Developing an implicit price deflator for New Zealand's health sector.
In all countries, a significant gap in empirical information concerns the price and quantity components of health expenditures. Whether the discussion is about short- or long-run shifts in health expenditures, the extent to which the shifts represent monetary versus real changes is typically not known, and yet this is crucial for designing optimal policies defining resource allocations to the health sector. The New Zealand Ministry of Health addresses this deficiency in empirical information by deflating all expenditures relating to health activities by the consumer price index (CPI), a procedure justified on the basis that the CPI measures the value of goods and services given up in order to spend on health (Muthumala and Ellis, 1995, p. 27). The "opportunity cost" justification for deflating health expenditures by the CPI is persuasive, however, only to the extent that marginal dollars spent on different types of health activities, and at different times, generate outputs of equivalent market value. In general, an assumption of equal output health dollars is not empirically plausible.
A conceptually superior procedure to deflating health expenditures by the CPI is to deflate them through a sectoral approach comparable to that used in deflating gross domestic product (GDP), where own-price deflators are identified for each type of expenditure. In the sectoral approach, the overall health expenditure price index becomes a weighted average of the various own-price indices. Here, the sectoral approach to deflating New Zealand's health expenditures is applied to 1961-1993 data, and an effort is made to establish the main implications of the sectoral deflation approach for interpreting health expenditure trends.
To this end, Sections 2 and 3 explain the details of how the sectoral deflation approach is applied, with Section 2 presenting the pertinent health expenditure data, and Section 3 presenting the available price indices relating to health Sections 4 through 6 then consider how the sectoral approach affects perceptions about health expenditures, with Section 4 presenting comparative estimates of health sector pricing trends, Section 5 presenting comparative constant-dollar health expenditure per capita estimates, and Section 6 analysing trends in health expenditures as proportions of GDP. Section 7 summarises conclusions and policy implications.
2. New Zealand Health Expenditures
Application of the deflation procedure defined in the Introduction requires a comprehensive set of sectoral health expenditure data covering a reasonably long period of time. In the case of New Zealand, the set is of better quality and more complete if public sector expenditures only (as opposed to public plus private) are considered. Table 1 indicates that a reasonably comprehensive set of public sector health expenditures is available, with health expenditures separated into five major categories: (1) operating expenditures of health care institutions (not including major capital outlays), (2) expenditures on community medical services (including expenditures on the services of fee-for-service health practitioners), (3) expenditures on dental services (including expenditures on dentists' services), (4) expenditures on drugs and prostheses, (5) expenditures on public health, and (6) expenditures on teaching and research in health. There is also a column labelled "other" expenditures that involves zero entries up to 1993-1994, when the Ministry of Health started reporting expenditures not allocable to one of the above five categories (the unallocable expenditures are identified as "Ministry of Health non-public health expenditures" and as "other payments").
Table 1. New Zealand Public Sector Health Expenditures, by Type of Expenditure, 1960-1961 to 1994-1995 ($000,000)
