Telecommunications and cost savings in health care services.

I. Introduction

Public concern and government attention has focused on the explosive rise in health care costs. Analysts in various academic and professional fields have investigated the ways in which health care costs can be controlled while improving the quality and availability of medical services. While several policy- and technology-based options have been examined, testimony presented at recent congressional hearings attests to the potential for telecommunications services, in particular, to broaden the availability and quality of health care services while moderating the costs of those services.(1) Practicing physicians, medical researchers, appointed officials, and other medical specialists have extolled telecommunications' role in enabling doctors to provide better service. All parties familiar with the ways in which telecommunications services have benefited health care urge policy makers to recognize these contributions and ensure that telecommunications, as a relatively low-cost means of improving quality and access, continue to be exploited in the provision of health care services. As summarized by J. Michael McGinnis, M.D., Deputy Assistant Secretary for Health,

As we look for ways to improve our effectiveness in reaching the public and health professionals alike with timely, relevant, and accessible health information, we look to new communication technologies not to solve our problems, but to provide us with the opportunity to develop new, creative solutions to improving the public health [8].

The health care sector is not unlike other sectors in its attempts to improve productivity and quality through increased usage of telecommunications services. Aggregate and sector-specific research conducted by DRI/McGraw-Hill has revealed a rapid increase in almost every industry's real usage of telecommunications services between 1963 and 1991, owing partly to steady declines in the price of telecommunications relative to other inputs. In each industry, as telecommunications displaced less efficient and more costly inputs, productivity increased: since 1978, about 25% of total direct and indirect aggregate productivity gains resulted from advances in telecommunications production and enhanced consumption possibilities for end-user industries [5]. These productivity gains saved the 1991 economy $102 billion in primary resources, excluding the savings enjoyed by the health care services and educational services sectors.

How, then, have doctors, hospitals and researchers used telecommunications services to control costs and improve access and quality? And, given the staggering growth in health care costs, what cost savings associated with the health care industry's usage of telecommunications can be calculated? This paper will focus on presenting the statistical methods that researchers at DRI/McGraw-Hill have used to answer the second question. In the first section, we assess patterns of telecommunications usage in the health care sector. In the second section, we describe the econometric modeling techniques used to estimate cost savings associated with telecommunications usage. In the third section, we present the results of this research and we detail statistical findings that help substantiate hypotheses that the health care services industry may have foregone substantial cost savings by failing to more aggressively employ telecommunications technologies. Emphasis will be placed on using the modeling procedure to integrate and expand existing projections of health care costs, and to estimate the ways that telecommunications services might reduce health care costs through the year 2000.

II. Cost Savings Due to Telecommunications Advances: The Issue in Context

Limited Access and Growing Expenditures

Expenditures on hospital care, physicians, drugs and other medical nondurables, nursing home care, and other medical goods and services rose from $250 billion in 1980 to $666 billion in 1990, and expenditures are forecasted to reach $1.679 trillion by the year 2000 [7]. From 1965 to 1990, the proportion of Gross Domestic Product (GDP) devoted to health care doubled from 6% to 12%; health care is projected to comprise 18% of GDP by the year 2000 [7]. An important part of this shift in expenditures is the sharp growth in the real price of health care services. During the 1980s, health care prices rose at twice the average rate of inflation, and, consequently, "rapid increases in the real price of health care have contributed to the overall rise in spending" [4]. Indeed, by 1991, health care costs indexed to 1964 were almost twice as high as the general price level. The change in the general level of consumer prices is compared with the change in consumer health care prices in Figure 1.

In conjunction with spiraling costs, U. S. health care delivery faces a problem of limited access. With health insurance a critical means of financing health care, the Congressional Budget Office reports: "According to the March 1992 Current Population Survey, more than 35 million people--14 percent of the total population--had no health insurance [11]. People in single-parent families and in families with an unemployed worker, young adults, and black Americans are disproportionately represented among the uninsured" [7]. At the same time, access is diminishing as critical care facilities are being concentrated at more limited sites. As governmental bodies are increasingly drawn into the mechanisms of health care service, either as regulators, planners, policy makers, or providers of funding, these bodies will have to weigh measures designed to meet the often conflicting ends of promoting access, advancing quality, and containing costs.

