5. Social cost of taxation
5.3 ESTIMATES OF THE MCF
5.4.4 MCF and Case Management Programs
A question that is often raised in the context of mental health programs is whether to include maintenance costs (food, housing, clothing, transporta-tion, etc.) as part of the evaluation. The literature includes examples where all of the maintenance costs are treated as costs – see Rice (1992) – as well
Social cost of taxation 131
Table 5.6: Weighted costs of systems of psychiatric care in Massachusetts
Cost source Boston Central Western
Unweighted costs
State funded $3421 $8415 $1762
Medicaid $10 542 $9742 $12 235
Total costs (all sources) $13 963 $18 157 $13 997 State allocation of total costs: CS $8692 $13 286 $7880 Federal allocation of total costs: CF $5271 $4871 $6118 Total costs (all sources) $13 963 $18 157 $13 997 Weighted costs
(MCFS) CS(MCFF) CF $17 149 $22 247 $17 215
Source: Brent (2000b)
as examples where none of the maintenance costs are considered relevant, as in Clark et al. (1994).
Maintenance costs are center-stage in attempts to move patients out of inpatient psychiatric care because the main objective itself is to bring about independent living in the community. Some of these maintenance costs are financed privately by the families of the patients and the rest come from government sources. Jerrell and Hu (1989) carried out a detailed cost-effectiveness analysis of three case management programs in California designed to substitute community costs (what they called ‘support costs’) for inpatient hospital costs (what they called ‘intensive costs’). The support costs consisted of medical, legal, family and other.
A case management program in this context refers to interventions that:
(1) integrate all the medical and supportive services that psychiatric patients would require in the community, and (2) ensure continuity of care.
The three case management programs considered differed greatly in the level of support they provided for the patients. The three programs were:
1. A Program for Assertive Community Treatment (PACT) that had a multidisciplinary team who met daily to report on their cases. The average caseload was 15–20 to 1. Most services were provided outside of the office setting;
2. A ‘Clinical Team’ program that had mainly social workers who acted as generalists. The caseloads were high, varying from 35 per therapist to 40–45 when the patient was thought stabilized. Teams met weekly and services were provided primarily in a clinic / office setting; and 3. An ‘Intensive Broker’ program whereby the case managers provided
the link (i.e., were brokers) between the clients and the services they required. What made the program ‘intensive’ was the fact that the staff provided a relationship with the patients and helped with socialization, unlike most broker models that supplied linkage-only services. Case managers were para-professionals, operating largely in field-based set-tings, meeting weekly with their clinical supervisors, and having case-loads of 18 to 1.
The data in the Jerrell and Hu study were converted by Brent (2000b) into a cost–benefit framework by using the standard CM assumption that a benefit was a negative cost. The benefits then of the case management pro-grams were the inpatient hospital costs that were avoided by incurring the community costs (including the case management costs).
The issue then was how to treat all the transfers associated with the case management programs. Brent split the transfers into two: private (RP) and public (RG). The private transfers (e.g., family costs) can initially be
132 CM and CBA
assumed to have no net impact when the beneficiaries and the donors are considered to be the same group (the family). The public transfers were treated differently according to the beneficiaries and financers. The public source of the finances meant that the MCF would be applied. So $1 of public transfers were assigned a cost of $ (MCF)1.
The recipients of the public transfer were the psychiatric patients. $1 of their income was judged to have a social value greater than $1 because their income (if any) was way below the average and they were regarded as a socially deserving group. As will be explained fully in Chapter 13, the premium given to the income of the beneficiaries (the patients) can be cap-tured in a CBA by a money income distribution weight am, which typically has a value greater than 1. The benefit of $1 transferred to the patients then became $ (am)1. The net social impact of each dollar transferred was deter-mined by the difference: am⫺ MCF. Brent derived a value of 2.28 as the dis-tribution weight and he chose a MCF⫽1.3 (the upper limit of the Californian MCFs whose average was given in Table 5.2 as 1.206). The public transfers therefore had a positive impact of 0.98 per dollar transferred, i.e., a dollar transferred amounted almost exactly to a dollar of benefits.
