The Economics of Welfare

Part III, Chapter XX

FIXED AND FLUCTUATING WAGE RATES IN PARTICULAR INDUSTRIES

§ 1. WHEN in practice it is decided to interfere with wages anywhere, either because they are “unfair” or for any other reason, there is at once presented a new problem. Effective interference involves either the authoritative award of a new wage rate or encouragement to employers and employed to agree upon a new wage rate. Equally whether an award or an agreement is made, it would be ridiculous that the terms laid down should be fixed for ever. The industrial situation generally, no less than the circumstances of particular trades, is in a continuous state of flux. Every award and agreement, therefore, must be restricted, either explicitly or implicitly, to a short period of time. How short the period shall be before revision can be called for depends entirely upon the practical difficulties that revision would have to face. Apart from this, it would seem, on the face of things, that, since conditions may change fundamentally at any moment, revision should be permitted whenever either side desires it. In practice, however, considerations of convenience alone would make it imperative that some minimum interval should be provided for. But there are also other considerations. Except where the relations between employers and employed are exceptionally good, it is dangerous to reopen fundamental wage controversies more frequently than can be avoided. In view of this it often happens that the governing decisions are given a currency of not less than, say, two years from the date when they are launched. For the purpose of this discussion we will suppose that that practice is adopted
generally. It does not, however, follow from this that, whenever a governing award or agreement about wages is made, the rate must, thereafter, be fixed rigidly for at least two years. For it may be possible to devise methods by which the governing award or agreement shall provide for variations in wages in response to temporary changes in demand—we may provisionally regard the conditions of supply, from the standpoint of this sort of period, as given—during the period that it covers. A choice, therefore, has to be made between a rigid arrangement and a plastic arrangement, and our investigation is not complete until the comparative effects of these have been ascertained.

In the first place, let us consider a rise in the demand for labour. If a fixed wage system is adopted, this implies that the nominal rate remains unaltered. Hence we might expect that the amount of labour provided will also be unaltered, and, consequently, that the aggregate amount of work done will be less than if the wage fluctuated. It must be remembered, however, that, though the wage per man remains the same, the wage per efficiency unit of labour is, in effect, raised for new employees. An adjustment is brought about, either by employers taking on inferior men at the wage formerly paid to good men only,
*86 or by resort to overtime at special rates.
In either event more is paid for the new labour units than for the old. It is conceivable, if the rise in demand is small, that the same addition will be made to the aggregate amount of labour employed as would have been made had the general wage rate been raised in an equal proportion. The difference is that the employer, by fixing what practically amounts to two prices as between his new labour and his old, preserves for himself a sum of money, which, under a one-price system, would have been added to the remuneration of the latter. This result is best illustrated under the method of overtime. Suppose that the normal working day is six hours at a shilling per hour—a shilling being the equivalent payment for the disenjoyment caused to the workman by the sixth hour’s work. Suppose, further, that the disenjoyment of an extra hour’s equally efficient work, to a man who has already worked six hours and received six shillings pay, is measured by fifteen pence. Then the employer can obtain seven hours’ work from that man, either by raising the general rate per hour to fifteen pence, or by paying the same as before for a six hours’ day and offering fifteen pence for one hour of “overtime.” The amount of work done is approximately the same on either plan;
*87 the only difference is that, if the former is adopted, the employer pays over to the workman an extra eighteen pence, which, under the latter, he retains for himself. This point is of some importance. As a general rule, however, particularly if the rise in demand is large, not all the extra labour that employers would like to have can be got by working overtime and taking on inferior men. Though, therefore, there will be some expansion in the dividend under the fixed wage plan, the expansion is not likely to be so big as it would be under the fluctuating plan.

In the second place, let us consider a fall in the demand for labour. If the wage rate remains at the old level, the quantity of labour which it will pay the employer to keep at work will be diminished. If the rate is lowered, it may still be diminished, but not in so high a degree. The point is well illustrated by a comparison, made shortly before the war in a Report of the British Board of Trade, between conditions here and in Germany. “Trade Union standard rates of wages do not prevail in Germany to the same extent as in Great Britain. In consequence workpeople have greater liberty in accepting work at wages lower than those at which they have previously been employed, especially in bad times. A more speedy return to employment of some kind and a consequent reduction in the percentage of trade union members unemployed results from this.”
*88 That is to say, in bad times more work is done under the plastic than under the rigid form of wage system.