Year Total Institutional Community Dental medical 1961 100.0 65.0 10.0 4.8 1962 107.0 69.0 11.0 5.0 1963 116.0 75.0 12.0 5.2 1964 123.0 82.0 12.0 5.3 1965 135.0 91.0 14.0 5.5 1966 150.0 103.0 15.0 5.6 1967 164.4 114.0 16.0 5.8 1968 170.0 118.0 17.0 6.0 1969 181.0 126.0 18.0 6.2 1970 202.0 139.0 20.0 6.4 1971 241.0 167.0 24.0 7.3 1972 296.0 205.0 25.0 9.6 1973 352.0 240.0 31.0 10.6 1974 410.0 278.0 36.0 11.8 1975 503.0 337.0 46.5 16.5 1976 624.0 414.0 61.3 18.2 1977 728.0 481.0 62.4 18.2 1978 855.0 567.0 68.1 19.8 1979 1,033.0 697.0 86.2 21.7 1980 1,216.0 851.0 89.7 28.5 1981 1,387.0 918.0 102.1 30.9 1982 1,695.0 1,224.0 110.6 33.0 1983 1,762.0 1,226.0 140.3 35.0 1984 1,953.0 1,396.0 140.9 37.2 1985 2,072.0 1,465.0 154.6 37.7 1986 2,519.0 1,745.0 201.6 41.0 1987 3,169.0 2,215.0 239.6 48.6 1988 3,681.0 2,548.0 259.6 61.3 1989 3,997.0 2,721.0 325.6 61.8 1990 4,255.0 2,863.2 463.9 30.8 1991 4,429.0 2,952.2 601.9 29.6 1992 4,473.0 2,921.0 644.7 32.8 1993 4,464.0 2,973.0 566.3 27.8 1994 4,776.0 2,889.9 728.8 35.0 1995 4,953.3 2,900.2 932.1 28.1 Year Drugs Public Teaching and Other health research 1961 13.6 5.0 1.6 0.0 1962 14.0 6.1 1.9 0.0 1963 15.0 6.7 2.1 0.0 1964 16.0 5.8 1.9 0.0 1965 16.0 6.4 2.1 0.0 1966 18.0 6.4 2.0 0.0 1967 20.0 6.5 2.1 0.0 1968 21.0 6.1 1.9 0.0 1969 22.0 6.7 2.1 0.0 1970 25.0 8.8 2.8 0.0 1971 27.0 11.9 3.8 0.0 1972 31.0 19.2 6.2 0.0 1973 33.0 28.3 9.1 0.0 1974 40.0 33.4 10.8 0.0 1975 45.0 43.9 14.1 0.0 1976 56.0 56.4 18.1 0.0 1977 70.0 72.9 23.5 0.0 1978 85.0 87.1 28.0 0.0 1979 98.0 98.4 31.7 0.0 1980 132.6 86.4 27.8 0.0 1981 147.0 143.0 46.0 0.0 1982 176.2 96.9 54.3 0.0 1983 196.0 110.2 54.5 0.0 1984 221.5 102.8 54.6 0.0 1985 255.4 101.0 58.3 0.0 1986 347.1 116.5 67.8 0.0 1987 443.4 138.9 83.5 0.0 1988 507.3 195.9 108.9 0.0 1989 558.5 211.3 118.8 0.0 1990 520.7 184.8 191.6 0.0 1991 560.7 129.5 155.1 0.0 1992 570.3 142.1 162.1 0.0 1993 595.9 148.9 152.1 0.0 1994 657.2 177.4 143.4 144.3 1995 692.9 109.4 154.8 135.8
Notes: Years are fiscal year ending. Numbers in bold italics are imputed to be consistent with the published statistics. The imputed dental statistics for 19621968 represent an interpolated constant annual growth rate, and the imputed data for public health and teaching and research are estimated as a residual for each year. The proportion of the residual going to public health is assumed to be 75.7%, which is the actual proportion for 1979-1980.
Sources: Griffiths (1986); McKendry and Muthumala (1991, 1993); Muthumala and Howard (1995); Annual Report of the New Zealand Department of Health (1961, 1966, 1971, 1976, 1979 and 1985 issues); Sutton (1988).
Despite the limited amount of sectoral disaggregation in Table 1, not all entries represent official estimates of amounts spent. Forty-nine entries are missing from the government published data: 7 for dental services, 21 for public health, and 21 for teaching and research. Imputed values for the missing entries are provided, highlighted in bold italics. The imputation procedures, described at the bottom of the Table, incorporate the constraint that the various types of health expenditures should add up to total expenditures for each year if all items are measured accurately. Because there are few gaps in the published statistics, adherence to the adding-up constraints makes the imputed estimates reasonably closely reflect the orders of magnitude involved; however, the figures are presented in such a way that analysts preferring to rely on government published figures only can do so.
When total rather than public sector health expenditures are considered, the gaps in the government published statistics are more numerous. Table 2 summarises the available data. It can be seen that the official figures constitute a reasonably comprehensive data set after 1980-1981, but before this point many of the sectoral figures are provided only on an occasional basis. Before 1980-1981, the most comprehensive sectoral information exists for 1960-1961, 1965-1966 and 1973-1974, because there are a number of articles attempting to provide complete structural profiles for these years (Brown, 1977a, b; Smith, 1979; Ward and Tatchell, 1972). For the 1961-1991 years overall, there are 120 missing observations in a total data set of 238, and 106 relate to 1961-1979 (Table 2).
