Economic Evaluation of an Investment in Medical Websites and Medical Web-Based Services

9 Willingness to Pay (WTP)

According to Rose et al. (2009), the application of such methods spans a wide scope from environmental science, marketing, political science, to health economics and transportation sector. Moreover they state that the usual setting of a Stated Choice experiment is to present participants with various scenarios (and their alternatives) that are different in a number of attributes and to ask them for their preference or to ask them for a measure of trade-off between two attributes. Such trade-off measure can be the Willingness to Pay. Johannenson (1996) states that the approach to use surveys in order to estimate the WTP over a selection of commodities, services or attributes is referred as “contingent analysis” and like the WTP approach it was first developed and applied in the environmental economics field.

Louviere and Islam (2008) state that “… the Willingness to Pay is a concept derived from welfare economics” and it can be referred as “… the marginal rate of substitution of particular attributes/levels for money”. In other words, this definition of Willingness to Pay (WTP) implies that Willingness to Pay is a method that can be used to address valuation issues and provide valid measures to be used in a concept of “welfarism” to evolve better services a social benefit. As it was afore-mentioned, Willingness of Pay can be regarded as an exchange value: i.e. what the public considers as value of a service that will be equal to the value of the benefits that the society will accrue from the implementation and the outcomes of that service.

I addition to this, it is stated that it is the main method to assess benefits in a Cost-Benefit analysis framework and does not only reflect the value of a change but also the utility that the actors of the change will have from its outcomes (Donaldson, 1999; Ryan et al. 1997). It is expected that higher WTP will reflect higher utility (ceteris paribus) and the Pareto criterion can be transformed in the case of WTP as “if the maximum amount that the gainers would be willing to pay for a change is greater than the minimum amount that the losers would be prepared to accept in compensation, then the change should go ahead, whether or not the compensation is paid” (Little, 1957; Donaldson, 1999).

According to Louviere and Islam (2008), the greater advantage of the WTP approach is that enables the analyst to compare different attributes that are normally measured in different, scales.

Luzar and Cosse (1998) in their research they suggest that the descriptive and predictive ability of the contingent valuation and the WTP can be enhanced if WTP is considered more as a behavioural indicator for the willingness of the people to accept, use and pay for a change, service or commodity. According to the Ajzen –Fishbein’s (1980) model of the attitude behaviour is constituted by:

• 
  beliefs
  
• 
  intentions
  
• 
  attitudes
  

Moreover, they supported the argument that except the fact that WTP can be influenced mainly by the beliefs of the participants concerning the consequences of their own choices and the belief of what kind of behaviour they expected by the others, their own behaviour and the attitudes, depending on the evaluation targets the WTP of the participants, might be influenced by:

• 
  The socio-economic characteristics of the participants
  
• 
  The age of the participants, since the participants, as grow up they on their own economic life cycle.
  
• 
  The family status of the participants, since the presence of children in the family made the respondents to report a higher WTP concerning issues that can affect the quality of life of the next generation, such as environmental issues or healthcare issues.
  
• 
  And the kind of participant’s occupation.
  

Yoder R. (1989), in his work endeavours to investigate the effect that a fee increase on the health services, curative and preventing, can have on the willingness and the ability pay of the patients. Yoder R. (1989) tries to compare the magnitude of the fee increase effect between the curative and preventing healthcare interventions, investigates how this fee increase affected governmental and missionary healthcare facilities through a case study in Swaziland. These findings have shown that in the Swaziland case the demand of preventing or curative healthcare interventions is not as inelastic as it was found by other studies around the world and he disproved the widely held views on the establishment of healthcare service patient fees “people are willing and able to pay for healthcare services” (De Ferranti, 1985; Chaplin and Earl, 2000) and “..in general the price of services does not matter, having a minimal if any effect on the decision to seek healthcare” (Leweis, 1985)

Yoder R. (1989) first of all analyses the methods of financing healthcare interventions distinguishing them into two categories: a) those which seek finance from various activities and sources excluding the financing directly from the patients/ users of these interventions and b) those, alternative to the first category financing methods, that use more the direct financing from the patients/ users of the healthcare services. The latter category of financing methods involves providing insurance and risk coverage programs to the patients, charging patients involved in healthcare interventions, decentralizing healthcare services, and finally using non-governmental resources in a more efficient and cost-effective way.

