CORRESPONDENCE
Pharmacoeconomics 1998 May; 13 (5 Pt 1): 566-569 1170-7690/98/0006-0563/$03.50/0 © Adis International Limited. All rights reserved.
The autho r ’s reply: How to measure or estimate indirect costs (i.e. individuals’ loss of production of goods and services due to their illness) in economic evaluations has recently become a hot research topic within health economics. Three different approaches seem to be the predominant in the literature; the human capital–cost approach[1] (defined as the lost gross income during the time absent from work), the friction-cost approach[2,3] (which implies that the long term indirect costs are defined as the cost for replacing a sick worker) and the QALY approach[4,5] (which states that a large proportion of the indirect costs are included in the patient’s reported QALY weight). In a recent article[6] in this journal, I argued that neither of these 2 latter approaches are generally consistent within the framework of economic theory from a societal perspective. The reason for this was that the friction-cost approach was shown to exclude many parts of the indirect cost components as well as resting on very strong assumptions about individuals’ valuation of leisure and about the labour market, whereas in the case of the QALY approach, parts of the indirect costs (such as tax and insurance) will not always be included in individuals’ QALY weights. The human capital–cost approach was also shown not to include all indirect cost components but a simple way of how to correct this was suggested. In their comments regarding my article, Brouwer and Koopmanschap argue for the use of the friction-cost approach and state that my description of the friction-cost approach was partly incorrect and that the results of my comparison to the human capital–cost approach, at occasions, were flawed. In response to Brouwer and Koopmanschap’s comments, I will examine their critique as well as suggest what I think is the way forward for the friction-cost approach in the measurement of indirect costs. However, my main criticisms remain: the friction-cost approach does not have a foundation in economic theory, the utility changes of individuals not directly affected by the healthcare intervention would have to be included in the economic evaluation and QALYs have to be used
as the effectiveness variable in order to correctly use this approach. Furthermore, their statement that the human capital–cost approach cannot be used to value production losses from a societal viewpoint is incorrect. The rest of this reply is structured as follows: section 1 contains an examination of the issues on the friction-cost approach and its comparison to the human capital–cost approach brought forward by Brouwer and Koopmanschap, section 2 further discusses the categorisation of indirect costs and section 3 holds some concluding remarks. 1. Critique of the Friction-Cost Approach One of the criticisms of the friction-cost approach in my previous article[6] was that it lacks a theoretical foundation. When analysing complex situations, as economic problems often tend to be, it is very useful to start with a theoretical model of the research question. If the research question is analysed by performing an economic evaluation, a theoretical model might even be necessary to assure the researcher that the analysis has a foundation in economic theory. Brouwer and Koopmanschap are, however, incorrect in their statement that I merely argue for the use of the neoclassical theory of economics. My critique was rather aimed at the fact that the friction-cost approach has not been given a theoretical foundation whatsoever, whether it is the neoclassical theory of economics or some other economic model. If the neoclassical theory of economics (which the human capital–cost approach is based on) is believed to be incorrect with regard to the labour market, it would, in my view, be logical to start with developing a new, consistent economic model of how the labour market works (or to use some of the alternative economic models that take market imperfections into account). Howitt,[7] for example, has developed a theoretical model of the labour market in which search and recruitment costs (both of which are important aspects in the friction-cost approach) are fundamental parts, and the book by Blanchard and Fisher[8] holds a number of competing models of the labour
567
market – all with a foundation in economic theory. From an alternative economic model, an estimation technique of how to calculate the indirect costs could be derived, rather than to somewhat arbitrarily try to correct the neoclassical theory of economics with empirical estimation methods as the friction-cost approach seems to do.
