Characteristics of a practical market index

The technique of market forecasting is concerned very largely with methods of singling out market indices, determining the effect each index really has, and combining the pertinent indices into a single measure, giving each component part its correct weight so that a reasonably accurate measure of sales possibilities may be obtained. The first of these problems should be handled on some basis other than mere seeming relationships between the apparent index and actual sales. A good market index should possess the following general characteristics :

• It should be based upon facts that are connected actually in some way with the given company's product line
• It should be sufficiently selective to eliminate the non-users of the company's products
• It should consist of a set of statistics that are accurate
• It should consist of up-to-date figures
• It should present the possibility of an economical compilation.

Need of testing proposed indices

The process of finding actual relationships between tentative indices and the company's sales is, in most cases, a process of trial-and-error experimentation. Apparent connections between an individual set of statistics and sales must be tested and analysed. In many cases coincidental market indices may seem to maintain a certain ratio to the company's sales, but these apparent connections may break down when they are examined carefully. In order to find factors which correspond with variations in sales by districts or trading areas it is desirable to use statistical methods of comparison. By such methods it can be discovered whether the suggested market index or factor actually moves along parallel with the variations of sales, or whether the two sets of data move in opposite directions.

Method of testing market forecast indices

Methods have been developed for testing forecasting indices. Some of these methods involve mathematical calculations. The most convenient as well as the most accurate statistical method of testing and expressing the connection between two sets of data is by means of a coefficient of correlation.

By working out the coefficient of correlation between a proposed index, such as number of income tax returns, for example, and the company's actual sales, trading area by trading area, the proposed index can be scientifically tested. If it shows a high percentage of agreement with sales it can be adopted and used in combination with other indices which have been tested in the same way. If it does not show a close agreement with sales, it should be discarded as an indicating possibility.

The various formulas for finding the coefficient of correlation between two sets of measuring data are presented and discussed in practically all of the standard works on statistics. They are too specialized for presentation here. As a matter of fact, any business concern which wishes to set up a forecasting system should call upon an experienced market consultant to work out the technique of the system.

Important considerations in testing indices

In addition to this testing of suggested indices to establish their connection with the company's sales, these suggested indices should be examined to discover whether or not they meet the other requirements. The selectivity of the index may be evaluated qualitatively and may indicate that more than the particular index is necessary. Income tax statistics, for example, are probably of little selective value in excluding non-users in the case of a product line that sells over the retail counter at a price of less than one dollar.

Some attention, too, should be given to the probable accuracy of a set of figures which it is proposed to use as an index. If the source and method of collecting the statistics that make up a single possible index are at all open to question, it must be used with the utmost caution.

The question of frequent revisions and the current accuracy of the index factor must be taken into consideration. Many governmental compilations are almost out of date by the time they are issued. Finally, the cost of preparing and compiling periodically the several index factors must be estimated. If the process of forecasting becomes too complicated and costly, there may be a real question as to its value to the company.

Commonly used market indices

Some of the most commonly used market indices are :

• Population
• Income tax returns
• Value-of-products indices
• Number of automobiles
• Number of telephones
• Trade indices such as bank deposits, building permits, etc.
• Magazine and newspaper circulation
• Number of retail outlets

This list is, of course, by no means exhaustive and represents only a few of the more commonly used indices. In every case, the individual seller will have to test these factors as suggestions and discover whether one or more can be used for his particular business. If these indices are of no value a search must be made for others that present an abundance of data and which are connected with sales. Each one of these indices listed will have quite definite limitations when used alone. A few words of comment are necessary with respect to each.

Population

In the case of product lines that have a universal appeal, population is frequently considered to be an important market index. But unadjusted population figures used alone are misleading. Differences in wage rates and buying habits over the various trading areas make such an index an inaccurate guide.

For example, if the Pacific Coast areas are accustomed to buying the product line more freely than is the case in other areas, the proportion of sales on the Pacific Coast is definitely affected. Likewise, population statistics will give too high an estimate for areas where there are only a few people with high incomes and too low an estimate where there are many people with high incomes.