In Table 2, imputed estimates, again in bold italics, are provided for these missing observations. The imputation procedures, described at the bottom of the Table, are designed to make the imputed values merge with the government ones at points where imputed and government figures are contiguous, and to make the trends in imputed total expenditures mirror the trends in respective public sector expenditures (as presented in Table 1). The latter procedure is preferable to that of applying long-run growth rate trends to annual changes, because the only regularities it builds into the total health expenditure estimates are those already in the public sector figures. It also generates reasonably accurate estimates in contexts where the ratios of public to total expenditures are high, and in this regard it is of note that, for New Zealand health expenditures in the aggregate, the publicly financed proportion varies from 76 to 88% during 1961-1993.
The above comments should not be construed to imply that the imputed estimates in Table 2 are as accurate as those in Table 1. Because there are so many more gaps in the total expenditure figures, it is difficult to bring the adding-up constraints to bear. The imputed estimates in Table 2 represent an effort to generate the best estimates possible on a series-by-series basis, based on their public sector components. Generating best estimates in this way means that the adding-up constraints are formally ignored and are often not satisfied. In Table 2, the magnitudes of the adding-up inconsistencies existing in the pre-1984 data are reflected in the non-zero entries in the final column. Thus, the negative entry of $11.7 million for other health expenditures in 1965-1966 means that adding up institutional expenditures, community medical expenditures, dental expenditures, drugs and prostheses expenditures, public health expenditures, and teaching and research expenditures generates a sum for total health expenditures that is $11.7 million larger than the government's estimate (of $179 million).
[TABULAR DATA 2 NOT REPRODUCIBLE IN ASCII]
It must be emphasised that, in developing the imputed estimates in Table 2, a great deal of effort was initially made to adjust the figures in Table 2 so that the adding-up constraints would be satisfied. The problem was that all adjustments attempted made one or more of the series implausible, by implying shares for public and private funding inconsistent with non-statistical descriptions of the situation, or by implying annual changes in some types of health expenditures of unreasonable magnitudes. In general, the effort to apply the adding-up constraints generated the impression that in the early years the amounts of private expenditures incorporated into total health expenditure estimates are not large enough to yield sectoral estimates consistent with aggregate ones, a measurement bias that could easily exist given the data collection and collation processes in place when the statistics were generated. This judgement can not be validated, however, which is one reason why all existing official expenditure figures are incorporated into Table 2. Analysts unhappy with the imputed estimates have the options of working only with the government figures or developing imputed estimates of their own.
3. Price Deflators for Health Expenditures
Limited data availability constitutes a problem not only with respect to the health expenditure figures but also with respect to the price indices available for deflating the figures. For the 1960-1993 period, there are only five major price indices relating to the health sector, and these indices are presented in the first five columns of Table 3. Two of these indices are developed directly by the government: the hospital price index measuring hospital costs including salaries, and the medical and health services index which is the component of the CPI measuring consumer price changes for services provided by health care providers, hospitals and insurers. Two other indices, the physician incomes index and the dentist incomes index, are based on government estimates of the average net incomes of fee-for-service doctors and dentists, but the figures are not presented by the government as indicators of pricing trends concerning practitioner time, as is being done de facto in this paper. The final index pertaining to the health sector is the drugs (or Pollard) index. It measures changes in drug prices at the wholesaler level, and is developed by the pharmaceutical industry [specifically, by IMS(NZ) Ltd].
[TABULAR DATA 3 NOT REPRODUCIBLE IN ASCII]
As can be observed, Table 3 presents nine price indices. Two of the additional indices are well known to economic analysts: the CPI and the implicit GDP deflator. These indices are not specific to the health sector but are nevertheless sometimes used for deflating health expenditures. As noted in the Introduction, the New Zealand Ministry of Health deflates health expenditures by the CPI, while up to 1996 the Canadian Department of Health and Welfare used the implicit GDP deflator (Health and Welfare Canada, 1983, p. 4). In its most recent publication on health expenditures, however, Health Canada has introduced a new "one size fits all" deflation procedure, which involves deflating private health expenditures encompassed in each health expenditure series by the health care component of the CPI, and the public expenditures encompassed in the series by the government goods and services implicit price index (Health Canada, 1996, part 3, pp. 18-19). None of the deflation procedures adopted by either the New Zealand Ministry of Health or by Health Canada are well designed to measure inflation specific to the health sector, even though for many kinds of empirical analysis this is what is required.
