A 6-10 dossier articles

132 The Journal of Risk and Insurance

way of financing care and also to expand the amount of medical care received by some parts of the population.

The final consensus of the conference may be stated in the words of one of the participants, "When I came into the con- ference the other day I said We are going to come out of here with a recommenda- tion that the situation be further stud- ied.'"^ With the unresolved questions concerning this type of program still be- fore us, it is hoped many of these studies will be completed before the politicians make their decision.

This is a most useful book for any person interested in the implications of a national health insurance program. Many changes have taken place since November 1970, but the conference pro- ceedings provide a most helpful source of information.

INFLATION, TECHNOLOGY AND GROWTH: POSSIBLE LONG RANGE IMPLICATIONS FOR INSURANCE. By Robert I. Mehr and Seev Neumann. Grad- uate School of Business, Bloomington, Indiana: Division of Research, Indiana University, 1972, $15.00.

Reviewer: J. D. Hammond, Professor of Business Administration, The Pennsyl- vania State University.

The general title of this new book sug- gests a rather traditional macro level re- view of the insurance industry as it is beset by economic and technological forces. Such is not the case. Professor Mehr, the senior author of the book, and Professor Neumann have employed the Delphi technique in an attempt to iden- tify various characteristics of the insur- ance industry in the year 2000. Although the cynic may suggest this to be an easy task for the insurance industry, the Mehr

'Page 259.

and Neumarm approach is a serious at- tempt to apply a relatively new forecast- ing device (the Delphi Technique) to a particular set of questions about the in- surance industry. As such, it deserves seri- ous attention.

The volume was written as a part of the 1970 Sesquicentennial celebration of Indiana University. The Mehr-Neumann volume is one of four companion pieces representing the School of Business con- tribution to the celebration. The three other works are not identified. Financial assistance for the series came from sev- eral grants from insurance companies. The stated purpose of the book "is to make some cautious, documented speculations about the long-range effects of infiation, technology, and growth on private insur- ance in the United States." Its objective, we are told, "is to identify both the pres- ent characteristics that are hkely to pre- vail until the end of the century and any new characteristics that are Hkely to emerge sometime between now and then."

A statement by a University executive in the foreword gives added scope. Mr. George Pinnell, Vice President and Treas- urer of Indiana University states: "I fully anticipate that in the years to come these volumes will be increasingly useful to planners and will clearly demonstrate the insight and vision of the authors. Whether time will corroborate their projections and prophesies is a matter that we will watch with fascination." Thus, there is the hope by at least one person that the Mehr- Neumann book and its companion vol- umes will be of use to planners in the in- surance world. It is a fair assessment of the most likely use of the book.

The book contains 319 pages of text with an additional 184 pages of support- ing material in several appendixes. The authors have assembled 111 tables, 88 of which contain data generated by the study. Graph lovers will be disappointed

Publications 133

to find only one graph. Labor economists will be pleased, however. It has a Phillips curve.

The entire findings of the research rest upon the use of the Delphi method. So far as the reviewer knows, this is the first application of the Delphi method in any- thing which might be called the insurance literature. Basically, the technique pro- vides for a systematic method of eliciting expert opinion. It was developed by the Rand Corporation as a device to be used for long-range forecasting, a situation where extrapolation of statistieal series is of doubtful value. The procedure calls for a group of experts to be polled repetitively concerning their opinions on a particular forecast. For example, such a group might be asked their opinion about various ef- fects of say, women's liberation, preemp- tive nuclear strikes, or the ecumenical re- ligious movement. In general, past use of the Delphi Teehnique has centered upon those questions where the use of statistical data is not possible or inappropriate. In any event, the opinions are compiled and are fed back to the panel for another round of opinion response. The feed-back procedure is then repeated until consen- sus is apparent. The technique is thus characterized by the need to develop con- sensus through a series of iterative exer- cises and by the use of experts.

