Knowledge that Transforms

John Lintner, J. Keith Butters, The Lintner-Butters Analysis of the Effect of Mergers on Industrial Concentration, 1940-1947: Rejoinder to Drs. Blair and Houghton, The Review of Economics and Statistics, Vol. 33, No. 1 (Feb., 1951), pp. 67-71

rrHE Census of Manufactures, taken decen1 nially from I809 to I899 and quinquennially thereafter to I9I9, became a welcome biennial event in the two interwar decades. It was relied upon as the source of the basic data, or as the statistical rudder, of many important manufacturing measures, like the Day-Thomas and Federal Reserve Board production indices, the National Bureau of Economic Research (Mills and Fabricant) production and employment series, the WPA National Research Project production, employment, and labor productivity series (later revised and extended by the Bureau of Labor Statistics), and the Bureau of Labor Statistics monthly chain indices of employment and payrolls. But the Census had to be bypassed in I94I and in the remaining war years. The statistical consequences of the suspension of the Censusand of the wartime instability of both the product assortment and industrial classification of manufacturing establishments -were serious indeed. Many measures had to be discontinued. Some, such as the Federal Reserve Board indices, were kept current by improvisations which could hardly have gained universal approval (like the introduction of man-hour series, generally with pseudo-productivity adjustments). The Bureau of Labor Statistics sought new benchmarks for its employment and payroll indices in the data of the Old Age and Survivors Insurance and State unemployment compensation programs. The quality of labor productivity measures deteriorated at a time when minor movements were interpreted as more significant for wage adjustments than ever before. In the absence of meaningful or reliable manufacturing statistics, traffic in conventional and mislabeled measures flourished. In I948, new legislation (Public Law 67I, 8oth Congress) authorized a quinquennial Census of Manufactures and annual and other interim surveys. These provisions and others concerning quinquennial nonmanufacturing canvasses contributed to the realization of the long-discussed objective of rationally scheduling the recurrent activities of the Bureau of the Census.2 The first Census of Manufactures undertaken in accordance with the new law pertains to I947; subsequent canvasses will refer to I953 and every fifth year thereafter. The first interim surveys pertain to I949 and

M ANY attempts have been made to estimate the national income of economically backward countries. The Statistical Office of the United Nations has assembled estimates of 39 countries, roughly half of which are countries with underdeveloped economies.1 And several projects are underway for making other and better estimates. But it may be asked to what extent these efforts are justified. In making estimates of Japanese national income some years ago, the writer frequently found difficulties in applying national income and product concepts to backward areas of the economy. The problem is not so much the degree of accuracy of the estimates involved, since these estimates are usually presented to convey orders of magnitude and nothing else. Rather, the problem is: Do national income and national product exist in the backward areas? (In the sense analogous to the question: Did national states exist in Western Europe before mercantilism?) One national income economist, very much puzzled by the results of the estimates for underdeveloped economies, has posed one aspect of the problem as follows: What meaning is to be attached to per capita income of the African population, including income in kind, in the neighborhood of ?2 a year? The reviewer has never been able to grasp figures of this magnitude, which are common for countries with much of the world's (italics added).2 It should be noted that the difficulty in this case is not due to inadequacies of exchange rate conversion; the currency unit of the African colony happened to be the British pound. The ?2 per capita income is an average. This means that perhaps half of the population lived on incomes between zero and ?2, while a substantial portion lived on incomes between zero and Li throughout the year. These strange results point to the possible existence of certain difficulties in the concepts, if the possibility of inadequate estimates and exchange rates is ruled out. To avoid confusion, it is necessary at the outset to distinguish for present purposes two of the major uses of national income estimates. First, there is the use of the ratios of the various national income components to the total national income of a given year (or to other components); for example, such ratios as one state's income to that of the nation (or another state); of agricultural income to the national income (or to manufacturing); of wages to the national income (or to property incomes); of investment to national income (or to consumption). The second use is the measurement of changes over time (or inter-spatially) of either the total national income or the various composition or structural ratios mentioned in the first use. For brevity and convenience, these uses may be designated as the compositional or structural use, and the index number use, respectively. This paper deals with the problems of the first or compositional use. The problems of the second use are essentially no different from those encountered in the construction of index numbers of physical production and as such they are best discussed under that topic. One of the reasons for the extensive and peculiar usefulness of national income statistics stems from the rough equivalence of actual (or accrued) net income to net value product. The national income totals estimated for the second use do not have this quality. But to the extent that the thesis of this paper has relevance for the second use, some brief comments are made at the end of this paper.