As a result of this inadequacy of unadjusted population figures, negroes and foreigners are often excluded and the figures for the native white population are used. But even such an exclusion, while it will reduce the figures for New England, the Middle Atlantic, and Southern states and will increase slightly the other sections, will not make any material changes in the distribution figures. Certainly, the native white population index carries no guarantee of interpreting and measuring buying habits. Nor will the distribution of the literate population over 21 years of age prove to be a factor of any greater value as a general rule.

Illiteracy is unevenly distributed and there is a marked tendency of the larger cities to attract members of the age group above 21. It would be folly, for instance, for a manufacturer of linoleum to rely to any considerable extent upon literacy statistics as a market index.

Income tax returns

The most obvious guide to purchasing power is that of income tax returns. But income tax returns are only a partial guide. Only about 6 per cent of the population have been paying an income tax. Large groups of the population do not earn incomes high enough to be required to make returns and other groups make their incomes from tax-exempt securities and other items of a similar nature. In fact, the market for most consumer's goods lies to a great extent among people whose incomes are not represented in income tax returns.

Moreover, an income totalling $3,500 in a city like New York is not the same as an income of $3,500 in Walhalla, North Dakota. The differences in the cost of living is all in favour of Walhalla. A comparison of income tax returns and population figures indicates that a much heavier percentage of income tax returns comes from the north-eastern section of the United States and from the Pacific Coast than the population figures seem to warrant.

The number of income tax returns alone, therefore, is hardly a selective or accurate market measure. If, on the other hand, some technique of adjustment can be worked out for average income, income range, and cost of living, area by area, the results should show the "length, width, and depth of per capita purchasing power."

Value-of-product indices

Among the most important of varied value-of-products indices are:

• The value of manufactured products
• The value of farm crops
• The value added by manufacture
• The value of mineral products
• The value of fishery products

Most of these statistics are obtained from governmental departments such as the United States Census Bureau, in its Census of Manufactures, the Department of Commerce, the Geological Survey and the Bureau of Fisheries. Like the income-tax statistics, those value figures are often out of date before they are made available. Furthermore, the value indices are limited to various sections of the country and tend to overemphasize geographical centres of production.

In a given section, too, the value of manufactured products may be high because they are turned out by power-driven machinery under mass production methods, while the workers may be relatively few and earning only low wages. In another section where less of the work is done by machinery a larger number of workers may be employed and the total purchasing power of the area may be large, while the value of manufactured products may be low. Moreover, the value indices do not show necessarily that the value added has been accumulated in the particular area or how it has been added. Profits may go to few or to many inside or outside of the particular area in which they are earned.

Automobile registrations

Automobile registrations are also frequently used as a market index. It is possible to obtain statistics on the total number of automobiles sold and on the number of each make sold in each county. But unless the product line has a very close relationship to the sale of automobiles, it is not apparent that automobile registrations constitute a good market index. When compared with income figures the automobile registrations for New England and the Middle Atlantic states are low while through the east and west North-Central states the figures are astonishingly high. The Curtis Publishing Company in its "Sales Quotas" said :

Automobile registrations emphasize rural markets. In rural districts an automobile is generally recognized to be a necessity ; in the city many who have the means to buy are deterred by lack of garage facilities and the congestion of city traffic.

It is difficult if not impossible to distinguish between automobiles that are purchased as necessities and those bought for pleasure. Nor do the statistics make any distinction between new and used cars. Moreover, it is difficult to know which areas are thoroughly "worked" by the automobile manufacturers' selling organizations and which are not. In general, it is perhaps safe to say that because of instalment buying, the ownership of automobiles is much less reliable as a market index to the manufacturer of a product line unrelated to automobiles than formerly was the case.

Telephones

Statistics with respect to the number of telephones are frequently considered of value as market indicators. They can be obtained easily but they are confined to the basis of the state and the city of 50,000 population or over. Even if they were obtainable by counties, it is doubtful whether or not they would serve as accurate measures in many instances.

Their ability to measure purchasing power is impaired by the effect of variations in rates, habits, size of cities, homogeneity of population, and other factors. Telephone rates are relatively lower in small towns than in large cities because the limited number of subscribers cuts down the service demands. In small towns, however, because of the homogeneity of the people, many families will not feel the need of telephones. In the large city, the lack of many personal acquaintanceships, the demand for rapid communication and the habit of ordering goods by telephone will promote telephone installations.