The resulting costs, benefits and revenues of the three case management programs are presented in Table 5.7. The top part of the table has all the CBA ingredients separately. The unweighted net benefits correspond to a traditional CBA where all transfers are ignored (and so are distributional issues and the excess burden caused by public funding). The PACT program was clearly the most beneficial (it was also the most cost-effective program in the Jerrell and Hu study).
In order that we can appreciate the independent and joint impacts of the MCF and the distribution weight, three versions of the weighted net benefits are given in Table 5.7 (based on Brent’s Table 2 and weight set 3 of Table 3 – see section 13.4.1 below). Note that the distribution weight, when it applies, is not only attached to the public transfers; it also magnifies the benefits (the sums spent on the programs are on behalf of the patients).
Similarly, the MCF attaches to the (government financed) costs as well as to the public transfers.
With just the MCF included, the absolute size of the numbers is smaller than for the unweighted amounts. This is because the MCF is attached to the costs, making weighted costs larger – along the lines of criterion (5.5) – and hence the net benefits are lower. When the distribution weight is applied, the benefits are multiplied by 2.28. Benefit magnitudes dominate cost magnitudes, so the absolute size of the net benefits more than doubles.
When both the MCF and distribution weights are incorporated, amounts still double the unweighted values due to the larger size of the distribution weight compared to the MCF figure.
Social cost of taxation 133
No matter how the absolute numbers are transformed, the result is that the relative sizes of the net benefits are not altered. The PACT program is the most beneficial irrespective of which weighting scheme is adopted. This is simply because the PACT program has the highest gross benefits B, as well as the highest net benefits, and so it has the most weighted benefits.
This study also uncovered an additional dimension of transfers. Apart from the excess burden and income distribution effects of transfers, there is a third way that transfers impact an economic evaluation. This is via a ‘pro-ductivity’ effect. The transfers themselves cause benefits and costs to be different. For the various case management programs, the transfers were a necessary ingredient to obtain the hospital cost savings. In fact, a dollar of public funds RGtransferred had a higher correlation with declining hospi-tal costs than even the expenditures on case management programs.
Typically, for each $1 of RG, the hospital (intensive) costs, which form the benefits, fell by $1.60.
5.5 FINAL SECTION
We close with the problems, summary and looking ahead sections.
5.5.1 Problems
In the theory section we explained how the excess burden and the MCF depended on the size of the elasticities. There is a second factor that is
134 CM and CBA
Table 5.7: Evaluation of case management programs
CBA category Intensive Clinical PACT
broker team adaption
Benefits: B 14 560 15 157 16 437
Costs: C 2326 2670 2769
Private transfers: RP ⫺45 ⫺123 ⫺141
Public transfers: RG 295 248 616
Unweighted net benefits: B⫺C 12 200 12 487 13 668 Weighted net benefits:
With MCF⫽1.3 11 448 11 612 $12 653
With distribution weight of 2.28 31 248 32 205 $35 496 With MCF⫽1.3 and distribution weight of 2.28 30 462 31 331 $34 481
Source: Based on Brent (2002b)
important that we have so far ignored, that public finance specialists emphasize. The neglected factor is the size of the tax rate itself. Typically, see for example, Browning and Browning (1987), where an equation is derived that quantifies the effect of the tax rate. Instead, in the following problems, we will establish the importance of the tax rate effect by re-examining the Wildasin case study, which provided implicit estimates of the MCF that varied with the tax rate.
Table 5.8 is based on a subset of the information in Table 5.3. The focus is just on population projections using alternative III and the case 2 set of assumptions. In column (2) of Table 5.8, the tax costs of Table 5.3 are reproduced and relabeled ‘tax rates’. This is because the tax cost is per dollar transferred and so is the equivalent of a tax rate. Column (3) of Table 5.8 is derived from Table 5.3 by dividing its column (5) by column (2).
Recall in the discussion of Table 5.3 that one column is C and the other is C.MCF; so dividing the latter by the former would derive an estimate of the MCF. Column (4) of Table 5.8 has to be filled in as one of the problems.
