Mrs. Romanis Braun, Senior Economist in the Current Studies Division of the Research Department, is a graduate of Cambridge University. She was formerly on the staff of the Oxford University Institute of Statistics; the Programmes and Plans Division, Ministry of Production, London; the Economic Directorate, Organization for Economic Cooperation and Development, Paris; and the Economic Survey Division, United Nations, New York. She has published several articles under the name Anne Romanis.
The concept of a curve relating the rate of increase in money wages to the level of unemployment was first used by A. W. Phillips in “The Relation Between Unemployment and the Rate of Change of Money Wage Rates in the United Kingdom, 1861–1957,” Economica, New Series, Vol. XXV (1958), pp. 283–99. The term “Phillips curve” is now also commonly used to cover the relation between unemployment and the rate of increase in prices in general.
See the Appendix for a description of the data.
In the early 1960s a number of detailed studies of the wage statistics were published by Professor E. H. Phelps Brown, L. A. Dicks-Mireaux, K. G. J. C. Knowles, J. R. Shepherd, and others relating to several of the problems treated here in connection with the experience of the 1950s. The effects of changes in the distribution of employment and of changes in overtime upon the published series of changes in weekly earnings have been analyzed by several writers: G. Penrice, in two articles in the Bulletin of the London & Cambridge Economic Service (which is published in The Times Review of Industry), “Earnings and Wage Rates Since 1938” (September 1952), pp. iv-vi, and “Earnings and Wage Rates, 1948–1955” (December 1955), p. xi; L. A. Dicks-Mireaux, in the same Bulletin, “Wage Earnings and Wage Rates, 1954–1957” (September 1958), pp. x–xi; L. A. Dicks-Mireaux and J. R. Shepherd, “The Wages Structure and Some Implications for Incomes Policy,” National Institute of Economic and Social Research, National Institute Economic Review (November 1962), pp. 38–48.
See the Appendix. Women regularly employed for less than 30 hours a week are not included in the calculation of average earnings. Since the proportion of women working part time has risen steadily, the earnings series increasingly understates the proportion of hours worked by women.
In the studies by Penrice and Dicks-Mireaux (cited in footnote 3), changes in the industrial distribution of employment were found to have little effect on the rise in average weekly earnings over and above the effect of changes in the demographic composition of employment.
See the Appendix.
This amounts to assuming that half the men in industry usually work one overtime shift a week. The fact that the margin between the standard workweek and average actual hours worked by men fell below four in October 1958 is not inconsistent with this treatment, because at that time some enterprises were working less than the standard workweek.
The assumptions chosen would imply that almost all the overtime work by men in the late 1950s formed part of the customary workweek (1955 being the only year of substantial overtime proper). Overtime proper would rise to about one hour a week in the early 1960s, following the first major reduction in standard hours and customary hours in 1960, and to about two hours a week in the late 1960s, following the second major round of reductions in 1965.
The narrower definition of overtime has the effect of raising the increase in wages considerably in 1954 and 1955, and somewhat in 1960 and 1964, and of reducing the increase in some years.
For a description of the nature of wage drift, see Anne Romanis, “Cost Inflation and Incomes Policy in Industrial Countries,” Staff Papers, Vol. XIV (1967), pp. 179–82.
This diagram is similar to the chart in Samuel Brittan’s article, “Wage Inflation and the Labour Market,” in the Financial Times (January 22, 1970, p. 17), except that hourly earnings are related to percentages of unemployment in June instead of April. It seems preferable to use the June figures because the level of unemployment in April is liable to be affected by special factors, such as an early or late spring and the timing of the Easter holiday.
This form of presentation was used very effectively by F. W. Paish in “The Limits of Incomes Policy,” Part I of Policy for Incomes and in Rise and Fall of Incomes Policy (Hobart Papers Nos. 29 and 47, published by the Institute of Economic Affairs, London, 1966 and 1969).
The reason for believing that substantial drift was to be expected in 1966 is discussed later.
Phillips took account of the influence of large increases in import prices on the cost of living in the original “Phillips curve” study. In “Incomes Policy: A Re-appraisal,” Economica, New Series, Vol. XXXVII (1970), R. G. Lipsey and J. M. Parkin take the rate of increase in wages to be a function of the percentage of unemployment, the movement of the general price level, and an index of union aggressiveness. The movement of the price level is assumed to reflect changes in unit costs; that is to say, the change in unit labor cost implied by the change in wages and the change in output per capita, and the change in import cost per unit of output implied by the change in import prices, assuming that the quantity of imports per unit of output is constant. No distinction is suggested between imports of primary products and imports competing with domestic industrial output. (See especially pp. 115-18.) Similar relationships were assumed in the National Board for Prices and Incomes, Third General Report, August 1967 to My 1968 (Cmnd. 3715, London, July 1968), pp. 63-67, and by David C. Smith, “Incomes Policy,” in Britain’s Economic Prospects, ed. by Richard E. Caves and Associates (The Brookings Institution, Washington, 1968), pp. 104–44.
