Search

The author comments on a paper by economist Carl Christ concerning properties of nine econometric models. He states that Christ made a number of comparative statements about the econometric models that may result in casual readers misinterpreting the comments. He mentions that the comparative statements could be interpreted to concern the statistical properties of the models, the models theoretical quality, or the forecasting ability of the models. He states that these property statements can be independent from each other, with a model being a poor forecaster but having excellent statistical properties, and that they imply nothing about the model's degree of confirmation.

In a critique of the paper, "Severity of Formal Sanctions As a Deterrent to Deviant Behavior," in the 1970 issue of the journal "Law and Society Review," scholar Paul E. Meehi has questioned the interpretation of the data in a particular section. On the basis of that evidence, it is concluded that severity of formal sanctions show no direct deterrent effect on the deviant behavior under consideration, but do appear to have an indirect deterrent effect through their impact on the normative climate of the campus. In challenging this interpretation of the data, Meehi expressed concern that non social science readers of the periodical "Law & Society Review," might be misled by the use of "causal language" in the interpretation of these tabulations from cross sectional data. In this extension of earlier analysis, this article attempts to explicate some difficulties in interpreting findings from cross sectional research and how these difficulties can be dealt with. Specifically, it is shown how cross sectional data can serve not only to test predictions of a causal model, but also to evaluate its causal assumptions.

The article presents abstracts of articles related to research and forecasting appearing in various periodicals. Articles abstracted include "New Research in Marketing," edited by Lee E. Preston, published by the University of California at Berkeley's Institute of Business and Economic Research, "The Home Building Market," published in May 1966 by Sweet's Construction Catalog Services, and "A British Look at Market Research in the U. S. A.," by Harold P. Hodge, appearing in the December 1965 issue of "Profit Evolution No. 8."

This paper is part of our continuing analysis of common stock price fluctuations, which we call volatility patterns. Here, our focus is upon industrial stock price indices. We investigate (1) whether or not an industry can be classified according to its stock price volatility--i.e, whether a series of volatility statistics, each computed for a relatively short time span, will be fairly constant for individual industries over time, and (2) if inter-industry volatility differentials remain constant over time. Part I presents a brief discussion of general issues related to the topic of volatility. Part II summarizes the computational models, and Part III presents the empirical results yielded by the industry study. Part IV summarizes our findings. [ABSTRACT FROM AUTHOR]

This is the first of a series of papers designed to report the various stages of a longitudinal research project on the study of managerial effectiveness. The purposes of this article are: (1) to set forth the various dimensions of the overall research effort; (2) to present a review of the research on the prediction of managerial success or effectiveness; and, (3) to establish the focus for a second paper subsequently to appear in the Academy of Management Journal describing the results and implications of an intermediate research effort. [ABSTRACT FROM AUTHOR]

Presents a paper on the relative occupational success of blacks and whites in the United States. How blacks get less in the job market; Goal of the paper to present the results of an extensive study of data on the employment of young men; Definition of occupations and measurings of earnings within occupations; Average wages rates of black and white males; Occupational success; The model for weighing the information; Comparison of blacks and whites with equal endowments; Comparison of random groups of blacks and whites; Comparisons with actual distributions; Projections to 1973; Discussion by the panel on the findings.

This paper applies residual analysis to build up improved forecasts from exponential smoothing models. The resulting model is not unknown, but the method of achieving it is more straightforward and intuitive than existing methods. The paper also shows that the model is identical to IMA(k, k) forecasting models. [ABSTRACT FROM AUTHOR]

This paper derives the class of nonstationary time-series representations for which exponential smoothing of arbitrary order minimizes mean-square forecast error. It points out that these representations are included in the class of integrated moving averages developed by BOX AND JENKINS, permitting various procedures to be applied to estimating the smoothing constant and determining the appropriate order of smoothing. These results further permit the principle of parsimony in parameterization to be applied to any choice between exponential smoothing and alternative forecasting procedures. [ABSTRACT FROM AUTHOR]

Information on the subjects discussed during the second annual meeting of the Operations Research Society of America held in Chicago, Illinois on May 21 and 22, 1954 is presented. An invited address entitled The Basic Processes of Operations Research and Their Relation to Management was delivered by M. L. Hurni of General Electric Co. A retiring presidential address was presented by R. F. Rinehart. Wroe Alderson of Alderson and Sessions presented four invited papers on applications to marketing and forecasting.