Table 5.8: The relation between the excess burden and the level of the tax rate
Year Tax rate MCF Marginal excess
burden
(1) (2) (3) (4)
2000 0.219 1.416
2025 0.376 1.941
2050 0.514 3.794
2060 0.577 6.153
Source: Based on Table 5.3
1. Calculate the marginal (additional) excess burden per dollar trans-ferred from the MCF figures for each tax rate in column (3) and insert them as column 4.
2. How would you summarize the relation between the tax rates and the marginal excess burdens? (For example, as the tax rate changes, does the marginal excess burden increase proportionally, less than propor-tionally, or more than proportionally?)
3. On the basis of your answer to question 2, would it always then be valid to draw a fixed difference between the MC.MCF and the S curves (like that in Figure 5.1) as the scale of a health care intervention expands and additional tax finance is required?
Social cost of taxation 135
4. On the basis of your answers to questions 2 and 3, examine the appro-priateness of assuming a fixed MCF to apply to extra funds required to finance each of these health care evaluations: (a) buying 10 more wheelchairs for a hospital; and (b) providing universal health insurance for all citizens.
5.5.2 Summary
Typically, health care evaluations, and health economics journals, have ignored the role played by the MCF in making economic evaluations.
Presumably the view was that the MCF was a public finance topic and thus not applicable to health economics. In this chapter we have argued that the excess burden of the tax revenues used to finance a health care project was just a particular type of negative externality. Since including externalities makes economic evaluations more comprehensive and responsive to all the consequences that individuals care about and value, there is no good reason to ignore this consideration just because it is attached to non-health effects.
In order for the effects of a health expenditure to exist at all, resources must be withdrawn from other activities. In the process of withdrawing inputs from elsewhere, consumers have to forgo more satisfaction than is directly reflected in the tax revenues passed over. This extra satisfaction forgone is the excess burden of financing the tax and this, when added to the revenue effect, amounts to the MCF. The main determinant of the size of the MCF is the elasticity of demand.
Of course, not all health care projects are financed by public funds and it is only for these types of expenditures that a non-unitary MCF is required. The public sector is obviously involved when it undertakes the health services itself, i.e., public ownership. Hence when public ownership is transferred (privatized) the MCF must be attached both to funds gained as well as given up by the public sector.
A second way that public funds are involved is where the private sector carries out the production, but cash transfers are required to ensure that the scale of private operations are not too low from a social perspective. Cash transfers are not directly resource effects. But they usually will entail distri-bution effects, whether between the private and public sectors, or between rich and poor groups. The MCF is required for the first type of distributive effect and this chapter has focused primarily on this aspect. Distribution weights are required for the second type of effect and the last case study introduced this particular dimension to the evaluation process (to be devel-oped further in later chapters).
The implicit assumption used in economic evaluations is that the finance for a new health care expenditure would be from the same set of taxes that
136 CM and CBA
generated past government expenditures. So we referred to a comprehen-sive study of MCFs for US federal taxes by Ballard et al., and took the average of these estimates to give us an idea of the magnitude of the MCF.
Since different states have different tax mixes, a state’s MCF is simply a weighted average of the federal tax MCFs, where the weights are the state tax shares coming from the various federal tax categories.
The applications stayed within the confines of the CM approach by dealing with cases where benefits were the same across alternatives, or where benefits were ignored, or where benefits were regarded as negative costs. The applications illustrated the many different ways that transfers could impact an economic evaluation. These case studies added an alterna-tive justification for why the MCF should be included in health care evalu-ations. That is, they showed that including MCFs made a difference to the desirability of certain health care projects. Some types of hospital privati-zation were judged worthwhile only when the MCF was set equal to 1; while future social security transfers changed from being worthwhile to being undesirable when a MCF that rose over time was incorporated.
5.5.3 Looking Ahead
With this chapter, we have completed the examination of the cost side of evaluations. From here we proceed to a consideration of the consequences.
The next part covers CEA.
Social cost of taxation 137