From a combination of these results it follows that, over good and bad times together, a wage system fluctuating on both sides of the mean level, in accordance with temporary movements of the demand for labour, means more work and, therefore, a larger national dividend than one permanently fixed at that level. This gain arises directly out of superior adjustment between demand and supply. It is the fruit of improved organisation, and is similar to the gain produced by improved machinery. It is not retained for long as an exclusive possession of the industry which first secures it, but is distributed over the community as a whole, with the result that a new general equilibrium is established somewhat more advantageous than the old. The interest of the national dividend thus requires that the wage should not stand at the mean level for periods as long as two years, but should undergo short-period oscillations about this level, in such wise as always to make the demand for labour and the supply of it equal.

§ 3. To this conclusion there is an objection, the limits of whose validity require careful investigation. It has been urged that fluctuations in the wages of individual workpeople
tend indirectly to impair both their moral character and their economic efficiency. Thus Professor (now Sir Sydney) Chapman writes: “It may be argued that there is far more chance of a somewhat steady wage, which varies infrequently and by small amounts only, contributing to build up a suitable and well-devised standard of life, than a wage given to sudden and considerable alterations.”
*89 If this is true, it follows that the direct advantages of a wage rate fluctuating with fluctuations of demand may be more than counteracted by indirect disadvantages. For, though the national dividend will, indeed, be enhanced for the moment, it may ultimately be diminished, in a more than corresponding degree, through the injury done to the quality of some of the nation’s workers.

In examining this argument we at once observe that the term “wages” ought to be deleted in favour of the term “earnings.” It is stability of earnings that enables a well-devised standard of life to be built up, and not stability of wages. Hence, if, for the moment, questions connected with distribution between different individuals are left out of account, we may put aside, as unaffected by Sir Sydney Chapman’s argument, all occupations in which the earnings are not liable to be pushed to so low a point under a fluctuating wage as they are under a fixed wage. The occupations thus excluded comprise all those in which the elasticity of the demand for labour is greater than unity. It may, indeed, be objected that, though, in these occupations, the workpeople collectively earn more in bad times under the fluctuating system, yet the particular workpeople who are so skilful as always to command employment earn less, and that it is stability in
their earnings rather than in those of others that is of especial importance. Since, however, these superior workpeople are presumably better off than their less skilful fellow, this latter statement is highly disputable. For there can be little doubt that fluctuations in the income of a poorer man cause more suffering, and, hence, more loss of productive capacity, than fluctuations of the same size in that of a richer man of similar temperament. Hence the rejoinder fails, and
it follows that, where the elasticity of the relevant part of the demand for labour is greater than unity, Sir Sydney Chapman’s argument is of no force against a fluctuating wage.

In occupations where the demand, from a short-period point of view, is highly inelastic, the result may be different. In these conditions the total earnings of the workpeople will touch a lower level in bad times under the fluctuating system. If it were the fact that in good times adequate provision were regularly made for bad times, the consequent evil effects on capacity, and, therefore, on the national dividend, would not be great. But, as everybody knows, the ordinary workman does not “conform his expenditure and his savings to the standard wage, and regard what he sometimes gets above that standard as an insurance fund against what he will at other times get below it.”
*90 Hence it is probable that there is a considerable net evil effect on capacity. Against this, however, there has to be set the fact that, under a fixed wage, unless the demand is perfectly inelastic, the available employment in bad times will be smaller, and more workpeople are likely to be thrown out of work altogether. It is not, of course, certain that this will happen. In some industries—most notably the cotton industry—a constriction of employment is met by short time all round instead of by a reduction in the numbers of the staff, and, in others, by a sharing of work more or less in rotation. But, in general, the actual number of unemployed persons in bad months will be greater under a fixed than under a fluctuating wage system; and the capacity of totally unemployed persons is likely to suffer in a very special measure. Hence, even in occupations where the demand for labour is highly inelastic, so long as it is not absolutely inelastic, the evil effect on capacity due to a fluctuating wage is matched by another due to a fixed one. This does not, of course, show that circumstances can never arise in which fluctuating wages are, on the whole, more injurious to capacity, and so indirectly to the national dividend, than fixed wages. It does, however, throw the burden of
proof upon those who maintain, in any particular instance, that such circumstances have arisen. For the two evils noted above are so vague and indefinite that it will often be practically impossible to weigh them against one another. In the absence of special detailed information our decision must then be based upon the one fact which is known, namely, that a wage fluctuating with fluctuations in the demand for labour has the better
direct effect upon the national dividend. In general, therefore, the defence of a system of rigid wage rates examined in this section must be adjudged to have failed.