The change in the unit value of manufactured imports should not be interpreted literally as implying the same change in the average price of manufactures competing with British goods (i.e., weighted according to shares of British output), since some manufactures that are important in imports but unimportant in domestic production are especially liable to price fluctuations. There is, however, a compensating factor in that the apparently greater stability of prices for certain products conceals effective changes in prices in the form of discounts, changes in sales services, changes in credit terms, etc.
In doing so employers may take into account the movement of foreign competitors’ prices in their export markets as well as the home market. However, the movement of import prices for a certain range of industrial products in the United Kingdom and the movement of foreign prices for those goods in the United Kingdom’s export markets will not be radically different in most years.
Since most wage negotiations take place in the winter, the expected change in earnings from October to October of the year indicated is related to the change in import prices from fourth quarter to fourth quarter of the preceding year.
It appears that, but for the impact of the Suez crisis on employment, the discrepancies would have shown a gradual transition from a higher-than-expected wage increase in 1956 to a lower-than-expected wage increase in 1959. This can be explained as follows. Most wage negotiations take place during the winter. The Suez crisis and the resulting shortage of fuel and power hampered production during the winter of 1956–57, and this led to an exceptional temporary reduction in the level of unemployment during the summer of 1957. Consequently, the increase in wage rates could be expected to be low relative to the level of unemployment that June. Unemployment was rising throughout 1958 but was declining very rapidly in the spring and summer of 1959. Thus, employment conditions at the time of wage bargaining were more favorable in 1958, and less favorable in 1959, than the unemployment figures in June would indicate. If some correction could be made for the marked effects upon unemployment of severe or mild winters, the average level of unemployment in December to March would probably provide a better index of wage bargaining conditions than the June figures.
This point is discussed later.
The increases in the import price index for foodstuffs, from second quarter to second quarter, were 6.5 in 1970, 3.3 in 1969, and 5.3 in 1968.
Because the influx of school leavers occurs between June and October, the decline in the number of new entrants to the labor force in the year ended October 1966, referred to above, here appears in the year ended June 1967.
Self-employed persons in agriculture and commerce constitute only a small proportion of the working population in the United Kingdom. Consequently, shifts from the self-employed sector to the labor force, as defined here, are less important than for most industrial countries.
In 1951 and 1966, both years of intense pressure when the number of vacancies was at its peak, the active labor force grew by only 120,000-130,000. One can assume that in these years, and also to a lesser extent in 1965, the relatively slow growth of the labor force was determined by supply rather than demand factors. In certain years (1953, 1956, and 1962), although employment increased, the growth of demand for labor was apparently lower than that of the potential supply, resulting in a fall in vacancies and also, in 1956 and 1962, a rise in unemployment, despite the increase in employment.
In these years, the rise in vacancies was half as large as the increase in employment, compared with one fifth of the increase in employment in 1960–61 (see Chart 8).
See Employment & Productivity Gazette, Vol. LXXVII (1969), p. 1009.
The rise in the number of vacancies in 1968 strongly suggests that the reduction in employment was due to the decline in the labor force, rather than vice versa, and that there was a tightening of the labor market rather than an easing as the unemployment figures would suggest.
E. H. Phelps Brown has emphasized that the Phillips curve association between changes in wage rates and the current level of unemployment arises because the latter is an indicator of the current stage of a more or less regular cyclical fluctuation in activity. See E.H. Phelps Brown, Pay and Profits (Manchester University Press, 1968), p. 13, and E.H. Phelps Brown and Margaret H. Browne, A Century of Pay (London, 1968), especially Chapter IB, Section 2, “The relation between trends of money wages and the trade cycle,” pp. 72–85.
See L. A. Dicks-Mireaux and J. C. R. Dow, “The Determinants of Wage Inflation: United Kingdom, 1946–56,” Journal of the Royal Statistical Society, Series A (General), Vol. 122, Part 2 (1959), pp. 162–64.
J.C.R. Dow, The Management of the British Economy, 1945–60 (Cambridge University Press, 1964), p. 356. The interpretation given above does not, however, agree with Dow’s view, that “this suggests that the effect of demand is to affect the climate of negotiations in a semi-political manner, and is not a purely economic response in the strict sense.”
For example, if one could translate the expected level of demand in the coming year into employment terms and, after allowing for the influence of demographic and other factors upon the supply of labor available for employment, arrive at an estimate of expected unemployment and change in unfilled vacancies, one could forecast the increase in wage rates and earnings to be expected on the basis of past experience. Alternatively, one could assess how much the level of demand would need to be curbed in order to keep the increase in wage rates in the coming year down to a certain percentage.
