Harvard. Report on statistics and national income courses. Crum and Frickey, 1945

 

William Leonard Crum and Edward Frickey taught Harvard’s economic statistics courses in the 1930s and 1940s.  Paul Samuelson recounted his second semester (Spring 1936) as a graduate student following his previous semester’s worth of Crum: “…I was able to learn genuine modern statistics from E. B. Wilson, bypassing Edwin Frickey (who with Leonard Crum taught at Harvard courses against modern statistics!)” [On this, Roger E. Backhouse’s Vol I: Becoming Samuelson, 1915-1948, p.101].

Reading the following intradepartmental report on economic statistics courses and how to integrate national income and product accounting into the graduate curriculum that was written by a committee of two (Crum and Frickey), one discovers that even a decade after Samuelson’s experience, the proper preparation of “ink charts” was a subject that warranted faculty discussion.  Harvard Ph.D. Robert Solow later went to Columbia to play catch-up ball with respect to statistical analysis before starting his M.I.T. contract.  Harvard economics was a full generation behind the times with respect to statistical method at mid-20th century.

A 1947 Crum/Frickey  joint memo regarding preparation for taking the comprehensive field exam in statistics has been posted earlier.

______________________

6 March 1945

Report on the course offerings in Statistics, and in National Income

At the Department meeting of 13 February, 1945, the undersigned were named a committee to study course offerings and proposed offerings in Statistics and in National Income, discuss their findings with the Chairman, and report to the Department. Attached are the two reports: I, on Statistics; II, (page 10) on National Income.

W.L. Crum
Edwin Frickey

*  *  *  *  *  *  *  *  *  *  *  *  *

I. The Offering in Statistics

At the meeting of the Harvard Economics Department on 13 February, 1945. W. L. C. and E. F. sought opinions from colleagues as to additional instruction needed in statistics and as to changes needed in existing instruction. The following is in part a report of the informal discussion, in part an indication of what W. L. C. and E. F. think can advisedly be done. The present statement is preliminary; a more definitive report will be prepared, after consultation with H. H. B., for submission to the Department at a later meeting.

1. Opinion was expressed that many of our graduate students show conspicuous lack of ability to present statistical material in the form of chart or table, for example, in theses. Instruction in statistics here has for several years relied upon capacity of students to learn by emulation—they have abundant opportunity to acquaint themselves with good statistical presentation, both tabular and graphic, in our courses in statistics and in the source materials of other courses. In course 21a, some instruction is incidentally given in orderly tabulation of limited sorts, but we make no attempt to teach students to prepare ink charts. Apparently, something more is needed; and three suggestions, perhaps all to be followed together, are made:
   1. By compressing some other parts of the work, we can include a small amount of instruction on presentation in course 21a. This should help put those graduate students who are required to take that course here on the right track.
   2. For students not required to take 21a, because they have had the “equivalent” elsewhere, one possibly helpful device is to require in course 121b a written report involving presentation in tabular and graphic form. Such report could be graded if sufficient funds are available to cover the grading, and the instructors could make a moderate effort to advise particular students about defects in their reports. The reports would presumably be required of all students in 121b, whether or not they had had 21a.
   3. The Department’s specialists in statistics could advise any graduate student, whose thesis involved matters of statistical presentation, concerning such matters. When the Department acquires a general research laboratory, with a regular supervisor, the supervisor could give such advice. In the meantime, the instructors in statistics could stand ready to give such advice in appropriate cases. The undersigned emphasize that this advice should be understood to concern presentation of statistical materials: they do not feel but they should be called upon ordinarily to advise such as student about sources of statistics for his thesis, or about the methods of analyzing the statistics, or about their interpretation. They have often given advice on such matters in certain cases, and will continue to do so, but take the stand that they should not be regarded as under the obligation to give such advice to all comers. The point is that; if the candidate proposes to write a statistical thesis in any field of economics, a vital part of his job is to obtain, analyze, and interpret his data. We see no reason why faculty specialists in statistics should make an extraordinary contribution to a thesis which happens to have quantitative aspects.
2. Little emphasis appeared, in any opinions expressed, on the need for laboratory instruction in statistics in our graduate offering. Some suggestion was advanced that the “homework” type of problem task could helpfully be employed. W. L. C. and E. F. have a little faith that much could be accomplished in this way – the great advantage of the supervised laboratory is that the supervisor can get students actively started on the task and can catch and clear away difficulties as they arise. (We assume, of course, any problem work of this sort, in graduate courses, should be on an advanced – not elementary – level.) To meet this suggestion, we propose only that point A2 above be put into effect, and that the following change in present operations be considered. At present, course 121a includes two home-work problems, which stretch over several weeks, but are not graded and are not used as bases for specific advice to individual students. The proposed change is that these problems be handed in, and treated like the problem described in A2. (In these cases, as in that case, grading of the reports would be feasible if funds are available for the purpose.)