§ 4. If then it is agreed that a wage fluctuating with fluctuations in the demand for labour is desirable, it becomes necessary to determine how frequently adjustments should be made. In pure theory it would seem that adjustments should be made continually from day to day or even from moment to moment. To this sort of adjustment, however, there are insuperable practical obstacles. Time is required for the collection and arrangement of the statistics upon which the changes must be based. Considerations of book-keeping and ordinary business convenience come upon the stage, and fix a lower limit, beyond which the interval between successive adjustments must not be reduced. Of course, this limit is not always the same. In a small local industry, for example, it will probably be lower than in a great national one. But in every industry it must lie considerably above the infinitesimal level which pure theory recommends. So far as it is possible to judge from the practice of those industries in which the interval is determined, as under a sliding scale, by considerations of convenience alone, it seems as though this interval should not be less than two or three months.
*91

§ 5. Having thus satisfied ourselves that the interest of the national dividend requires a wage-system fluctuating within the normal period of agreement or award, and having determined the intervals that should elapse between successive fluctuations, we have next to determine on what plan the fluctuations themselves should be organised. It is evident that, other things being equal, the larger any fluctuation of demand is, the larger the corresponding change of wage should
be. This principle has now to be worked out in the concrete; and to that end the chief factors upon which the magnitude of demand fluctuations depends must be described. These fall into two divisions: (1) movements in the employers’ demand schedule for the commodity which the labour we are interested in is helping to produce; and (2) movements in the supply schedule of the other factors that co-operate in production with that labour. These two sets of influences will now be reviewed.

§ 6. Movements in the employers’ demand for the commodity the labour is helping to make are derived directly from movements in the public demand for the commodity. The liability to oscillation of that demand is, of course, different for different classes of commodities. The most obvious distinction is between articles which are desired for immediate personal use “for their own sake” and articles which are desired largely as means to distinction through display. The demand for articles of the former sort is likely to be the more stable, because, as Jevons suggests, people’s desire for them is generally steady for a longer period. For example, we may notice the stability of the pinafore industry: “No clothing trade (in Birmingham) suffers so little from short time.”
*92 On the other hand, commodities that are largely display articles are liable to fluctuations of desire, as opinion transfers the distinction-bearing quality from one thing to another. Thus the demand would seem to be less variable for common objects of wide consumption than for luxuries. To these special considerations should be added the more general one, that an industry which supplies a wide market made up of many independent parts is likely to enjoy a steadier demand than one which supplies a narrow market. This is merely a particular application of the broad proposition, familiar to statisticians, that “the precision of an average is proportionate to the square root of the number of terms it contains.”
*93 It is well illustrated by the interesting study of M. Lazard on
Le Chômage et la profession. Using figures from the French census of 1901, he takes the percentages of unemployment
there recorded in a number of industries, and sets them alongside of the average number of men (
moyen effectif) per establishment in the several industries, and finds, on a method of his own, an inverse correlation. Connecting large unemployment with variable demand, he explains this correlation by the relation, which he believes to subsist, between a large
moyen effectif and
l’extension des débouchés cammerciaux. “The connection between this latter phenomenon and the size of the personnel is evident. Large establishments exist only when the markets to be served are considerable. Now, a large market must also be a relatively stable market, because in it considerable decreases in the consumption of some customers have a chance of being balanced by increases in the consumption of others; and this stability, implying, as it does, stability of production, implies at the same time the absence or, at all events, a diminution of unemployment.”
*94 In like manner, he argues: “If unemployment seems to grow as we pass upward from primary towards finishing industries, this circumstance is explained by the fact that the industries at the top of the scale, being more specialised, have narrower markets. On the other hand, the industries that deal with raw products provide the material needed by numerous other industries, and, therefore, enjoy the advantages which a multitude of outlets confer.”
*95 The same principle may, of course, be invoked to explain the stability of the demand for railway transportation, as compared with the demand even for such things as
coal, sugar, or iron. In like manner, an industry, which sells largely in foreign markets as well as in the home market, is likely, other things equal, to have a more stable demand than one which sells in the home market only—except, indeed, when it is disturbed by changes in the rate of duty imposed on its products by an important foreign customer.

