90
Chapter 11
(d) Other factors. There are also other factors at work which account for some of the deterioration in the unemployment-inflation trade-off. Of these we have been able to identify only the greater militancy of workers after the Paris events of 1968 and rising benefit replacement ratios in many countries at various times up to around 1980.
The main interest is in the policy implications. The clear message is that benefits, labour market policy, and bargaining structure play an important role in affecting the course of unemployment.
12         __________
How Can Unemployment be Reduced?
By bringing together all we have learned, we can now draw significant policy conclusions. Unemployment is not determined by an optimal process of allocation. Though it does perform a vital role in the redirection of labour, ks^ level is subject to a host of distorting influences, tending to make it higher than is „economically.,efficient. The most obvious of these distortions are
1.  the benefit system, which is subject to massive problems of moral, hazard (unless administered well), and
2.  the system of wage determination, where decentralized unions and employers both have incentives to set wages in a way that generates involuntary unemployment, and where bargained wages create a mismatch between the pattern of labour demand and supply.
Both these systems generate negative externalities. While there may be some positive search 'externalities from unemployment, it is hard to suppose that these are of the same order.
However, the negative distortions do not mean that unemployment is too high in every country. This depends on how much else the country has done to offset them.
Policy-makers have to apply a cost-benefit approach to each possible policy option open to them in their existing circumstances. They inevitably operate in the world of the
92                                    Chapter 12
second-best and most of the forms of intervention that are proposed introduce other distortions. Even so, they may improve the welfare of millions and make an economy thrive rather than limp.
We shall begin by looking at policies towards the unemployed, including policies on benefits, since the lessons here are clearest. We shall then look at the issue of bargaining and incomes policy. Then we shall discuss the role of employment subsidies.
All these kinds of policies can help a lot. We end by discussing ones that are unlikely to help—profit-sharing, work-sharing, early retirement, and reduced employment protection.
Policies for the unemployed (benefits and active manpower policy)
(i) Benefits
The unconditional payment of benefits for an indefinite period is clearly a major cause of high European unemployment. This possibie"effect of the welfare system was never intended by its founders. For example, the architect of the British welfare state, Lord Beveridge, proposed in his Report (1942) that 'unemployment benefit will continue at the same rate without means test so long as unemployment lasts, but will normally be subject to a condition of attendance at a work or training centre after a certain period . . . The normal period of unconditional unemployment benefit will be sixmonths.' He believed that, after that, 'complete idleness even on an income demoralises'.
Yet somehow this simple truth got overlooked. The unconditional welfare system worked so well in the booming 1950s and 1960s that people failed to realize that it gravely weakened the economy's self-correcting mechanism in the face of adverse shocks.
The obvious lesson is that unconditional benefits must be
How Can Unemployment be Reduced?            93
of limited duration. But then, what after they run out? One approach is nothing, as in the USA. This is a harsh route, in which some people end up on the scrap-heap. It also ignores the fact that benefits of even limited duration are subject to 'moral hazard' and liable to encourage an inefficient degree of unemployment. The other approach is active manpower policy.
(ii) Active manpower policy: the Swedish example
The classic example of an active manpower policy is the Swedish system. In the 1960s most foreign economists (including some of us) thought the Swedes had gone over the top. But the wisdom of their approach was proved by the fact that, even after two oil shocks, the Swedish unemployment rate never lingered over 3 per cent; long-term unemployment was never allowed to emerge, and unemployment"quite soon came down to under 2 per cent. So it is worth describing the essential features of their system of manpower policy.
Benefits for the unemployed run out after 14 months, but linked to this are labourf market policies to make sure that people find productive Work. These have four main ingredients.
(a)   The placement services (employment exchanges). These go into intensive operation from the moment a person becomes unemployed. Case loads are low—onľy 35 unemployed people per member of staff, compared with at least five times more in Britain. And the exchanges have excellent information on the labour market both locally and elsewhere, based on the compulsory notification of vacancies.
(b)   Retraining. Hard-to-place workers are sent on high-quality training courses—in some cases, as soon as they become unemployed. Thus, economic change is welcomed as an opportunity to provide experienced workers for the
94                                    Chapter 12
industries of the future. Generally about 1 per cent of the workforce are on courses of this kind.
(c)  Recruitment subsidies. If workers have not been placed within six months, employers recruiting them can be offered a 50 percent wage subsidy lasting six months. The numbers taken up under this scheme peaked at 0.3 per cent of workers in 1984.
(d)  Temporary public employment and the right to work. If all these measures fail, the public sector (mainly local authorities) acts as the employer of last resort. It provides work for up to six months, mostly in construction or the caring services. Provision is highly counter-cyclical, covering some 2 per cent of the workforce at the peak and under 0.5 per cent by 1988. Anyone whose benefit entitlement has run out is entitled to such work by law.
Šuch policies are expensive, and the Swedes spend nearly lper cent of national income on them. But, by keeping down unemployment, the programmes (reduce unemployment benefits, which in the EC cost 1.5 per cent of GNP compared with 0.7 per cent in Sweden. In the long term the Swedish programmes may be largely self-financing to the Exchequer. In terms of social cost-benefit analysis, they almost certainly pass the test.16 )
By any criteria, the Swedish labour market has performed extremely well (during the 1980s). The employment-population ratio,, already the highest in the!" world, has gone on rising, while it has fallen in all the main EC countries. However, in the last year or so, the system has been placed under extreme pressure. As a consequence of various adverse demand shocks and problems with the banking system, unemployment in Sweden has risen to unprecedented heights (over 5 per cent). Whether or not the Swedish system can survive intact under this pressure remains to be seen.
