RBA Annual Conference – 1990 Discussion

1. Richard Blandy[*]

Just over a decade ago I was a discussant at an analogous conference hosted by the Reserve Bank. My task was to discuss a paper by Bob Gregory and Ron Duncan titled The Labour Market in the 1970s. In Rip van Winkle fashion I have now re-emerged to discuss that paper's successor: Bruce Chapman's The Labour Market.

I say in Rip van Winkle fashion advisedly because just as the work of my colleagues and myself at the Institute of Labour Studies in the 1970s was not very prominent in Bob and Ron's paper, so the work of my colleagues and myself at the National Institute of Labour Studies in the 1980s is not very prominent in Bruce's paper, either. Ten years' work on the Australian labour market – and not a mention! No Judy Sloan, Mark Wooden, Bob Drago, Peter Dawkins, Roy Kriegler, Sue Richardson or Dick Blandy. No Keith Norris, Charles Mulvey, Ray Petridis, Margaret Nowak, Peter Kenyon, or Peter Dawkins (again) from the West Australian Labour Market Research Centre. No Rimmer, Howard, Deery, Creedy, Niland, Plowman, Lansbury or others from the labour centres in Monash, Melbourne, New South Wales and Sydney. No Australia Reconstructed. No Enterprise Bargaining Units: A Better Way of Working. I could go on.

Bruce's paper does contain a large number of excellent references, of course. Most of these are either to international sources (often with a connection to ANU) or to national sources associated with Canberra (and especially with people at or connected with ANU).

I have made the preceding observations because the situation they capture is of some scientific interest and leads to my first proposition about Bruce's paper. Bruce's paper is a major contribution from what should be termed (in Lakatosian style) the “Canberra labour market research program”. This research program is driven by the perspectives of national government and its public servants and advisers, by macro-analysis of “big pictures”, by econometrics on big data sets, and by international macro comparisons. Its underlying theories of micro-behaviour and institutional arrangements in the labour market are very limited, and often ad hoc. Hence, Bruce concentrates mainly on such matters as the share of wages in GDP, overall employment and unemployment, overall labour force participation, overall vacancies, overall earnings, overall productivity, overall real unit labour costs, and so on. “Obviously econometric work is necessary to make [a] story strong”, he says (pp. 39–40). The only institutional arrangement examined is a macro-institution from a macro-perspective: the Accord. This is also referred to as “a public interest wage system” and “a consensual incomes policy”.

This research program has its merits and its weaknesses. But the point is that there are obviously a number of non-Canberra labour market research programs in existence in Australia represented by the work to which Bruce makes no reference. I shall mention just one: the NILS research program. The NILS research program is driven by the perspectives of organisations as producers, by econometrics on within-enterprise and between-enterprise workplace data, by data on individuals, by case-studies, and by theoretical and empirical analysis of institutional and organisational arrangements replete with “examples”, “anecdotes” and other valid sources of micro information.

May I digress momentarily on the significance of the NILS research program for understanding the Australian labour market. As Lester Thurow said in Dangerous Currents, labour market analysis has been “a Sargasso Sea of economic shipwrecks”. In Dangerous Currents, and later in The Zero-Sum Solution, Thurow put his finger on the problem: economists have not understood production. When we write Q = f(K,L,…), specify some mathematical form and run some econometrics we have not really got close to understanding production. Our idea of production is machinelike, abstract and artificial. It is very misleading. If we do not understand production properly we cannot properly understand the demand for labour, the productivity of labour, skill formation, turnover, absenteeism, management, unions, or much else that “matters” (as Bruce himself often puts it) about the labour market. Nor can we sensibly analyse the effects of different labour market institutions and arrangements (union structures, management methods, industrial relations, arbitration, and so on) on production, employment and wages. This is why “industrial relations” has been allowed to slip off into a field of its own and why labour economists often say things like “that is industrial relations” when confronted with a question of evaluating institutional arrangements.

As Australia internationalises, growth in Australian prosperity and employment will depend on our gaining a far clearer understanding of production and of the institutional arrangements in the labour market that facilitate production in individual workplaces. Only individual workplaces produce. Macro policies that do not pay regard to their likely consequences on individual workplaces are unlikely to yield preferred labour market outcomes.