§ 7. When the oscillations of the public demand for any commodity are given, it is natural to suppose that the oscillations of the employers’ demand for it will be exactly equivalent to them. As a fact, however, the employers’ demand usually oscillates through the smaller distance of the two. The reason for this is the common practice of making for stock. In bad months the employer is glad to acquire and warehouse more goods than he wishes, for the moment, to sell, while in good months, because he has these goods to fall back upon, his demand for new ones rises less for them than the demand of the public whom he supplies. His demand this month is, in short, derived from the anticipated public demand of a considerably longer period, thus divesting itself to some extent of temporary oscillations. Naturally the extent to which making for stock takes place is different in different industries. The practice is less attractive to employers, the greater is the cost of carrying a unit of the commodity affected from one point of time to another. This cost depends, of course, in part, upon a circumstance affecting all commodities equally, namely, the rate of interest. For all carriage across time implies a loss of interest through holding commodities unsold. It also depends upon a number of circumstances which differ for different commodities. Of these the most obvious is the expense of storage. One important determinant of this expense is the resistance that the commodity makes to
physical wear and tear in transit across time, or, more broadly, its durability, in respect both of decay and of accidental breakage. In this quality the precious metals and hard materials, like timber, are specially favoured. As we should expect, things that are extracted from the earth are, in general, more durable than things that are grown on it. It is interesting to note that, in recent times, the development of refrigerating and other preserving processes has
rendered a number of commodities, chiefly articles of food, much more durable than they used to be. The Committee on Hops, for instance, wrote in 1908: “At the time of the previous inquiry in the year 1856, attention was called to the fact that ‘the deterioration which hops suffer when kept prevents the superabundance of one year from adequately supplying the deficiencies of another.’ The advent of cold storage has effected an adjustment between years of plethora and years of scarcity, with the resultant effect upon prices.”
*96 It may be added that such things as the direct services rendered by missionaries, doctors, teachers, train-drivers, and cab-drivers are wholly incapable, and such things as gas and electricity are in great measure incapable, of being stored. A second important determinant of the expense of storage is the resistance that the commodity makes to
psychical wear and tear in transit across time, or, more broadly, its steadiness of value. The contrast I have in view is between staple goods of steady demand—Charles Booth once cited philosophic and optical instruments as instances of this class of goods
*97—and fashion goods of unsteady demand. Clearly, there is a higher cost of carriage and less inducement towards storage for a commodity, which, next week, nobody may want, than there is for one for which a constant market is assured. Thus the turned and pressed parts of bicycles, which are much the same whatever type of frame is in vogue, are largely made for stock, whereas this is not done with completed bicycles, the form of which is liable to fashion changes. An extreme instance of unfitness for stock-making is afforded by commodities, such as ball-dresses, which every purchaser will wish to have constructed to her own special order, and which, when “readymade,” have practically no appeal for her. It is possible that things at one time customarily made to individual order may, subsequently, become more generalised; and
vice versa. Houses are sometimes built to the order of would-be private owners, and sometimes as a speculation. The boot industry has developed from an earlier stage, in which the individual order method predominated, to its present condition, in which most boots are ready-made. In certain textiles, on the other
hand, there is evidence of a movement in the opposite direction. A recent Report on unemployment in Philadelphia observes: “Twenty years ago a manufacturer made carpet or hosiery or cloth and then went out and sold
that carpet or hosiery or cloth. To-day the order comes in for a particular design, with a certain kind of yarn or silk and a certain number of threads to the inch, and the manufacturer makes that particular order. Formerly a manufacturer produced standard makes of his particular line, and simply piled up stock in his warehouse in the off-season…. To-day manufacturers make, as a rule, very little to stock and run chiefly on orders.”
*98 It is evident that every development towards the standardisation of products renders making for stock more practicable, while every development away from standardisation renders it more difficult.