This paper centers on what happens to an exponential smoothing forecasting system when the smoothing parameters are change. It gives sufficient conditions under which several proposed models give numerically identical results, and notes a surprising symmetry in one basic model between the smoothing constants for the smoothed average and the smoothed trend. The paper defines smoothing-constant values that cause oscillatory behavior, discusses the difficulties caused by oscillation, and finds that commonly used values of the smoothing constants lie in a region of oscillation of varying period. It uses these results in an anlytical study of the tradeoff between responsiveness and variance; some dominant choices emerge. Finally, on the basis of these results, the paper examines several sample sets of data from the literature, and discusses implications for the choice of smoothing constants in complex situations. [ABSTRACT FROM AUTHOR]

The article presents comments of the author on the article "Is Prediction Feasible in Social Work? An Inquiry Based Upon a Sociological Study of Parole Records," by Ernest W. Burgess. I am to discuss Burgess' paper with respect to its general methodology. A personal reference, which will disclose my standpoint of interest, will perhaps be pardoned. In a publication of 1914, I endeavored to find a basis for predicting the future political behavior of farmers and working men with respect to each other. For this purpose I developed certain indices based on the previously recorded votes of individuals of both classes in state legislatures.' In what I conceive to be their essentials, my method and that which Burgess has used in the paper of today are the same. We both have taken account of simple alternatives in behavior, an aye or no vote in one instance; conformity with parole or its violation in the other. In neither case, that is, has the behavior measured been regarded as variable. We both have employed the percentage of individuals in a general group who comply with one of the alternatives as a normal expectancy, or probability ratio by which to compare the corresponding behavior tendencies of subgroups, as expressed by similar percentages within the latter.

In this article the author comments on the research paper related to the prediction of business failures using accounting data by Jarrod W. Wilcox. The author remarks that in evaluating the present work, one could discuss the model at the conceptual level based on strictly paired observations and tests in which a failed firm is paired with a nonfailed firm with similar total assets from the same industry. Perhaps more important, the strict pairing in the tests means that classification errors also occur only in pairs in Wilcox's tests. According to the author, any alternative prediction method based on more input data and more calculation should at least be clearly better than these success rates, and should warrant the extra cost involved.

Assume that you have been asked by the president of a large firm to advise him on the most desirable levels of decentralization in order to control the activities of different departments within his organization. Upon what factors would you be likely to base your decision! A recent investigation indicates that to establish appropriate levels of decentralization, one's decisions should be, at least partially, based on: the types of productive tasks workers perform in general; and, the degree of predictability of work demands, in particular. This paper provides a partial explanation of several non-administrative cause for differences in the degrees of discretion workers exert in performing their tasks. Specifically we attempt to trace the influence of two factors upon discretion. One is the predictability of work demands; the other is the professionalization of jobs. Before attempting to analyze these relationships, let us take a brief view of the research design and the components of discretion. [ABSTRACT FROM AUTHOR]

The article focuses on economic forecasting when the subject of the forecast is influenced by the forecast. It has been argued that when forecasts influence economic behavior they may serve as an instrument of control, and that in such instances, the numerical accuracy of forecasts loses its relevance. What is relevant in such instances is the self-reinforcing power of the forecast, that is, the power of the forecast to make the variable in question move in the direction of the forecasted change. But surely all that really matters, from the point of view of control, is the discrepancy between the desired or target value of the forecast variable and its actual or recorded value. A forecast which is successful in terms of the above mentioned discrepancy may be self-nullifying, self-reinforcing, or neither. Whether the forecast is self-reinforcing or sell-nullifying is, in fact, quite irrelevant to the problem of economic control. Economists Emile Grunberg and Franco Modigliani have demonstrated that, if upper and lower bounds are imposed on the forecast variable, an accurate forecast is always possible.