The results of his research have shown a rapid decrease of public attendance in governmental healthcare units and facilities, while there was an increase in the percentage of attendance at the mission facilities since the fees in mission’s facilities remained the same. By analysing in numbers the rapid decrease in the attendance in the healthcare facilities that were established by the state, Yoder suggested that this decrease indicates that for every 1% increase applied on the healthcare intervention fees utilization of the healthcare units tend to decrease by 0.32%. Moreover, according to Yoder observations, when the government increases the healthcare service fees for the public then the patients that find the increase unaffordable according to their money allowance or those that consider that their health issues are not serious and that they do not affect their life are more prominent to avoid attendance in the healthcare facilities. Furthermore, patients that tended to have multiple attendances in healthcare units, they will also reduce their visits to these units.

In addition to this, what is remarkable in Yoder’s research is that cases of not so serious or life threatening conditions such as myoskeletal conditions presented a lesser decline in average compared to more serious conditions, something that might can be explained if the annoyance of the symptoms for the patient/sufferer is taken into consideration.

Ryan and San Miquel (2000) state that as the popularity of methods like the WTP increases and is more and more used in research the necessity to validate this method increases also. They evaluated the consistency of the answers of the participants concerning their WTP in studies related to healthcare by examining the responses of the participants according to a predefined stated preference by them. The participants that answered giving a WTP much smaller for their preferred commodity, service or attribute or a large WTP for a less preferred choice, can be characterised as inconsistent. After the identification of the inconsistent answers, they followed Sen’s (1993) argument that there might be a rationality behind inconsistent answers that must be investigated, rather than to simply exclude these inconsistent answers. Their investigation for the rationality behind the inconsistent answers brought in light the issue that although patients prefer more conservative medical interventions, they perceived in some scenarios, as in the case of hysterectomy, these interventions to be more expensive. So they answered that they prefer a more conservative treatment, but when the question about their WTP risen, they answered in a cost-reduction base and that caused the inconsistency. Even if the respondents haven’t been provided with cost information, their pilot survey had shown that they are answering in a cost–reduction base. Their results were consistent to other studies that highlighted this cost reduction rationale of some participants, when they are asked for the WTP of a preferred and an alternative scenario (Schkade and Payne, 1994; Donaldson et al., 1997). In order to solve this problem the US National Oceanographic and Atmospheric Administration (NOAA) Panel on contingent valuation proposed a referendum style WTP questions where the participants are asked if they are keen on paying a specific amount of money for a commodity, service or attribute usually answering with a yes or no. Although this was proposed as a solution to the bias of cost-based questions this kind of bias may still result from this approaches other studies have shown (Diamond and Hausman, 1994).

In order to solve this problem, other studies proposed the use the marginal WTP approach in which the participants are asked for their preference over a commodity, service or attribute and then are asked to state their maximum WTP (Donaldson et al., 1997). On the other hand, Ryan and Hughes (1997) proposed as an alternative methodology to solve the problem of cost-based answers by using a conjoint analysis discrete choice model. In this setting the participants are presented with various discrete-choice models with differences in attributes and at first place they state their preferences. Then by the use of a deference model and taking also into account the cost of an attribute, Ryan and Hughes (1997) state that the analysts can estimate marginal WTP values of the change from one preference to the other.

Johannesson (1996) states that there are two main approaches used to measure the WTP of the changes in healthcare interventions. The first requires the researcher/analyst to observe the actual choices of the patients concerning the health provided healthcare interventions, risks, costs and potential benefits. The second approach is the use of surveys (contingent analysis) to estimate an expected behaviour to be expressed in WTP.

Based on the works of Gafni (1991) and O’Brien (1994), Johannesson (1996) argues that WTP approach is mainly used in the healthcare field ex-ante valuing services from an insurance perspective and a more predictive point of view, it might be easier and produce more valid results if WTP is assessed as an expected WTP among patients that for sure need a particular kind of healthcare intervention and/or its alternatives. In a CBA the benefits are a priori defined as the willingness to pay of the participants, considering this willingness of pay as automatic transformations of the benefits that a specific commodity, service or attribute offers to them. The main challenge of CBA in health is to monetize the benefits of a healthcare intervention or change and not of various tradable commodities and services and according to Johannesson WTP have served well towards that direction and can offer even more abilities to the researcher/analysts in the future.