A third additional index found in Table 3 is the drugs and health services index. For the period from 1974-1975 to 1994-1995, this index is the Pollard drug index, with the base year shifted from 1974-1975 to 1989-90; from 1960-1961 to 1974-1975, it is the medical and health services index, with the 1974-1975 value shifted from 0.118 to 0.336.
The final additional index found in Table 3 is an implicit health-sector price deflator derived from the other indices. The derivation reflects the effort to deflate each type of health expenditure by an own-price index. The own-price criterion dictates the optimal deflation index among the indices available in respect to institutional expenditures, community medical expenditures, dental expenditures, and drugs and prostheses expenditures, with the best indices being respectively the hospital prices index, the physician incomes index, the dentist incomes index, and the drugs index (as it is incorporated into the drugs and health services index). Deflating these four categories of health expenditures by these indices generates constant dollar estimates for 85 to 95% of all health expenditures, depending on the year in which the expenditures are deflated.
Choosing the best deflation procedure for the remaining 5 to 15% of health expenditures where own-price indices are not available is problematic. As implied in the above paragraph, the procedure opted for in relation to drug expenditures prior to 1975 is to deflate by the medical and health services index, as it is incorporated into the drugs and health services index. The justification for doing this is that the medical and health services index is comparable to the drug index ?in emphasizing purchaser prices rather than provider costs. Other deflation procedures with weak justifications are the following. Teaching and research expenditures are deflated by the physician incomes index on the assumption that earnings rates of medical teachers and researchers are correlated with earnings rates of medical practitioners, and public health expenditures are deflated by the implicit GDP deflator on the assumption that a neutral deflation procedure is, optimal when there is no criterion for deflating in a non-neutral way (deflating by the implicit GDP deflator is neutral in the sense that the estimated proportion of GDP allocated to public health is the same, whether current or constant dollars are considered).
While the own-price criterion dictates which indices are best for health expenditure price deflation among the indices that are available, the deflation procedure is complicated by the fact that the New Zealand government introduced a Goods and Services Tax (GST) of 10% on October 1, 1986, which was increased to 12.5% on July 1, 1989. Because the Ministry of Health deflates health expenditures by the CPI, which reflects movements in consumer prices including those caused by changes in GST, it has been expedient for the Ministry to measure all health expenditures with the GST component included. However, the hospital, medical, dental and drug indices in Table 3 do not reflect movements in the GST, the first three indices because they measure movements in production costs rather than purchaser prices, and the last index because it measures movements in drug prices at the wholesale rather than retail level. If these indices are used to deflate the Ministry's health expenditure estimates, either the GST components of the expenditures must be excluded, or the indices themselves must be adjusted to incorporate the effects of the GST. The latter approach is adopted here, because it results in an implicit health-sector price index that can be directly applied to official health expenditure data.
The procedure adopted to modify the hospital, medical, dental, and drug indices in Table 3 is to multiply all index values by 1 + g, where g is the rate of GST, and equals zero for all years prior to 1986-1987, 5% for 1986-1987, 10% for 1987-1988 and 1988-1989, and 12.5% for years after 1988-1989. Each index is then adjusted so that its 1989-1990 value equals unity.
Overall, the above deflation procedures mean that the implicit health-sector price index presented in Table 3 is an average of (1) a GST-adjusted hospital care prices index weighted by expenditures on institutional services, (2) a GST-adjusted physician incomes index weighted by expenditures on community medical expenditures plus expenditures on teaching and research, (3) a GST-adjusted dentist incomes index weighted by expenditures on dental services, (4) a GST-adjusted drugs and health services index weighted by expenditures on drugs and prostheses, and (5) the implicit GDP deflator index weighted by expenditures on public health. Because the weights change each year, the implicit health-sector price deflator is a Paasche price index; in this regard, it is comparable to the implicit GDP deflator, which is also a Paasche price index, but not to the CPI which is a Laspeyres price index.