Correspondence
celling of less important work, I never stated that these costs were excluded in their empirical work. However, the more important question is whether they should or should not be included in the friction-cost approach, and according to Koopmanschap and colleagues’ earlier methodological work, it seems as if it is the latter that is the suggested approach.[2,3]
1.1 Short Term Absence 1.2 Long Term Absence
With regards to short term absences from work, Brouwer and Koopmanschap argue that firms do not behave according to neoclassical economic theory and that the value of the employees’ marginal productivity is lower than their marginal cost (i.e. their marginal wage). They suggest that only 80% of the costs for absent individuals should be accounted for since that would better resemble the average value of the production lost (they thereby, implicitly, seem to assume that the employees are paid according to their average productivity rather than their marginal productivity). However, I believe that, at least until a better theory is suggested, neoclassical economic theory could be used as a ‘decent’ approximation (and that firms probably act as expected profit maximisers). According to this theory, labour is paid in accordance with its marginal productivity. Therefore, a worker’s hourly, constant wage rate equals the value of his or her production in the last hour of work. However, due to diminishing returns to labour (i.e. a negative elasticity of time vs output), the value of a worker’s production is greater than his or her constant wage rate for the hours of work before the last hour. Hence, for a nonmarginal healthcare intervention, the average value added per employee might be larger than his or her marginal productivity and the indirect costs would, in this case, be underestimated. As long as we study the effects of a marginal healthcare intervention (which is usually assumed – see Sugden and Williams[9] for a discussion), we do not need to take potential differences between average and marginal production losses into account. Regarding the costs for internal labour reserves, making up for lost work after illness and the can© Adis International Limited. All rights reserved.
For long term absences, I believe that the friction-cost approach, if corrected for as suggested in my article[6] (e.g. multiple friction periods and differences in utility of leisure between the absent employee and the former unemployed individual), might conceptually have something to offer to the human capital–cost approach (i.e. some substitution of labour will naturally occur). However, my suggested corrections might substantially increase the indirect costs, as compared with the original friction-cost estimates, and might also be difficult to perform (see section 1.3). Taking these aspects into account might result in no significant differences between the 2 approaches. Furthermore, the statement that the human capital–cost approach only represents individual behaviour and is not intended to be used from a societal perspective is very unclear since society is assumed to be the aggregated number of individuals living in it, according to the neoclassical economic theory. If the absent employee is not replaced by an unemployed individual, a series of friction periods must occur (still making the assumption that there would be an unemployed individual at the end of the series of friction periods, which might not always be the case). Brouwer and Koopmanschap are correct in stating that in the labour market, it is often the case that the employee has to inform the employer when he or she is leaving the firm within a certain time period before it actually happens. However, sometimes the employee also has to leave the firm the very same day that he or she resigns, but even if this is not the case, it would only result in shorter friction periods and the series would still remain. This series of friction periods Pharmacoeconomics 1998 May; 13 (5 Pt 1)
Correspondence
will of course not be indefinitely long, but would not necessarily have to be nonproportional to the length of the series, contrary to what Brouwer and Koopmanschap state. The question of retirement (which, from a theoretical perspective, could be seen as a change from a paid to an unpaid job) and mortality is somewhat more difficult, and readers who are interested can refer to the article by Meltzer.[10] 1.3 Leisure Time
I agree with Brouwer and Koopmanschap that differences in the value of leisure time can be incorporated into the QALY weight (which is also related to some of the issues they bring up in their discussion on long term absences from work, disability and mortality). It is important to note, however, that the lost utility of leisure for the unknown unemployed individual at the end of the series of friction periods will not be directly affected by the healthcare intervention (since he or she is not ill) and will therefore usually not be included in the economic evaluation. This is, thus, another reason why the long term indirect costs will be underestimated with the friction-cost approach. Also, this means that it is implicitly assumed that QALYs are used as the effectiveness measure when calculating the indirect costs with the friction-cost approach, something that must be kept in mind. In a way, this assumes a perfect market for health-economic evaluations where everyone is using QALYs as the effectiveness measure, something that has been proven far from always being the case.[11] 1.4 Direct Labour Costs
Assuming that the opportunity cost of labour for direct healthcare costs is close to zero is indeed quite unrealistic and was merely used as an extreme example. However, using the friction-cost approach in an economic evaluation does imply that we would have to make rather strong assumptions on the possibility of substitution of labour, the number and duration of the friction periods, and the utility of leisure of the unknown unemployed individual at the end of the series of friction © Adis International Limited. All rights reserved.