§ 8. There is another point to be made in this connection. Granted that the practice of making for stock causes the oscillations of employers’ demand for any commodity to be less than the associated oscillations in the public demand, it might be thought by the hasty reader that the relation between these two oscillations can be expressed by a constant fraction, different in different industries, but the same in any one industry whatever the size of the oscillation. This is a mistake. If, in response to a given percentage elevation or depression of the public demand, the employers’ demand is elevated or depressed through a percentage five-sixths as great, it is not to be expected that the same proportion will hold good for larger changes of public demand. As a rule, making for stock is carried up to a certain point for a small inducement, and, after that, is extended only with great reluctance. Hence the public demand is apt to undergo slight oscillations without producing on the employers’ demand any appreciable effect. But, after a point is passed, further oscillations in it tend to be accompanied by further oscillations in the employers’ demand, the magnitude of which rapidly approaches towards equality with theirs. These considerations justify the provision, which is found in most sliding scales, that
alterations in the price of the commodity must exceed some definite amount before any alteration takes place in wages.
*99 They also serve as a ground for the rule, which in practice is universal, that, when wages are conjoined to prices under a sliding scale, the percentage change in wages shall be smaller than the percentage price change to which it corresponds.
*100

§ 9. Let us now turn to the second determinant of fluctuations in the demand for labour in any occupation, which was distinguished in § 5. Among the co-operant agents whose supply schedule is liable to move the most obvious are the raw materials used in the occupation. In extractive industries, such as coal-mining, these do not play any significant part; but in the majority of industries they are very important. In accordance with what was said in § 6, it is evident that, when they are obtained from a large number of independent sources, the oscillations of supply are likely to be less than when reliance has to be placed on a single source. For this reason the imposition of high protective duties upon any material, with a view to ousting foreign sellers, may be expected to bring about increased fluctuations. In addition to raw materials, the co-operant agents include the services of auxiliary labour and the services of machines. Mechanical improvements, for example, involve, in effect, a lowering of the supply schedule of the services rendered by capital invested in the occupation that is affected by them. Moreover, in some industries, Nature herself, as represented by the light and heat received from the sun, is a very important co-operant factor in production. Thus there is a high degree of seasonal variability in the demand for labour in the building trades, because the advent of frost in winter seriously interferes with brick-laying, masonry and plastering, while the shortening of the hours of daylight, by necessitating resort to artificial illumination, adds to the costs and further handicaps such work. No doubt, recent developments, such as the substitution of cement for mortar, are doing something to lessen the influence of climatic changes upon this industry,
*101 but their
influence is still very important. The same remark applies to the industry of discharging cargoes at the London Docks, which is liable to serious interruptions by frost and fog. On the other hand, indoor trades and trades little dependent on weather conditions, such as engineering and shipbuilding—dress-making is obviously not relevant here—display a relatively small amount of seasonal variability. Thus, according to a study by Sir H. Llewellyn Smith extended over some years, the mean difference in the percentage of unemployment between the best month and the worst month was 3¼ per cent in the building trades and only 1 1/3 per cent in the engineering and shipbuilding trades.
*102

§ 10. When the demand for labour in any industry fluctuates in a given measure, the appropriate consequent fluctuation in the rate of wages is not, of course, determined by the size of the demand fluctuation alone. It depends also on how elastic the demand for labour in the industry is and how elastic the supply. The more elastic the demand is, the larger the appropriate wage change will be: the more elastic the supply (from the short-period standpoint here relevant), the smaller it will be. The former of these propositions does not call for special comment. But of the application of the latter to practice something should be said. The supply of labour is elastic in occupations so situated that a small change in the wage rate offered in them suffices to divert a considerable quantity of labour between them and other occupations. In general, therefore, the following results hold good. First, when, as with the Scottish shale and coalminers,
*103 a small industry is neighbour to a kindred and very large one, the wage change corresponding to a given oscillation of demand should be less than in an isolated industry. Secondly, a set of circumstances, which would justify a given change in the wages of workmen specialised to a particular industry or locality, would justify a smaller change in those of labourers, whose lack of skill, or of managers, the generalised character of whose skill, renders them more mobile.
*104 Thirdly,
among workmen trained to a particular job, the wage fluctuations corresponding to given changes in the demand of one industry employing them should be smaller when there are other industries in which their services are required. Thus, in a boom or depression in the coal trade, the wages of mechanics employed in the mines should fluctuate less than those of hewers. Fourthly, when wages are paid by the piece, the percentage fluctuation corresponding to a given fluctuation in demand should be smaller than when they are paid by the time. For, since, under the latter system, a rise of pay has not the same effect in inducing workmen to pack more labour into an hour, the elasticity of the supply of efficiency units is lower. Fifthly, broad changes, such as the spread of information, improvements in communication, or an increased willingness on the part of workpeople to work at a distance from their homes, will tend to increase the elasticity of the labour supply and so to diminish the wage change appropriate to any given fluctuations of demand. Lastly, in industries subject to regular seasonal fluctuations, many of the workpeople will have prepared themselves for the slack periods by acquiring some form of skill for which the demand at these times is apt to increase. Hence, within the normal limits of seasonal fluctuations, the supply of labour will be fairly elastic, and, therefore, seasonal demand changes should not seriously alter wages. Thus, despite the great fluctuations in the demand for gas stokers between summer and winter wages should not, as indeed they do not, fluctuate much, because these workers are often also employed in making bricks, the demand for which expands in the summer.

