At that time I was concerned not with current stocks but with the possibility that a dangerous surplus might develop by July 1. Mr. Brown's figures indicated that all was not well, and although I was impressed by them I hesitated to overrule the divisional people who had the responsibility for selling. There will always be some conflict between the figure men and the salesmen, since the salesmen naturally think they can do something about a statistical situation, as they often can. I got in the middle here—between Mr. Brown and the divisions—as I often did in the presence of conflicting representations of reality. But then in May 1924 Mr. Brown and I made a trip into the field to discuss distribution problems with the dealers in their places of business, and on that trip I came to know beyond doubt that the March cutbacks had been inadequate and that overproduction was not just a possibility for July but already a certainty. It is not often that the chief executive of a large corporation himself discovers visible overproduction by a physical check of the inventory. But automobiles are big units easy to count. In St. Louis, my first stop, in Kansas City, and again in Los Angeles, I stood in the dealers' lots and saw the inventories parked in rows. The figure man in this instance was right and the salesmen were wrong. Everywhere the inventories were excessive.
I then issued one of the few flat orders I ever gave to the division managers during the time I served as chief executive officer of General Motors. This order directed all division managers to curtail production schedules immediately—the over-all total reduction for the corporation amounted to about 30,000 units a month. By cutting production schedules drastically we were able to reduce dealer stocks to manageable proportions in a few months' time, but not without considerable economic hardship to the employees of the corporation who were laid off.
On June 13, 1924, I was taken to task by the Finance Committee for not having anticipated and prevented this overproduction. The committee adopted a resolution requesting me to explain how our production schedules were developed and who was responsible for the excessive number of General Motors cars in dealers' hands during the spring and summer, and what we proposed to do to avoid a repetition of such an event. The committee stated its questions as follows:
First. What plan of procedure has heretofore been followed in developing production schedules?
Second. What was the justification of a production schedule of 101,209 cars for April as indicated by the February 25th forecast recognizing that the stock of cars in the field was approximately 236,000 as of the end of February?
Third. Why were steps not taken earlier by the Operating Divisions to drastically curtail their production more in line with the unsold stocks of cars in the field and the consumer demand?
Fourth. What steps will be taken to assure effective control of production schedules in the future to guard against over-production?
Fifth. In what way will the Finance Committee be informed as to the general aspect of the situation enabling it to determine whether monthly forecasts are predicated upon a volume of sales to consumers which is in harmony with the Committee's judgement of general business conditions?
In my reply to the committee on September 29, I censured some of the divisions, particularly Chevrolet and Oakland. I pointed out that Cadillac was the only division that had been guided in any way in its production planning by sales to ultimate consumers. The other divisions had developed their production schedules in various ways, generally following the notion that the responsibility from the sales standpoint ceased with the delivery of the product to the dealer or distributor and that the corporation had no further interest in the situation. Our response to the events of 1924 formed a turning point in the development of our procedures for control of our production schedules. I described the situation at that time in a report to the Finance Committee:
First: That up to the first of July, 1924, or thereabouts, production schedules were developed in various ways, based largely on the theory that the responsibility from the Sales standpoint ceased with the delivery of the product to the dealer or distributor, and that the Corporation had no further concern in the transaction, and as long as dealers and distributors could be forced to take cars, the situation was regarded as entirely healthy and constructive.
Second: That no real study of the fundamentals had ever been made. By this is meant that although information as to the sale of cars to consumers — manifestly the real index — had been available in a more-or-less satisfactory form during the past two years, this fundamental data had never been developed and capitalized as a guide in the preparation of production schedules.
Third: That July 1st, 1924, a procedure was developed, which, it is believed, places production schedules on an entirely scientific basis, and founded on a real fundamental index. Accountability for the development of such schedules is now definitely established between the Operating Divisions and the Corporation itself, the latter satisfying itself as to the constructiveness of same. A copy of this Procedure entitled "Monthly Forecast of Deliveries to Consumers, Production, Inventory and Sales", has already been submitted to your Committee, but to make this report complete, that section of it dealing with an analysis of production requirements based on forecast of deliveries to consumers, is appended as "Exhibit A".
Fourth: The lack of a proper and fundamental development of production schedules was in no sense limited to General Motors Divisions, for, as a matter of fact, the same method was being followed by the entire industry. This situation is one of the causes contributing to the present economic condition of the average automotive dealer, which condition was reported to your Committee by the undersigned following the field trip taken during the month of May.
Fifth: After due consideration, the undersigned, on behalf of the Corporation, issued a statement which, as evidenced by comments received from dealers and distributors, and by editorials in the automotive press, has been recognized as performing a valuable service and establishing a precedent which other automotive manufacturers will be inclined to follow in future.