An emphatic suggestion was made that graduate students have the use of laboratory equipment, and be made welcome in the laboratory. We do not believe this can be managed with the laboratory facilities of course 21a. We note, however, that a moderate chance now exists that the University will presently provide the Department with a research laboratory in statistics, adequately equipped, and under competent supervision. If and when this is done, no difficulty will arise in making ample place for work by graduate students on any statistical tasks in which they may properly be interested. We remark that the arguments in favor of a general research laboratory in statistics are much more likely to bring conviction in responsible quarters that the argument, however strongly put, in favor of facilities merely for the occasional use of graduate students.

3. Supposing we are to give an additional half graduate course in statistics, opinions pointed toward three alternatives:
   1. A course in theory, intermediate between course 121a and Prof. Wilson’s course 122b. This does not appear a good use of our manpower, for the election in such a course would inevitably be small, especially as the mathematics prerequisites would necessarily be much more severe than those – almost nil – on which we now limp through 121a.
   2. A further course was suggested – beyond 121b and perhaps alternating with it – in topics in the application of statistics to economic fields. Economics 121b now includes a selected list of such topics, which varies moderately from year to year; but it is by no means a comprehensive coverage of all even of the major possibilities. We could readily prepare an additional half course to be called 121c of further topics in the applied fields, and many students would probably like such a course. Such a course can be described as follows:

Economics 121c will be a half-course which might be entitled Topics in Applied Economic Statistics. Economics 21a or its equivalent will be a prerequisite. Properly qualified undergraduates may, with the consent of the instructor, be admitted to economics 121c.

Economics 121c will deal with statistical problems arising in connection with the use of basic statistical data in a selected list of economic topics. (As compared with 121b this course will lay more emphasis on the basic material and less emphasis on statistical theory.)

On each topic each student will be expected to familiarize himself with the immediate and the basic sources of the main materials, through actual examination of such materials, and to present a critical appraisal of these fundamental statistics. The instructor will give a succinct historical background – an outline of the principal work which is already been done on the topic. The instructor and the class will work out together conclusions as to what are the leading issues involved, and will consider what it is that statisticians are trying to measure and what they should be trying to measure.

Such topics as the following will be included:

Consumption
Commodity prices
Cost-of-living
Employment and unemployment
Wages
Money and Banking
Production and Trade (certain phases)
Balance of international payments
Public Finance

1. 1. 1. 1. The subordinate suggestion that, in this case, basic preparation for the oral exam and also the write-off field might consist of 121a and either 121b or 121c, was advanced. A strong objection to this appears in the fact that 121b, although made up largely of topics in applied statistics, now includes – and should continue to do so – certain topics which need to be covered by every general economist (we do not here have in mind the statistical specialist) who is to have “literacy” in the field of economic statistics today. Several of the “applied” topics now in 121b include in fact fundamental matters of statistical theory needed by all economists, and not elsewhere covered in our instruction. These include, for example: the theory of index numbers, statistical deflation, secular trends in business cycles, the basic theory of measuring production and income, and at least demand and cost curves not to mention more sophisticated matters of econometrics. These essentially theoretical topics in statistics should remain part of the basic graduate year course in statistics. (This goes also for our present topic of national income: even if the Department offer a course in that subject, the course will not be taken by all students, and all should have at least the brief survey now in 121b). For the foregoing reason, we emphatically urge that 121a and 121b stand as the basic year course in the field, and that the new course 121c be regarded as an additional – but not an alternative–half course.
         2. The subordinate suggestion at 121b and 121c be given in alternate years appears to fall for the same reason given in C2a.