The longer-term Phillips curve, relating the increase in wage rates to alternative average levels of unemployment maintained for several years, is likely to be S-shaped—i.e., almost flat over a range of moderate unemployment levels, rising steeply at very low levels of unemployment when wage rates are in effect bid up by employers in different industries competing for labor, and falling steeply at some high level of unemployment, where the unions cease to be able to negotiate effective increases in money wage rates and negative wage drift causes actual wages to fall below negotiated rates. The Phillips curve observed in the short-run (SS) and the long-run curve (LL) would then compare as shown in the diagram. (The increase in wages is indicated on the vertical axis, and the percentage of unemployment on the horizontal axis.)
The inflexion points on the LL curve would almost certainly represent the outer limits of politically tolerable unemployment levels. But it is unlikely that the minimum level of unemployment tolerable in the long run would be attained, if the economy has habitually experienced fluctuations around a higher level of unemployment and the electorate regards the rate of wage and price inflation associated with temporary unemployment of x or less on the SS curve as intolerably high.
See A. W. Phillips, “Employment, Inflation and Growth,” Economica, New Series, Vol. XXIX (1962), p. 11.
For a fuller discussion of this point, see Romanis, op. cit., pp. 186–88. For a more strongly worded criticism of the Phillips hypothesis along similar lines, see John K. Gifford, “Critical Remarks on the Phillips Curve and the Phillips Hypothesis,” Weltwirtshaftliches Archiv (Band 102, 1969, Heft 1), especially pp. 79–81.
In industries where piecework is important, there is likely to be positive wage drift whenever productivity expands rapidly.
See, for example, E.H. Phelps Brown and M.H. Browne, “Earnings in Industries of the United Kingdom, 1948–59,” The Economic Journal, Vol. LXXII (1962), pp. 517–49; Dicks-Mireaux and Shepherd, op.cit.; C. Gillion, “Wage Rates, Earnings, and Wage-Drift,” National Institute of Economic and Social Research, National Institute Economic Review (November 1968), pp. 52–67.
The averages relate to full-time manual workers in manufacturing industries; construction; gas, water, and electricity; mining other than coal and quarrying; and a few services (laundries, dry cleaning, garages, shoe repairing; public administration, certain transport services but not railways, London Transport, nationalized Road Transport services before October 1966, or most dockworkers before October 1967). The data do not cover changes in earnings of white-collar workers in those industries (clerical, technical, and administrative personnel and salaried persons generally are excluded from the returns). Manual workers in agriculture, coal mining, railways, docks, distribution, hotels, catering, and other services are excluded.
See Ministry of Labour Gazette, Vol. LXXVI (1968), p. 106.
Employment & Productivity Gazette, Vol. LXXVIII (1970), p. 77, Table 126.
Indices of both these series were formerly published regularly in the Department of Employment and Productivity’s Statistics on Incomes, Prices, Employment & Production; see the final issue, dated June 1969, p. 68.
A similar, but usually minor, understatement of overtime hours may occur because some workers are on short time, working less than the normal workweek.
The differences observed in 1954 and 1955 may be less valid than those for other years because the published data on women’s earnings do not distinguish full-time from part-time women workers, and the figures used in the calculated series may not be exactly consistent with those included in the published series.
A simple numerical example will make this clear.
Assuming that overtime is paid at 50 per cent above the standard rate, average hourly earnings would be 104.54 per cent of the standard rate initially. If the amount of overtime remained unchanged at 4.2 hours, the differential would increase to 104.65, the standard rate remaining unchanged. Any increase in the standard rate would result in a
The principal differences are that coal mining is included, and services and public administration are excluded, from the Census of Production figures. The effect of this difference is probably to make women’s share of employment in the Census of Production in industries markedly lower and slightly more volatile than their share of employment in the wider grouping.
The number of men working part time is insignificant.
The exclusion causes some slight overstatement of average hourly earnings and may impart a slight upward bias to the change in such earnings when the proportion of women working part time increases markedly from year to year. Rough estimates suggest that part-time work by women probably accounted for about 5 per cent of total hours worked by women in 1953-54, and for at least 11 per cent in 1968.
The figures have been arrived at as follows. The percentage of youths in the total of male employment is assumed to be the same as in the Census of Production industries; the hours worked by men and youths are known, so the relative weights of men and youths are known. Similarly, the percentage of girls in total female employment is assumed to be the same as in the Census of Production industries; the number of women is reduced by the proportion of part-time workers in manufacturing; the hours worked by full-time women workers and by girls are known, so the relative weights of full-time women workers and of girls are known. The relative employment of men and women is not, however, known, owing to the difference in the industries covered. The proportion of male to female employment is adjusted by trial and error to produce an average hourly earnings series that shows the same percentage movement as the published figures.
In October 1968 men’s hours were 21 per cent longer on average than those of women working full time; in the late 1950s the difference was 17 per cent.