The fact that the hospital, medical and dental price indices measure movements in production costs rather than purchaser prices is a conceptual limitation of the implicit health-sector price index as derived, because cost-oriented price indices do not reflect increases in productivity in the health sector (with increases in productivity reflecting a combination of increased quantity and quality of output per input used). In empirical terms, however, the shortcomings of the implicit health-sector price index in adjusting for productivity changes are no greater than the shortcomings of the price deflators applied to most service industries by statistical agencies like Statistics New Zealand and Statistics Canada. In all cases, the shortcomings exist because the measurement problems are intractable with current measurement technologies. In the case of health care, the hope is that improvements in the ability to empirically measure quality-adjusted life years (QALYs) will eventually solve the problem of measuring productivity change. Until such improvements materialise, it is necessary to ,keep in mind that existing constant-dollar estimates of health expenditures are of more value in generating insights about relative resource movements between the health and non-health sectors than they are in measuring movements in health outputs. If constant-dollar figures based on current measurement technologies are used to estimate growth in real health outputs despite the limitations of the data, then it must be realised that real output growth will be underestimated by the amount of the unmeasured productivity increase.
While the above qualifications are important in ascertaining what can be properly gleaned from use of the implicit health-sector price index, they must be tempered by two caveats. The first is that information about movements in resource allocations between sectors has value even if not coupled with knowledge about relative output changes. The second caveat is that, post-1974, measured increases in health outputs per capita are larger when based on the implicit health-sector price index than on the CPI (Table 4), even though a part of the attraction of relying on the CPI has been that it is considered to measure productivity changes better.
[TABULAR DATA 4 NOT REPRODUCIBLE IN ASCII]
It is tempting to conclude this section with an overview comment suggesting the superiority of the sectoral deflation approach, and it presumably is the case that most analysts will find deflating health expenditures by an implicit health-sector price index preferable to deflating these expenditures by the CPI, the implicit GDP deflator, or by a combination of the health sector component of the CPI plus an implicit price deflator for government goods and services. However, in making such an assessment one must be aware of what is involved. If trends in both nominal and real health expenditures were directly observable at least some of the time, it would in principle be possible to ascertain the empirical accuracy of specific deflation procedures used to derive real expenditure estimates from nominal ones. Unfortunately, in applied work, real expenditures must always be defined in terms of the deflation procedures used, which means that confidence in the constant-dollar expenditure estimates and confidence in the deflation procedures creating the estimates amount to the same thing. It is possible to reject the implicit health-sector price index on the basis that it is conceptually and for empirically flawed since, for example, it does not measure productivity changes (a conceptual failing), and trends in practitioners' incomes are not unbiased estimates of hourly earnings (an empirical failing). However, in applied work the preferred criterion for rejecting a procedure is that it is not the best one available rather than that it is not perfect. In policy applications, the alternative to developing an implicit health-sector price index has been that of relying on extremely aggregated deflation procedures, and this is the alternative that has been chosen in most countries most of the time. Aggregate deflation procedures adjust for general inflationary changes but do not measure price changes specific to individual sectors of the economy.
4. Movements in the Implicit Health-Sector Price Deflator
While consideration of movements in the implicit health-sector price index is not a way of assessing measurement accuracy, it is a way of ascertaining how reliance on the index affects perceptions about pricing trends. Figure 1 presents the trends in the implicit health-sector price index during 1961-1993, with the index values presented as ratios of the CPI, and as ratios of the implicit GDP deflator. In general, the trends indicate that the implicit health-sector price index increased faster than either the CPI or the implicit GDP deflator between 1960-1961 and 1974-1975, and slower between 1974-1975 and 1992-1993. The trends also suggest more comparability between the price movements measured by the CPI and the implicit GDP deflator than between either of these aggregate price deflators and the implicit health-sector price index.
[Figure 1 ILLUSTRATION OMITTED]
During 1961-1975, the implicit health-sector price index increased at an average annual rate of 7.65%, the CPI increased at an average annual rate of 6.15%, and the implicit GDP deflator increased at an average annual rate of 5.55%. During 1975-1993, the respective annual growth rates were 10.22% for the implicit health-sector price deflator, 10.90% for the CPI, and 10.50% for the implicit GDP deflator.
5. Movements in Constant-Dollar Health Expenditures per Capita
The movements in constant-dollar health expenditures per capita implied by deflating health expenditures by the implicit health-sector price deflator, and by the CPI, are presented in Table 4. The estimates based on the implicit health-sector price deflator suggest that constant-dollar health expenditures per capita increased by 2.23% annually during 1961-1975, and by 2.21% annually during 1975-1993, while the respective CPI-based estimates are 3.68% and 1.60%. The estimates based on the implicit health-sector price deflator generate an impression of a relatively constant secular growth rate in health care utilisation, while the CPI-based estimates generate an impression of a growth rate that has declined significantly over time. Neither of the constant-dollar estimates supports the widespread view that the growth rate in health care utilisation has been increasing, but can be reconciled with the view by assuming appropriate magnitudes of productivity growth. Analysts making such assumptions must keep in mind, though, that, although their assumptions can not be rejected on the basis of current measurement technologies and resulting empirical evidence, they can not be confirmed either.