mHE literature dealing with industry-wide or multi-employer bargaining associates this bargaining with various characteristics of industries. Factors that have been emphasized as conditioning the form of collective bargaining which has evolved include relative labor costs, nature of the product market, extent of competition, number of firms, nature of production process, structure of the union, etc. Many writers stress the significance of relatively high labor costs as a factor influencing the adoption of that type of collective bargaining.' Relatively high labor costs (as measured by the relationship of wages and salaries to total sales or value added by manufacture) are cited as a factor impelling employers to agree to multi-employer bargaining in order to equalize their competitive cost situation. The recent availability of the I947 Census of Manufactures makes it possible to check the accuracy of this widely held assumption, as well as other economic characteristics of industries which practice multi-employer bargaining. These Census data can be applied to the industries which the U. S. Bureau of Labor Statistics has classified as bargaining through associations or by groups of employees.2 The present article proposes to relate relative wage costs in terms of value added by manufacture and value of product shipped to various types of multi-employer bargaining and to company-wide bargaining. Table i shows the experience in I947 for industries with multiemployer bargaining; and Table 2 shows the relationships for industries using company-wide bargaining. Do the industries with multi-employer bargaining have higher relative wage costs than those with company-wide bargaining or than the average for all manufacturing industries where such over-all data are available? The BLS tabulation of industries with multiemployer bargaining includes both manufacturing and nonmanufacturing industries. Census data, however, have not been made available for nonmanufacturing industries since I939. Hence, for those industries, I939 data had to be used. These industries were:

IN the symposium on the controversy over monetary policy, which I was privileged to read in proof, I was impressed by the great difference in practical approach to vital problems which derives from relatively small differences in emphasis among the writers. With much of what the distinguished participants in the symposium have said I am in agreement. There is also a great deal on which I differ, but I shall not take space here to discuss my disagreements. I firmly believe, however, that an airing of conflicting views in a professional publication is a valuable way to develop an area of agreement among economists, which can then be safely trodden by legislators and administrators. I agree emphatically with Friedman that economists must think things out on their merits without regard to political feasibility. Considerations of expediency not only tend to interfere with processes of thought (of the writer as well as of the reader) but also diminish the degree of reliance that legislators and administrators are likely to place on the experts' recommendations. When economists venture into politics, they are apt to cease being good economists and yet to fall far short of being good politicians. Paul Douglas may be an exception which confirms the rule. I wonder how many of the participants would accept the following propositions to which I subscribe. i. Monetary policy has some effect on economic stability, though the magnitude of this effect and the economic price that must be paid for it are controversial. 2. Limitations on the effectiveness of monetary policy are equally applicable to fiscal policy. In fact, in broader terms, they are two aspects of the monetary approach and must both contend with powerful non-monetary factors. 3. Direct controls do not attack the causes of inflation but only moderate its effects. If promptly imposed and effectively administered, they are useful in holding the line while methods of attacking inflation at its source are being adopted and put into operation. 4. size and urgency of government outlays in a modern war and in preparedness are such as to subject a free market economy to a strain that it may not be able to support without losing its character as a market mechanism. Among the participants in the symposium there are vast differences in approach, in analysis, in emphasis, and in proposed programs of action, but I am most interested in determining what area of agreement could be established. In the light of the four propositions stated above, what would be a feasible program for monetary authorities to pursue? It is clear that in an inflationary period they must not contribute freely to a growth in the money supply. Whatever the extent of its influence, it is a factor that cannot be disregarded, and it is the direct (and principal) field of responsibility of monetary authorities. They can have only an indirect and relatively minor influence on the intensity of use of the existing supply of money and must look to other institutions and groups for most of what can be done to influence it. In regulating the supply of money monetary authorities must operate principally, almost exclusively, through bank reserves. This point, in my judgment, is not sufficiently emphasized in the symposium. rate of interest as such is not effective in controlling money creation over the short run in an inflationary situation. Shortage of reserves at the disposal of banks, on the other hand, results in much more restrictive lending practices. Lerner does not like this, and many other economists think that this argument evades the real issue, which they believe revolves around the interest rate. It is true that money is practically always procurable at some rate (though not necessarily for all applicants). Monetary policy, however, deals not with absolutes but with day-to-day realities. A bank having no 1 S. E. Harris, L. U. Chandler, M. Friedman, A. H. Hansen, A. P. Lerner, and J. Tobin, The Controversy over Monetary Policy, this REVIEW, XXXIII (I95I), PP. I79-200.

