) 0.010" (0.005) 0.435

(0.549)

(0.512)

-0.614" (0.313) 0.677 (0.515)

-0.282* (0.306) 0.345 (0.483) 3.101 * (0.407) 1.330* (0.425)

Notes: 1Standarderror in parentheses 2Modelwithout attitude variablesAdjustedR2 Model 1 =.08 3Modelwith attitudevariables * = Significant at 90%confidencelevelAdjustedR2 Model2 - .19

reported in order to compensate for an increase in the R 2 resulting from increasing the quantity of independent variables. Typical for these models, a relatively low percentage of the variation in willingness to pay was explained by each base model without attitudinal explanatory variables. The attitudinal models were estimated by adding attitudinal explanatory variables to the initial specification. Based on the f-test, both models are statistically significantly, suggesting a relatively good overall model fit. In addition, a number of explanatory variables are statistically significant with the hypothesized signs. As hypothesized, the attitudinal variables significantly enhanced the explanatory power of the willingness-to-pay models, but more so for the individual models than the state level models. This appears consistent with hypotheses suggesting that more salient attitudes more directly influence behavioral intentions (Mitchell

Willingness or Intention to Pay

441

& Carson, 1989). The attitude variable constructed according equation 1 is positive and statistically significant in each model. In addition, the subjective norm variable, representing an individual' s normative beliefs multiplied by their motivation to comply, is positive and significant in both models. Other statistically significant explanatory variables include the variable representing efforts to treat water quality (TREATED) through the purchase of filters, bottled water, or other actions. TREATED is negative, suggesting that individuals who are currently addressing their water quality are not willing to pay for additional improvements in water quality. Ownership of a private well (almost half the respondents), indicated by the 0-1 indicator variable PRIVATE, is positive and significant in both models. Ownership of a private source of drinking water is positively associated with a willingness-to-pay for improvements in water quality. The presence of young children (KIDS) in the respondent's household is positively and significantly associated with a willingness to pay for changes in water quality. Respondents' income is positively and significantly associated with WTP, suggesting that higher incomes are not only willing to pay, but able to pay for changes in water quality. The two variables representing age, AGE and AGESQ, are statistically significant with negative and positive signs respectively, supporting the hypothesis of a nonlinear relationship between an individual's age and willingness-to-pay for improvements in water quality over a life cycle. Education displays the hypothesized negative sign at the level of lower education (EDUC 1) and the hypothesized positive sign for higher level of education (EDUC2). Both education variables are statistically significant. Not statistically significant are explanatory variables indicating farming as an occupation (FARMER), gender (FEMALE), or variables identifying the oil industry (OIL) or agriculture (AGRIC) as primary sources of local water quality problems. SUMMARY AND CONCLUSIONS

Efforts to improve the descriptive or predictive ability of contingent valuation studies eliciting willingness-to-pay may benefit from considering willingness to pay as a behavioral intention. As a behavioral intention, the variation in WTP may be influenced by salient environmental attitudes which can be elicited for inclusion in nonmarket valuation. The Ajzen-Fishbein model of the attitude-behavior relationship is composed of three main components: beliefs, intention, and attitudes. Beliefs are the base from which attitudes, intentions, and behavior are eventually molded. In predicting or explaining behavioral intentions such as WTP, two types of beliefs are important: beliefs about the consequences of acting out the behavior, and normative beliefs, i.e., beliefs about how others think one should behave. Attitudes are a subjective judgement and are considered to be a function of beliefs about an object. While behavioral intentions are a joint function of subjective

442

THE JOURNAL OF SOCIO-ECONOMICS Vol. 27/No. 3/1998

norms and attitudes, they are the immediate determinants of actual behavior. Behavioral intentions are the component of the attitude-behavior relationship measured by willingness-to-pay. This research has used the attitude-behavior framework in an empirical analysis of willingness-to-pay for improvements in rural water quality. Using primary data collected by a mail survey of rural Louisiana residents, the variation in WTP for changes in rural water quality was analyzed with and without attitudinal explanatory factors. Two models were evaluated: WTP for changes in individual water quality and WTP for changes in state level water quality. Consistent with hypotheses suggesting that more salient attitudes more directly influence behavioral intentions, the attitudinal variables significantly enhanced the explanatory power of both willingness-to-pay models, but more so for the individual model than the state level model. If the attitude-behavior relationship presented here is thought to be relevant in the case of predictive or explanatory environmental behavior modeling, omission of appropriately specified attitudinal variables suggests a specification bias that in extreme circumstances can result in biased and inconsistent estimators. One result of this is that the usual confidence interval and hypothesis testing procedures are likely to give misleading conclusions about the statistical significance of the estimated parameters. While applied economists are encouraged to err on the side of parsimony in model specification, given the widespread acceptance of social-psychological attitude measures, it seems inappropriate for economists to simply relegate their influence to the error term. In conclusion, the existence of salient environmental attitudes among consumers of environmental goods and services suggests that researchers are remiss when omitting appropriately constructed and measured attitude variables in contingent valuation studies. An improved understanding of the attitude-behavior relationship may add much to our understanding of the intentions underlying individual' s willingness-to-pay.