In case there are any doubts that the NILS research program “matters”, I would like to note, first, that the LSE has recently established a Centre for Economic Performance which will subsume its Centre for Labour Economics. Its brief is:

“… to develop understanding of what makes people productive within organisations, relating this to individual motivation, the adoption of new technology, the supply and development of skills and the system of industrial relations

The new centre will focus largely on the performance of firms

The main factors to be investigated are the firm's use and development of trained manpower, its adoption of new technology, its work organisation and its industrial relations. The researchers will, therefore, be drawn from the full range of social sciences …”[1]

Secondly, Dertouzos, Lester and Solow from MIT also share the research agenda that NILS subscribes to. I recommend Made in America: Regaining the Productive Edge to you. When Robert Solow writes a book like this, based essentially on case studies, is it not time that Australian labour economics paid more regard to this labour market research agenda and its methods?

The Japanese regard the secret of their productive success as what they call (somewhat forbiddingly) their “humanware technology”. I recommend Haruo Shimada's paper in the Japan Labour Bulletin, 1 January 1990. The example of Japan, has, of course, been very influential in the development of the labour market research program that NILS – and a number of far more significant others – are subscribing to.

This long digression has been necessary in order to explain my second proposition about Bruce's paper without, hopefully, being offensive to an excellent economist and friend, who has written a most valuable and interesting paper on the Australian labour market: the paper does not get to grips properly with a major issue in the Australian labour market – reforming relations at work to be more productive. He has not done so, in my opinion, because this is not a major part of the Canberra labour market research program to which he subscribes. I would warmly welcome him – and anyone else who cared to join – to switch to what I have called the NILS research program.

That having been said, let me turn to the matters Bruce has addressed in his paper. In shorthand, what Bruce says is that employment went up a lot more than in other OECD countries (and more than might have been expected), while real wages, real unit labour costs and the wages share came down more than in other OECD countries (and than might have been expected). Unemployment (including disguised unemployment) fell very substantially and the labour force participation rate is found to be strongly procyclical, i.e. labour supply has responded swiftly to changes in labour demand. Much of the change in employment was in part-time jobs. Productivity improvement is also found to be procyclical (because of labour hoarding effects), although Bruce is not quite so confident about this result for the 1980s as the 1970s, if I read him correctly.

His major effort is devoted to an analysis of the impact of the Accord which he finds, by reducing real wage outcomes by 70 cents per hour (i.e. about 9 per cent, or $27 per week, or $1,400 per year), to have increased employment by 312,000 (or 4.2 per cent), reduced unemployment by 148,800 (or 2.3 per cent), the unemployment rate by 2.1 per cent, the long term unemployed by 78,000 (7 per cent as a proportion of the unemployed) and labour productivity growth by 0.25 per cent per annum (i.e. by about 1.9 per cent in 1989 – about $6–7 billion).

Time does not permit me to consider in detail all of Bruce's interesting canvas. I shall, therefore, devote my attention to Bruce's major contribution: his analysis of the Accord.

I do not think anyone doubts that reductions in real wages have a strong impact on employment in Australia. Indeed, if the Accord has been of benefit in Australia, everyone agrees that that benefit has been an expansion of employment as a result of a real wage reduction. However, it cannot be real wages per se that matter for employment, but real unit labour costs (with which real wages will be strongly correlated in the short term). If this were not so, it would be impossible to have increasing real wages and increasing employment on a secular basis. History shows that you can have those things together. What matters is productivity growth outstripping real wage growth.

In this regard, it is proper to note that the turnaround in the employment/population ratio (Figure 1) coincides with the fall of real unit labour costs below 100 in Figure 22. Real unit labour costs have continued to fall throughout the Accord period to levels not before recorded over the past generation. It would be extraordinary if such a circumstance had not been associated with rapid employment growth.

The real test for the Accord (or some other institutional arrangement) is not whether it can deliver falling real wages and increases in employment (which is not a very palatable long-term approach to improving labour market outcomes) but whether it (or some other arrangement) can deliver rising real wages and increases in employment. We do not yet know how successful an Accord might be in achieving this quinella.