Promoting Quality and Access via Telecommunications

In recent years, technical advances and innovative applications have generated an important role for telecommunications services in enhancing health care access and improving service quality. Advanced fiber-optic networks facilitate the dissemination of medical tests, records, research data, and findings, and the transfer of patient information involving medical imaging technologies. Telecommunications also enhances the abilities of medical professionals to communicate personally with each other and for both professionals and patients to access medical information via electronic networks. At the Mayo Clinic, "telemedicine" links the clinic with experts at other locations by means of two-way broadband television [9]. In Memphis, Tennessee and Bowling Green, Kentucky telecommunications is enhancing access to equipment and the dissemination of test results, including results from Medical Resonance Imaging (MRI) equipment and EKGs, among medical facilities and practitioners, using fiber-optic links and fax machines [10]. Hospitals in St. Louis and Houston have conducted tests in which X-ray and other images are shared among professionals both inside and outside of the facilities using standard telephone lines. Other trials of telecommunications applications for image transfer, data transfer, and consultation are being conducted or planned in Massachusetts and North Carolina [10].

The ability to transfer medical information, including images, and to engage in consultation is encouraging the application of telecommunications to the remote practice of medicine. This enhances access to advanced medical care for many Americans in medically under serviced areas, such as inner cities and rural areas [10]. In Texas, three networks, KARENET, INTERNET, and MEDNET, use various communications technologies to distribute information, conduct conferences, and link specialists with patients and medical practitioners in remote rural areas [10]. In North Dakota, fiber-optic systems are used to retrain displaced farmers to find new careers in the field of medical services [10].

Controlling Costs via Telecommunications

Not only have telecommunications been involved in improving the quality and availability of U.S. medical care, but also in improving service efficiency and controlling costs. Many such applications involve improvements in administrative efficiency. In Connecticut, a subsidiary of Blue Cross and Blue Shield deployed the first state-wide electronic insurance claims and medical management network in the United States. Currently receiving 20% of its claims electronically, the subsidiary has consequently cut $300,000 in annual costs [10]. The practice of remote medicine also cut costs by using specialists and facilities more effectively. In suburban Philadelphia, a clinic transmits X-ray images via satellite to University of Pennsylvania Hospital radiologists. In enhancing access, the transmissions also reduce diagnostic costs to the clinic, which does not need to maintain a radiologist on site.

Aggregate Telecommunications Usage in the Health Services Sector

The overview above indicates that the medical services sector has exploited the increased availability of advanced telecommunications services. We now summarize patterns in the health care sector's overall usage of telecommunications services from 1963 to 1991. Over this interval, the health care services industry increased its real (1991 dollar) telecommunications usage, expressed as total expenditures on telecommunications services, more than sixfold, increasing from $482 million in 1963 to $2.97 billion in 1991, representing an annual rate of growth of 6.71%. More important than the growth in the level of telecommunications usage, however, is the growth in that usage relative to the total output of the health care sector. In 1963, for every dollar of health care output, the industry spent 0.81 cents on telecommunications services. This usage increased to 1.33 cents by 1991, representing an annual growth rate of 1.77%. While this increase is substantial, related research conducted by DRI has shown that the average U. S. industry increased its relative usage of telecommunications by 3.0% per year with a standard deviation of 2.2 percentage points across a total of 36 individual sectors examined. In this context, health care's usage of telecommunications increased at a statistically reasonable rate. The health care sector has thus utilized telecommunications services for a number of innovative purposes, yet the growth in overall telecommunications usage lags, to some degree, economy-wide telecommunications usage growth. We next examine the input structure of the health care services industry and begin to account for production decisions that have increased levels of telecommunications usage.

III. The Calculation of Cost Savings Due to Advances in Telecommunications Services Production and Use: Methodology and Data

In this section we quantify the direct savings to health care providers attributable to reductions in the relative price of telecommunications and its substitution for more costly inputs. To conduct its analysis, DRI constructed a "translog" econometric model of health care costs covering doctor and dentist services, hospitals, nursing and personal care, and other medical services.(2)

The Translog Function

A wide variety of functional forms have been applied to the analysis of production and cost relations. Among the most popular forms used in current research is the "translog," a function that may be taken as a second-order approximation to a large family of forms and that imposes no a priori restrictions on estimated elasticities. A translog cost function is typically specified as:

[Mathematical Expression Omitted]

where C represents optimized cost, y is output, t signifies a time variable, and [g.sub.ij] = [g.sub.ji] and [w.sub.i], [w.sub.j] represent input prices.