§ 11. So far our discussion has been confined to the
general nature of the relation between demand changes and the wage changes that should be associated with them. We have still to inquire whether the wage fluctuation that corresponds to a given change in the demand for labour in any industry should be related to it in the same way, whatever the amount and direction of this latter change may be. The answer must be in the negative, since the supply of labour will not have the same elasticity for all amounts and both directions. We may, indeed, presume that the elasticity will not vary greatly for changes in demand fairly near to the mean. The entrance and the exit to most trades are about equally open. In the North of England mining districts, for example, “the migratory miners include a large number of skilled mechanics, who divide their time between mining and their other handicraft according as either industry offers a better chance of profit.”
*105 A moderate upward movement in demand should, therefore, in general be met by about the same percentage of wage change as an equal downward movement. But for large upward and downward movements this symmetry no longer obtains. When the demand for labour falls considerably, there is a limit beyond which the wage cannot be reduced without reducing the available amount of the labour in question to zero. This limit will be determined, for unskilled men, by the conditions of life when they are unemployed altogether and subsisting on unemployment insurance and so on, and, for skilled men, in the last resort, by what they can earn in unskilled occupations. Thus, if the demand for labour has fallen in more than a moderate degree, a further fall should be accompanied by a less than proportionate fall, and eventually by no fall, in wage.
*106 On the other hand, when the demand for labour rises considerably, the effect upon unskilled wages should be proportionate to that produced when it rises a little. For skilled labour the percentage of wage increase should be even greater. For, while the power of those already in a trade to work extra hours, and the probable presence of a floating body of
unemployed, will enable a moderate addition to the supply of labour to be made fairly easily, these resources will be ineffective when a large addition is required.
*107

§ 12. In the light of this general analysis, we have now to inquire how far it is possible to provide, in the terms of a governing award or agreement, automatic machinery for adjusting the wage rate to changes in demand occurring within the period covered by it. The best-known instrument intended for this purpose—an instrument at one time widely used in the coal industry, and still popular in the iron and steel trades
*108—is a sliding scale connecting the wage rate of an industry with the price of its finished product. Changes in the price of the product are regarded as indices of changes in the demand for the labour that produces it; and a definite scheme relating different amounts of price change to different amounts of wage change is set up. This scheme varies, of course, with the particular conditions of different industries, and is, or should be, based on the considerations adduced in the seven preceding sections of this chapter. The hope is that a skilfully devised scheme will cause the rate of wages to vary in the way in which those considerations show that it ought to vary.

§ 13. To any one looking critically at this machinery it will naturally occur that, since it makes wage changes depend, not upon contemporaneous, but upon antecedent, price changes—
e.g. price changes as recorded in the preceding quarter—the adjustment made under it must necessarily be incorrect. This, however, is not so. For the connection between the employers’ and the public’s demand for any commodity always bridges an appreciable interval. Oscillations in the employers’ demand lag behind the primary oscillations to which they correspond. It is generally only after prices have remained up for some little while that employers think seriously of expanding their business, and they hesitate in a similar manner about reducing production
when a depression sets in. Their demand for labour at any time is thus derived from the public demand for the commodity which existed at an earlier time. It follows that the supposed defect in sliding scales, that they fix future wages by past prices,
*109 is really an advantage. It is, indeed, sometimes objected that an intelligent anticipation of events before they occur is coming to influence more and more the conduct of industrial concerns; and that, so far as this tendency prevails, the adequacy of past prices as an index of future demand necessarily diminishes. “Why, then, should wages automatically fall when the leaders of industry have cast their eyes over the future, and proclaimed the need of an enlarged output and more hands? Or why should wages rise when employers see that good trade is behind, and are preparing for a period of marking time?”
*110 The answer to this argument is found in a closer analysis of the phrase “public demand.” In the present connection it signifies the demand, not of the ultimate consumers, but of those intermediate dealers who buy from the manufacturers, and whose operations are the proximate cause of changes in wholesale prices. Where such persons are present, it is extremely improbable that prices will fall when the anticipation of the leaders of industry are roseate, or rise when they are gloomy. For these anticipations will generally be shared by the dealers, and, if so, will be reflected in their present demand and, hence, in present prices. The foregoing objection is, therefore, only relevant on occasions when the forecasts of manufacturers and of dealers are at variance. Since, however, the former forecasts are, in the main, based upon the latter, these occasions will be exceedingly rare.