Bruce runs tests of the Accord by comparing outcomes since 1983 with predictions based on pre-Accord data (starting in the late 1960s). The crucial piece of his analysis involves five models whose outcomes are presented in Figures 24–28. Eyeballing the graphs, NIF88 and Murphy don't show much difference in average rates of change of nominal wages with or without the Accord, although on average, the Accord looks to have a lower average rate of change. Russell shows a different pattern of change in real hourly wages, but the end points are very similar (1983 and 1989). Mitchell and Lewis, on the other hand, show prodigious differences in rates of nominal wage change, and real weekly wages, respectively. Which of these is to be believed?

This poses an interesting test of Bruce's proposition that “obviously econometric work is necessary to make a story strong” (to which reference was made earlier). In fact, he acknowledges that the “so-called Accord effects are not statistically significant in the NIF88 and Russell estimations” (pp. 44–45). “As far as real wage levels are concerned, [the Russell Accord coefficient] is statistically insignificant”. “In the NIF88 and the real wage level equation of Russell there is a statistically insignificant negative effect”. Bruce then concludes, on the one hand, that “strong and unambiguous conclusions concerning the role of the Accord in restraining wages are unwarranted” but, on the other (with Fred Gruen), “that the case for the Accord having restrained wages is made more easily than is the case for it having had no effect”, and, on a third hand (with Dowrick and Junankar), “that it is plausible and conservative to argue that the Australian incomes policy effects were to reduce wage inflation by three percentage points per year, and to cut real wages by 10 per cent from 1983 to 1989”! If one is going to play econometrics as a science, I believe one should stick to the strict rules. The last conclusion does not seem to follow easily from these rules, except in relation to two of the models: Mitchell and Lewis.

If the projected outcomes from Mitchell's model are to be believed, by 1989, without the Accord, nominal wages would be increasing by about 23 per cent per annum and would have been increasing by more than 15 per cent per annum since 1985. Can this be seriously entertained as a plausible outcome for the Australian economy as a piece of history? I would not like to believe that any government, Treasury, Reserve Bank or Arbitration Commission would let the projected outcomes from Mitchell's model actually eventuate. Either the values of the right hand side variables in Mitchell's model could not be those under which the Accord has operated; or else the structure estimated from data up to 1983 would have to have been altered in some other way (e.g. Thatcherite-type industrial relations legislation). The comparison with the Accord based on Mitchell's model may well suffer, therefore, from “a limitation of reduced-form modelling” (p. 45).

The projected outcomes from the Lewis model suggest (eyeballing the graphs) that, by 1989, real weekly wages would have been about 7 per cent greater (although Bruce claims a 9–10 per cent effect, (p. 44), confirmed as 10 per cent, on p. 45). With inflation at 8 per cent per annum, say, this would imply a nominal rate of wage increase on average of maybe 10 per cent per annum without the Accord – a far cry from the projected outcomes of Mitchell's model.

Therefore, interesting, skilful and ingenious as Bruce's exercise undoubtedly is in trying to track down the effects of the Accord – and personally I am willing to believe that the Accord has had the directions of effect postulated – the exercise rests on econometric quicksand right from the start of his modelling chain of effects. I do not understand how five different models giving ostensibly different, and in some cases statistically-insignificant, results can form strong econometric backing for the precise wage effects claimed for the Accord. What would the 95 per cent confidence intervals be around the projected wage effects postulated by Bruce?

Similar difficulties emerge with estimated effects “downstream” from the wage effects, for example employment. A range of employment effects, assuming the postulated wage effects to have been already decided, are presented (p. 52). The range is between 3.7 per cent and 10 per cent, i.e. between 289,000 extra jobs and 782,000 extra jobs. Bruce decides to take 310,000 jobs as his estimate. But does not his suggestion of an alternative procedure of “averaging across the wage-employment models” (p. 53) suggest that the econometric results might be regarded as so uncertain that an equiprobable weighting of them (which implies complete ignorance as to their relative reliability or quality) might be appropriate. The mean for the sample is 519,000 jobs and the standard deviation is 190,000. A 95 per cent confidence interval for the true value with 4 degrees of freedom would be approximately 519K +2.776 × 190K, i.e. − 8,000 jobs to +1,046,000 jobs. This estimate assumes that the wage effects of the Accord are known with certainty, furthermore. A confidence interval incorporating uncertainty about the wage effects of the Accord would be larger, of course.