For a cost function specification to be consistent with microeconomic theory, costs should be increasing in output and factor prices. Cost functions must also be linearly homogeneous in factor prices. For the translog cost function, this implies:

[summation of] i[b.sub.i] = 1, [summation of] i[g.sub.ij] = [summation of] i[g.sub.ii] = [summation of] i[g.sub.iy] = 0. (2)

Estimating the Parameters of the Translog Model

With reference to the equations above and to the material to follow, let:

(a) [w.sub.1] = [w.sub.k] = the price of capital,

(b) [w.sub.2] = [w.sub.1] = the price of labor,

(c) [w.sub.3] = [w.sub.m] = the price of materials,

(d) [w.sub.4] = [w.sub.p] = the price of telecommunications services.

The translog cost model may be estimated using single equation procedures, but gains in efficiency may be achieved through further manipulations. Following the steps outlined by Berndt [2], consider a differentiation of the cost function:

d ln C/d ln [w.sub.i] = ([w.sub.i]/C)(dC/d[w.sub.i]). (3)

But the derivative of the optimized cost function with respect to factor price is the corresponding factor demand, [x.sub.i]. Substituting this value establishes:

d ln C/d ln [w.sub.i] = [w.sub.i][x.sub.i]/C = nominal cost share of factor i. (4)

Cost share equations for k, l, m, and p can then be established:

[S.sub.k] = [b.sub.k] + [g.sub.kk] ln[w.sub.k] +[ g.sub.kl] ln [w.sub.l] + [g.sub.km] ln [w.sub.m] + [g.sub.kp] ln [w.sub.p] + [g.sub.ky] ln y + [g.sub.kt] t(5)

[S.sub.l] = [b.sub.l] + [g.sub.lk] ln [w.sub.k] + [g.sub.ll] ln [w.sub.l] + [g.sub.lm] ln [w.sub.m] + [g.sub.lp] ln [w.sub.p] + [g.sub.ly] ln y + [g.sub.lt] t (6)

[S.sub.m] = [b.sub.m] + [g.sub.mk] ln [w.sub.k] + [g.sub.ml] [ln.sub.wl] + [g.sub.mm] ln [w.sub.m] + [g.sub.mp] ln [w.sub.p] + [g.sub.my] ln y + [g.sub.mt] t(7)

[S.sub.p] + [b.sub.p] + [g.sub.pk] ln [w.sub.k] + [g.sub.pl] [w.sub.n] [w.sub.l] + [g.sub.pm] ln [w.sub.m] + [g.sub.pp] ln [w.sub.p] + [g.sub.py] ln y + [g.sub.pt] t. (8)

Such a system of cost and share equations may be efficiently estimated using "seemingly unrelated regression" methods proposed by Zellner [12]. To perform the estimation, it is necessary to remove one factor share equation, expressing the prices in the remaining share equations as price relatives with the denominator being the price of the factor whose share equation has been removed. The parameters of the removed equation may be recovered through the use of the homogeneity restrictions.

Data Used in Estimating the Translog Cost Function

Data were drawn from private and nonprofit health care providers and establishments and thus does not include data on costs from government operated facilities. Percentage cost savings attributable to telecommunications technological advances, price reductions, and enhanced usage of telecommunications realized by private/nonprofit health care providers were imputed to the health care segment of government activities. This was done on the assumption that governmental health care providers produce services in a manner similar to those in the private sector, or at least realize gains from telecommunications advances at a similar rate. (Public health care delivery amounts to roughly one quarter of the total output of the private and nonprofit health care service sector.)