 

1. 3. Instead of the course described under C2 suggestion was made that we introduce a course in administrative (we use this word provisionally, for want of a better) statistics – mainly, but not exclusively, governmental statistics. We have not outlined such a course in full, but can suggest its nature by indicating that it would emphasize the problems encountered in actually doing statistical work in government or private agencies. Such topics as the preparation and use of index numbers of prices and production; the compilation and use of data on employment and the labor force; statistics of farm production and operation; the gathering of and analysis of facts concerning trade, both foreign and domestic; financial data such as are developed by the treasury, the S. E. C, the F. R. B., and private agencies; statistics used in the analysis of particular enterprises; the rapidly developing field of quality control in industry, suggest themselves for inclusion. The nature of the course can also be indicated by somewhat loose contrast with the course described under C2 above: in that course, the point of view is of the user (economist, or other analyst) of statistics, and attention is given to the origin of the statistics only in so far as it is needed to guide and inform the user. In this course, the point of view is of the maker of statistics, and attention is given to the use of the statistics only in so far as it is needed to guide the maker in his work. This course would go far toward meeting the contention that our students, while well founded in statistical theory, are not ready to handle the kind of statistical tasks which they encounter in government or other research agencies.

At the moment we are not ready to choose between the courses described under C2 and C3, the former (and obviously the latter) being understood as in addition to, and not alternative to, 121b.

 

4. No opinion was expressed concerning course 122b, and we think it should continue to be given in alternate years.

No opinion was offered concerning the content of course 121a. We have in mind some compression of one of the topics know given. This, plus the longer term under the peace-time schedule, will enable us to give more satisfactory attention to the topic of small samples.

We were commissioned to report also on national income. This is covered in a separate memorandum.

 

5. We layout now, in tentative form and subject to revision by the Department, our recommendation as to the entire offering in statistics in the early post-war years.

21a. Substantially as at present, but with the change outlined in A1.

121a. Substantially as at present, but with the change outlined in B and the change noted in D.

121b. Substantially as at present, but with the change outlined in A2.

(Courses 121a and 121b to be regarded as the core of the preparation in the field of statistics, and to be recommended to the candidates for the general oral in statistics as the most helpful unit in their preparation.)

121c. A new half course, either that described under C2 or that under C3. To be open to graduate students who have had 21a or by consent of the instructor to those who have had the equivalent of 21a, and by consent of the instructor to properly qualified undergraduates who have had 21a.

122b. Substantially as at present, and to be given in alternate years as at present.

*  *  *  *  *  *  *  *  *  *  *  *  *

II. Offering in National Income

The suggestion is made that a half course, at the graduate level, in National Income be offered. The main purpose of such a course would be to give our students an extensive factual basis for their use of national income concepts and data in a wide range of our theoretical and applied fields. The course by itself could provide only a beginning for specialization the subject of national income for its own sake, and we do not understand that the Department contemplates recognizing the field in that subject.

While the course should be concerned primarily with the facts of national income, we understand that some attention could properly be given to the interpretation of those facts into their economic and social implications. Moreover, even to handle properly the factual side, the course would need give much attention to matters of definition and concept, matters which actually stand at the root of most of the “problems” of measuring national income and its chief constituents.

The core of the course would consist of the presentation, discussion, and criticism of the existing statistical facts on the national income and its constituents. These materials would presumably be limited to the United States; although some of the critical portions of the course, dealing with concepts and the like, would necessarily make large drafts on studies in certain other countries. Emphasis would be on the problems of measurement, the effectiveness and validity of the methods used, and the appropriateness of the results obtained as answers to questions posed by the economist.

In addition to the over-all aggregate of national income, viewed in real and money terms and in its variations over time, the course would examine the chief constituents of national income. These would include:

1. Contributions to national income by various types of economic activity.
2. Contributions from various geographical regions (much less is known on this.)
3. Allocation, so far as it is known, to the several factors of production.
4. Distribution according to size of income (money income) received by individuals.
5. Distribution of income according to use: consumption expenditures of individuals (perishable, semi-durable), consumption through government, savings (by individuals, by enterprises, by government).
6. Capital formation, and its relation to savings.
7. Relation of taxes and public expenditures to the flow of income.

Your committee makes no recommendation as to the personnel to be assigned the task of conducting such a course. It does recommend: that the course be limited to graduate students, and to those advanced concentrators who receive permission from the instructor(s); that all students who take the course be required to have completed one half year course at the graduate level in economic theory and in statistics; that the course be given each year, rather than in alternate years; that the course be considered as a pro-seminar in statistics for the purpose of excuse – under our existing rules for reducing the oral examination to three fields – from the oral examination in statistics.

 

Source: Harvard University Archives. Department of Economics, Correspondence & Papers, 1902-1950. Box 23. Folder “Course Announcement 1945-46”.

Image Source: Crum and Frickey in the Harvard Class Album, 1942 and 1950.