This paper describes a planning model designed to be used by managers of family-planning systems to improve understanding, forecasting, and planning. The macro-flow model describes the patient movement through post-partum and nonpost-partum programs. The flows model the phenomena of: outreach recruitment, continuance, post-partum checkups, switching methods, referral, migration, contraceptive-use experience, private protection, method effectiveness, advertising response, follow up, abortion, and medical services. Strategic variables can be linked to the flow parameters to produce capacity requirements and budgetary implications. The model output includes benefit measures of total active patients, couple years of protection, 'births protected,' and unwanted births prevented. The fertility aspects of births prevented are modeled through a nonstationary Markov process submodel that considers demographic phenomena without burdening the basic flow structure. The input procedures used to process patient-visit, outreach, clinic-survey, and experimental data are discussed and some empirical results are reported. The combination of data-based estimates and subjective judgment is done by 'fitting' the model to past observed data. Testing and control are done by 'tracking' model performance through conditional prediction, diagnosis, and updating. The model is implemented in an on-line, conversational program that facilitates evolutionary model building by allowing the user to specify his model options. The application and testing of the model in the Atlanta Area Family Planning System are discussed and the experiences of managers in using the model to gain new insights, forecasts, budgets, and plans are reported. [ABSTRACT FROM AUTHOR]

A comparison of the final league tables of the 1971-72 football season, with forecasts made by Goal before the season began, shows significant positive correlation. This seems to indicate that football results am not pure chance (although them is obviously a considerable dement of chance). The paper questions how the data would have been handled by statisticians who do not approve of significance tests. [ABSTRACT FROM AUTHOR]

A brief discussion of Statistical Quality Control Charting procedures is first presented with special reference to the relevance of the objectives and assumptions. An approach to the design of discrete feedforward and feed. back control schemes, which are of great importance for example, in the chemical industry, is then given. This approach to control employs discrete stochastic and dynamic models discussed in Part I of this paper (Box and Jenkins, 1968) and has a close !ink with the forecasting problems discussed there. The control algorithms obtained are ideally suited to discrete digital computer control. However, for common simple situations the Algorithms may be represented by suitable charts or nomograms which may be employed to obtain improved manual control. The paper ends with a discussion of a problem typical of that arising in the parts manufacturing industry. Here, attention must be given to the cost of making an adjustment to the machine as well as to the cost of being off target and to the stochastic nature of the disturbance. An example is given where the appropriate form of action is like that required by Roberts's modification of a Shewhart chart. However, the justification required to make such action appropriate is very different from that previously given. [ABSTRACT FROM AUTHOR]

In a recent article Baruch Lev analyzed 29 peak-demand forecasts (consisting of 1917 monthly forecasts) made by Commonwealth Edison Company between 1955 and 1966. The conclusions made by Lev from analysis of these actual forecasts bear a close resemblance to the kinds of conclusions that we derived from an application of time series analysis to historical data to produce forecasts. It is the purpose of this paper to show how the conclusions which Lev made about forecast errors would also be true of the errors resulting from forecasts derived solely from analysis of historical data. Although in no way definitive, such a demonstration of consistency between actual forecasts made by businessmen and statistical forecasts made only from historical data yields further evidence of the practical applicability of time series analysis to forecasting problems in real life. In particular Lev concluded that the major factor affecting the distribution of forecast errors was the month of the forecast and that the "… interval between prediction and realization has only a slight effect on the prediction errors." Mr. Lev's analysis of these observations showed that the expected value of forecast errors was about one quarter of one percent and independent of the time interval of the forecast. In addition he pointed out that the variance of the log forecast errors increased slightly with the prediction interval but was affected more directly by the month which was forecasted. [ABSTRACT FROM AUTHOR]

Refunding is the operation by which the issuer of a bond calls it before maturity in order to replace it with another issue. The chief reason for calling is the interest saving which a new bond may make possible. In deciding whether to call a bond at any given time, the issuer must consider not only the potential savings from the current refunding, but also from future refundings up to some horizon fixed by financial policy. These savings must also be compared with those resulting from the decision to postpone refunding. This paper formulates the refunding problem in terms of a dynamic programming model in which the term structure of interest rates is utilized to obtain information about the future course of interest rates. [ABSTRACT FROM AUTHOR]