In my statement to the Finance Committee then, I summed up my personal feelings:
(a) That it is rather a reflection on the General Motors Corporation, and equally so on the industry as a whole, that nothing of this kind has been accomplished before, nevertheless, like many other important considerations which have not yet been developed as logically as they might be, this should be looked upon as a natural happening in an industry which is not yet thoroughly stabilized.
(b) There is no question in my mind but that General Motors now has absolute control of its production schedules. I feel further that the policy inaugurated by General Motors, as evidenced by the Corporation's Dealer policy, and by similar statements on the part of other manufacturers, cannot help but assist the economic position of dealers and thus be a great help to the industry of which General Motors is such an important part.
I have recounted this episode of 1924 because of its consequences, for it marked the beginning of reasonably effective production control in General Motors. In a certain very important sense, this involved the reconciliation of the work of two kinds of persons in General Motors—essential, I should think, in any corporation with a nationally distributed consumer product. One kind is the sales manager with his natural enthusiasm, optimism, and belief that he can, by his efforts, influence total sales. The other is the statistical person who makes his analyses objectively on broad general evidence of demand. Resolving the conflict between these two viewpoints would give one an idea, for example, of how many cars one should expect dealers to stock. This conflict was especially acute in those days when we had not yet solved the problem of reconciling production levels with seasonal peaks in sales. And of course behind this was the basic problem of controlling production.
Two things were involved: first, the art of forecasting, and second, shortening the reaction time when a forecast proved wrong, which can be expected to happen even in the present day of complex mathematical forecasting technique.
&
nbsp; Since we at headquarters had begun to develop techniques of fact-finding and analysis, we were in a better position than the divisions to forecast total industry demand and total sales of our products for the entire model year. And since production, the level of dealer inventories, and general financial planning all depend to a large extent on the outlook for the entire model run, we decided in 1924 to establish an official corporation-wide estimate of consumer demand— that is, an estimate of the number of cars in each price group likely to be sold to the public during the coming year by the entire automotive industry—and to have this correlated with the division managers' forecasts, keeping in mind the percentage of each price group that General Motors might reasonably be expected to obtain. This corporation-wide estimate was based upon actual experience of sales during the past three sales years and an appraisal of the general business outlook for the coming year.
We took the first actual step in putting limits on the divisions in the spring of 1924. Mr. Brown and I worked out along the above lines an estimated volume of business for the second half of the year for the corporation as a whole and for each division. We called this expectation of sales volume the "index volume," that is, the volume to be regarded as the pointer for a twelve-month period. After the index volume had been approved by the Operations Committee, I issued a general letter to the division managers on May 12, 1924, asking that their forecasts for the last six months of 1924 be based on that index. This letter read, in part:
Heretofore in asking for these estimates [the divisional sales forecasts] the basis upon which they have been considered i.e., the volume of business, has been left to the individual judgment of each Operating Division. This time, and dealing with the second half [of the year], I believe we are taking a constructive step forward. I mean by this that the Operations Committee has determined, as a group, the probable trend of business for the manufacturing year beginning July first . . . That being the case, we are able to supply specific information which will assist our Divisions in more accurately forecasting their operating results. . . .
We have in General Motors, I believe for the first time and dealing now with the Corporation as a whole, a definite and logically expressed viewpoint as to what the probabilities are a year hence. Of course the trend may change. It may improve, and I personally believe it will. It may decline but I hardly think that possible. If either takes place an adjustment will be made from month to month in such manner as to eliminate the extreme peaks and declines that have heretofore been characteristic of the industry and of General Motors.
Now what did this tale of internal conflict over statistics come down to? Essentially it was a matter of statistical controls versus salesmanship, which was brought to a head in 1924 by a recession in the general economy following directly upon the boom year of 1923. At that time the salesmen and the general managers were caught in the illusion of riding the wave. In our then excessively decentralized scheme I let them ride. Actually, however, this was not a mere bias in favor of the salesmen, for I had no convincing information with which to counter their intuitions. The information, as I have said, was weak and late. The information was weak because it was neither accurate nor comprehensive enough. It was arrived at by inference from dealer stocks and unfilled orders. This was good enough over a period of time, but the critical trouble was precisely the length of the period. We knew nothing about the most recent five or six weeks of our car sales, and this gap therefore was filled with the speculations of the protagonists—the statisticians with their trend lines on the one hand, and the salesmen with their optimistic intuitions on the other. I was, as I have said, in the middle without any means to judge the contending claims—not a comfortable position for a chief executive officer.