Investigating the relationship between the ex-ante and expected Willingness to Pay, Johannesson applies both approaches in various scenarios before and after a change in a healthcare intervention, emphasizing the fact that if the utility is supposed to be constant before the change, the analysts should investigate the WTP for the improvement in healthcare while after the intervention they investigate the WTP to avoid any kind of deterioration or unwanted effects. He concludes that expected WTP can be taken from actual observed WTP if the ex-ante WTP is multiplied by the probability of the presented in the scenario healthcare state to happen. Both are monetary measures of the same changes in the utility function and their difference is the use of different marginal utility income to monetise the change and their actual relationship depends on the variations of that income with the actual income and health status of the participants. In addition to this, Johannesson presents two examples in which the first describes participants risk neutral with respect to income and constant utility income, irrespectively of what the health status of the participants is, having as a result the ex-ante and the expected WTP always to be equal, and a second example, with only deference compared to the first a risk averse relationship of the participants with the income, then the expected WTP will be the lower bound of the ex-ante WTP. On the other hand, in the case that the marginal utility income increases with respect to better healthcare states and does not remain constant, then the expected WTP might overcome the actual WTP in the case that the intervention does not restore the full health of the patient, because if restores the full health then again the expected WTP is a lower bound of the actual WTP.

Olsen et al. (2004a) in a similar view with Johannesson (1996), after examining the ex-ante and ex post approaches in estimation of WTP, state that there is no right or wrong approach and the choice depends on the case which is under investigation, e.g. in a case that the WTP for a non-emergency healthcare intervention or in general in a case of elective care then ex ante approach can be a more reasonable approach to be used. Moreover they examine the difference of the participant’s stated WTP in an insurance premium and a tax contribution concept. They believe that that respondents can be influenced and state a different WTP when the survey question/scenario refers to a insurance premium or a tax contribution to the society. Their results have shown that the public interprets differently the WTP questions and they state greater WTP in the version that offers more to the community.

After examining the attitude and the WTP of women from various ethnical backgrounds towards a specific female examination (mammography), Wagner et al. (2001) have shown that WTP can be influenced also by the ethnicity/race of the participants with the non-Hispanic white, Latino and African American women to be reluctant to pay more for predictive examinations in contrast to the Chinese and Filipino women.

Olsen et al. (2004b) examine the existence of scope insensitivity in WTP surveys in healthcare. The state that there are continuously empirical evidences that participants’ WTP value for a programme is not influenced by the magnitude of the outcome that this program will have. This is called scope insensitivity and is defined as “…..the finding that respondents’ WTP values are not sensitive to differences in the number of units produced” (Olsen et al., 2004b). This comes in contrast to the expected consumer behaviour that is supposedly to look for the greater outcome according to the neo-classical theory for consumer behaviour. Especially as long as this method is used by decision makers in the healthcare sector it is questioned how it can assist them in source allocations using evaluation methods sensitive to the magnitude of the intervention’s outcome. Olsen et al. (2004b) state that this might happen because in a WTP survey which presents scenarios with lot of attributes that enhance participant’s utility then participants are not so strictly cognitively focused on the outcomes of the program/intervention. As healthcare programmes and interventions are complex there is a similar issue of cognitive overload that can be avoided only by limited description to the participants, something that is not considered to present valid results. An alternative rational behind the scope insensitivity is that the WTP value is only indirectly attached to the outcomes of a particular programme/intervention and participants use the WTP as an approval value for the programme/intervention, rather than an evaluation for its outcomes. Although in their research there was an increase in mean and media WTP as the outcomes of a health programme doubled, the participant were sticking to their initial statement.

Protiere et al. (2004) investigated the impact that additional, and not necessarily strictly health-related information can have on the WTP that participants express for multiple healthcare programmes. Based on psychological literature and literature concerning the consumer options they examined the well-established fact that any additional piece of information provided to the participant in a transaction can influence the participant’s decision in a great way. Their results confirmed that distribution of more information can lead to different WTP results but and the mean WTP they estimated was increasing with the presentation of more neutral information in contrast to past studies that reported that the mean WTP was lower for the more informed participates suggesting a negative relationship between the information provided and the mean WTP.