§ 14. There is, however, a much more serious objection to be urged against sliding scales. Changes in the price of the finished product only constitute a good index of changes in the demand for the labour that makes it, on condition that other things are equal. But in real life other things are often not equal. For these other things include the supply conditions of raw material, of the services of auxiliary workpeople,
and of the services of machinery: all of which supply conditions are liable to vary. It is evident that a given oscillations in one direction of the supply schedule of any one of these things causes the employers’ demand for the labour, whose wages we are considering, to oscillate in exactly the same manner as an equal oscillation in the opposite direction in his demand schedule for the finished commodity. Hence, in order to infer the oscillations in labour demand from those in the employers’ commodity demand, we need to subtract from the latter whatever oscillations occur in the supply schedules of the other factors of production. It is true that no provision has to be made for corrections under this head (1) when the supply schedules of the other factors are certain not to oscillate, or (2) when the part they play in the cost of the commodity is so small that their oscillations can be neglected without serious inaccuracy. It is not easy to imagine an industry in which the former of these conditions could be postulated; but the latter holds good in extractive industries, such as coal-mining, where nearly the whole cost of production is labour cost. Except in these industries the index afforded by price with changes is seriously defective. A fall in price will occur in consequence of a fall in the demand for the commodity, and also in consequence of a cheapening in the supply of the raw material. Thus there are two routes connecting changes in price with changes in labour demand. A price movement caused in one way indicates a fall; caused in another way, a rise. If, for example, the price of iron goes up on account of an increase in the public need for iron, there is a rise in the demand for iron-workers’ services; if, however, it goes up because a strike in the coal trade has rendered one of the constituents used in making it more expensive, there is a fall in this demand. It is obvious that, in the latter event, wages ought not to follow prices, but should move in the opposite direction.

§ 15. As a way of escape from this difficulty, it is sometimes proposed that the index should be, not the price of the finished commodity, but the margin between its price and that of the raw materials used in making it. “Margins” are utilised with apparent success by the officials of the Cotton Workers’ Union, who obtain them by “subtracting the price
of raw cotton (calculated from the five leading sorts) from the price of yarn (of eleven kinds) or of calico (of twenty-three kinds),”
*111 and order their wage negotiations accordingly. This index has the advantage of moving in the same way in response to a fall in the demand for the commodity and to an increase in the expense of obtaining raw material. The solution it affords is not, however, perfect. Among the contributory factors to the production of the finished article raw material is only one. The conditions of supply of auxiliary labour and of the services of machines are also liable to vary, but their variations are not reflected in any change in the “margin.” Mechanical improvements, for example, mean, in effect, a cheapening of the help rendered by machines. When such improvements are occurring, margins are liable to mislead in the same manner as, though in a less degree than, crude price statistics. Furthermore, margins, equally with prices, and because prices enter into their construction, are subject to a serious practical inconvenience. They are not likely to afford a good index in industries where the general level of elaborateness and so forth in the goods produced is liable to vary. In these industries an apparent change in price may really indicate nothing more than a change in the kind of article manufactured. This difficulty is specially likely to occur when prices are deduced from the quantities and values of exports, since there is reason to expect that the cheaper varieties of goods will gradually yield place in foreign trade to the finer and more valuable varieties.

§ 16. Yet another plan, and one which avoids some of the above difficulties, is to base a sliding scale, neither on prices nor on margins, but on “profits.” The report of the wage negotiations in the cotton trade in 1900 makes it clear that, in that trade at all events, it is exceedingly hard to arrive at a satisfactory estimate of what “representative” or aggregate profits are.
*112 What reckoning, for example, is to
be made for firms which have failed altogether and disappeared? It is interesting to observe, however, that the agreement, which settled the great coal strike of 1921, provided what is in effect a sliding scale based on profits.
For each district standard wages and standard profits (aggregating 17 per cent of the cost of the standard wages) were set up; and, every month, the balance, after the costs other than wages had been met, was to be divided between profits and wages in the proportion of 17 to 83. The coal industry is, perhaps, better adapted for this kind of arrangement than the majority of other industries would be.