Data on government expenditures, purchases, and receipts for health care services were drawn from the National Income and Product Accounts of the United States [3]. Health care costs are expressed as a function of prices, output, time, and interactions among these variables. The analysis implicitly embeds information about potential alternative technologies of applying inputs to obtain health care service, and how these technologies change over time. As prices of the inputs change, total costs change not only because prices differ but also because when input prices change relative to one another, the mix of inputs used by producers also tends to change. The cost equation determined by DRI captures the combined immediate effect of price changes on costs and the subsequent effect of those price changes on input mix and the influence of such shifts on costs.

Information on average labor compensation was taken from the U. S. Department of Commerce, Bureau of Economic Analysis (BEA) and the U. S. Department of Labor, Bureau of Labor Statistics (BLS). Capital usage is defined as depreciation on all capital assets and is taken from the BEA "Detailed Investment by Industry" database. Telecommunications share figures were drawn from an interindustry model and databank maintained by DRI.

Analysis of the cost structure of the health care services industry reveals that in nominal terms, an increasingly greater share of overall costs have been devoted to labor costs, which rose from 46% of total costs in 1963 to 62% in 1991. Capital and materials costs each declined over the same interval, falling 27% and 28% respectively. Significantly, the nominal share of telecommunications costs fell nearly two-thirds from 1963 to 1991. As will be discussed later, this is primarily the result of large relative telecommunications price declines. When these cost shares are expressed in real terms, however, we note greater stability in the share of total costs accounted for by each input. Therefore, while each input's real share of total cost has not changed at all since 1963, increases in the relative price of labor have increased labor's nominal share of total costs by 33%, or 1% per year. Labor costs have crowded out health care services' incorporation of capital, materials, and telecommunications. This observation will be discussed at greater length and in the context of cost savings due to telecommunications usage in section III below.

IV. The Calculation of Cost Savings Due to Advances in Telecommunications Services Production and Use: Results

Table I, below, presents the parameters as statistically estimated by the translog system of equations described above. The table presents the estimated coefficients and their level of statistical significance.(3)

In evaluating goodness of fit statistics, recall that the coefficients of the factor share equations of translog systems are constrained to equal associated parameter values in the jointly estimated cost function. Also, note that the estimation process produces the parameters associated with minimum system-wide error, not necessarily those associated with minimum error for any individual equation in the system. In spite of these restrictions, each of the four equations (one cost and three share) estimated for the health services industry have associated [Mathematical Expression Omitted] statistics of at least 0.92. In addition, the parameter estimates typically exhibit strong t-statistics and provide an excellent foundation for the general analytical applications discussed in this article.

Table I. Translog Model Parameters

Equation Coefficient                Capital Share   Labor Share    TC Share

Constant                               0.0867(**)   -0.1512         0.0472(**)
Price of Capital                       0.0065(*)    -0.0075(**)    -0.0008
Price of Labor                        -0.0075(**)    0.1593(**)    -0.0090(**)
Price of Telco.                       -0.0008       -0.0090(**)     0.0101(**)
Log of Output                         -0.0196(**)    0.2016(**)    -0.0106(**)
Time                                   0.0002(**)   -0.0017(**)     0.0001(**)
[Mathematical Expression Omitted]        0.92          0.95           0.98

Calculation of Actual Savings Realized from 1963 to 1991

Once the cost equation and associated input share equations were statistically determined, a simulation was conducted in which health service providers in every year from 1963 to 1991 were constrained to confront the telecommunications relative prices that held in 1963. The percentage difference between costs determined by actual historical prices and simulated costs determined by historical and constrained telecommunications prices were then applied to historical costs to determine health care providers' direct cost savings attributable to reductions in price and improvements in the technology of telecommunications services.

DRI concludes that if the telecommunications technology had not advanced since 1963, direct health care costs would have been measurably higher in every year since 1965. Annual savings in health care expenses due to telecommunications advances from 1963 have reached or exceeded 0.5% of total costs since 1974, and have reached or exceeded 1% of total costs since 1980. By 1991 the total costs of private and nonprofit health care service provision (the sum of purchased materials, labor compensation, depreciation of capital assets, and telecommunications expenses) reached $413.5 billion; public health care services purchased by governments amounted to an additional $115.1 billion. Had telecommunications technology not advanced since 1963, DRI determines that those combined costs would have been $7.4 billion greater. In addition, DRI has calculated the cumulative direct cost saving in health care service provision due to advances in telecommunications production and health care consumption for each year from 1963 to 1991. These savings totaled $71.0 billion in 1991 dollars.