The truth is that a very large variety of results can reasonably be held to have followed from the Accord, even holding to Bruce's model and assuming that “econometrics makes stories strong”.

When Bruce concludes that the Accord is an improvement he really means compared with Australia's experience in the decade or so preceding the Accord. This is a restricted perspective, but it is certainly damaging to the underlying institutional structure of the labour market in the 1970s. So far, the Accord has simply served to put a strait-jacket around this structure. Most countries do not have Accords. And many of them appear to have secularly better labour market outcomes than Australia does. When Bruce says that “a fruitful area of research would be in international comparisons of the role of institutional arrangements on wage outcomes and productivity research” (p. 50) he has failed to acknowledge the work already done in this area in Australia and elsewhere under research programs other than the Canberra research program. The fact is that incomes policies have not been necessary in most countries to yield superior labour market results compared with Australia over the past 20 years or so. The Australian incomes policy is an improvement in some dimensions, mainly because the underlying institutional arrangements are extremely inadequate. Our trend productivity growth is mediocre and our labour market institutions are a cause.

It is clear that major players in the labour market recognise this. “Government-union accord is now useless – Keating”; “Keating tips more flexible wage system”; “PM forecasts enterprise wage deal”; “Unions reveal strategy for flexible wages” are headlines from January in Australian newspapers. It is very probable from research under programs other than the Canberra research program that an enterprise- or workplace-focused institutional system would yield superior results; and that this requires fundamental change in underlying union structures, arbitral arrangements and management practices. The Accord has served a useful purpose as a short-term strait-jacket on potentially inferior outcomes until a more rational underlying institutional structure can be arrived at. The real issue in the Australian labour market is to build that structure so the strait-jacket can be removed.

Bruce regards the Accord as a solution to “insider-outsider” problems in the labour market and as providing a “public interest” wage system. There are other solutions to the insider-outsider “problem”, as Australia's low unemployment rate history before the 1970s demonstrates (not necessarily the best solution). Obviously, terms under which those presently employed are willing to accept additional employees are fundamental. An incomes policy is only one way, not necessarily the best, to evolve appropriate contractual terms which solve this problem. Competitive product markets combined with seniority rules and performance-based pay are likely to provide the sort of appropriate incentives to solving the problem on a micro-plane. A restructuring of union coverage which concentrates the costs of excessive wage settlements on, and only on, the members of the unions concerned, is also likely to be appropriate. The “flow-on” problem implicit in our present structure of unions and awards which underlies the usefulness of the Accord is central to the “insider-outsider” problem in Australia.

I also believe Bruce's characterisation of the Accord as a “consensual” incomes policy leaves an impression of beneficence which may not always obtain. The “nonconsensual” pilots' pay claim, for example, was resolved with great damage to the economy (probably preventing even greater damage to the economy from a “wage breakout”), by the use of civil law writs for damages, by use of the armed forces, by use of international suppliers, by use of foreign “scab” labour, by outlawing negotiations with a legitimately-registered union, by insistence on individual employment contracts, and by exclusion of “ex-employees” from the ambit of the ILO's Freedom of Association Convention. These are heavy costs in terms of freedom of association. Judges, top civil servants and parliamentarians have had legitimately-recommended pay increases halved. Persons working outside the “consensual” system have received pay increases far in excess of those within it. I'm not sure that I am all that “consensual” myself, about a fall in Professors' earnings (relative to average weekly earnings) of 30 per cent over the past 15 years. Presumably the impending major shortage of academics forecast by VPSEC should also be regarded as “consensual”; not to mention the loss of economists from Treasury and other economics areas of the public service. The bases, methods of enforcement and outcomes of “consensus” deserve greater attention than Bruce has developed in his paper.

In summary, I have focused on my reservations about Bruce's paper, and this may have obscured the fact that, in my overall assessment, I consider that Bruce has given us a very scholarly, ingenious and interesting account of the Australian labour market in the 1980s. It is an incomplete account by my lights; but this is understandable given the research program to which Bruce subscribes. I don't believe that that research program offers the answers for those labour market improvements that Australia needs. I believe that the Accord has improved outcomes in the Australian labour market in some important respects, compared with the 1970s experience. But I also believe that far more thorough institutional reconstruction is required which may not sit well with our present, Accord-style arrangements. Perhaps they will. I would like to believe so.