Donaldson C. and Shackley P. (2002), aiming to investigate how useful can be the use of the WTP values to assess the priority of the healthcare programmes, showed that there is an inconsistency between the ranking that the participants preferred and the allocation of their higher WTP, something that they couldn’t manage to avoid even by using the marginal approach. These results raise questions about the validity of the use of this method to assess medical programs priorities.

Finally, Cookson (2003) states the WTP approach for assessing the benefits of healthcare interventions tend to be insensitive towards the actual benefits of the intervention and inflate the value used in the evaluation process.

9.1Willingness to Pay vs. Human Capital Approach vs. Quality Adjusted Life Years

Comparing the WTP approach with the human capital approach, Johannesson (1996) stated that in the healthcare field, in the early days, economic evaluations were based on the human-capital approach, according to which the value of the improvements in healthcare interventions is defined as “the decreased in consumption in healthcare and the increased production”. Soon after strong criticism this approach was replaced by Cost-Benefit, Cost- Effectiveness and Cost-Utility analysis, which are based more on the Willingness to Pay concept. Johannesson (1996) has shown that the human-capital approach used to estimate the value of the improvement in healthcare programmes/services or interventions does not necessarily indicate the WTP of the public towards that improvement. Although the human capital approach does not reflect the individual WTP, Johannesson states that in systems with public or private insurance that is not taken into account as an estimate of the expected costs, this approach can produce a valid estimate of these costs, since it uses the difference between the increase in consumption and the increase in production to estimate the increased survival from the improvement in healthcare.

With the human capital approach heavily criticised as an evaluation approach, QALYs and WTP became the most preferred approaches to assess the value of the healthcare outcomes. Bala et al. (1998) compared these two approaches. The QALYs strives to transform the outcomes of the improvement in the healthcare interventions into a specific duration and quality of life (Bala et al., 1998; Hammit, 2002). As Ryan et al. (1997) state, QALYs and Cost Effectiveness analysis is mainly used of clinical outcomes, while WTP can be used for assessing value from the health process itself rather than only from its outcomes. On the other hand, WTP transforms the outcomes of the improvement in healthcare interventions into a universal monetary value much easier to be used by the evaluation actors. While the QALYs approach first estimates various weights for different healthcare states and then combines these weights in order to end up with a final value for the healthcare intervention, the WTP approach estimates directly the final value by asking the preference of the patients/users. Furthermore, QALYs follows specific restrictions and assumptions in order to be considered as utility “measures” (Liljas et al., 2000), while WTP is an assumption-free approach although the validity of its results are questionable concerning how accurately it estimates the real WTP of the public. QALYs in order to present an accurate impression of the utility they must follow the conditions below (Hammit, 2002):

• 
  There must be mutual utility independence, which refers to the fact that “preferences between lotteries on health states do not depend on the remaining years of life, keeping the lifespan constant or preferences between lotteries on lifespan do not depend on health state” (Hammit J., 2002).
  
• 
  The percentage of the rest of the duration of his/her life that a patient is willing to sacrifice to move from a healthcare state to another does not depend on the remaining lifespan
  
• 
  If a healthcare state is constant, then patients prefer the lottery on longevity that enlarges their lifespan.
  
• 
  The choice that a patient make in a specific period expressing their preference does not depend on their healthcare status in other periods.
  

After comparing the two approaches using various scenarios, Bala et al. (1998) concluded that WTP and QALYs results are not significantly correlated, something that in simple words means that the two approaches produce different results and they must be considered equivalent approaches without further investigation. Moreover, Wagner et al. (2001) stated that QALYs are not as sensitive as WTP in measuring value in cases of acute illnesses and this approach is more a measure of effectiveness rather than a measure of exact benefit. Finally, Hammit (2002) states that QALYs are estimating value following the assumption that the probability of a healthcare outcome is the main factor that influences the preference of the patients/ users with respect to the health risks and health interventions. Furthermore Hammit (2002) emphasizes also the fact that QALYs estimations present lesser variance that the WTP outcomes and thus supports also the opinion that these methods cannot be considered that produce equal results and the choice of the approach to be used is based in the end on the personal preferences of the analyst/ researcher.