Table II. Health Care Cost Savings from Telecommunications Advances (Billions
of current dollars)

                                              Additional Costs had
                                              Telecommunications
                             1991 Costs          not Advanced

Private Health Care Costs      $413.5                 $5.8
Public Health Care Costs        115.1                  1.6
Total                           528.6                  7.4

Source: Bureau of Labor Statistics. Bureau of Economic Analysis, various
publications.

Governments directly purchase health care services, primarily through construction and operation of health care facilities. Increasingly, however, governments are financing health care consumption through transfers, grants, and subsidies. Total net government expenditures--the sum of direct purchases, transfer payments, grants-in-aid, and subsidies, less service receipts and surplus from government enterprises--have been rising rapidly both in absolute terms and relative to other government expenditures. The proportion of these expenditures compared with total government expenditures is shown in Figure 5. Assuming government expenditures would be proportionately affected by cost savings in a given year, DRI calculates that had there been no advances in telecommunications production or in consumption of telecommunications services by health care services from 1963 to 1991, government health care expenditures would have been $42.1 billion greater.

Telecommunications and Unrealized Prospective Cost Savings in Health Care

The preceding analyses determined historically realized cost savings from realized advances in telecommunications production and the applications of telecommunications as a cost-effective substitute in health care services. However, it is likely that the full potential of telecommunications as a substitute for more expensive inputs and processes has not yet been realized.

As described above, the cost savings due to telecommunications advances arise for two reasons. First, technological change in the telecommunications services sector causes absolute telecommunications prices to be lower than otherwise. Because telecommunications has been an input in the provision of health care services, the fact that this input may be purchased more cheaply than would have been possible without progressive technological change creates cost savings. Technological change in telecommunications, however, also caused telecommunications prices to fall relative to other input prices. In general, a reduction in the relative price of an input will induce substitution of the input for more costly inputs. This price-induced substitution of telecommunications for relatively more expensive inputs is the second reason for historical savings in health care costs as a result of advances in telecommunications production and consumption. Let us assess in greater detail these substitution phenomena.

The real substitution of telecommunications inputs for other inputs is reflected in the growth, over time, of the proportion of real telecommunications consumption to real health care output. The movement of this ratio reflects the movement of telecommunications' share of health care costs. As described briefly in section I above, while the national average growth in "telecommunications intensity" for all industries from 1963 to 1991 was 3.0% per year, the health care services industry increased its intensity of usage by only 1.77% per year. The cause of this slow rise in the telecommunications intensity of health care services is not clear, but may result from three related matters: First, the particular characteristics of the health care production process may limit the extent to which telecommunications may be employed as an input. Second, the health care sector may be somewhat slow to adapt technologies not immediately connected with patient care. Third, provider reimbursement policies may discriminate against the recovery of telecommunications costs.

It seems likely that health care providers have not responded as quickly as they might have to cost signals in adopting cost-reducing telecommunications technologies. We have touched on only a few of the possible factors that have driven this trend. Clearly, additional inquiry into the dynamics shaping the health care sector's adoption of advanced telecommunications would be useful to policy makers. Testimony presented before the U. S. House of Representatives Subcommittee on Telecommunications and Finance in March 1993, supports the importance of such prospective research. During those hearings, government officials and health care specialists agreed that the health care sector could and should exploit new telecommunications technologies by underwriting dozens of cost-reducing, quality-enhancing projects. In addition, as discussed in greater detail below, researchers have shown that health care providers have foregone a range of possible telecommunications applications that might have saved the 1990 economy over $36 billion, yielding net savings of $20 billion.

As we pursue an explanation of the relatively slow substitution of telecommunications for more costly inputs, we examine the translog-derived Allen partial elasticities of substitution between pairs of inputs. Unlike the ratio of substitution between two inputs, which describes the amount of one input that can be saved as a result of investing one additional dollar in another input, the partial elasticity of substitution describes the rate at which a ratio of real input pairs will change in response to changes in the relative prices of the two inputs. High elasticities of substitution between a pair of inputs will indicate the presence of high substitution possibilities.

It is significant that the partial elasticities of substitution between each input (capital, labor, and materials) and telecommunications are much closer to zero for health care than for almost all other sectors of the economy. In health care, the elasticity of substitution between labor and telecommunications, -0.2, while indicative of a slightly complementary relationship, is also relatively low, and describes a production technology that does not create rapid substitution possibilities between inputs. In particular, the elasticity of substitution between capital and telecommunications in health care is estimated at 3.7, a value that is relatively high, but fully 77% lower than the economy-wide average of 16.2. This indicates, for example, that the average industry responds to a 1% change in the price of capital relative to telecommunications by making a 16% substitution of telecommunications for capital, while the health services sector makes only about a 4% substitution. By contrast, two similarly structured service industries that have been investigated in other DRI research, business services and personal services, have been found to have elasticities of substitution between capital and telecommunications of 17.9 and 10.9, respectively. Therefore, although the substitution of telecommunications for other inputs has generated substantial cost savings, the relative inelasticity of substitution between capital and telecommunications has forestalled even greater cost savings.

The specific characteristics of technologies applied in health care provision, then, may be the reason that telecommunications intensity, while growing recently in health care services, has grown more slowly in this industry than in others. Evidence from detailed health care operations research suggests, however, that there are cost-effective technologies that have not been applied in the provision of health care services. DRI research results, presented below, evaluate the independent findings of previous analysis to determine potential unrealized savings from unapplied but available technologies involving greater substitution of telecommunications for other inputs.

A recent study has concluded that telecommunications could substitute for specific and substantial amounts of costly inputs [1]. The value of inputs telecommunications could yet effectively substitute for was calculated to be $36 billion (1990 dollars), of which almost $30 billion would arise from electronic management and transmittal of patient information, nearly $6 billion from electronic submission and processing of health care claims, about $600 million from electronic inventory management and controls, and some $200 million from video conferencing for professional training and remote medical consultation [1]. The study did not estimate the telecommunications costs that would be required to substitute for the replaced resources [1].

However, the DRI translog health care cost model presents a mechanism for analyzing interrelated factor input demands in the provision of health care. DRI estimates that in 1990 a dollar of telecommunications service input could be substituted for about $2.30 in combined labor, capital, and materials (goods and services). Consequently, the telecommunications costs associated with the substitution of $36 billion in other inputs would be $15.65 billion. Therefore, the net savings associated with proposed applications of existing telecommunications technology would have reduced 1990 health care costs by $20.35 billion. According to the Congressional Budget Office (CBO), health care costs excluding prescription drugs and other nondurables was $611 billion that year [7]. The $20.35 billion in potentially realizable savings thus amounted to 3.33% of health care expenditures (excluding prescription drugs and nondurables). These findings are summarized in Table III.

Table III. Potential Savings if More Telecommunications Had Been Substituted
for More Costly Inputs (1990)

                                            Billions of current dollars

Health Care Costs (CBO)                                $611.0
Unrealized Potential Substitutions                      $36.0
Net Savings Had Substitutions Occurred                  $20.35
Percent of Total Costs                                    3.33%

Source: Lemieux and Williams [7]; Arthur D. Little, Inc., [1].
Table IV. CBO Projections of Health Care Costs and DRI Estimates of Savings,
Assuming Full Adoption of Existing Technology: Savings Rate of 3.33% (Billions
of current dollars)

                        1995         2000

U.S.
Health Care Costs      $994.0      $1,568.0
Potential Savings       $33.1         $52.2

The CBO forecasts health care expenditures to rise to $994 billion (current dollars) in 1995 and reach $1.568 trillion (current dollars) by the year 2000. Thus, on the assumption that savings in the future will be in the same proportion as in 1990, potential applications of existing telecommunications technology could save the U. S. economy $33.1 billion in 1995 and $52.2 billion in 2000. These results are summarized in Table IV.

These calculations assume that further advances in telecommunications will not further reduce telecommunications prices relative to the prices of other inputs. This is a very conservative assumption, given that over the last three decades telecommunications prices have grown much more slowly than almost all other input prices.

If, however, the assumptions above were modified slightly to reflect recent historical trends in telecommunications relative prices and real telecommunications usage in health care, estimates of cost savings that could be achieved through more intensive use of existing telecommunications technologies would be substantially higher. In order to perform this calculation, it would be necessary to establish: 1) the rate at which nominal expenditures for labor, capital, and materials could be replaced per nominal dollar expenditure in telecommunications services by the year 2000, and 2) the rate at which health care costs would grow through the year 2000.

First, due to telecommunications' steadily declining relative prices, the substitution ratio between telecommunications and other inputs rose steadily between 1982 and 1991. If this trend is extended in a linear fashion to the year 2000, the amount of inputs for which a dollar of telecommunications expenditure could substitute would be 37% higher than the 1991 level. Second, if cost forecasts projected by the CBO apply to the inputs for which telecommunications may potentially be substituted, applications of telecommunications technology (which could have substituted for $36 billion in 1990 health care costs) would substitute for $92.4 in health care costs by the year 2000. On this basis, the application of existing telecommunications technologies could reduce U. S. health care costs in the year 2000 by $63.05 billion, or 4.02%.

V. Conclusion

We have described DRI's application of econometric techniques to estimate telecommunications services' contributions to cost savings in the health care sector. While the cost savings have been substantial, evidence suggests that further cost savings may be possible. The research in this paper provides policy makers with a better understanding of health care costs and the extent to which further adoption of telecommunications in the provision of health care services might affect industry costs in the future.

1. Ideas presented at a series of hearings conducted before the U.S. House of Representatives Committee on Energy and Commerce, Subcommittee on Telecommunications and Finance, 19 January to 31 March 1993.

2. For more detailed discussion of the specification and properties of the translog cost function, see Cronin, et al. [6].

3. The following symbols are used to denote statistical significance: "*" indicates significance at the 95% confidence level, and "**" indicates significance at the 99% confidence level. Because the cost equation is specified with over 20 explanatory variables, to conserve space, parameter statistics are not displayed; r-bar squared for the cost equation, however, is estimated at 0.94.

References

1. Arthur D. Little Inc., "Can Telecommunications Help Solve America's Health Care Problems?" Arthur D. Little Inc., Reference 91810-98, July 1992.

2. Berndt, Ernst. The Practice of Econometrics. Boston: Addison-Wesley, 1991.

3. Bureau of Economic Analysis. National Income and Product Accounts of the United States, Vol. 2, 1959-88. Washington: U.S. Government Printing Office, September 1992, Tables 3.9A, 3.9B, 3.16, 3.17.

4. Council of Economic Advisors. Economic Report of the President. U.S. Government Printing Office, February 1991, p. 137.

5. Cronin, Francis J., Elisabeth K. Colleran, Paul L. Hebert, and Steve Lewitzky, "Telecommunications and Growth: The Contribution of Telecommunications Infrastructure Investment to Aggregate and Sectoral Productivity." Telecommunications Policy, November 1993.

6. Cronin, Francis J., Mark A. Gold, Paul L. Hebert, and Steve Lewitzky, "Factor Prices, Factor Substitution, and the Relative Demand for Telecommunications Across U.S. Industries." Information Economics and Policy, No. 5, 1993.

7. Lemieux, J. and C. Williams. Projections of National Health Expenditures. Washington: Congressional Budget Office, October 1992 page xii.

8. McGinnis, J. Michael, M.D., Public Health Service, U.S. Department of Health and Human Services, 23 February 1993, testimony before the House Subcommittee on Telecommunications and Finance, 19 January 1993.

9. National Telecommunications and Information Administration. NTIA Telecom 2000. Washington: U.S. Government Printing Office, October 1988, p. 108.

10. National Telecommunications and Information Administration. Telecommunications in the Age of Information. Washington: U.S. Government Printing Office, October 1991, p. 67.

11. Weisbrod, Burton, "The Health Care Quadrilemma: An Essay on Technological Change, Insurance, Quality of Care, and Cost Containment." Journal of Economic Literature, June 1991.

12. Zellner, A., "An Efficient Method of Estimating Seemingly Unrelated Regressions and Tests for Aggregation Bias." Journal of the American Statistical Association, 1962, no. 57, 348-68.