The article discusses forecast evaluation in accounting. Accounting data are often used in the development of forecasts and accountants are often called on to make forecasts. Development of a variable overhead rate per direct labor hour may, for example, be viewed as development of a forecast--whether it is used in a standard cost system, in a budget, or in a product mix decision. Similarly, establishment of a depreciation policy requires a forecast of service life. Furthermore, the ability to forecast some specific outcome has been offered as a criterion for selection among accounting alternatives. But, whatever its ultimate purpose, development of a forecast constitutes a decision--a conscious choice among alternatives. The concern in this paper is limited to this decision aspect of forecasting. In particular, in the first section it is demonstrated that selection among forecast alternatives is a subset of the accountant's general problem of selection among information alternatives. Following this, characterization is done of prevailing methods of selecting among forecast alternatives as a simplification of the basic information choice problem. The paper then concludes with two illustrative simulations.

The paper draws attention to the profound changes that are taking place in the proportions of the population of Great Britain that are single, married and widowed at various ages. These changes are the net result of a number of demographic, economic and sociological developments, such as the decrease in mortality rates and the changing balance of the sexes. After discussing an abortive attempt to forecast the future progress of the proportions, based on changes that have taken place in the recent past, the paper describes a method of estimating the pro- portions of each sex that are likely to be found in each marital status at the various attained ages in future years if the experience of the years 1950-52 in Great Britain as regards rates of marriage, divorce, widowhood and re-marriage continue indefinitely, subject to minor modifications to ensure that the numbers of marriages, etc., for the two sexes are consistent, and that the total number of married men equals that of married women. The variations in the proportions in each marital status in Great Britain during the past thirty years and those to be expected during the next thirty years, and ultimately, if the experience of the years 1950-52 persists, are set out and analysed. It is shown that, whereas the proportion of men never marrying has up to the present been about half the proportion of women never marrying, the continuance of present trends would result in the proportion of men never marrying becoming double that for women. This notable change is to be attributed, in the main, to the effect of improved vitality in childhood in replacing the excess of women at the younger adult ages by an excess of men. It is pointed out that, in the event, this change may not be fully realised, because a net emigration of men or a net immigration of women may mitigate the relative dearth of potential brides in future; but the proportions never marrying set out in the paper show the directions towards which current conditions are tending. The consequential changes for other marital statuses are also set out and discussed. [ABSTRACT FROM AUTHOR]

The article presents comments of the author on the article "Is Prediction Feasible in Social Work? An Inquiry Based Upon a Sociological Study of Parole Records," by Ernest W. Burgess. The paper by Burgess will be discussed from three points of view,its general methodological significance, its specific statistical methods and its limitations for practical purposes. Heretofore behavior has been studied principally in relation to specific factors, each factor being treated as an entity in itself and considered apart from any other factor. The results of this method have been largely negative, for the studies have shown that, when other things are equal, the difference between criminals and non-criminals on any one item is comparatively small. The next step evidently was to devise a method by which these small differences could be massed so that their combined effect could be seen. The significance of Burgess' paper is that it takes this next step, providing a method by which the massing or accumulation of factors can be measured or demonstrated. At the same time he has included a larger number of items than had been included in any published report and has used the composite rate as a basis of prediction. He has invented a scientific tool of very great utility both in the explanation of behavior and the administration of agencies which attempt to control behavior.

Abstracts of research and forecasting articles appearing in various periodicals. Articles abstracted include "Sales Forecasting Techniques Used by Selected Life Insurance Companies," by James P. O'Grady, Jr., appearing in the March 1965 issue of "The Journal of Risk and Insurance," "Life Insurance Fact Book, 1965" published by the Institute of Life Insurance, and "The Real Truth about Forecasting," by H. G. Hunt, appearing in the July 1965 issue of "Manager."