REFERENCES Arcury, T.A. Johnson, T.P., & Scollay, S.J. (1987). Ecological worldview and environmental knowledge: the new environmental paradigm. Journal of Environmental Education, 17, 3540. Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50, 179-211. Ajzen, I. & Driver, B.L. (1992). Contingent value measurement: On the nature and meaning of willingness to pay. Journal of Consumer Psychology, 4, 297-316. Ajzen, I. & M. Fishbein, M. (1980). Understanding attitudes and predicting social behavior. Englewood Cliffs, New York: Prentice Hall. Ajzen, I. & Peterson, G.L. (1988). Contingent value measurement: The price of everything and the value of nothing? In G. L. Peterson, B.L. Driver, & R. Gregory (Eds.) Amenity Resource Val-

Willingness or Intention to Pay

443

uation: Integrating Economics with other Disciplines, (pp. 65-75). State College, PA: Venture. Allen, C., Machleit, K., & Kleine, S.S. (1992). A comparison of attitudes and emotions as predictors of behavior at diverse levels of behavioral experience. Journal of Consumer Research, /8, 493-503. Belsley, D., Kuh, E., & Welsch, R. Regression diagnostics, New York: Wiley, 1980. Bishop, R. & Heberlein, T. (1990). Economic valuation of natural resources: Issues Theoo', and applications. Boulder, CO: Westview Press. Coursey, D., Hovis, J., & Schulze, W. (1987). The disparity between willingness to accept and willingness to pay measures of value. Quarterly journal of economics. John Wiley and Sons, Inc. Dawes, R. M. (1972). Fundamentals of attitude measurement. New York: John Wiley and Sons. Dillman, D. (1978). Mail and telephone surveys. New York, NY: John Wiley & Sons, Inc. Dillman, D. (1991). The design and administration of mail surveys. Annual Review ~?(Sociology 17, 225-249. Fazio, R.H., Powell, M., & Herr, P. (1983). Toward a Process model of the attitude-behavior relation: Accessing one's attitude upon mere observation of the attitude object. Journal of Personali o" and Social Psychology, 18. Fazio, R.H., Powell, M.C., & C.J. Williams. (1989). The role of attitude accessibility in the attitudeto-behavior process. Journal of Consumer Research. 16, 280-8. Fishbein, M. (1967). A behavior theory approach to the relations between belief about an object and the attitude toward the object. In M. Fishbein (Ed.) Readings in attitude theory and measurement (pp. 389-400). New York: John Wiley and Sons, Inc. Heberlein, T.A. (1989). Attitudes and environmental management. Journal of Social Issues, 45, 3757. Henderson, M.E., Morris, L.L., & Fitz-Gibbon, C.T. (1987). How to measure attitudes. California: Sage Publications. Ladd, E.C. & Bowma]l. K. (1996). public opinion on the environment. Resources~br the Future, 124, 5-8. Lynne, G., Shonkwiler, J.S., & Rola, L. (1967). Attitudes and farmer conservation behavior. American Journal of Agricultural Economics, 70, 12-19. Lynne, G., Shonkwiler, J.S., & Wilson, M. (1988). Water permitting behavior under the 1972 Florida Water Resources Act. Land Economics, 67, 350-35 I. Manfredo, M., Yuan, S., & McGuire, F. (1992). The influence of attitude accessibility on attitudebehavior relationships: Implications for recreation research. Journal e~fLeisure Research. 24. 157-170. Millar. M.G. & Tesser, A. (1990). Attitues and behavior: The Cognitive-affective mismatch hypothesis. Advances in Consumer Research, 17, 86-89. Mitchell, R.C. & Carson, R.T. (1989). Using surveys to value public goods: The Contigent valuation method. Washington D.C. :Resources for the Future. Mueller, D.J. (1986). Measuring social attitudes. New York: Teachers College Press. Napier, T., & Forster, D.L. (1982). Farmer attitudes and behavior associated with soil erosion control. In H.G. Halcrow, E.O. Heady, & M i . Corner (Eds.) Soil conservation policies, institutions, and incentives, (pp. 137-150). Ankeny, IA: Soil Conservation Society of America. Nowak, P.J., & P.F. Korsching. (1983) Social and institutional factors affecting adoption and maintenance of agricultural BMP's. In F. Schaller & G. Bailey (Eds.) Agricultural management and water quali~, (pp. 349-373). Ames, IA: Iowa State University Press. Randall, A. (1987). Resource economies. New York: John Wiley and Sons. Summers, G.F. (1970). Attitude measurement. Chicago. IL: Rand McNally.

444

THE JOURNAL OF SOCIO-ECONOMICS Vol. 27/No. 3/1998

Thurstone, L.L. (1967). The measurement of social attitudes. In M. Fishbein (Ed.) Readings in attitude thory and measurement (pp. 14-25). New York: John Wiley and Sons. Wilson, T.D., Lisle, D.J., & Kraft, D. (1990). Effects of self-reflection on attitueds and consumer decisions. Advances in Consumer Research, 17, 44-47.