The number of telephones may indicate families rather than individuals, but in the apartment-house cities such will not necessarily be the case. Cities will differ considerably one from the other in the number of telephones as compared with either the number of families or of individuals.

Indices of trade activity

There are numerous groups of statistical indicators of trade activity that are sometimes used as market indices. Such figures as bank deposits, bank clearings, savings bank deposits, and building permits may be classed as trade indices. Total bank deposits present their particular weaknesses as market indicators. In the first place, the figures for bank deposits cannot be secured for all cities. In the second place, bank deposits include both checking accounts and savings accounts.

Even the checking accounts are distorted by corporate and unit store deposits. Finally, banks receive deposits from wide surrounding areas particularly in rural and suburban sections. The received deposits cut across county lines—an individual county may show considerably more bank deposits than actually originate within its boundaries. Some of the same objections arise in the use of bank clearings as a market index. The corporate and unit store distortion is present as in the case of bank deposits. Moreover, duplication exists because the same check may be cleared at a number of different clearing points.

Savings bank deposits indicate something with respect to purchasing power. But they actually show the frugality of the depositors as well as their net purchasing power. Once a savings deposit is made it is seldom withdrawn for the purchase of convenience or shopping goods. Thus, in a sense, savings bank deposits show where money is rather than where it is being or will be spent.

So far as building permits are concerned it is evident that such figures may be important to manufacturers of product lines which are used in new building construction. But in the case of a product line which is not directly related to the building trades it is hard to see any real service that this index may render of itself.

Magazine circulation statistics

The circulation statistics of various magazines are also used frequently as market indices. But circulation figures are hardly fundamental market factors. They are merely imperfect reflectors of market possibilities. When magazine circulation is used in combination with population figures, the influential population factor is duplicated in each index. When used alone, circulation figures present the question of what particular magazine circulation data should be used. In the various territories there is a marked variation in the circulation of different magazines.

If circulation data really indicates buying power and buying tendency, there is always remaining the difficult question of selecting the magazine which best indicates these matters. The experiments in combining the circulations of several leading national magazines have not resulted in the formulation of an accurate market index. One such index which combines the circulation of 30 national magazines shows only five per cent of circulation in New York State.

Yet New York has at least ten per cent of the population and an even higher percentage of the purchasing power. The fact that illiterate or foreign born people do not subscribe to American national magazines does not bar them from the market as purchasers of staple and fancy foods. In fact, in the case of butter, the very diet of groups of people like these may cause them to use more butter per capita than similar sized groups of literate and native-born consumers.

Number of retail outlets

Occasionally, the number of retail outlets on a per capita basis is used as a market index on the theory that such statistics reflect the demand for goods. However, the figures for retail stores are far from accurate and they do not measure accurately the amount of business actually done at retail in particular territories. It is evident that one area may produce sales 100 per cent greater than another area and yet not support any more retail establishments. Nor do retail outlets indicate at all directly the net purchasing power of the ultimate consumers in those areas.

Necessity for several indices

From this discussion of the limitations of some of the most commonly used market indices it should be clear that in almost every case accurate market forecasting will necessitate the use of more than one index. When several market indices are found which have a decided relation to the territorial variation in the sales of a particular commodity, the next task is to give each one of these market factors its proper weight and to combine them all into one single index. This task is by no means an easy one. In practice, the weighting of individual indices is accomplished mainly by estimating the relative weights.

Suggested methods for weighting and combining indices

One interesting method of weighting indices arbitrarily was explained by Nelson Seubert, consulting statistician, in the Editor and Publisher. Mr. Seubert discussed three market indices: The percentage distribution of population, the percentage distribution of the number of all income tax returns, and the percentage distribution of income tax returns on incomes of $10,000 and more. The first of these market indices he called the "quantity factor," the second, the "quality factor," and the third, the "luxury factor." To indicate the methods of weighting these indices, Mr. Seubert set up a six-fold classification of products having a universal appeal and suggested the proper weighting for each. They appear as follows :