Can demand and consequently the market for consumers' durable goods be reduced to a mathematical model? Yes, these business economists contend, provided the right framework is chosen and proper statistical correlations are made to compensate for those eternal marketing variables -- price, credit, income, and use-facilities. They illustrate their point by constructing an econometric model, applicable to any durable good, and demonstrate its use and accuracy by analyzing sales of washing machines. [ABSTRACT FROM AUTHOR]

In this article, the author tests the hypothesis about the relationship between the level of short interest in equities and changes in the prices of these equities. Standard regression techniques on aggregate and firm data demonstrate that the short interest level is probably not a significant predictor of stock prices, even though regression techniques are difficult to apply here. A relationship between equity prices and short interest has implications for the profitability of short trading activities. A method for measuring the performance of investors by making conventional present value calculations is used to compare those with values simulated by a chance model. The results lend support to the view that there is no relationship between short trading and equity prices, for short traders do neither better nor worse using a random selection process.

The article focuses on the random walk theory of stock price behavior which posits that past stock-price movements cannot be used to predict future market prices that would allow traders to profit from the predictions. If stock price movements were to become predictable enough that a profit were possible, proponents of the random walk theory argue that enough market participants would quickly recognize and exploit the pattern. These market participants compete enough to make all non-random fluctuations so small that they cannot be exploited for a profit. As a result, the trader would be as well off with a buy and hold strategy as he would with a mechanical decision rule based upon past price changes.

This paper presents the optimal design of tests for demonstrating the reliability of a system, assuming that all the subsystems have a very high reliability so that no failure is expected to occur during the tests. [ABSTRACT FROM AUTHOR]

This paper develops a method for representing prognosis data and determining victim survival rate as a function of victim category, time, and treatment. Such a determination allows the evaluation of a community's emergency care system using the criterion of patient survival rather than some secondary evaluation criterion used in the absence of prognosis data. Typical prognosis curves are developed that express the probability of death from a specific illness or injury as a function of time and care received. Finally, examples are provided that show the probabilistic relation of mortality to time under varying assumptions of treatment and victim category. [ABSTRACT FROM AUTHOR]

The characteristics of seasonally adjusted, exponentially smoothed forecasts are studied through the frequency response and impulse response functions. The appropriateness of using exponential smoothing for updating the seasonal base series is called into question, due to its tendency to store random noise for long periods. Exponential smoothing with trend is also examined. It is concluded that Brown's formulation has the property of being critically damped, so that its response to changes in the time series under study is the most rapid possible without overshoot. [ABSTRACT FROM AUTHOR]

This paper presents a comparison between similar types of American and Canadian forecasts of annual gross national product for an identical period (1960-69) using the same error measures. It was concluded that on mean, mean absolute, and mean square error bases, American forecasts were significantly more accurate than were the Canadian ones in predicting relative changes in their respective annual gross national products during the period 1960-69. Forecasters in both the United States and Canada were likely to have made the same types of errors on average in forecasting any given year. The general conclusion from the analysis of the errors seemed to suggest that a bias toward underestimating change may be less evident in the American gross national product forecasts than in those made in Canada. The negative mean errors were smaller, the percentage of underestimates was smaller, the covariance components of the mean square errors were larger, and the regression and test statistics were less suggestive. One of the more important points is that when "adjustments" of the predictions for differences in the basic nature of the economies are made, the conclusions as to differences in the "accuracy" of aggregate forecasts made in various countries may change a great deal. That is, conclusions as to differences in forecasting ability based solely on the comparison of "absolute accuracy" could be seriously misleading if the economies are not of approximately similar natures. The method of "adjustment" used here assumes that information as to the nature of the economies is contained in the past history of the variables to be forecast, so that a more worthwhile comparison is obtained when the respective judgmental sets are first "adjusted" by extrapolative predictions from these past actuals. Comparisons of the "relative" accuracy of forecasts showed that the American predictions of annual gross national product percentage changes had larger relative mean square errors than did the Canadian sets. That is, the American sets were only about as accurate as, and probably less accurate then, relative to extrapolative predictions. the forecasts made in Canada. The predictive efficiency of the Canadian gross national product forecasts relative to that of the extrapolative predictions was generally higher than that of the American predictions; the amount of autonomous information and available extrapolative information included was also somewhat higher. However, the extrapolative information actually used in forecasting was quite low in both countries, probably because of the essentially trendless nature of percentage changes in gross national product. [ABSTRACT FROM AUTHOR]