568
periods. This would indirectly affect the way that the part of the direct costs that consists of labour costs are being viewed (see Johannesson and Karlsson[12]). 2. Categorisation of Costs The categorisation of indirect costs can naturally be done in many ways as long as all indirect cost components are included in one of the headings. Brouwer and Koopmanschap further suggest that absence from unpaid work can be valued by the wage rate of a professional housekeeper. This issue of how to value paid and unpaid work in case there is no information available on the affected individuals’ gross income or valuation of leisure time is, however, something that I just briefly mentioned in my article.[6] Using the wage rate of a professional housekeeper for absence from unpaid work is again somewhat arbitrary, but might be an acceptable approximation. As for the indirect costs accrued by family or friends, it is true that neither the human capital–cost approach nor the frictioncost approach includes these costs, but when using the human capital–cost approach, I do stress that they should be included in order for this approach to be theoretically correct. How to value these costs is indeed difficult and is an important issue for future research. 3. Conclusions I agree with Brouwer and Koopmanschap that neither the human capital–cost approach nor the friction-cost approach are optimal for calculating indirect costs. However, I do believe that the existence of a theoretical foundation (and not necessarily neoclassical economic theory) of the chosen estimation method of how to calculate the costs and effects of a healthcare intervention is of great importance when performing an economic evaluation from a societal perspective. The human capital– cost approach has a foundation in neoclassical economic theory which might not reflect the changing labour market in all aspects, but might be an acceptable approximation. Furthermore, some additional indirect cost components must be added (as Pharmacoeconomics 1998 May; 13 (5 Pt 1)
569
mentioned earlier) for this approach to be theoretically correct. Some of the limitations with the friction-cost approach are that: (i) it implicitly assumes that QALYs are used as the effectiveness measure (which often is not is the case); (ii) it rests on strong assumptions of the labour market; (iii) it has to include changes in utility for individuals not directly affected by the healthcare intervention; and (iv) it lacks a theoretical foundation. As it is currently defined, I believe that this approach will underestimate the indirect costs (especially long term cost). Even though the friction-cost approach attempts to give some insights on how the labour market currently works, it needs to be embedded in an economic theory in order to be regarded as a competing method to the traditional human capital–cost approach of how to estimate indirect costs in economic evaluations from a societal perspective. Bengt Liljas Department of Economics Lund University Lund, Sweden and Astra USA Inc.
© Adis International Limited. All rights reserved.
Correspondence
Westborough Massachusetts, USA References 1. Weisbrod BA. The valuation of human capital. J Pol Econ 1961; 69: 425-36 2. Koopmanschap MA, Rutten FFH, van Ineveld BM, et al. The friction cost method for measuring the indirect costs of disease. J Health Econ 1995; 14: 171-89 3. Koopmanschap MA, Rutten FFH. A practical guide for calculating indirect costs of disease. Pharmacoeconomics 1997; 10 (5): 460-6 4. Siegel JE, Torrance GW, Russell LB, et al. Guidelines for pharmacoeconomic studies. Pharmacoeconomics 1997; 11 (2): 159-68 5. Gold MR, Siegel JE, Russell LB, et al, editors. Cost-effectiveness in health and medicine. New York: Oxford University Press, 1996 6. Liljas B. How to calculate indirect costs in economic evaluations. Pharmacoeconomics 1998; 13 (1): 1-7 7. Howitt P. Business cycles with costly search and recruiting. Q J Econ 1988; 103: 147-66 8. Blanchard OJ, Fisher S. Lectures on macroeconomics. Cambridge: The MIT Press, 1989 9. Sugden R, Williams A. The principles of practical cost-benefit analysis. Oxford: Oxford University Press, 1978 10. Meltzer D. Accounting for future costs in medical cost-effectiveness analysis. J Health Econ 1997; 16: 33-64 11. Backhouse ME, Backhouse RJ, Edey SA. Economic evaluation bibliography. Health Econ 1992; 1 Suppl.: 3-7 12. Johannesson M, Karlsson G. The friction cost method: a comment. J Health Econ 1997; 16: 249-55
Pharmacoeconomics 1998 May; 13 (5 Pt 1)