It is easy to get the impression from evaluating the constant-dollar health expenditure estimates found in Table 4 that, empirically, the choice of price deflation procedure is of little consequence. Consider the following simple regression, which makes per capita expenditures based on the implicit health-sector price deflator ([E.sub.1]) a function of per capita expenditures based on the CPI ([E.sub.2]), and where the numbers in parentheses are Student's t-statistics relating to the regression coefficient values:
[E.sub.1] = 70.1226 + 0.9186[E.sub.2]
[R.sup.2] = 0.9654
N = 33
The high [R.sup.2] value could give the impression that [E.sub.1] and [E.sub.2] have similar implications for empirical analysis.
However, when the estimates are analysed in terms of annual differences, the impression changes. Consider the following regression, defined in terms of proportionate rates of change (and based on the statistics in the last two columns of Table 4):
[Delta][E.sub.1]/[E.sub.1] = 0.0088 + 0.5384 [Delta][E.sub.2]/[E.sub.2]
[R.sup.2] = 0.4100
N = 32
The relationship between [E.sub.1] and [E.sub.2] is still statistically significant at the 1% significance level, suggesting the acceptability of using one series as a proxy for the other if the index preferred on conceptual grounds is not available. However, insofar as only 41% of the variation in [E.sub.1] can be explained by changes in [E.sub.2], the implication is that the prediction error involved in using the proxy series is quite high.
6. Movements in Health Expenditures as Proportions of GDP
While constant-dollar health expenditures per capita are the preferred statistics for measuring health care utilisation patterns and trends, they are less useful for indicating the social economic burden that the utilisation represents. The most often used statistics for the latter purpose are health expenditures as a proportion of GDP, with the current dollar ratio considered a good indicator of the financial burden implied by health care utilisation, and the constant-dollar ratio considered a good indicator of the resource (or real) burden.
In distinguishing between the financial and real burdens of health activities using health expenditure to GDP ratios, the choice of deflation procedure is critical. In general, procedures that apply aggregate price indices (measuring general inflationary change) to the health sector promote the perspective that the financial and real burdens of health care utilisation are similar. Indeed, if health expenditures are deflated by the implicit GDP deflator, the financial and real burdens are procedurally constrained to be identical, because both the numerator and the denominator of the health expenditure to GDP ratio are deflated by the same price index. When reliance is placed on the CPI, the result is not so extreme, but it nevertheless remains the case that CPI deflation downplays price movements that are health-sector specific (Figure 1). By way of contrast, the implicit health-sector price index facilitates identifying health sector price changes that are different from those occurring in the economy as a whole.
Figure 2 presents the movements in health expenditures as proportions of GDP measured in both current and constant dollars, with the constant-dollar ratios reflecting deflation through the implicit health-sector price index. The statistics suggest a significant shift in secular trends in the mid-1970s, with 1973-1974 representing the pivotal year. Prior to 1973-1974, the financial burden of health care utilisation was increasing but the real burden was declining slightly. Health expenditures as a proportion of GDP in current dollars increased from 4.47% in
[Figure 2 ILLUSTRATION OMITTED] 1960-1961 to 5.36% in 1973-1974, but in constant dollars declined from 5.56% to 5.44%. After 1973-1974, with the slow down in New Zealand's economic growth fostered by the energy crisis and by Britain's entry into the Common Market, the burden of health activities started increasing in real as well as in financial terms. During 1974-1993, New Zealand's constant-dollar ratio of health expenditures to GDP increased from 5.44% to 7.44%, and its current dollar ratio increased from 5.36% to 7.62%. These figures give the impression that, for 1974-1993 overall, the growth rate in health care prices was just a little lower than the growth rate in prices generally, and that the increase in the financial burden of the health sector was mainly attributable to the sector absorbing a higher proportion of the economy's resources.