N the May I950 issue of this REVIEW, Professor Roose concludes that Federal Reserve policy had an important responsibility in precipitating the I937-38 recession, because (i) the second raise in reserve requirements in the spring of I937 pinched central reserve city banks, forcing them to unload large amounts of government securities with a resultant fall in bond prices in March and April I937, and (2) thisfall in bond, and subsequently stock, prices in turn dampened the market for new security issues.' Professor Roose argues that the peak in new security issues in June I937, the fifth month after the announcement of the increase in reserve requirements and the second month after bond prices had reached a low point, was the result of special circumstances and, in fact, hid weaknesses of the capital market at that time. There is no doubt that the 3313 per cent increase in reserve requirements, half of which took effect on March I and half on May I, forced some banks to sell government securities, although they did so at a profit. Total member-bank holdings declined by $856 million, or about seven per cent, during the first six months of I937; the effect of about one eighth of these sales on the bond market was nullified by open-market operations of Federal Reserve Banks in March and April.2 But for the Board of Governors to be at fault, it must be proved that the decline in bond prices, which was arrested by May I, was effective in diminishing new investment through the capital issues market, for bank loans continued to increase through June I937 SO that investment was not hindered through a blocking of this source.3 Professor Roose adopts three pieces of evidence to show that there was a worsening of conditions on the capital market in the spring of I937 in spite of the peak in new security issues in June, and argues that the high cost of borrowing as a result of declining bond and stock prices was the cause of this deterioration. But his evidence is based on two unusual months, December I936 and June I937; when the fourth quarter of I936 is compared with the second quarter of I937, there is no clear-cut indication that the capital market was in a shaky position. Roose's evidence is: (i) There was a substantial shift from stocks to bonds, 54 per cent of the new capital issues being comprised of bonds in December I936 and 70 per cent of new issues being bonds in June I937. But actually there was no such marked shift even into the third quarter of I937; bonds comprised 6i per cent of new issues in the fourth quarter of I936, 67 per cent in the first quarter of I937, 65 per cent in the second quarter, and 65 per cent in the third quarter.4 (2) The peak in new securities in June I937 resulted from an affluence of special, highgrade issues during that month. This is true, but there were no such high-grade issues in May. Of the II companies floating bonds for new financing that month which had a Moody's rating on existing issues, only two had a minimum rating of Aaa (Cincinnati, New Orleans and Texas Pacific Railroad, and Central Illinois Light Company), two had A, one had Baa, one had Ba, two had B, and three had some form of C rating. None of the stockissuing firms had a high rating.5 (3) The number of stock issues for new and untested companies in June I937 was considerably smaller than in December 1936. But the number of new issuances of common stock representing new financing during the period was as follows: October I936 (20), November (I4), December (32 ), January I937 (IO),

DATA relating to the size distribution of income in the United States, fragmentary as they are,2 support the thesis that income is more equally distributed today than it was during the prewar period. Although families at the very lowest income levels do not appear to have gained appreciably in the redistribution of income, there can be little doubt that the middle-income families have gained at the expense of the wealthiest families. The share of the income received by the top fifth of the nation's families and single persons decreased by I2 per cent between I935-36 and I948, whereas the share received by the middle three-fifths increased by I4 per cent during the same period (Table i). What factors account for this

John M. Blair, Harrison F. Houghton, The Lintner-Butters Analysis of the Effect of Mergers on Industrial Concentration, 1940-1947: A Reply, The Review of Economics and Statistics, Vol. 33, No. 1 (Feb., 1951), pp. 63-67