• Class 1. Cheap necessities and bulk staples. Weighting: Quantity factor 90 per cent. Quality factor 10 per cent. The great weight is logically given to population, but the amount of money available for purchases always affects the situation; hence the 10 per cent weighting given to income.
• Class 2. Moderate-priced necessities and packaged staples. Weighting: Quantity factor 70 per cent. Quality factor 30 per cent.
• Class 3. Quality-priced necessities and staples. Cheap luxuries and low-priced nonstaples. Weighting: Quantity factor 50 per cent. Quality factor 50 per cent.
• Class 4. Fancy-priced necessities and staples. Moderate-priced luxuries and nonstaples. Weighting: Quantity factor 30 per cent. Quality factor 70 per cent.
• Class 5. Quality-priced luxuries and nonstaples. Weighting: Quantity factor 10 per cent. Quality factor 90 per cent.
• Class 6. Fancy-priced exclusive luxuries of high unit value. Weighting: Quantity factor 10 per cent. Exclusive luxury factor 90 per cent.

Weighting and combining basic indices

Another equally interesting method of combining and weighting indices in the case of a forecasting procedure used in estimating the wholesale and jobbing market for electrical supplies was reported by Edmond E. Lincoln, while economist for the International Telephone and Telegraph Company, in an article in the Management Review. In this instance basic and variable indices were used.

The basic market factors are those that have a steady growth in given territories, "so that within a reasonable period of time relationships between different territories, based on these factors, will probably not change appreciably." The particular basic indices which were selected were : Urban white population, population in territories served by central stations, value of production added by manufacture ( all industries) , central station capacity, residence meters, total horsepower of industrial plants, capitalization of industrial plants manufacturing electrical material or products incorporating electrical materials, number of telephone stations not owned by Bell System, railroad mileage and passenger cars.

These market factors were combined by arbitrary weightings into a composite basic factor index.

Weighting and combining variable indices

The next step was the selection of "variable" market indices, those factors that are very sensitive sales indicators so far as the market for electrical goods in the various districts is concerned, "currently and from year to year." The particular "variable" factors chosen were: Building permits as compiled by Dun & Bradstreet, production of electrical energy and bank debits outside of New York. The next bit of procedure was to weight and combine the two composite indices into final factors. The weightings decided upon were two for the composite basic factor and one for the composite variable factor.

Weighting often a matter of judgment

Throughout these procedures, the weighting of the various factors is a matter of judgment based on past experience and on sales possibilities relative to various classes of customers. In many cases such judgments are relatively sound. For instance, in the make-up of the composite variable factor a 50 per cent weighting is given to the index of building permits. Obviously, a very large per cent of the sales of electrical supplies are directly or indirectly influenced by new building activities.

New experiments in weighting

Although in present practice indices are weighted and combined on the basis of empirical judgment alone, a considerable amount of experimentation is directed toward the use of mathematical and statistical methods to give each market factor its proper weight. The so-called statistical Method of "multiple correlation" is growing in use because it actually accomplishes an automatic and proper weighting in combination.

Adjusting for competition

Finally, it should be said again that although the total sales of a given product line, territory by territory, can be determined by a reliable composite market index, the percentage of this market which can be obtained for a particular company depends upon the strength of the competition.

Once the total potentialities are estimated, it is necessary to examine the particular competitive share of the business which has been obtained by the company in the past and to rely upon such data in predicting the sales possibilities for the individual company. Of course, the percentage share which has been secured in the past may be combined with a reasonable "percentage of encroachment" on competition. But this latter percentage, although arbitrarily set, must be determined with great caution.

It must be determined in the light of the conservative opinion of the company's sales executives, the efficiency of present and proposed sales methods in the various territories, and the expected efficiency of competitors' sales methods in the same territories. Obviously, the last factor is exceedingly hard to predict. Market forecasting is a particularly precarious undertaking when it comes to making adjustments for competition.

Public Relations
Activities for public relation
Budgets for public relations
Changing happenings into news
Community relationships
Consumer relationships
Costs for public relations
Effective areas of PR
Functions public relations
Government relationships
Labor relationships
Magazines public relations
Methods of communication
Newspaper public relations
Prestige achievements
Public field relations
Public utilities
Radio, TV and PR
Stockholder relationship
Techniques public relations