Examination of Figure 2 indicates that it may be more appropriate to treat 1974-1993 as a set of three periods (1974-1981, 1981-1985 and 1985-1993) than as one, however. During 1974-1981, both the nominal and constant-dollar ratios of health expenditures to GDP increased, the former by 4.19% annually, and the latter by 3.37%. The fact that the financial burden increased by more than the real burden suggests that the secular trend of rising relative health prices in evidence during 1961-1974 continued during 1974-1981, even though the country's economic circumstances worsened significantly (New Zealand's average annual growth rates in constant-dollar GDP declined from 4.02% during 1961-1974 to 1.12% during 1974-1981).
After 1980-1981, New Zealand's worsening economic conditions impacted on trends in the health sector. From 1980-1981 to 1984-1985, i.e., up to the Point of New Zealand's foreign exchange/international credit crisis, both the financial and real burdens of the health sector declined; the current dollar ratio of health expenditures to GDP declined at an annual rate of 4.50% during the 4 year period, and the constant-dollar ratio declined at an average annual rate of 3.02%. The fact that the financial burden declined by more than the real burden indicated that relative health care prices declined, a significant deviation from the 1961-1981 upward trend.
In the aftermath of the foreign exchange/international credit crisis, when government policy was directed at creating a more market-oriented economy and at reducing the size of the public sector, relative health care prices ceased declining, and both the financial and real burdens of the health sector resumed the upwards trends in place before 1980-1981. During 1985-1993, the average annual growth rate in current-dollar health expenditures per capita was slightly higher than the growth rate in constant dollars, at 2.97% versus 2.53%.
While it is beyond the scope of this paper to analyse in detail the movements in specific types of health expenditures, it is of note that such analysis substantiates and augments the picture generated by analysing the aggregate health expenditure statistics. Of particular note is that constant-dollar dental and public health expenditures increased faster than other kinds of constant-dollar health spending during 1961-1974, declined in relative importance during 1974-1981, and declined absolutely after 1980-1981. In the crisis years of the early 1980s, New Zealand addressed its cost problems in health partly by lowering the priority attached to dental and public health services, and the redefinition of priorities did not change in the period succeeding the crisis.
7. Assessment and Conclusion
The analysis in this paper indicates that a sectoral approach to deflating health expenditures is feasible, although for historical data the approach requires reliance on some imputed statistics, and it requires reliance on price indices that do not account for productivity changes. Insofar as there is interest in analysing past trends, these data problems are and will remain unresolvable, which is not to say that analysts interested in historical issues must adopt sectoral deflation procedures identical to the one presented here. Alternative procedures might involve different ways of compensating for gaps in the statistics, and they might involve generating an implicit health-sector price deflator from a subset of health expenditures where acceptable own-price indices are available, not to mention other options. The statistics are presented in this paper so that the alternative procedures are easily applied, if preferred. However, all alternative sectoral deflation procedures that have been experimented with as a backdrop to this paper generate minor changes in empirical results compared with the procedure of relying on CPI-deflated constant-dollar figures. They all suggest a major shift in the secular trends in the mid-1970s, and minor shifts around 1980-1981 and 1984-1985. The critical question is whether the shifts are real or artefacts of the sectoral deflation procedure. My judgement, based on a consideration of how the shifts relate to New Zealand institutional history, is that the shifts reflect real changes in the supply and demand conditions that occurred in health care markets.
Because the development of an implicit health-sector price index is feasible, this does not necessarily mean that it should be done. The main objections to developing such an index are partly pragmatic (it should not be developed because crude and cheap deflation processes provide almost as good information' empirically) and partly conceptual (it should not be developed because most intra-sectoral and own-price indices do not adequately account for productivity changes, whether the changes result from increases in quantities or qualities of outputs produced). The pragmatic objection suffers from the weakness that, if the optimal indices are not available, it is impossible to know when the crude deflation procedures yield erroneous insights. As has been shown, the New Zealand Ministry of Health price deflation procedure often yields quite different empirical results from the implicit health-sector price index approach developed in this paper. Moreover, the Ministry of Health's approach militates against differentiating between the concepts of financial and real health sector burdens, a bias which, over time, promotes the view that financial and real economic problems are one and the same.
The conceptual objection is more obtuse. The Ministry's view that the CPI is a good index on conceptual grounds, while the hospital care prices index is unacceptable because it measures input costs rather than output values, was a factor in the Ministry's decision to cease generating the hospital index in 1994. The Ministry does not seem to understand that, even if the CPI is better designed for deflating consumer expenditures than the hospital index is for deflating hospital expenditures, it does not follow that the CPI is better than the hospital index for deflating hospital expenditures.