§ 17. The preceding discussion, while it has revealed explicitly or implicitly many difficulties in the way of constructing an effective sliding scale, has, nevertheless, made it plain that scales adequate to take account of all changes in the demand schedule for labour are theoretically possible. In favourable circumstances, when, for example, as in coal-mines, labour is by far the most important element in the cost of production, there is no reason why a fairly close approach to the theoretical ideal should not be made. Obviously, when this can be done, an award embodying a scale is much superior to one embodying a single fixed wage for the whole period covered by it. But, while this is so, we have to recognise the very serious disability from which even a perfectly constructed scale must suffer. A given rise or fall in the price of a commodity or in the profits of an industry covered by a scale may be brought about either by a change in the real demand for the commodity or by an expansion or contraction of money or credit, which affects the general level of money prices but leaves real conditions substantially unaltered. It is plain that, if the price of coal or the profits of coal-owners go up, say 50 per cent, as a part of a general 50 per cent rise due to a purely monetary cause, the proper response in the wages of coal-miners is a 50 per cent rise. But, if the price of coal or the profits of coal-owners go up 50 per cent in consequence of an increase in the real demand for coal, the proper response, as indicated in §8, will be a rise of considerably less than 50 per cent. It follows that any scale, which provides for the right adjustment of wages
when the price of coal or the profits of coal-owners change from a cause special to coal, must provide a wrong adjustment when these change from a general monetary cause. Something might be done to remedy this defect by making wage changes depend
both on changes in the price of coal, or the profits of coal-owners,
and on changes in general prices, or, if we prefer it, in the “cost of living.”
*113 Thus there might be a scale of the type described in previous sections, referred not to absolute changes in coal prices, but to differences between the changes in coal prices and in general prices; and, superimposed upon this, there might be a second scale making wages vary in the same proportion as general prices, or, alternatively, as the cost of living. Thus, if coal rose 50 per cent while general prices rose 20 per cent, wages should rise 20 per cent
plus whatever fraction of 30 per cent (the rise peculiar to coal) the special coal scale might decree. This arrangement, however, though it would work satisfactorily when changes in general prices were due to monetary or credit causes, would not, as was shown in Chapter XVII. § 4, give a right result when these changes were due wholly or in part to such causes as the destruction of capital in war, bad harvests, general improvements in transport methods, and so on.

§ 18. There is yet another disability, even less open to remedy, from which any form of sliding scale necessarily suffers. These scales, being designed to relate changes in wage rates to changes in the demand for labour in the particular industries they cover, cannot from their nature recognise or in any way allow for changes in the supply of labour to these industries, such as might be brought about, for example, by the development or decay of some other industry employing the same type of labour. Plainly, however, wage rates ought to be adjusted to meet changes of this type equally with changes in the demand for labour. The failure of scales to do this has sometimes led to results so plainly unreasonable that it has been necessary for one party
to a scale agreement voluntarily to concede to the other terms more favourable than those which the scale decreed.

§ 19. From these considerations it appears that, though we may expect from scale agreements better adjustments than could be got from fixed wage agreements covering periods of equal length, better adjustments still will be obtained if the relations between employers and employed are good enough to allow the two-monthly or quarterly variations of the wage rate, which take place during the currency of the governing agreement or award, to be based, not exclusively upon the variations of a mechanical price or profits index, but also upon any other relevant considerations that may present themselves. There are several examples of this type of settlement. In the Scottish coal agreement of 1902, the relevant portion of which remained unchanged till 1907, it was provided “that the net average realised value of coal at the pit bank for the time being,
taken in conjunction with the state of the trade and the prospects thereof, is to be considered in fixing miners’ wages between the minimum and the maximum for the time being, and that, in current ordinary circumstances, a rise or fall of 6¼ per cent in wages on 1888 basis for each 4½d. per ton of rise or fall in the value of coal is reasonable.”
*114 In like manner, in the agreement entered into in the Federated Districts in 1906, it was provided that “alterations in the selling price of coal shall not be the sole factor for the decision of the Board, but one factor only, and either side shall be entitled to bring forward any reasons why, notwithstanding an alteration in the selling price, there should be no alteration made in the rate of wages.”
*115 Under schemes of this type reasoned and agreed action, instead of automatic action, is required every two or three months. The successful introduction of them is, therefore, only practicable in industries where cordial relations prevail between employers and employed.