National Institute of Labour Studies, Flinders University. [*]

LSE News, March 1990, p. 34. [1]

2. George Fane[*]

I think that a survey of The Labour Market should cover the following topics: (1) labour supply; (2) labour demand; (3) the Accord; (4) competition and regulation in the labour market, including the role of the unions and the Arbitration system in setting minimum wages and restricting competition, and the Hancock Committee's defence of the institutional status quo; (5) the government's (abysmal) policy on training and apprenticeship; and (6) affirmative action. Chapman's paper covers (1), (2) and part of (3), but not (4), (5) or (6). I am critical of his treatment of both (1) and (2) and I think it is essential to mention topics (4) and (5) when discussing (3): the Accord document makes it clear that part of its price was a commitment by the ALP to reform industrial relations law only in ways congenial to the ACTU, such as abolishing the IRB and facilitating union mergers (Accord, Feb. 1983, pp. 15–16). One consequence of this decision not to deregulate the labour market was the preservation of a wage structure which inhibits training and apprenticeship.

(a) Participation and hidden unemployment

Chapman's Figures 6 to 10 provide a useful summary of what happened to participation, employment, the labour force and unemployment. Figures 11 to 13 (part-time and full-time employment) and 16 and 17 (long-term unemployment) are also interesting.

Chapman constructs an “estimated potential labour force” (EPL). EPL is set equal to the actual labour force in 1966, when unemployment was very low, and in 1989, when the participation rate was at an all-time high. In the years between, EPL is the linear trend shown in Figure 9. The excess of EPL over the ABS's estimate of the labour force is Chapman's “hidden unemployment”. Chapman then constructs an “estimated unemployment rate” by adding his estimate of hidden unemployment to the ABS's estimate of unemployment. Whereas unemployment as estimated by the ABS fell from a peak of 10 per cent in 1983 to 6 per cent in 1989, Chapman's estimated unemployment rate fell from 16 per cent to 6 per cent over the same period. (Table 1 and Figure 10).

I have two criticisms of this exercise. First, the emotive term “hidden unemployment” appears to have implications for policy which Chapman should either repudiate or make explicit: does he agree with me that the only useful thing the government can do to alleviate all forms of involuntary unemployment is to attempt to reduce the power of the unions and the IRC to set excessively high wages? Does he also agree that, “second best” arguments aside, there is no reason why the government should try to encourage people to shift from voluntary non-employment (leisure, house-work, care of children, study etc.) to employment? Second, his concept seems to be metaphysical: suppose I were to set hidden unemployment at 100,000 in 1966 and at 500,000 in 1989 and to connect these points by means of a relatively smooth time trend to derive an alternative series for the estimated “potential labour force”. Is there any empirical test which would show whether the resulting series for “hidden unemployment” was better or worse than Chapman's? To give an affirmative answer to this question one would have to say what hidden unemployment is supposed to measure, but Chapman has not done this. What characteristics distinguish those who are part of the “hidden unemployed” from those who are not? Without a list of these characteristics the concept is empty.

The ABS uses a number of concepts intended to measure hidden unemployment which are not empty: given information on a particular person one can tell whether he, or she, is a “discouraged job seeker”, or is “marginally attached to the labour force”. None of the ABS's measures is at all closely correlated with Chapman's measure and none of them gives such a glowing picture as his of the reduction in joblessness during the period 1983–1989: whereas Chapman's measure falls by 400,000 over this period, the corresponding fall in the ABS measure of discouraged workers is only about 30,000. The ABS estimate of discouraged workers is usually only about a quarter or a third of Chapman's measure of hidden unemployment. To get numbers of the same order of magnitude as his for the period 1979–1988 one would have to include a large fraction of the people who gave “other” reasons for not actively seeking work. It seems hard to argue that any significant fraction of these people is involuntarily unemployed since these other reasons are: “ill-health etc.”; “attending an educational institution”; “had no need to work”; “give others a chance”; “welfare payments/pension may be affected”; “moved house/holidays”; “unable to find suitable childcare etc.”; “other family considerations”; “other reasons”; and “did not know”.