This paper investigates the welfare implication of changes in the transportation system in two special areas. The first is when the origin-destination demand for transportation may be assumed to be fixed and the second case is where, considering only work trips, origins may vary but destinations may not. A technique is described that could be used to forecast the new origin-destination demand resulting from a change in the transportation system. The technique also provides an appropriate measure of the welfare implications. A particular objective of the paper is to point out the pitfalls of using transportation (generalized) cost saving as a welfare measure whenever origin-destination demand may not be assumed fixed. In this situation it is shown that the welfare measure must take into consideration the benefit derived from the increased choice in available housing sites. [ABSTRACT FROM AUTHOR]

This paper describes a case study of selected inventory accounts maintained by a large manufacturer of rocket engines. It was hypothesized that a significant improvement in average inventory levels, number of purchase orders, and number of stockouts could be obtained by augmenting existing methods of inventory management with certain OR based techniques. These techniques included exponential smoothing of demand forecasts and statistical determination of optimum safety stock levels. Computerized simulation was used to determine: (1) the most effective parameter (alpha) values for use in the exponential smoothing formulas, (2) an estimate of the potential dollar savings obtainable if the proposed methods were implemented, and (3) a means by which these methods could be implemented. A simulation model was developed for two particular inventory accounts, its correspondence with the real world was tested, and the alpha values and anticipated dollar values were determined. Additionally, a set of curves were developed for use by inventory management to assist in determining an optimized policy for varying values of purchase order cost, inventory carrying rate, and cost of stockouts. [ABSTRACT FROM AUTHOR]

This paper considers the use of exponential smoothing to forecast time series which are compound processes containing both deterministic and random components. Recognition is given to the fact that modifications made in the simple smoothing formula to correct for one assumed component in the input series, such as a trend, interacts with the other series' components, and the total forecast error may be increased. The mean squared forecast error, using exponential smoothing, is compared when the random inputs are independently distributed and when they are autocorrelated. The latter case being commonly encountered in sales series The characteristics of the smoothing procedure is examined when the input time series is geometrically autocorrelated and the range of values of the smoothing constant to minimize the forecast error is calculated. [ABSTRACT FROM AUTHOR]

In a recent article [5] published in the Cahiers économiques de Bruxelles Professor J. Waelbroeck analyzed world trade flows in a very interesting way. He took total exports to and total imports from each region as given and applied the RAS method—originally constructed for input-output analysis—to analyze the extent to which flows known from an earlier table can account for flows in a later year. The approach has some resemblance to certain aspects of bivariate information theory. The present paper takes this theory as a starting point to provide an alternative method of analysis. The empirical part of the paper is based on the same data as those which are used by Waelbroeck. It turns out that the two methods yield virtually equivalent results, which is not really surprising, but which sheds an interesting light on the nature of the RAS method. Section II considers international transaction shares from the standpoint of information theory; section III considers some empirical results, and a third prediction method; section IV the RAS method; section V, the information inaccuracy of forecasts of import and export distributions for individual regions; and section VI traces the E.E.C. and Communist policy effects. [ABSTRACT FROM AUTHOR]

Economists' forecasts influence business behavior and, therefore, values of variables being forecast. Forecasts of national income, for example, influence private investment outlays which, in turn, help to determine the level of national income. It is sometimes suggested that economists' forecasts are self validating, implying that a forecast will be more nearly correct if it is made public than if it is kept a dark secret. Almost at the other extreme, there is the view that publicly announced forecasts must inevitably be invalidated by the response of the business world to the forecast. On a slightly different plane, it is sometimes argued that the instability of the economy can be traced, at least in part, to faulty forecasting and would be reduced by an improvement in the quality of forecasts. This paper will examine the validity of these three points. It will be shown that when the variable to be forecast is itself influenced by the forecast the announcement of a forecast does not necessarily validate, or tend to validate, the forecast, accurate forecasting may be impossible and even perfectly accurate forecasting, if it were possible, may add to the instability of the economy.