Mehr and Neumann have adhered strictly to the Delphi procedure. Invita- tions were sent to a group of 70 experts to participate in the study. Of this num- oer, 64 accepted and 58 eventually com- pleted the project. It is not unreasonable to think that the six drop-outs r^ulted from exhaustion. After receiving detailed inputs of background information on the American economy and possible techno- logical developments (panelists were also 'fee to develop additional background in- formation in these areas), each panel ' ' b received a 25-page questionnaire

containing 73 questions about various as- pects of the insurance business. A sum- mary of these first round responses was then compiled and sent to eaeh panel member. Each panelist had the chance to reconsider and revise his first round re- sponse and was asked to explain why his judgments deviated from the norm of the round one responses.

Second round responses were then cir- culated again to eaeh panel member, to- gether with a summary of the reasons underlying the deviating opinions. Mem- bers were asked to reconsider their sec- ond round opinions in light of the new information and again to revise their re- sponse to the question if that was felt necessary. For the atypical third round responses, members were asked to explain why they were unimpressed with the stated reasons underlying such responses. These responses were again summarized and returned to the members where each had a final opportunity to modify his re- sponse. At this point, the median of the fourth round response was taken to be the consensus of the panel.

The 58 finishers represented a cross- section of expert opinion. The oracles rep- resented universities, government bodies, corporate insurance buyers, journalists, and executives from both property-liabil- ity and life insurance.

Two of die first three chapters of the book are devoted to the presentation of background material on technology and the economy. The first chapter discusses the difficulties of long-range prediction and a discussion of the Delphi method. The remaining eight chapters are devoted to the presentation of the research re- sults. Here, we are able to learn the panel responses to sets of questions dealing with the entire industry, life insurance, health insurance, the property and liability in- surance industry, automobile insurance, property and liability insurance lines ex-

134 The Journal of Risk and Insurance

capt auto. A summary is presented in the final chapter.

The general tone of most of the ques- tions asked of the panel can be seen from a sample of the responses. We learn that the panel consensus sees social insurance to be the dominant insurance form in the year 2000; that the purchase of life in- surance policies characterized by high and moderate savings wQ! decline; that the percentage share of health insurance pre- miums written by private insurers wHI in- crease (from 53.7 to 60 percent); that the premiums to policyholder surplus ratio for property and liability insurers will in- crease only slightly; that the percentage of total auto premiums written by the top ten insurers wiU increase; and that direct- writing insurers will further increase their share of the market.

All responses are given in terms of a point estimate but the authors have also provided a statement of the response vari- ance about the estimate. For example, panel members were asked to forecast the premiums to policyholder surplus ratio and the 1966 value of that ratio was taken as the starting point—about 1.4. The con- sensus forecast value was 1.7. The 95 per- cent confidence interval presented in the results is 1.65 to 1.91.

So much for the content and the ap- proach of the book. Though the approach is innovative for the insurance literature, it is not without some Hmitations.

While the Delphi Technique is gen- erally recognized as a useful forecasting device, the value of using experts has been subject to question. Stated differ- ently, if one were to use any reasonably intelligent group of people, the consensus answers finally arrived at may be little different than those generated by the ex- perts. It is an interesting possibiHty and one which has some support in the Delphi Hterature.

A second problem concerns any fore- cast for the year 2000. The rate of change in all things affecting any institution—in-

cluding insurance—is so high that any forecast by any method must be suspect. Most Delphi research has dealt with ques- tions not amenable to traditional statis- tical analysis and where long-range pre- dictions deal more with shifts in values rather than time-series projections. For example, Delphi studies have dealt with anticipated changes in American values in the year 2000 and with changes in the goals of educational institutions. The Mehr-Neumann work does not deal with those or similar phenomena directly. In- stead, panelists were asked to forecast a particular point value for several eco- nomic projections deaHng with insurance. Although considerations of value changes and similar shifts within the economy were considered in arriving at forecast values, the consideration was not syste- matic. The resulting forecast for various time series for a point 30 years in the fu- ture is an exercise requiring more faith in judgment than even actuarial science.

The investment in time by panel mem- bers precludes the asking of questions to satisfy every reader. Still, some areas were omitted from consideration. For example, there is no direct consideration of lapse rates nor of the distribution costs in life insurance. The related major problem of turnover among life insurance agents was not included. If one is interested in panel consensus, on such problems, he must in- fer them from questions dealing with gen- eral operating efficiency or the prospec- tive growth in group coverage. Such questions were more directly considered for property and Hability insurance than for life insurance. Still, it is diBBcult to fault a 73 item questionnaire for a Delphi process for errors of omission.

It would be very helpful to know th« identity of the panelists. We are assured they are experts but nonetheless one would like to make his own assessment of such quaHfications. Further, the number of eJi' perts from each of the categories repiC' sented is not given. Thus, we do not know

Publications 135

if all of the areas are equally represented or whether one group might have a dis- proportionate impact on the process. Since the Mehr-Neumann questionnaire is so comprehensive, one wonders whether each of the experts is expert in all of the aspects of the insurance covered in the investigation. One suspects not.

The book is interesting to read and in- tellectual curiosity is stimulated by the large number of questions and the re- sponses of the panel. The reader cannot help but project his own responses and compare them with those of the panel. Herein lies the chief value of the book. While the panel projections for a point nearly 30 years distant are simply too speculative for use by executives or regu- lators, one would hope that sueh groups would study the research. They may dis- agree with the projections or feel insulted at not being consulted, but a serious read- ing of the book where one role-plays the panel may be for insurance executives, policy-makers,—and educators too—a unique thinking experience.

Professors Mehr and Neumann have provided us with a thorough application of a relatively new research tool which has not previously appeared in the insur- ance literature. The research methodology is detailed and sound and its presentation clear and concise. The projected values of the research will not likely serve as diiect inputs to corporate planning models is insurance (there may be none) but it cannot help but make planners better thinkers.

^DAMENTALS OF RISK AND IN- SURANCE. By the late Curtis M. Elliott ind Emmett J. Vaughn, John Wiley and 5ons, Inc., 1972, x and 703 pages.

^viewer: William M. Howard, Professor f Finance and Insurance, University of

fundamentals of Risk and Insurance is i for use in a college-level survey

course in risk and insurance. The stated intent of the authors has been to create a text that is consumer oriented. The types of consumer the authors apparently have in mind are individuals and families. For example, there is an entire chapter of 24 pages on general liability insurance for the individual. Chapters on property and liabiHty insurance for business firms, surety bonds and credit insurance are largely independent of other chapters and may be omitted.

The section on life and health insur- ance, 7 chapters, seems to be aimed al- most exclusively at individuals and fam- ilies. Only two and a half pages are allotted to forms of group life insurance and group annuities. Group health insurance is men- tioned casually in a paragraph on meth- ods of marketing health insurance.

The authors have recognized the prob- lem of handling the subject of risk management in an elementary text and have chosen to avoid extensive treatment of statistical techniques and utility theory. A 15-page chapter entitled "Risk Manage- ment" describes the nature and function of risk management. It appears to be ade- quate for individuals and families; it pro- vides an introduction of the subject to those who may pursue it more deeply, and is consistent with the stated purpose of the book.

What knowledge may the authors of insurance texts reasonably assume their readers bring to the subject? Can they assume a knowledge of elementary prob- ability, principles of statistics and busi- ness law? EUiott and Vaughn assume no knowledge of probability and statistics. They include just enough on these sub- jects to allow the reader to understand the nature of insurance. A chapter on "Negligence and Legal Liability" makes one wonders again why teachers of insur- ance (including this reviewer) seem to feel that students must understand the causes of liability losses but not neces- sarily of property losses. Most of us—in-