(b) Employment, unemployment and wages

The stylized facts are summarized in Chapman's Figures 9 (employment) and 20 (real hourly earnings). My criticism of his survey of the links between real wages and employment is that it follows Gregory and Duncan, 1979, (GD) in emphasizing the importance of relative factor prices, whereas I think that, in empirical tests of the neoclassical and Keynesian theories of labour demand, it is much more helpful and interesting to focus on the unconditional short-run demand for labour, as a function of the real wage, the state of technology and the capital stock. Using the conditional demand for labour, which is a function of the level of output and relative factor prices, is an unreliable way of getting nowhere. Unreliable because it involves both horrendous measurement problems (GD, 1979, Figures 10 and 11, pp. 277–278) and the extremely unrealistic assumption that capital can be costlessly adjusted in the short run; and it gets nowhere, because the Keynesian labour demand function still has to be estimated to explain how the firm determines the profit-maximizing level of output.

There have been some attempts to estimate properly-specified unconditional short-run demand functions for labour using Australian data. As far as I am aware, the most extensive of these attempts is by an ANU Ph.D. student, Jim Hagan. One of Hagan's findings is that current and lagged real wages have statistically significant negative coefficients in a regression explaining aggregate employment in manufacturing industry which also includes the price of inputs relative to the price of output, the capital stock and a time trend. Hagan obtains a long-run elasticity of employment with respect to real wages, holding the capital stock constant, of about minus one. Measurement errors and identification problems would lead one to expect that this estimate may be too low.

(c) The Accord

Chapman's primary defence of the Accord is that it helped to reduce unemployment by holding down real wages. He reports that he, Dowrick and Junankar (1989) have argued that, on the basis of the econometric work reported by Chapman and Gruen (1990) (CG) “it is plausible and conservative to argue that the Australian incomes policy effects were to reduce wage inflation by 3 percentage points per year, and to cut real wages by 10 per cent between 1983 and 1989”. Since 3 per cent per year cumulated for six years comes to almost 20 per cent, CG presumably believe that, without the Accord, monetary policy would have been more expansionary by enough to add about 1.4 percentage points per year to the rate of price inflation. However, to make this argument it would seem to be necessary to estimate a monetary policy reaction function, which CG have not done.

The econometric work which Chapman reports is the outcome of five separate attempts to see whether the Accord shifted the Phillips curve. It is potentially interesting, but I think the numerical estimates are even more uncertain than is usual in econometric work. During the twenty years before 1983 the stylized facts were the opposite of those predicted by the theory: there were secular increases in both inflation and unemployment. The only way to reconcile the facts with the theory is to assume an unexplained upward trend in the natural rate of unemployment. As a result, no one has ever estimated a convincing Phillips curve, and a list of all the variables ever included in published attempts at finding one would be very long. There has been a correspondingly wide range of equation specifications.

Chapman suggests that the Accord worked by shifting the non-accelerating inflation rate of unemployment (NAIRU), but none of the five models used by CG is compatible with the existence of a NAIRU; not even with one which is shifted by changes in incomes policies. Lewis's model fits the data by using time trends and by replacing unemployment by GDP. All the rest appear to imply the existence of a long-run trade-off between inflation and unemployment. And why is it wrong to conclude from Figure 28 (Russell's model) that the Accord served a useful purpose for five years, but had finally broken down by the end of 1988?

If one cannot find a stable Phillips curve before 1983, it is hard to put much trust in estimates of the amount by which the Accord has shifted it. Nevertheless, the wage restraint since 1983 has been quite remarkable (Chapman's Figure 20), and I am ready to accept that the attempts by the ACTU leadership to restrain the wage demands of individual unions have helped to hold real wages below the levels which would otherwise have been produced by Australia's existing labour market institutions. However, there are also other partial explanations for wage restraint, including the deterioration in the terms of trade; the high level of unemployment; and the fact that events in the UK and US must have suggested to the ACTU leadership it had better start behaving if it knew what was good for it. An attempt at labour market deregulation might have produced failure (as in the UK under Heath) or, after an initial confrontation, greatly diminished union power (as in the UK under Thatcher). Labour market deregulation might well do more to raise the variance of wages and earnings than to reduce their average levels. It is excessive real wages at the bottom of the wage distribution which are important causes of unemployment and inhibitors of training and apprenticeship in the neoclassical model. Correspondingly, it would be very interesting to compare changes in the wages of the lowest paid with changes in average wages during the last seven years; the restraining powers of the Accord might look less impressive if changes in the more relevant wage measure were substituted for changes in average wages.

The claim that the great benefit of the Accord is that it has depressed real wages creates two paradoxes. First, any anti-neoclassical economists who now plug this view, would seem to be forced into the novel situation of having to concede that the ability of unions to set real wages at above market-clearing levels is a cause of unemployment. Second, the Accordions must explain why the union movement – and the political party which it finances – should be so enthusiastic about a policy of holding down real wages? Aren't union leaders paid to raise wages? And why has all this suddenly happened in the last seven years?

Chapman's answer is “insider-outsider” theory: the insiders, i.e. the unions, now care about the outsiders, i.e. the unemployed, and show their care by the wage restraint which has led to the fall in unemployment. In my view all this explains nothing. Had it wanted to, the leadership of the ACTU could always have shown its concern for the unemployed by restraining its demands at national wage cases. What explains the new-found concern of the ACTU leadership for the unemployed is first, that unions around the world have been on the defensive, and second, that in exchange for wage restraint, the legal immunities of Australian unions have been preserved, and the labour market has been made the one no-go area for the deregulatory policies which have challenged every other vested economic interest in the last decade. This is the price of the Accord and to justify it one must argue that its benefits are worth this price.

(d) Competition and regulation in the labour market

The issue of labour market reform was the subject of the major inquiry by the Hancock Committee, whose 1985 report defended the status quo.

Deregulation of the labour market would involve abolishing the de jure and de facto minimum wages which result partly from the explicit actions of the Arbitration system in setting legal minimum award wages and partly from its more covert role in providing unions with a wide enough range of legal immunities that they can set their own minimum wages. It is important to understand the source of these legal immunities, since many of the strong-arm tactics adopted by unions in strikes and picketing are technically illegal under the common law. In Britain, the Trades Disputes Act, 1906, gave unions explicit immunity from tort; Mrs. Thatcher's reforms have considerably limited this immunity. In Australia, no equivalent of the British Trades Disputes Act was ever passed; instead, the Arbitration system made unions judgement-proof in most situations: delays are normally so long as unions are unlikely to be able to pay any significant fraction of the damages for which they are technically liable. The situation is very different whenever temporary injunctions against union lawbreaking are readily available to employers: Sections 45 D and E of the Trade Practices Act, which were introduced in 1977 and 1980 respectively, have been quite successful in outlawing secondary boycotts, which were always technically illegal under the (normally unenforceable) common law.

I think the most promising way of reducing the de facto legal immunity for union bully-boy tactics is probably to amend the Trade Practices Act to extend the availability of temporary injunctions against all forms of picketing, coercion and intimidation. One could, however, make a logical, if cynical, case for preserving the status quo as follows: both equity and efficiency would be enhanced by dismantling the legal immunities which unions currently enjoy; however the unions are sufficiently powerful that any attempt to dismantle their existing privileges would probably fail – remember Ted Heath – and would therefore have costs in excess of its benefits. The best policy is therefore appeasement: we can have a piece of paper guaranteeing peace in our time, if we promise not to interfere. If Chapman's defence of the Accord – and his apparent acceptance of the labour market regulatory status quo, which is surely the price of the Accord – differs from this cynical position, I hope that he (and the other Accordions) will speak out loud and bold to tell us just what the differences are.

3. General Discussion

While there was no unanimity, there was a degree of consensus that:

  • the Accord had done something to restrain real and nominal wages and that this had allowed employment to grow faster without inflation worsening;
  • there have been costs associated with the Accord – e.g. inflexibility and narrow dispersion of wages; union immunity from some requirements of restrictive practices legislation; over-reliance on monetary policy; giving away tax cuts and more generous social service spending. But the Accord should be seen as part of a broader process (not simply as an incomes policy) which facilitated evolutionary changes to wage-setting practices and helped change attitudes of the parties;
  • many felt that the Accord was not a stable system and would collapse or evolve to something else.