This paper discusses the methods and the results of a forecast for the demand for operative and obstetrical procedures and the supply of anesthesiologists in Cuyahoga County, Ohio. The techniques and results of a ten-year forecast for the demand for and supply of anesthesiologists and auxiliary personnel in greater Cleveland are discussed. Several regression models were used to forecast supply based on population, number of physicians, and the income per capita. The demand models were based on population, age and sex distribution projections, and historical data regarding operative and obstetrical procedures. The results of these "objective" models were then compared to forecasts under uncertainty generated by a panel of experts using the Delphi Method. Alternative states of health care delivery were investigated and implications for future anesthesiologist manpower requirements detailed. [ABSTRACT FROM AUTHOR]

This paper is a case study in the development, implementation and evaluation of the delphi technique, which systematically makes use of expert judgment in generating manpower forecasts. The study was conducted in a large national retail organization on professional manpower. The results of the delphi technique are compared with results generated by conventional regression based models and the actual experience of the organization, which serves as the criterion. The study also analyzes the informational elements used by experts during the delphi procedures and develops a model based on these elements. The usefulness of the delphi in generating manpower forecasting models is also discussed. [ABSTRACT FROM AUTHOR]

Seasonal-style-goods inventories are characterized by the substantial losses associated with those merchandise items which are unsold at the end of a season. If the probability distribution for sales for the entire season were known, the decision problem would reduce to determining the optimal lot size, as in the classical newsboy problem. In reality, however, it may be possible to consider sales estimation as a multi-period inventory problem because there may be opportunities during the course of the season to review the earlier sales performance to reestimate forecast errors made earlier in the season, and to take appropriate action based on the new estimates. In this paper Kalman-Shaw linear feedback filtering is used to estimate previous forecast errors. The model is formulated in such a way that, given an initial forecast, the subsequent reestimation of the initial forecast error requires only the actual sales experience to date. [ABSTRACT FROM AUTHOR]

A sequential decision problem is partitioned into two parts: a stochastic model describing the transition probability density function of the state variable, and a separate framework of decision choices and payoffs. If a particular sequential decision problem is a recurring one, then there may often exist human forecasters who generate quantitative forecasts at each decision stage. In those cases where construction of a mathematical model for predictive purposes is difficult, we may consider using the forecasts and forecast revisions provided by the existing forecasting mechanism as the state variable. The paper considers a specific class of problems in which improved forecasts of some unknown quantity are available before each decision stage. A small number of actual forecasters are studied through analysis of historical data to see whether the data-generating process for forecast changes is quasi-Markovian. The data are generally, although not entirely, consistent with the hypothesis that ratios of successive forecasts are independent variates; their distribution appears to be conditionally Lognormal. Some possible reasons for these results are explored. In cases where the hypothesis holds, a dynamic programming approach to the sequential decision problem may be used to provide optimal decision rules. The usefulness of the approach b illustrated with a numerical example involving crop planning, and the example is extended to explore the effects of using the methodology when the required assumptions do not hold. [ABSTRACT FROM AUTHOR]

This paper illustrates how individual forecasts and forecasting techniques may be evaluated by the use of established decision theory. Given the probability distribution of the forecast error, we first find the optimal strategy for a decision process, i.e., how to make the most efficient use of a forecast. After expressing the expected profit of the optimal strategies in terms of the probability of a correct forecast, we illustrate how to find (1) the value of an imperfect forecast, and (2) the value of a forecasting technique. With this information, we can then determine when to use a forecast, the maximum amount to pay for both a forecast and a forecasting method, and the conditions under which it is worthwhile to attempt to improve the accuracy of a forecasting method. [ABSTRACT FROM AUTHOR]

This section presents abstracts of research papers on business fluctuations which were published as of June 1967. Price Images, Inflation and National Incomes Policy examines the problem of learning prices and price-judgments, and of re-learning them when people are exposed to long periods of inflation. The Existence of Golden Ages and Stability in the Two-Sector Model examines the relationship between the existence and the stability of Golden Ages. The paper proves that although the two-sector model may be unstable and may possess multiple equilibria, the behavior of per capita consumption across alternative Gold Ages is not perverse. Forecasting Errors and Business Cycles found that aggregated sales forecasts systematically underestimate actual changes in sales. One explanation is to be found in macroeconomic theory. In equilibrium, actual sales equal expected sales. An Econometric Model of Mild Business Cycle presents an econometric study of the three U.S. business cycles in the period 1951-1961. The model used is based on the inventory cycle hypothesis of Metzler. In Do Regional Business Cycles Exist?,the influence on regional cycle differences of dissimilar response of firms in the same industry was tested for with the Analysis of Variance technique.