9.2Biases and Other Issues Encountered While Using WTP Approach

Donaldson (1999) suggests that WTP cannot be used to enhance comparisons of utility between people with different levels of income and ability to pay. In order to solve this issue, Donaldson (1999) suggests that one solution is the “do nothing” solution assuming that this does not affect a social valuation. Then he presents Little’s (1957) “decision making approach” that is based on two conditions to identify the appearance of a welfare gain:

• 
  The gainers are able to compensate the losers without risking their welfare status
  
• 
  The redistribution of income/benefits/resources is assumed to be “good”
  

Another approach that Donaldson (1999) proposes is the use of distributional weights derived, for example, from marginal taxation rate to convert utility to monetary values.

Frew et al. (2004) pinpoint the fact that one of the first problems that the researchers have to face when dealing with WTP is to ensure that the participants understand how the method works and the nature of the problem they are asked to solve. This can be solved by piloting a questionnaire and ask participants how difficult they had found it or ask them to compare the good/service under evaluation to a recent purchase they made of equal value, in order to achieve more actual-behaviour valuation.

Secondly, they analysed the problems of the starting point bias referring to the fact that the valuation is heavily influenced by the first/initial value that is proposed to the participants, and the interviewer bias which refers to the phenomenon where the participants tend to choose an answer that they think will please the interviewer. These according to them are issues often met while using the bidding game method to obtain the participants’ maximum WTP.

Ryan and San Miquel (2000) state that, as the popularity of methods such as the WTP increases and is more and more used in studies, the necessity to validate this method increases also. They were particularly concerned about the cost-based answers in a WTP question that can reflect the cost of the service rather than the maximum willingness to pay of the participants. This can cause an underestimated valuation if WTP is applied in the context of a Cost-Benefit analysis framework. They suggested that one way to solve the problem is to focus the attention of the participants on the benefits that the evaluation target has by firstly asking the participants to rank their choices and then to state their maximum WTP for their less preferred choice, or by using a conjoint analysis approach by presenting with various different in aspect of attributes commodities/discreet choices and asking participants to state their preference.

In the same wavelength as Johannesson’s (1996), Blumenschein et al. (2001) attempts to assess the relationship between the hypothetical and the actual WTP in the healthcare field. Since WTP is based on hypothetical questions asked to the participants there is a hypothetical bias associated with this practice (Liljas et al., 2000). They used contingent analysis with a dichotomous (yes/no) choice approach and only one price for the participant to choose if they accept it or not. They attempted to investigate the issue that has raised numerous debates between researchers and refers to the question if the estimated WTP mirrors the real behaviour of the participants. Moreover, they state that the dichotomous choice contingent analysis tends to produce overestimated results and this is usually referred as hypothetical bias (Liljas et al., 2000). In order to avoid that kind of bias that is rooted from the hypothetical setting of the survey

Cumming et al. (1999) proposed an initial small talk between the interviewer and the participant about hypothetical bias in order to make respondents be more careful that their answers express their true WTP. Another method to solve the hypothetical bias is to distinguish between the True and False yes answers by asking a follow up question concerning the level of certainty of the participant for the stated value and taking the definitely sure yes answers as the true yes (Johannesson et al., 1998; Bluumeschein, 1998).

Onwujekwe et al. (2005) opine that the main cause of divergence between stated WTP and actual behaviour of the participants is the presence of any kind of bias. One method that they suggested for the elimination of this phenomenon is the presentation of more realistic scenarios and paradigms to the participants. They also argue that there might be natural reasons for this divergence, because as time passes, consumers can change their opinion and accustom it to the new conditions of the environment in which they live in.

Blamley et al. (1999) suggested also another way to deal with the hypothetical bias phenomenon by having non-monetary choices to distinguish the participants that just support the cause and then with follow-up questions to investigate if they would be really willing to pay for a service or a commodity.

Foreit et al. (2003) stated that in their surveys they had to deal with the “yea-saying” phenomenon. This phenomenon occurs when the participants stated a lower WTP than they had accepted during a bidding game interview or open-ended payment card setting.

Finally, Stewart et al. (2002) examined if the ordering of the programmes presented to the participants for evaluation can influence their WTP decision. Their findings suggest that there is a percentage of ranking bias that can affect the WTP measurement, since the most participants feel free from their “obligations” after evaluating the first programme.

Download 1.95 Mb.

Share with your friends: