Acton 1975 - Nonmonetary Factors in the Demand for Medical...

Info iconThis preview shows pages 1–20. Sign up to view the full content.

View Full Document Right Arrow Icon
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 2
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 4
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 6
Background image of page 7

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 8
Background image of page 9

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 10
Background image of page 11

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 12
Background image of page 13

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 14
Background image of page 15

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 16
Background image of page 17

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 18
Background image of page 19

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 20
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Nonmonetary Factors in the Demand for Medical Services: Some Empirical Evidence Jan Paul Acton RAND Corporation Nonmonetary factors are expected to assume an increasingly important role in determining the demand for medical care as the out-of-pocket money price falls (due to spreading health insurance coverage or the enactment of the federal health insurance legislation). A utility maximization model is used to develop predictions for the demand for “free” and nonfree care. A simultaneous-equation system is estimated on a survey of users of New York City's “free” outpatient departments and municipal hospitals. The empirical results support the major predictions that nonmonetary factors such as travel distance will function as prices in discouraging demand and that earned and non- eamed income have different impacts. A number of implications for public policy are suggested, including the possibility of substituting income maintenance for the direct provision or insurance of medical care. I. Introduction The demand for health care would deserve close scrutiny by researchers if only because of the size of the health sector. In addition to its size, the health sector has been one of the most inflationary in recent years, causing increasing interest by the public and policymakers alike. Finally, the growing numbers of proposals for significant changes in the health care system, especially proposals for federal health insurance legislation, make This paper is a revised version of a RAND report, “Demand for Health Care When Time Prices Vary More than Money Prices,” which was delivered to the winter meetings of the Econometric Society, 1972. The report was prepared for the Oflice of Economic Opportunity and the Health Services Administration of New York City. Raymond Lerner, of the State University of New York at Stony Brook, provided survey data for analysis. William Butz, David Chu, Michael Grossman,John Koehler, Joseph Newhouse, Charles Phelps, Norman Shapiro, Finis Welch, and a careful reviewer made comments on an earlier draft. Without trying to implicate them, I am grateful for their assistance. [Journal of Political Economy, 1975, vol, 83, no. 3] © 1975 by The University of Chicago. All rights reserved. 595 Copyright © 2001 . All Rights Reserved. JOURNAL OF POLITICAL ECONOMY imperative a better understanding of the determinants of demand. This research is part of a growing literature that studies the demand for medical services using disaggregated data.1 It differs from previous literature in concentrating on alternative mechanisms that may arise to determine demand for medical care as money prices to the patient fall through insurance. The out-of—pocket money price of medical care has been decreasing as a proportion of total price in recent years, primarily because of the spread of health insurance and the rising opportunity cost of time (and perhaps increased time needed to receive care). There is every reason to believe that money prices will continue to fall in relative importance because of the secular trend in insurance coverage and opportunity cost of time and, perhaps even more important, the prospect of national health insurance. With the decreasing relative importance of money prices, it is reasonable to expect an alternative mechanism to control demand. A mechanism involving time is quite likely to assume this role since medical care usually requires a payment in both travel time and waiting time.2 Indeed a substantial increase in queues and waiting lists was one of the widely noted results of the enactment of the National Health Service legislation in Britain (see Harris [1951] and citations therein). This paper considers the effects of travel distance in determining the demand for medical services in New York City, an especially good “laboratory” in which to try to examine the effects of nonmonetary prices because of the long-standing availability of free care in the city’s municipal hospitals and clinics.3 After developing a formal model of the demand for medical services that includes a payment in money and in time for private care, the predictions are tested on a cross-sectional survey of about 2,600 users of city hospital outpatient departments (OPDs). Although limitations of the data base indicate cautious interpretation, the empirical results lend support to the model’s major predictions. Empirical verification of the conjecture that time is important in de- 1 See, among others, Grossman 1972b; Phelps and Newhouse 1972; Acton 1973a; and Rosett-Huang 1973. 2 The importance of time as a determinant of demand was suggested by Becker (1965), its application to the demand for medical care by, among others, Leveson (1970) and Holtman (1972). 3 In a related study, Acton (1973a) examined the role of travel time and waiting time in the demand for medical services using two surveys of poverty neighborhoods in New York. This study differs from the previous report in a number of ways. First, it is based on a survey of the users of outpatient departments (OPDs) rather than a survey of the general population in a neighborhood. Consequently, we know that everyone had nonzero utilization of the health sector. This allows the specification of a simultaneous set of equations to describe the demand at several different sources of care. Second, the chief measure of price paid for “free” public care is travel distance to the clinic. Further, travel distance is specified as endogenous in this system. In the previous study, the large proportion of observations on the limit value of zero health sector utilization forces the estimation of reduced-form equations using Tobit estimation technique. Copyright © 2001 . All Rights Reserved. u DEMAND FOR MEDICAL SERVICES termining the demand for care raises a number of important policy issues. These not only include the efiect on the distribution of services to re- cipients of care; they also indicate powerful policy instruments for increasing the medical access of target populations. Some of these policy options—which include the location of clinics and the substitution of income maintenance for subsidized care—are discussed at the end of this paper. II. Model of Demand for Medical Services The underlying theoretical model for this study is developed in detail in Acton (1973a, 1973b).4 The model concentrates on the role of money prices, time prices, and earned and nonearned income. For simplicity, the model is developed in terms of only one provider of services, but the implications for several providers can easily be drawn. In the theoretical and empirical sections to follow, I will be studying the demand for care at alternative providers—free city clinics, private physicians, and hospitals. Assume two goods enter the individual’s utility function: medical services, m, and a composite, X, for all other goods and services. Using an assumption of fixed proportions of money and time to consume m and X and the full income assumption, the model can be represented as follows: Maximize U = U (m, X) (la) subject to (p + wt)m + (q + ws)X g Y =y + wT, (lb) where U = utility, m = medical services, X = all other goods and services, ,0 = out-of-pocket money price per unit of medical services, t = own-time input per unit of medical services, q = money price per unit of X, s = own-time input per unit of X, w = earnings per hour, Y = total (full) income, y = nonearned income, and T = total amount of time available for market and own production of goods and services. 4 Similar models can be found in Becker (1965), Grossman (1972a), and Holtman (1972). Copyright © 2001 . All Rights Reserved. 598 JOURNAL or POLITICAL ECONOMY A more complicated specification of the model was not used because this simpler formulation yields most of the same predictions and because the data do not permit estimation of the unique implications of the richer specification. 5 Eject: of a Change in Price It can be shown that the assumptions sufficient to make money function as a price in determining the demand for medical services are also sufficient to make time function as a price, producing negative own-time—price elasticities of demand and positive cross-time—price elasticities. One of the chief interests in this study is the relative importance of money prices and time prices in determining the demand for medical services. If we let H equal the total price per unit of medical services (i.e., H = p + wt), then the elasticity of demand for medical services with respect to money prlCC 1s "mp = finmfli and the elasticity with respect to time (which equals the elasticity with respect to wt) is6 t n... = "’5 mm. <2b> Comparing these two elasticities yields the second prediction from the formal model, namely, that 11",, E 11"”, as wt 5 p. Clearly, as [2 goes to zero and wt does not, the time-price elasticity will exceed the money-price elasticity. In other words, as the out-of—pocket payment for a unit of medical services falls, because of either increasing insurance coverage or the availability of subsidized care, demand becomes relatively more sensitive to changes in time prices. Further, this implies that the demand for free medical services should be more sensitive to changes in time prices than demand for nonfree services, because time is a greater proportion of the total price at free than at nonfree providers. 5 Grossman (1972a) allows the amount of medical services, In, to influence the total amount of time, T, available for production. Phelps (1973) makes the selection of in- surance parameters, and therefore p, endogenous to the system. Some researchers (notably Grossman [1972a, 1972b] and his followers) have taken the Lancaster (1966) formulation of letting the argument “health” enter the utility fimction and then deriving a demand for medical services. The present data do not allow us to estimate the relation trans- forming medical services into health. 6 These elasticities are only approximate in the long run if insurance premiums are adjusted to reflect the changes in utilization. Copyright © 2001 . All Rights Reserved. DEMAND FOR MEDICAL SERVICES 599 Efikcts qf a Change in Income Exogenous changes in income can arise either from a change in earnings per hour or from a change in nonearned income. The two effects are not, in general, equal. The assumptions that are sufficient to make money function as a price are also sufficient to mean that an increase in nonearned income will produce an increase in the demand for medical services. The effects of a change in the wage rate cannot be determined a priori because of offsetting influences. An increase in earnings per hour produces an income effect, which acts to increase demand. It also raises the opportunity cost of time, which reduces demand for time-intensive activities. The net effect on the demand for medical services depends on the time intensity of the price of medical services relative to the time intensity of the price of all other goods and services. It can easily be established (see Acton 1973a or 1973b) that the substitution effect of a change in the wage rate on the demand for medical services is positive if and only if w: wt —— > —-— , (3) (q + w) (p + w) that is, if the time price is a larger proportion of the total price for the composite good, X, than it is for medical services, m. The substitution effect is necessarily negative for free sources of medical care since the condition in equation (3) will not be met as long as there is a nonzero monetary price for X. Of course, the net effect of a change in wages may still be to increase the demand for medical services if the income effect exceeds the substitution effect. Intuitively, however, the effect of a wage change on the demand for medical services is more like a price effect for free sources of care (and therefore more likely to be negative) and more like an income effect for nonfree sources of care (and therefore more likely to be positive). Predictions from Other Formal Models As noted above, the model developed here is deliberately simplified because it is adequate to produce testable hypotheses for the variables of primary interest. There are some additional hypotheses regarding the effects of education and age from the Grossman (1972a) investment model that can be tested with these data. Grossman argues that medical services are combined with other inputs the individual supplies to produce health and that it is health that enters the utility function. Now, if education raises health productivity (e.g., more highly educated people are more skillful in combining inputs to produce health) and if the price elasticity of health is less than one, then education will have a negative effect on Copyright © 2001 . All Rights Reserved. 600 JOURNAL OF POLITICAL ECONOMY the demand for medical care.7 The second implication of Grossman’s work involves the effect of age on consumption of health services. If the price elasticity of demand for health is less than one, then the demand for medical services will be positively correlated with the depreciation rate on health. In general, empirical evidence suggests that the depreciation rate increases over the life cycle, causing a positive effect of age on con- sumption of medical care. HI. The Empirical Base The data used for this study come from a 1965 survey of users of the out- patient departments of New York City municipal hospitals. Respondents were selected from a random sample of persons at the clinic; hence, the . probability of being interviewed is proportional to the frequency of use of the clinics. The Appendix defines and presents the mean values of the variables used in this analysis. In addition, the means are reported when observations are weighted by the inverse of the number of clinic visits. These weighted means indicate the mean characteristics of people who ever use the clinics, rather than the mean characteristics of the patient - loads at the clinics. The respondents were questioned about their previous year’s medical use and a number of sociodemographic characteristics from the previous year. The interviews at each of the OPDs were conducted in four waves spread throughout the year to cancel seasonality in usage.8 There are advantages and disadvantages to using survey data. One of the advantages is that disaggregated data provide a more precise description of individual behavior because individual rather than aggregate values can be used for explanatory variables. This overcomes the bias away from zero that . is frequently encountered in using average values in aggregate data.9 Other advantages include the much larger sample size typically available in surveys so that variance of coefficients can be reduced. The chief .. disadvantage of survey data in general is that it relies on recall by the individual. This frequently leads to an underreporting of some variables, particularly medical utilization and income.10 ‘ 7 In the consumption model, given a “neutral” effect of education on all household activities, the elasticity with respect to wealth must also be less than one to produce a . negative relationship between medical care and education (Grossman 197217, pp. 36—37). 3 Details of the study and selected analysis (chiefly analysis of variance) can be found ~ ’ in Lerner and Kirchner (1967) and Lerner, Kirchner, and Dieckmann (1967, 1968). 9 See Newhouse and Phelps (1974) for an elaboration of this point. . ‘0 Recall for ambulatory care seems to be extremely accurate for physician use in the last 2 weeks and for hospital use in the last year, according to Regina Loewenstein (personal communication, 1971). Underreporting will generally bias the coefficients of explanatory variables toward zero. The estimated elasticities, however, need not be , “WW.W. .. Copyright © 2001 . All Rights Reserved. .imm....-mmm_.mmmwmw.. We...“ M. .. l DEMAND FOR MEDICAL SERVICES 601 Some responses to the surveys were checked independently, increasing the validity of the data, while others were coded poorly for present analysis. The number of outpatient department visits (OPD) and hospitalizations (HOSP) in the preceding year were verified with providers. The distance (DIST) to the hospital of interview was calculated from the person’s home address. The number of private physician visits was coded in intervals and had to be calculated. This calculation and all important variables are discussed in detail in Acton (1973b). IV. Estimation Results The demand for health care by type of provider is estimated from a simultaneous-equation system using two-stage least squares. Four struc- tural equations are specified and 28 exogenous variables are used for estimation. All the equations are overidentified by several variables.11 After a brief overview of the model, this section describes first the structural equations and then the reduced-form equations. Overview of the Model The four structural equations describe the volume of ambulatory and inpatient services demanded and the distance traveled to the free am- bulatory care.1 2 The last year’s volume of CPD visits and the number of biased. If all people recall k proportion of their utilization in the previous year, then the estimated elasticity of demand for care with respect to, say, price is film I L __ 8m 9 """‘ av km 31; m’ which is the same as the elasticity that would have been estimated with full recall. By the same argument, the cross-price elasticities and the elasticity of substitution of one type of provider for another should be unbiased. There is further potential bias in this particular study because of the fee structure at the municipal hospitals. In order to receive free care, the individual was supposed to be unable to pay for private care (although I am told that it was well known at the time that anyone who asserted he was entitled to free care would receive it without challenge). This institution, however, might have caused some persons to underreport income or utilization of private facilities. There is no evidence that such an error was introduced and the interview opened with a statement of confidentiality, but in the absense of verified income data the possibility exists. 1‘ Since a number of the variables are dummy variables for health status or mode of transportation, they really contain less information than their number indicates. There— fore, in checking the number of excluded exogenous variables for purpose of identification, I counted a set of mutually dependent dummy variables as one variable. ‘2 Some of the “exogenous” variables in this cross-sectional survey are really endog- enous in a larger system that includes life-cycle behavior (such as labor-force participation and family size) and a broader set of economic decisions. Since we are specifying only a subset of these relations, it is possible that some of these exogenous variables are really proxies for common underlying theoretical variables. I tried to limit the possibility of such undesirable interference by specifying such variables as family health status only in the equations where their principal effects could be anticipated. Copyright © 2001 . All Rights Reserved. 602 JOURNAL OF POLITICAL ECONOMY visits to a private physician (PRIVE) are the two alternative sources of ambulatory care. They are technical substitutes for one another. The volume of private-physician visits and the number of hospitalizations are entered as right-hand endogenous variables in order to measure sub- stitution and complementarity efl'ects among different types of providers. We would prefer to enter the money price of private-physician visits and the cost of a hospitalization and examine the cross-price effects, but this information is not available in the survey. Since substitution or com- plementarity is important both theoretically and empirically in medical care, we wish to measure the effects as best possible and limit the bias in the remaining coefficients. One consequence of entering quantities on the right-hand side is to make price elasticities different from the usual elasticities, as discussed next. The distance (DIST) to the outpatient department where the interview was conducted is specified as an endogenous variable in this system. It is the chief measure of the price of an OPD visit and functions as a cross- price term for PRIVE and HOSP.13 Distance actually measures several things. It includes the physical distance one has to travel and the money and time costs of travel, and it is associated with higher informational costs. The informational costs represent the fact that patients generally have less difliculty in finding out about the quality and suitability of a close-by clinic (for instance, by asking neighbors or by having experienced the care themselves) than in finding out about a distant clinic. The money costs and informational costs will tend to be positively correlated with the distance traveled. However, those who previously lived near one clinic and now live farther away (perhaps nearer another) have a lower in- formation cost with the former, more distant, clinic. Therefore, the coefficient on HABIT should be positive in the structural equation for distance, but it will capture only part of the differences in information costs in this and other equations. DIST is endogenous in this system because the distance a person travels to the CPD is influenced by a number of the same underlying variables influencing demand for medical care, including the opportunity cost of time, the health status of the user, and his sociodemographic characteristics. Consequently, the only way to see the total effect of these variables is through the reduced-form coefficients. The result of this specification is that the elasticities with respect to DIST are not identical with the usual price elasticity of demand because they show the effect of distance holding constant the level of all other things, in- cluding other forms of medical care. With three or more goods, it is not pos- sible to sign bias without stronger assumptions about the utility function. 1 4 l 3 An imprecise measure of waiting time is included. If the person says he waits longer at an OPD than at a private physician’s office WAIT take the value one. 1‘ With only two goods, the own-cross-price elasticities estimated holding constant the quantity of the other good are lower bounds (in absolute value) for the normal Copyright © 2001 . All Rights Reserved. "Mam..www~wmmmw.w a. DEMAND FOR MEDICAL SERVICES 603 The important exogenous economic variables in this system are earned and nonearned income (GRWAGE and NWAGE), assets (LIQUID and EQUITY), and health insurance (INSF). The insurance variable is very imprecise and indicates only that at least one member of the family had some coverage (this was before Medicare and Medicaid). Additional important variables in this system include the mode of transportation used for the named sources of care (the suffix letter P indicates mode of private care), health status, and a number of sociodemographic variables.1 5 The Structural Equations The specifications of the structural equations and their estimated co- eflicients are given in table 1.16 In the absence of an explicit utility function, there is no unique set of structural equations. A linear specifica- tion is employed. The principal value of the structural equations is to show the effects of endogenous variables on one another. The net effect of exogenous variables is clear only from the reduced-form coefficients in table 2. The OPD Equation The structural equation for OPD visits (1) includes the other three endogenous variables. Private-physician visits are technical substitutes for OPD visits, producing an elasticity of — .l l. Hospitalizations appear to be significant technical complements to OPD care with an elasticity of .47. This complementarity is reasonable, first, because inpatient care frequently requires ambulatory follow-up. Second, hospitalizations elasticities. With three or more goods, we need stronger assumptions (such as, hospitaliza- tions are complementary to one form of ambulatory care and substitutes to the other form) because of offsetting effects to determine the sign of the bias unambiguously. ‘5 It can be argued that the method of transportation is an endogenous variable just as is the distance traveled to OPD care. For both conceptual and practical reasons, it is considered exogenous in this model. First, it can be viewed as exogenous if before the period of observation the person has already made a decision about the methods of transportation he will use for various types of trips. We can argue that he is unlikely to alter this choice significantly during the course of the year. A more practical reason is that making the method of transportation endogenous means that the normal assumptions about the distribution of the error term would not be satisfied, and either a Tobit or probit model would be more appropriate. Estimating a simultaneous Tobit system with 12 or so endogenous variables is probably unwarranted with this data base. 1‘ The t-values reported are the asymptotically normal values. By multiplying these by the t-adjustment factor, the Dhrymes (1969) t-statistics for ZSLS result. Copyright © 2001 . All Rights Reserved. . P 8. m3 3% :.r a.“ mad: 6... we 3% 8% 8..“ 5:7 8.1 3.0 $0.? 8. :.N +8.0 8. :2 :3 8. m3 god 8. 5: N85 2.: +3 30.01 :o. 3.0 :3 8.1 2: 8051 e... 9....“ new 2. cam «85 8. 92 "No.0 :. 3; wood e . e80 e N “80 EL .95 Ema :2 .amv mmom 8.: mg 32: 8...| ix wfi.._| mu. afifi «.2; 5. EN mwwd 5. mm; Sod 3. mm: mite 3. 9mm ~26 8. Now N23 8. mad omod am... new if “NI min unodl r g @000 AS .amv mag no. | No. l .3. 3. No. | No, I co. | +0. .3. if S. _._...l t no; em; omd and mud and E .0 $4 me; u mmo._| . . . . . . . . . . . . . . . . . . . . .. QOOO was; I . . . . . . . . . . . . . . . . . . . . . . . . . .Nm N36 . . . . . . . . . . . . . . . . . . . . .. $2.3m 5+6 . . . . . . . . . . . . . . . . . . . . . 28:5» Eqde . . . . . . . . . . . . . . . . . . . . . . . . mmzm . . . . . . . . . . . . mlA: X STOI . . . . . . . . . . .. «In: x QHDUHA mmmd . . . . . . . . . . . .. T2 x mO<>>Z o5.— . . . . . . . . . .. «12 x mO<>>MO "gonomcxm Nmmdl . . . . . . . . . . . . . . . . . . . . . . thHQ .v$.: . . . . . . . . . . . . . . . . . . . . . . .503 End! . . . . . . . . . . . . . . . . . . . . 1m>5~m . . . . . . . . . . . . . . . . . . . . . . . . . . . QmO "maoaomowcm c800 mmqm5m<> a: .amv Bo *zoF<Sme wqu ESE mzoF<de ASSHUDMFw mo Ezflofimmoo ~ “mng 604 Copyright © 2001 . All Rights Reserved. .m-wmmmw“hfiwum _ ., . 8.. .1 .32”. vuuswmozza we ESE of an coon—~5in 336335 .mutmumamfiu «o: “muflumoEvs 3:0 “mafia BOSE mam—mum a o3 $2 2&6 £4 mad OBS m5? mi: mmd an. wmm.~ nafiw mum—é wwmd oméV 2.0 owed HZ<BmZOU 21......mZHAO AUQNH mM<O . . mNmDm ...........m~mDm . .UQwH IM<O ZNmDm 353m . 22724222 MMO?» ...mA<2 ........MO<€E Tofime/w 88$ 3253??“ 5 O 6 Copyright © 2001 . All Rights Reserved. . . . 39° . . 0.36 NS: nmw m . . . . . . . . . . . . . . . . . . . . . . . HZ<HmZOO 8.| mmcdl _o.| Noodl ~o.| Bodl S. “No Q . . . . . . . . . . . . . . . . . . . . . . . . . . . . . mZHJU 8.| wwodl 8. X56 no. 285 oo.I «we; I . . . . . . . . . . . . . . . . . . . . . . . . . . . . mnxfih 8.| Sod] cc. wwod me. .vNNN oo.I mwndl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3330 8. | ewodl 5. Quad we. «find a. I 3;.— | . . . . . . . . . . . . . . . . . . . . . . . . . . . . . MNmDm 5.| muodl 5. mwcd E. mwfim 5.I mnoél . . . . . . . . . . . . . . . . . . . . . . . . . . . . . mamDm 8. m9: 5. 3.06 8. mid 8.I 9‘“de . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Egg 8. 8nd 5. +36 5. mmwd oo.l mohdl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . #35 mm. mufu mo. ~86 .3. Sad No. | .3de . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .NmDm mu. mac; *6. Sod no. 0— #6 No. | mmwdl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ~mDm S. | wocodl 5. 25.0 No. I noodl 0—. mid . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bug 5. mime co. Sod 5. R26 No.| mondl . . . . . . . . . . . . . . . . . . . . . . . . . . . . ZHEZ 5. god mo.| nmcdl No. NNNd Nc.| nowdl . . . . . . . . . . . . . . . . . . . . . . . . . . . . MMO?» co. | coodI 3. 53.0 no. I nmmdl 8. I scodl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ham; mc. :26 0—. :86 $.l Kgdl .vo.l vmwdl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . m3; 5. I Noodl mo. wood 8. mad 3. l awoél . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .UDQm me. | emu.ol 5. | moodl 8.| Ngdl 8. I $64 I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mm no. | E.de .3. god no.l mowdl we. | 59— I . . . . . . . . . . . . . . . . . .I . . . . . . . . . . MUEQ we. 28.0 e_.l iodl 5.! mocodI on. :36 . . . . . . . . . . . . . . . . . . . . . . . TS x WO< oo.| mooodl vo.| ovodI m_.l mmmdl mQI m:.ol . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. MFG mo. mwwd m_.l g—dl $.I Ava—AI afil monél . . . . . . . . . . . . . . . . . . . . . . . . . . . .. GOOD 8. mead 3. I midl we. I Gov; I me. | ~5le . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Nm 8. N86 5. wood 8. | 36.0! we. own; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333nm 5. | wgdl No. mwod MN. new; 8. mwod . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . P745 co. .806 he. mood mo. 8nd 5. 9.2.0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .meH co. 56.0 8. Sod 8. «bod 8. god . . . . . . . . . . . . . . . . . . . n L: x pan—Um 8. wood 8. 936 8. Rte 8. 8—6 . . . . . . . . . . . . . . . . . . . vie x QHDUHA No. 356 mo. I Eodl we. mm—d _o.| mwfidl . . . . . . . . . . . . . . . . . . . . n12 x NO<>>Z «O. :_.o we. | mmodl no. . . . . . . . . . . . . . . . . . . . uuoO Em .amv Ema {m *mzoFSudm Exchénopnmd no EEO—E00 AE .ua mmom “woo r «you :3 .95 ESE N mamflh mmmd S I n+wd I v :2 x HO<>>MO If # uvoU 5§> :3 .3: 80 Copyright © 2001 . All Rights Reserved. 606 DEMAND FOR MEDICAL SERVICES 607 produce a budget efl'ect that may increase the demand for less expensive forms of ambulatory care. Third, the municipal OPD clinics act as ports of entry to municipal hospitals. One of the important predictions of the formal model developed above is that distance to the CPD would function as a price in determining demand. Equation (1) supports this prediction, producing an elasticity of — .14. It is likely that the elasticity with respect to travel time is greater (in absolute value) than —.14- because people traveling farther will tend to take more rapid forms of transportation.1 7 The mode of transportation to OPD care has its major influence on the distance traveled and is included in the DIST structural equation. The PRIVE Equation The structural equation for private physician care complements the estimates from the CPD equation. Again, PRIVE and OPD are sub- stitutes and PRIVE and HOSP are complements. DIST functions as a cross price to PRIVE, producing an elasticity of about .07. The estimate of .07 is considerably smaller than the cross—time elasticities of about .6 estimated for private care in Acton (1973a). Again, it is likely that the elasticity with respect to travel time is greater. An additional downward bias in the coefficient may be caused by the omission of a measure of travel distance from the structural equation for PRIVE. All we have to measure travel distance are the dummy variables for method of trans- portation. In this structural equation, the partial effects indicate that walkers demand the least care, while taking the bus, subway, or taxi all have about the same effect. The statistically significant coefficients on mode of transportation suggest that distance is playing an important role in demand (this conclusion is supported by the reduced-form co- efficients reported in table 2). The Hospitalization Equation The structural equations for hospitalization reveal effects for the endog- enous variables consistent with the other equations. Clinic and private- physician visits are complements to hospitalization, with elasticities around .14 and .10. The coefficient on distance to an OPD has a positive sign (elasticity = .18) 17 For instance, it may not be worth taking a bus a few blocks, but it is worthwhile when crossing town; similarly, it may be worthwhile to switch to an express train for part of a longer journey by subway and then return to a local. The own-price elasticity of demand for public ambulatory care with respect to self-reported travel time is estimated between - .6 and — 1.0 in Acton (1973a). Copyright © 2001 . All Rights Reserved. 608 JOURNAL OF POLITICAL ECONOMY The Distance Equation The only included endogenous variable in explaining distance to the CPD is volume of CPD services, arguing that those who go more frequently will want to reduce the travel price they have to pay. The results of equation (4) show a coefficient not statistically significantly different from zero. The distance equation is specified as a function of a number of the exogenous variables that are common to the medical care equations— including earned and nonearned income, mode of transportation, health status, and sociodemographic characteristics. The most important theoretical prediction for this equation is that those with a higher oppor- tunity cost of time should travel shorter distances for OPD care. The coefficient on GRWAGE in equation (4-) does not support this prediction, although its t-ratio is only 0.76. The general effects of the opportunity cost of time follow the predictions as indicated by the coefficient on GRWAGE in the reduced-form equations for medical care. The Reduced-Farm Equations The chief value of the reduced-form equations is to indicate the net influence of the exogenous variables on the dependent variables. Con- sequently, the effects of a few key variables will be examined across equations, and the estimated coefficients will be compared with the predictions of Section II. The Role of Travel Distance and the Value of Time There were several predictions about the role of time on the demand for various types of care. The first prediction, that time (and travel distance) would function as a price, was supported by the structural equations. They reveal a negative (own-price) coefficient on the variable DIST in the CPD equation and a positive (cross-price) coefficient in the PRIVE and HOSP equations. Second, we expect those with a higher opportunity cost of time to demand less time—intensive care. Both working people and people with a higher opportunity cost of time demand less time-intensive OPD and hospital care and more private-physician care. A further hypothesis, that the higher wage rate would shorten travel distance, was not supported by equation Income and Assets Unless service from a provider is an inferior good, the nonearned income elasticity should be positive. Equations (5), (6), and (7) show a pos- itive elasticity of demand for private-physician services and a negative . “flew-«WWW», V, Copyright © 2001 . All Rights Reserved. .m.mW—~WMWWM ,,,,,,, n DEMAND FOR MEDICAL SERVICES 609 elasticity for OPD and hospital care, suggesting that private outpatient care is a normal good and municipal health care is an inferior good (in an economic sense). The effects of earned income depend on the time intensity of medical goods relative to other goods and services. The model predicted that earned income would function more like a price effect in the demand for free care and more like an income effect in the demand for nonfree care. The reduced-form coefficients show a positive elasticity of demand for private outpatient care with respect to GRWAGE and a negative elasticity of demand for public care. This supports the popular impression that OPD and hospital care are time intensive relative to private physicians’ care and relative to all other goods and services. Education and Age Two other predictions from the formal models involve the effects of education and age. Grossman suggested that, if those with more education were more efficient producers of health, education would have a negative coefficient (as long as the price elasticity of demand for health is less than one). Equations (5) and (6) show a negative effect of education on OPD visits and a positive effect on private-physician visits. However, the decrease in the number of OPD visits is only partly made up by increases in the number of private visits so that the net change in ambulatory visits produced by an increase in education is negative. This finding is compatable with the predictions of Grossman’s (1972b) model discussed above. The investment model also predicted a positive correlation of age and the depreciation rate on health. The present data on ambulatory utilization show a negative (but very small) elasticity for PRIVE and a positive net effect of age for OPD care. The OPD finding supports the hypothesis that depreciation in the health stock increases with age. The negative coefficient on age in the hospitalization equation may indicate older persons going to nursing homes. The remainder of the coefficients describing health status, insurance, and sociodemographic characteristics of the population generally conform to the expectations based on other studies. As found by most researchers, those in poorer health demand more inpatient and outpatient care. They also travel shorter distances to receive their OPD care. The effect of at least one family member’s having fair or poor health (FAMH) reinforces the health status effects. In general, the positive correlation of health status in the family and the budget effect of a very sick member should either increase utilization at each source of care or cause a shift to public care; FAMH is positive in the OPD equation (5). Although it is an imprecise variable, if anyone in the family has health insurance, the person will seek more medical services, both public and private. Blacks and Puerto Ricans generally receive less care of all types than do their Copyright © 2001 . All Rights Reserved. 6IO JOURNAL OF POLITICAL ECONOMY white counterparts, and they travel shorter distances to receive it. Finally, men seek less ambulatory care and more inpatient care than women. This finding provides further support for the suggestion of Acton (1973a) that males may let their health deteriorate further than females do before seeking medical care, so that when they go they require more intensive care. V. Conclusion This study supports the prediction that travel time (as measured by distance) functions as a price in determining the demand for medical services when free care is available. This survey of users of the municipal OPDs indicates negative own-price elasticities with respect to travel distance at free providers and positive cross-price elasticities for nonfree providers of care. Further, the estimated distance elasticity approaches or equals the money-price elasticities that have been estimated by a num- ber of researchers.18 The predicted negative effect of earned income on distance was not found, but persons with higher earned income are more likely to use the private sector, which is relatively less time intensive, than the public sector. / There are a number of policy implications of this study. Two of the most important involve (l) the redistribution in services that will be caused by a change in money and time prices and (2) the possibility of using income subsidies rather than direct provision of goods to meet public objectives. As money prices out of pocket are reduced, because of either the continued spread of private insurance or the enactment of a federal health insurance scheme, demand will become more responsive to time prices. In turn, this will permit persons with a lower opportunity cost of time to bid services away from those with a higher opportunity cost of time because they will face a price that is relatively less costly. This conclusion holds even if there is not differential coverage by income class and even if there is no supply side response increasing the time prices. It is likely that a shift in demand will be accompanied by an increase in the time needed to receive a unit of medical services.19 This will further 18 F eldstein (1971), Davis and Russell (1972), and Rosett and Huang (l973) have all reported money-price elasticities in the range —.5 to — 1.0, although there is reason to believe these estimates are biased upward (see Newhouse and Phelps 1974). Other more conservative measures of the money-price elasticity (using several data sources) place it around —0.15 or less (see Phelps and Newhouse 1974). ‘9 This supply response is likely for a number of reasons. First, it may be optimal from the point of view of the provider to have a queue to even out the variation in demand that he experiences without having to invest in significant excess capacity. A shift in demand will generally cause the optimal queue length to change (for instance, the oppor- tunity cost of an idle moment of the supplier’s facility is higher). Second, the suppliers may not be profit maxirnizers, so that they do not respond to a shift in demand by charging ._.m.mm.m.m..mwmflmm_.wwwaM ‘ Copyright © 2001 . All Rights Reserved. DEMAND FOR MEDICAL SERVICES 61! increase the relative shift in favor of those with a lower opportunity cost of time. The significant effect of travel distance on the demand for medical services suggests a policy instrument for delivering more services to target groups. By moving clients closer—either by improving transportation, locating clinics closer, or by establishing satellite clinics around central facilities—the consumption of selected populations can be increased. A second important policy implication is for alternative means of meeting the objective of increasing medical services consumed by target populations. In one form or another most proposals reduce to a sub- sidized provision of services, whether through social insurance schemes such as Medicaid or various federal health insurance proposals, or through direct provision of care as in neighborhood health centers or the requirement that Hill-Burton hospitals provide charity care. But as Davis (1972) has correctly noted, there is seldom a consideration of the extent to which changing the income distribution will alleviate the desire to subsidize the medical purchase. Even in the administration’s proposals for income maintenance (FAP) and subsidized medical care for the poor (FHIP) there was little discussion of the degree to which one can be substituted for the other. The proposed comprehensive health insurance plan (CHIP) now before Congress provides an opportunity to illustrate the tradeoffs. Income subsidy will not, in general, meet the objective of risk spreading for medical expenses (unless it induces a significant demand for health insurance), but it will raise the average demand for medical services in the subsidized population. The equations reported in tables 1 and 2 put us in a position to address this question of substituting income maintenance for subsidized medical care to achieve a given increase in medical services consumption. Since income maintenance is a nonearned source of income, the elasticity of demand for medical care with respect to changes in nonearned income is used. A hypothetical example, not based exactly on FAP and CHIP provisions, will serve to illustrate. The estimation results in table 2 indicate that a $1,100 increase in nonearned income for a family with a current nonearned income of about $400 and earned income of about $2,900 (in 1965) will produce an increase of about 11 percent in the demand for private practitioners’ care per person. If the money-price elasticity of demand for ambulatory medical services is around — .15 over the range under consideration,20 and the out-of-pocket the highest possible monetary prices but instead allow time prices to increase. In particular, physicians may be income satisfiers rather than maximizers. See Newhouse (1970), Frech and Ginsburg (1972), and Newhouse and Sloan (1972) for a discussion of physician pricing behavior. Third, there may be a conscious attempt to redistribute services by discriminating in favor of those with a lower opportunity cost of time. See Nichols, Smolensky, and Tideman (1971) for a discussion of the first and third points. 2° The actual money-price elasticity may be even lower than this. See Phelps and Newhouse (1974) for a discussion of the price elasticities in several published reports. Copyright © 2001 . All Rights Reserved. 612 JOURNAL OF POLITICAL ECONOMY expenditure is reduced from 25 percent of money price to 15 percent (the upper limit on CHIP’s coinsurance rate and the rate for an insuree with income of $3,000), then the demand for private care will increase by 7.5 percent. Clearly, one means of increasing private medical consumption by the poor is income supplementation, and the magnitude of the change may be comparable over the range of subsidy and income guarantee under consideration. Appendix Definition of Variables Used and Their Mean Values21 AGE = Age in years. Means = 35.6, 31.1. BLACK = Dummy variable equaling one if Negro or indeterminable, or other than Puerto Rican, Mexican-American, American Indian, or other white. Means = 0.38, 0.41. BUSl = Dummy variable equaling one if patient’s usual means of trans- portation to the clinic requires one bus or train. Means = 0.42, 0.42. BUSZ = Dummy variable equaling one if patient’s usual means of trans- portation to the clinic requires two or more buses or trains. Means = 0.23, 0.21. BUSlP = Dummy variable equaling one if patient’s usual means of trans- portation to private doctor requires one bus or train. Means = 0.07, 0.08. BUS2P = Same as above except two or more buses or trains. Means = 0.03, 0.03. CAR = Dummy variable equaling one if patient’s usual means of trans- portation to the clinic was by car driven either by individual or a friend. Means = 0.05, 0.06. ‘ CARP = Same as above except to private doctor. Means = 0.03, 0.03. CLINS = Total number of clinics used last year. Means = 1.71, 1.29. DIST = Distance in miles to the hospital of interview. Means = 2.14, 2.13. EDUC = Highest grade completed, in years. Means = 6.25, 6.23. EQUITY = Equity in home. Means = $207, $198. ERNRS = Number of earners in the family. Means = 0.83, 0.91. EX = Dummy variable equaling one if health status of patient is excellent. Means = 0.095, 0.13. FAIR = Dummy variable equaling one if health status of patient is fair. Means = 0.30, 0.27. FAMH = Dummy variable equaling one if all family members reported health as good. Means = 0.34, 0.33. GOOD = Dummy variable equaling one if health status of patient is good. Means = 0.46, 0.48. GRWAGE = Gross annual earnings from all wage earners in the family. Means = $2,929, $3,215. HABIT = Number of years patient has been coming to current clinic. Means = 5.03, 4.10. 21 The first mean value is for the unweighted data and the second is for the data weighted by 1/OPD to adjust for the probability of being sampled. Copyright © 2001 . All Rights Reserved. ..... ammmmmuwuwm . DEMAND FOR MEDICAL SERVICES 613 HOSP = Number of hospitalizations last year. Means = 0.32, 0.30. HSSIZE = Number of persons in individual’s household. Means = 3.67, 3.92. INSF = Dummy variable equaling one if any family member has medical insurance. Means = 0.35, 0.36. LIQUID = Liquid assets. Means = $177, $207. LNGWT = Dummy variable equaling one if patient had to wait a long time before being taken care of at the hospital where he was inter— viewed. Means : 0.59, —. MALE = One if male, zero if female. Means = 0.38, 0.39. NMAIN = Dummy variable'equaling one if main source of medical care is other than the same clinic as at time of interview. Means = 0.047, 0.064. NWAGE = Nonearned family income in last year. Means = $878, $802. OPD = Number of visits in last year to a physician in a clinic. 'Means = 7.65, 2.97. PR = One if Puerto Rican; zero otherwise. Means = 0.34, 0.35. PRIVE = Number of visits in last year to a physician in his private office. Means = 1.46, 1.63. TAXI = Dummy variable equaling one if patient’s usual means of trans- portation to the clinic was by taxi. Means = 0.091, 0.095. TAXIP = Same as above except to his private doctor. Means = 0.021, 0.020. WAIT = Dummy variable equaling one if patient had to wait longer in the hospital where he was interviewed than in private doctor’s office. Means = 0.28, —. WELF = Dummy variable equaling one if individual had some type of welfare assistance. Means = 0.24, 0.22. WORK = One if indivudal worked either full or part time. Means = 0.16, 0.20. References Acton, Jan Paul. “Demand for Health Care among the Urban Poor with Special Emphasis on the Role of Time.” Memorandum R-115l-OEO/NYC, RAND Corp., April 1973.(a) . “Demand for Health Care When Time Prices Vary More than Money Prices.” Memorandum R-1189-OEO/NYC, RAND Corp., May 1973.(b) Becker, Gary. “A Theory of the Allocation of Time.” Econ. J. 75 (September 1965): 493—517. Davis, Karen. “Health Insurance.” In Setting National Priorities: The 1973 Budget, edited by Charles Schultze et a1. Washington, D.C.: Brookings Institution, 1972. Davis, Karen, and Russell, Louise. “The Substitution of Hospital Care for Inpatient Care.” Rev. Econ. 5mm. 54- (May 1972): 109-20. Dhrymes, Phoebus. “Alternative Asymptotic Tests of Significance and Related Aspects of 2SLS and SSLS Estimated Parameters.” Rev. Econ. Studies 36 (April 1969): 213—26. Feldstein, Martin S. “An Economic Model of the Medicare System.” Q.J.E. 85 (February 1971): 1—20. Frech, H. E., and Ginsburg, Paul B. “Physician Pricing: Monopolistic or Com- petitive: Comment.” Southern Econ. J. 38 (April 1972): 573—77. Grossman, Michael. “On the Concept of Health Capital and the Demand for Health.” J.P.E. 80, no. 2 (March/April 1972): 223—55.(a) Copyright © 2001 . All Rights Reserved. 614 JOURNAL OF POLITICAL ECONOMY . The Demand for Health : A Theoretical and Empirical Investigation. New York: Columbia Univ. Press (for Nat. Bur. Econ. Res.), 1972.(b) Harris, Seymour E. “The British Health Experiment: The First Two Years of the National Health Service.” A.E.R. 41 (May 1951): 652—66. Holtman, A. G. “Price, Time, and Technology in the Medical Care Market.” J. Human Resources 7 (Spring 1972): 179—90. Lancaster, Kevin. “A New Approach to Consumer Theory.” J.P.E. 74, no. 2 (April 1966): 132—57. Lerner, Raymond C., and Kirchner, Corinne. Municipal General Hospital Outpatient Population Study: Social and Economic Characteristics of Patients in New York City Outpatient Departments, 1.965: Financial Eligibility under Medicaid and Potential Reimbursement to the City. Report no. 1. New York: School Public Health and Admin. Medicine, Columbia Univ., 1967. Lerner, Raymond C.; Kirchner, Corinne; and Dieckmann, Emil. Municipal General Hospital Outpatient Population Study: Social and Economic Characteristics of Patients in New York City Outpatient Departments, 1965 : Methodology. Report no. 2. New York: School Public Health and Admin. Medicine, Columbia Univ., 1967. . New York Municipal General Hospital Outpatient Population Study, 1.965: Data on Background, Medical Care Utilization and Attitudes of Outpatients, by Hospital. Report no. 3. New York: School Public Health and Admin. Medicine, Columbia Univ., 1968. Leveson, Irving. “Demand for Neighborhood Medical Care.” Inquiry 7 (December 1970): 17—24. Newhouse, Joseph P. “A Model of Physician Pricing.” Southern Econ. J. 37 (October 1970): 174—83. Newhouse, Joseph P., and Phelps, Charles E. “On Having Your Cake and Eating It Too: A Review of Estimated Effects of Insurance on Demand for Health Services.” Memorandum R-1149-NC, RAND Corp., April 1974. Newhouse, Joseph P., and Sloan, Frank A. “Physican Pricing, Monopolistic or Competitive: Reply.” Southern Econ. J. 38 (April 1972): 577—80. Nichols, D.; Smolensky, E.; and Tideman, T. N. “Discrimination by Waiting Time in Merit Goods.” A.E.R. 61, no. 3, pt. 1 (June 1971): 312—23. Phelps, Charles E. “Demand for Health Insurance: A Theoretical and Empirical Investigation.” Memorandum R-1054-OEO, RAND Corp., July 1973. Phelps, Charles E., and Newhouse,]oseph.“The Effects of Coinsurance on Demand for Physician Services.” Memorandum R-976—OEO, RAND Corp., June 1972. . “Coinsurance, the Price of Time, and the Demand for Medical Services.” Rev. Econ. and Statis. 56, no. 3 (August 1974-): 334—42. Rosett, Richard N., and Huang, Lien-fu. “The Effects of Health Insurance on the Demand for Medical Care.” J.P.E. 81, no. 2 (March/April 1973) : 281—305. Copyright © 2001 . All Rights Reserved. ...
View Full Document

This note was uploaded on 02/08/2012 for the course ECON 388 taught by Professor Chaka during the Spring '11 term at SUNY Albany.

Page1 / 20

Acton 1975 - Nonmonetary Factors in the Demand for Medical...

This preview shows document pages 1 - 20. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online