This paper is in two parts. The first section describes the relationship of a number of variables to political status and their relationship to each other. The second section deals with the kind of predictions that might have been made with these data and the relative efficiency of one type of prediction compared to another. [ABSTRACT FROM AUTHOR]

This article discusses the development of statistical methods especially useful in test construction and evaluation. The attempt to predict a dependent variable from one or more independent variables rested upon the use of both linear and curvilinear models were initiated by the author. A great deal of effort was also directed to ascertaining optimal sets of weights for the variables in linear composites when in the absence of a criterion it was desired that each of the variables satisfy certain condition such as contributing to the variance of the composite terms of its intended degree of importance.

An interest in the political strategy implications of the 1960 election polls led authors to collect a sample of forecasts from around the U.S. The project started quite innocently as a result of a penchant for clipping articles from the newspaper "New York Times." Once launched on clipping poll results authors also wrote a few letters to newspapers and research organizations to obtain clarification of the news reports or additional information. But the reader should be warned that, despite these efforts, this brief review of last fall's polls is somewhat capricious and unsystematic, as of September 1, 1961. In all, some data were obtained on 36 separate measurements of electoral preference or voting intention. Of these all except one were styled polls, straw votes, or opinion surveys. Of the other 35 measurements about which minimal information is available, 14 were conducted by newspapers, 12 by commercial or academic survey agencies, and 9 for Democratic or Republican campaign groups. The percentage errors for the polls ranged from to 16.8, with the average error being 2.9.

Any financial statement provides a partial basis for evaluating the results of current operations and current financial position. Financial statements relating to periods of less than one year have the same evaluation and forecasting objectives as annual statements. However they differ in a manner that they are also used by outside investors in forecasting the results that will be shown on the annual statements. But interim statements have never received the professional attention that has been devoted to annual statements. The purpose of this research paper is to explore the objectives of interim reporting and to make a tentative statement of concepts to be applied in the development of interim financial statements. Through this exploration it suggests that the usefulness of published interim income statements in predicting annual profit is likely to be impaired if these statements do not include adjustments for fluctuations in the timing of cost releases and concludes that the management has a responsibility for anticipating in income statements for the first three quarters the annual total of a wide variety of business expenses.

The article focuses on the extent to which predictions can be made about certified public accountants (CPA) examination results. Studies of the use of the Institute sponsored Orientation and Achievement Tests as a means of making such predictions have produced encouraging results. On the basis of these studies, which are of limited scope, it appears that reasonably accurate predictions can be made in many instances. There are many benefits which will accrue to the profession and to individuals if fairly accurate predictions can be made that a particular individual should be able to pass the CPA examination; that another individual is unlikely to pass, and that still another one has a slight chance to pass. Such predictions would aid appreciably in guidance work, and would be of value to employers. Predictive devices might be used by state examining boards in various ways. Boards might be aided in their efforts to encourage potentially able men to continue preparation and to discourage those who are obviously not ready to take the examination as well as those who may never qualify. Predictive devices might aid considerably in reducing failures.

The mere fact that certain stimuli are labelled "values" does not make them non-measurable, nor does the fact that responses to such stimuli are called value judgments prevent them from displaying empirical regularities which may enable social scientists to make predictions. This was the conclusion reached in an earlier paper which reported the results of three empirical tests of the hypothesis that qualitatively unlike values are incommensurable. All three tests resulted in rejection of this hypothesis, thereby indicating that "human values, including those which are regarded by certain authorities as being of infinite worth, become measurable relative to each other in exactly the same manner as other verbal stimuli. Although the critical ratios which resulted in rejecting the null hypothesis in the previous study were all quite large, it seemed desirable to repeat with a new group of respondents at least the third of the three experiments—the one dealing with "infinite" values. Denominationally stratified samples of the ministerial members of the Seattle Church Council had been used in the first run of this experiment.