Some saw micro-economic issues such as employer/employee relations, demarcation, and craft-based unionism (which were outside the scope of Chapman's paper) as being central. An emphasis on better work practices and productivity in general would allow higher employment and higher real wages. The main issue, in this view, was how to push the production frontier further out, rather than how to accommodate macro policy to the institutional constraints. The problems of incorporating long-term productivity growth into a structured wages framework like the Accord were recognised.

Chapman responded to Fane's view that Accord-type arrangements decrease the potential for productive changes in union restrictive practices. He argued that it was not inconceivable that an incomes-policy framework could be used to bring about improvements in this area (a possible example being “second-tier” tradeoffs of wage increase for a diminution in demarcation). He did concede, however, that if the incomes policy made it more difficult to achieve efficiency gains at the work-place, the case made for it on the basis of propitious macro-economic outcomes is overstated.

There was agreement that the big gains from the Accord are behind us, and that the new “wages agenda” should give more emphasis to flexibility and productivity, probably moving in the direction of more enterprise-based bargaining. The question here was whether the change should be evolutionary or revolutionary (i.e. a breakdown of the old system, as had occurred in Sweden). Any endorsement of past performance ought to be combined with some assessment of its future viability and what could be done to change or modify the current system if it is not, in fact, stable.

There was no agreement that the complexities of labour could be explained in the simple one-dimensional schema cited in Chapman's paper, with enterprise bargaining at one end of the spectrum and very centralised systems at the other end. International comparisons were made difficult by the need to put each country's wage-fixing system in its historical context – e.g. constraints of seniority were important in the US, and there were other country-specific constraints. However, many saw the key to substantial labour market reform in terms of moving the Australian system away from what they saw as an unhappy compromise between centralisation and decentralisation, towards a more enterprise-based system or, for some, towards a system where competition would play a greater role in wage setting through, e.g. reform of the Trade Practices Act.

Some were not so pessimistic about the future of the Accord and its stability, believing that unions could still get a good deal out of a wage-fixing system like this. So there was the prospect of evolutionary changes. There were already examples of the evolutionary process: to the extent that people had left the Accord framework and were negotiating outside it, there was a greater opportunity to match wage growth and productivity. This would allow greater dispersion of wages. Some of the fastest growth sectors (e.g. services) were, to a high degree, already operating on an enterprise basis.

Some felt that it was difficult to evaluate whether the Accord had been a success or not because the models were giving conflicting results and their econometrics was not very robust. For example, the Accord's flexibility and even ad hoc nature meant that it was difficult to capture its effects econometrically. While economics made an important distinction between real and nominal wages, the Accord had, at times, focused on one and, at times, on the other. A number of participants thought that the relationship between wages and employment was much more complex than the simple regressions (and the current conventional wisdom) indicated. The strong effects picked up in the econometrics may be reflecting, e.g. the degree of under-utilisation of labour within a firm.

A number of participants disagreed with Chapman's concept of hidden unemployment. Some of these differences were about measurement (see, e.g. Fane's comments), while others worried about the policy implication of the calculation, under which the authorities might be tempted to stimulate the economy in order to employ the hidden unemployed. Chapman maintained that its importance lay in its recognition of the endogenous nature of employment. When measured employment rises, the measured labour force rises.

Some disagreed with Chapman's handling of productivity. The figures are open to various interpretations, depending on the time period chosen and the concept used. Dowrick's figures (quoted by Chapman) adjust for a number of problems in the raw figures. There was a measure of agreement that Australia's productivity had been “mediocre but not abysmal”.

Productivity also depended on the stage of growth of an economy, with the spectacular German and Japan performances occurring at a time when income was still low and there was room for large improvement. There was recognition of the importance of adjusting for those sectors whose productivity was arbitrarily measured at zero. Concern was expressed not only about the change in the growth of productivity, but also the absolute level.

It was important to get the concepts of wages right: for employers, total costs (i.e. including all “on-costs”) were relevant, and for the labour supply response, the post-tax wage was the relevant one.


Australian National University. [*]