British Academy: The UK's National Academy for the Humanities and Social Sciences
Robert M Solow: 'What is Labour-Market Flexibility? What is it Good for?'
Copyright © The British Academy, 1998
To be printed in Proceedings of the British Academy, Volume 97
I think that the evidence just presented is very strong, but probably not conclusive. That is because the location of the Beveridge curve can not be a complete summary of the degree of labour-market rigidity at a given time and place. In particular, one of the factors underlying the generation of vacancies is the intensity of job-creation through the appearance of new firms and the expansion or transformation of old ones. It is certainly possible that job creation could be inhibited by apprehensiveness about the working of the labour market. As an example, limitations on their ability to discharge workers if sales expectations are disappointed would surely make employers less eager to create vacancies and hire workers. (Whether this mechanism is quantitatively important is another matter.) In principle this side-effect could be seen as an adverse shift in the Beveridge curve. But one must allow for the possibility that the same effect could be confounded with a movement along the Beveridge curve in the data for France and Germany.
To explore this possibility, I turn to an altogether different kind of evidence. A year or two ago the research branch of the McKinsey consulting firm (it is called the McKinsey Global Institute) conducted an extensive study of economic performance in France and Germany. A handful of academic economists (of whom I was one) participated in that study as an active advisory committee; but the part of the work that I need to emphasize here was done mainly by McKinsey's own people, each contributing experience and expertise in a particular industry. This part of the study consisted of six detailed industrial case studies, each comparing the performance of a particular industry in France and Germany with the same industry in whatever country was the world champion in productivity, usually the U.S. but Japan in one instance and the Netherlands in another. The industries studied were automobiles, housebuilding, telecommunications, retail trade, consumer banking, and computer software.
The notion of "economic performance" is not self-explanatory. In the McKinsey study a national industry got good grades for high productivity — compared with the benchmark — and for high employment. It is important that in most cases the two go together. The country with the highest productivity in any particular industry is usually the country that has created the most jobs in that industry, and this is true even in industries like telecommunications where it can not be thought that the benchmark country has drained jobs from others. No economist will be surprised at this finding; it is worth mentioning only because political opinion in Europe often seems to embrace a refined version of the old Luddite fallacy.
These remarks are just by way of introduction. The relevant and important part of the McKinsey case studies is that in each instance the team tried to think explicitly about causality. If a particular industry in France or Germany has been deficient in productivity or in job creation, why has it not done better? What factors in the institutional environment and what features of economic policy have caused the shortfall in productivity and in job creation? This part of the analysis was generally not mechanical or even quantitative; it relied on the experience and judgment of the consultants and their informants in each industry, but the advisory committee was able to insist that these judgments be backed up by concrete detail. We would not have tolerated arguments that violated economic logic. This is not the way I am used to arriving at conclusions. But I have to say that I found the process enlightening. The case I am trying to make can only benefit from finding that altogether different approaches lead in a common direction.
In the course of this and earlier studies, the McKinsey group has worked out a standard protocol for characterizing causal influences on a given industry in a given place. It is a sort of vertical checklist, beginning with the macroeconomic environment and descending through externally imposed product-market regulations and institutions, through the constraints and costs that stem from the organization of the labour and capital markets, including the incentives and limitations that arise from pressures on corporate governance, to the intensity of competition and the degree of exposure to industrial best practice, and ending with details of the production process and the organization of functions and tasks. In the end, each of these potential causal factors is classified as being important, merely secondary, or not significant at all in distinguishing the productivity and job- creation performance of this national industry from the corresponding benchmark. I want to report on these conclusions.
I will describe the conclusions case by case, before summarizing the overall message. In the automobile industry, the only mature manufacturing industry among the six, Japan is the benchmark. Comparing France and Germany with Japan, the group finds that restrictive work rules are a factor of only secondary importance, and differences in labour costs are negligible. The important causal factors come from the limited exposure of the European industry to competition, and from inferior management of operations.
In housebuilding, it is found that labour-market factors are essentially insignificant in explaining the productivity shortfall of France and Germany compared with the Netherlands. Differences in product regulations and in internal organization are far more important. Germany does have higher labour costs than the other countries, including France, and these might be a secondary factor if benchmark productivity were achieved.
In the telecommunications sector, the conclusion is that restrictive work rules are a secondary causal factor in productivity comparisons. The important causal factors are elsewhere. Independent of productivity differences, the French and German industries generate less output and employment than they might. This surely has nothing to do with the labour market and everything to do with competitive intensity, pricing, and marketing effort.
In consumer banking, the European productivity shortfall has little or nothing to do with the labour market, and much more to do with forces arising in product markets and in internal management. Nor does the labour market have a significant effect on the output of consumer-banking services, conditional on productivity. Differences in job-creation arise elsewhere.