Hence the need first to limit the divisions with a forecast of sales for the model year. But since this expectation could easily be upset by actual developments in the market, there was further need for a corrective device which would enable us to retreat from (or advance beyond) the expectation, and thereby also readjust the expectation. And remember this, that in the automobile industry you cannot operate without programing and planning. It is a matter of respecting figures on the future as a guide. The essential elements are the forecast and the correction, each equally critical. For upon the forecast for the model year, made several months before that year begins, depend the plans and outlays for tooling and other preparations for the level of actual production. After the model year begins, this forecast (the index volume), although often revised, is the guiding mark for six to eight months, after which the model year is closed out by a final unalterable decision on production. The tooling, of course, is unalterably settled ahead of time. But after the model year begins, we depend upon the accuracy of current information and the swiftness with which we receive it as the control mechanism with which to make other essential corrections. These were the lessons of 1923-24, and they led to the following actions.
We worked out in 1924 and 1925 a system of statistical reports to be sent by the dealers to the divisions every ten days. The core of these reports was the information on dealers' sales of cars and trucks to consumers during ten-day periods, together with deliveries of used cars to consumers and the number of both new and used cars on hand in dealers' lots. Used-car inventories were important because if they backed up in the hands of dealers they would block the sale of new cars. With this information in hand each ten days, the divisions thereafter had an up-to-date, comprehensive picture of the situation in the field. The divisions and the headquarters staff were then able to take corrective action and make new forecasts with greater accuracy.
As a further aid to sales forecasting, we used independent data on retail sales to supplement the ten-day reports from dealers. Since the end of 1922, we had obtained from the R. L. Polk Company regular reports on new-car registrations (these reports were also available to others in the industry). The whole procedure thus put production and scheduling on a more disciplined basis and clearly defined the accountability of the operating divisions and of the corporation management for the development of production schedules.
We have endeavored always to refine and improve our techniques in the area of estimating retail demand, and the Distribution and Financial staffs have achieved some success in the field of market analysis. In 1923 the Sales Section undertook a comprehensive study of the total automobile market, based on the then prevailing concept that there was a "pyramid of demand" (formulated by Mr. Bradley in 1921). The study attempted to provide information on the size of the total market for the next few years, the market potential in the various price classes, the probable effect of price reductions on the size of the market, the competitive relationship of new and used cars, and when the so-called "saturation point" would be reached in the market. The findings of this study underestimated the future growth in the market—but its comprehensive approach to the problem represented a significant advance in market-analysis techniques in the automobile industry. The analysis of market potential by price class, in particular, was an important concept that had not previously been developed to a satisfactory degree. Also, the 1923 study clearly demonstrated the relationship between potential automobile demand and the distribution of income in the United States. With this knowledge we were able to give more meaningful recognition to the pyramid of demand in planning our sales strategy and productive capacity.
The 1923 study failed to gauge accurately the future growth of the market largely because it underestimated the effect of two important factors on new car sales. One of these was the process of continuous product improvement which stimulated consumer demand by providing increasing values for the customer's dollar. The other was the continued growth of the economy and the effect of general economic conditions on the sales of the industry in any specific year. In this latter connection Mr. Bradley later introduced into the consideration of market potential the concept that there was a definite relationship between car sales and over-all economic activity. He an
d his staff continued their interest in the question of the ups and downs of the automobile business in relation to the business cycle and saw that when business, and hence national income, was on an ascending trend, car sales increased at an even faster rate than income; and when business was on a declining trend, sales decreased at a faster rate than income. As additional statistical material on the over-all economy became available, we were able to refine our techniques and demonstrate the remarkable close correlation between car sales and personal income, a correlation that still exists today for car sales and disposable income after taxes.
To return then to the problem of production control: Once the total year's production of a division had been forecast, the division manager's problem was to spread that production out over the year to maintain his output at as even a rate as possible while still allowing for seasonal fluctuations in sales. This was not easy. The automobile business is still seasonal to some extent, and it was very much so in the early 1920s before we had improved roads, the closed car, and such devices as financial incentives to dealers to increase their trade-in allowances during slack periods.
From the standpoint of the dealer's convenience and the most economical control of finished-goods inventory, the factory should have varied its output to conform to seasonal demand. Such a practice would have reduced the risk of obsolescence and the cost of storing finished products for both dealer and manufacturer. On the other hand, absolutely level production—or the nearest to it that could be attained—was ideal from the standpoint of efficient utilization of plant and labor and from the standpoint of the employee's welfare. Since considerations of economical distribution and economical manufacture were thus diametrically opposed, planning and judgment were required to find a reasonable balance between them.
My Years With General Motors Page 17