The above comments are not meant to imply that the current difficulties in measuring productivity increases in the health sector do not constitute problems. They do. What is being implied is that, until the point when QALYs can be satisfactorily measured empirically, "the baby should not be thrown out with the bath water" in terms of current measurement strategies. There is value in developing constant-dollar health expenditure statistics based on indices measuring input costs rather than consumer prices, particularly in relation to the goal of measuring and differentiating between real and financial burdens of health care utilisation.
A final point that should be made is that this analysis is not just about the measurement of past trends, even though it might become that if the Ministry persists in its strategy of considering the CPI to be the only deflation index necessary (not only has the Ministry recently ceased producing the hospital price index, but Statistics New Zealand has recently ceased collecting the statistics on physician and dentist incomes, on the basis that the information serves no purpose). What this analysis does is provide a broad outline of a procedure for deflating health expenditures that can be modified as improved measurement procedures evolve. The analysis constitutes a broad outline rather than an exact blueprint because analysts setting up deflation procedures for current and future data have options not open in analysing the distant past. These options include collecting and collating data such that reliance on imputed estimates is not necessary, engaging in more sectoral disaggregation, and developing sectoral own-price indices with fewer conceptual and empirical limitations than the indices used here. These options will become realities, however, only to the extent that policymakers recognise the inadequacies of the CPI deflation approach, and the benefits from having constant-dollar figures specifically designed to measure changes in resource allocations in relation to the health sector.
Brown, M. C. (1977a), "Comparative aspects of the New Zealand and Canadian health systems", Economic Record, 53, 182-187.
Brown, M. C. (1977b), The Financing of Personal Health Services in New Zealand, Canada, and Australia, Canberra: Centre for Research on Federal Financial Relations, Australian National University.
Griffiths, B. (1986), Trends in the Distribution of Vote Health Expenditures in New Zealand, Wellington: Health Research Development Unit, Department of Health.
Health and Welfare Canada (1983), National Health Expenditures in Canada, 1970-1982, Ottawa.
Health Canada (1996), National Health Expenditures in Canada, 1975-1994, Ottawa.
McKendry, C. G. and Muthumala, D. (1991), Health Expenditure Trends in New Zealand, 1980-91, Wellington: Department of Health.
McKendry, C. G. and Muthumala, D. (1993), Health Expenditure Trends in New Zealand Update to 1993, Wellington: Ministry of Health.
Muthumala, D. and Ellis, J. A. (1995), Health Expenditure Trends in New Zealand, 1980 1995, Wellington: Ministry of Health.
Muthumala, D. and Howard, P. S. (1995), Health Expenditure Trends in New Zealand, 1980-1994, Wellington: Ministry of Health.
New Zealand Department of Health (1985), Annual Report of the Department of Health, Wellington: Government Printer, 1961, 1966, 1971, 1976, 1979, and 1985 issues.
Smith, A. G. (1979), Health Expenditures in New Zealand: Trends and Growth Patterns, Wellington: New Zealand Department of Health.
Sutton, F. (1988), Trends in Pharmaceutical Benefit Expenditure Over the Last Decade, Wellington: Department of Health.
Ward, J. T. and Tatchell, P. M. (1972), "Health expenditures in New Zealand", Economic Record, 48, 500-517.
Malcolm C. Brown, Health Services Research Centre, Victoria University of Wellington, and the Wellington School of Medicine.
Thanks must be expressed to Miriam Laugesen, who did much of the data collecting pertaining to this paper, to IMS(NZ) Ltd who, acting on behalf of the Researched Medicines Industry, provided me with their Pollard drug index; to both Statistics New Zealand and to the New Zealand Ministry of Health, who provided data not normally published, and to two anonymous referees who made helpful suggestions concerning the paper. Responsibilities for remaining errors and weaknesses in the paper remain mine.
|Printer friendly Cite/link Email Feedback|
|Author:||Brown, Malcolm C.|
|Publication:||New Zealand Economic Papers|
|Date:||Jun 1, 1997|
|Previous Article:||Keynes, purchasing power parity and exchange rate policy in New Zealand during the 1930s depression.|
|Next Article:||How did the wealthiest New Zealanders get so rich?|