In retailing the productivity differences are small, but the underlying facts are more complicated. An important part of the picture is that high minimum wages in France and Germany induce stores to avoid high-service, high- employment formats. Thus the European industry does not provide the large number of low-wage jobs that it does in the U.S.
In the software industry, labour-market flexibility is an insignificant factor. All the action in job-creation is in product markets and scale effects.
I have compressed a detailed and sophisticated report by McKinsey into a couple of slogans. But I have to summarize even those. There are a few contexts in which labour-market factors are a significant influence on the number and kind of jobs created. But the bottom line is clearly that these case studies strongly confirm the inadequacy (to put it mildly) of the standard litany that places the blame for low employment in Europe squarely on the inflexibilities of the labour market. It turns out that practiced observers of the industrial scene, when they come down to careful, structured evaluation, do not classify labour-market rigidities as an important causal factor in the failure of (at least) these six industries to create more jobs. One might guess — naively, I think — that the observers in question, consultants and business insiders, might normally be disposed to single out the labour market for criticism. The fact that they do not do so lends a little more weight to the conclusion.
This kind of evidence reinforces the interpretation of the Beveridge curves suggested earlier. The likelihood is that France and Germany have moved to high- unemployment regimes by sliding along their Beveridge curves, and not as victims of adverse shifts in their Beveridge curves. The implied weakness in job creation is most likely the result of excessive and anti-competitive product-market regulation, restrictive macroeconomic policy, especially monetary policy, and inadequate discipline from the capital markets. This is quite different from the conventional picture.
None of this is to deny that European labour markets are in fact highly regulated and spotted with rigidities. Even if these deviations from a pure spot market are not the main source of the long spell of high unemployment, they may still be a source of real cost to the European economy. Anything that limits the ability of firms to adapt to changed circumstances is a possible source of inefficiency. So one might prescribe an attempt to achieve greater labour-market flexibility even if this were not expected to have any durable effect on the level of unemployment. There are some qualifications, however.
Flexibility also entails some costs of its own, and they are worth mentioning because they seem to be neglected in current debates, despite their obviousness. A job provides not only a regular wage but also some security of income. It is not far- fetched to simplify by imagining each job to be characterized by its wage and by some measure of its permanence, like its expected duration. Jobs have many other relevant characteristics as well, ranging from safety to sociability, but I would like to focus on security. Employees clearly value both aspects of a job (as who should know better than tenured academics). On the other side of the labour market, both aspects of a job represent costs to employers. The provision of job security is costly precisely because it limits the employer's freedom of action in adapting to exogenous changes in the market environment. One would expect both parties to an employment contract to be willing on some terms to trade off wages against security. One can see this happening in collective bargaining, with further by-play among differently situated workers on that side of the bargain. (Frank Hahn and I have made a first stab at modeling this situation; see pages 95-101 of our Critical Essay.)
One obstacle to this line of thought is that it is hard to contract for job security per se. The relevant characteristic of a job is its ex ante expected duration, not its ex post actual duration. It is almost impossible to say whether any concrete act of "downsizing" represents (a) the occurrence of a contingency whose possibility was foreseen in the original explicit or implicit contract, or (b) an attempt on the part of the firm to shift risk to its employees in violation of the explicit or implicit contract. Whichever is true, the firm will always plead necessity and the workers will resent the ex post application of a rule, whether or not it was contemplated ex ante.
There are no doubt sophisticated ways to achieve incentive compatibility in this context. In actual fact, however, one imagines that employment contracts try to meet this difficulty by imposing inefficient limitations on the firm's flexibility (for instance by regulating outsourcing). To repeal many or all of such arrangements suddenly by legislation amounts to a shift of risk from a firm to its employees. This is not self-evidently a reasonable thing to do. Workers are almost certainly more risk-averse than the firms that employ them, since they have practically no opportunity for diversification and no possibility to claim compensation through the tax system for a capital loss on job-specific human capital.
The welfare economics of job security is a complicated subject that deserves more analysis than it has had, and surely more than I have tried to provide here. I introduced the subject only to make a point about economic policy. If pure unadulterated labour-market reform is unlikely to create a substantial increase in employment, then the main reason for doing it is the anticipated gain in productive efficiency, however large that may be. But if we respect the wage earner's desire for job security, and it seems at least as respectable as anyone's desire for fast cars or fat-free desserts, then an improvement in productive efficiency gained that way is not a Pareto-improvement. More labour-market flexibility may still be worth having — and I think it is — but then the losers have a claim in equity to some compensation. The trick is to find a form of compensation that does not cancel the initial gain in labour-market flexibility. Some forms of job-protection, like those described in the insider-outsider literature, provide job security for incumbent workers at the expense of the excumbents. That is not what we are after. But a bit of ingenuity might get somewhere. I call attention to the fact that this is a far cry from your basic European central banker's folk-theorem.
Return to title page | Publications HomePage | Academy HomePage |