How Can Unemployment be Reduced?           95
(Hi) Policy towards the long-term unemployed
The lessons here are particularly obvious for the countries of Eastern Europe which have started from a position of zero unemployment. But, for a country with high unemployment, there is also the question of how to get from here to there. In high-unemployment countries around half the unemployed have been out of work for over a year. For such workers the chances of finding a job are very much less than for the short-term unemployed. And, for the same reason,, long-term unemployment is doing much less to restrain inflation than short-term unemployment. >
For these reasons, active help to the unemployed should be concentrated on the prevention.of'long-term unemployment. If we remove from unemployment a newly unemployed person, we are removing someone who on average would have left unemployment fairly soon anyway. If we remove a person at risk of long-term unemployment, we are removing someone who might otherwise continue much longer in unemployment. So the external benefit to the taxpayer from removing the second type of person is much greater than that from removing the first. Unless the costs are disproportionately greater, therefore, help should be concentrated on those at risk of long-term unemployment.
(iv) Displacement, substitution, and deadweight
But manpower policies are often criticized on two grounds. First, though they provide jobs for those helped, they may reduce employment for others (by/'displacing' labour in other firms, or/'substituting' for other workers in the same firm). This argument is often based on the notion that there is a limited demand for labour (arising fromlimited aggregate demand for products). If so, the argument is almost totally misconceived. For the aim of manpower policy is to improve the supply side of the economy, on the assumption that this isí the[main Hnn^ing factor,mot aggregate demands
96                                    Chapter 12
But there will almost certainly be some substitution and displacement for supply-side reasons. For example, if long-term unemployment is greatly reduced, there may need to (be some small increase in short-term unemployment in order to restrain wage pressure.) In principle, the magnitude of the total effect of a policy can be determined by rinding out hjowJt__affects not only the outflow rate from unemployment but also the inflow rate (the unemployment rate being determined by the ratio of the inflow rate to the outflow rate).17
The second charge against manpower policy is that it often pays money for things that would otherwise have happened ('deadweight'); for example, an employer is paid for hiring someone he would have hired anyway. Transfer expenditures of this kind are undesirable if they then have to be paid for by distorting taxes. But such elements are (' probably a smallish) issue in the overall social cost-benefit calculus of most active labour market pohcies.
(v) Pin-point targeting
The pohcies we have discussed have the major merit of being.targeted directly at the problem in hand. For example, general regional aid is often advocated because there are more unemployed in one region than another. But f/ rnuch_oX.jt .fails to relieve unemployment. In contrast,~the policies we have been discussing(aim directly at unemployment) They are Urns highly regional, but they are regional asa consequence of dealing with unemployment, rather 'than in order imperfectly to do so.) Likewise, these pohcies deal^directly with skillsjnismatch where it is identified, rather than by some more general intervention.
Policies on mismatch (employment subsidies and training)
This does however raise the issue of whether there is a case for more general action to combat the mismatch across
How Can Unemployment be Reduced?           97
regions and across skills. Suppose there are two markets (say North and South), with higher unemployment in the North. One could approach this problem by increasing labour demand in the North or by reducing labour supply there (by out-migration). But it does not make sense to attempt both; for subsidies to employment in the North will be paid for by higher taxes in the South. This policy is bound to discourage migration.
So which policy should be attempted? If better returns to migration do little to encourage migration, then (ignoring externalities) the correct^ policy is to subsidize employment in the high-unemploymenf area. But(suppose migration is very responsive, with all workers indifferent between regions at the prevailing rates of wages and employment. Then, even though there is job rationing, the classic principles of public finance apply: in the absence of externalities, there should be uniform taxation.
However, there are.....externalities. Migration into low-unemployment areas creates a demand for extra infrastructure, publicly financed. It may also damage the losing region. This argument, together with unresponsive migration behaviour, provides*' the foundation of the case for regional policy. But onetmust stress that (other distortions that reduce migration, such as housing policy,) do need urgent reform.
With skill formation, the case is somewhat different. Training suffers from the standard externality problem— that trainers are not able to trap the full social return, either because of 'poaching' or because of the tax wedge. Even though the supply response is again quite weak, this constitutes a case for favourable fiscal treatment for education and training.
As we have already said, direct policies affecting the unemployed should be judged by different criteria from those affecting the overall balance of supply and demand. This is because of the pin-point targeting which gives them their extra leverage.
98
Chapter 12
The reform of wage bargaining, and incomes policy
(i) Bargaining systems
We turn now to the other key issue: the reform of wage bargaining. Here we have discovered two main points. First, other things equal, unemployment is lower the lower is union coverage and the lower is union power in each bargain. This suggests the merits of limiting the power of individual unions. But, second, for a given union coverage and union power, unemployment is lower when employers coordinate their wage offers at an industry or national level, and likewise when unions co-ordinate their wage claims.
So there seem to be(two forms of organization that work well. One (as in the USA) has low union coverage—and preferably low union power. The other (as in Scandinavia and Austria, and to a lesser extent Germany) has high union coverage—with low union power again at the decentralized level, but with strong national unions dealing on equal terms with employers? The choice between these systems is clearly political and depends also on the size of country. But economic arguments are also relevant.
The issue is whether institutions can be created which overcome the externalities involved in decentralized wage-setting (whether by firms and/or unions). The ideal is that a consensus develops about an appropriate 'going rate' for nominal wages, which is then implemented without requiring unemployment to eliminate the wage-price and wage-wage spirals. In this context there is a role for
1.  an informed national debate about what rate makes sense;
2.  reports by respected bodies such as councils of economic advisers and research institutes;
3.  national talks between employers and unions.
If the climate of opinion is responsible, a kind of implicit contract may emerge, as often happens in Germany and
How Can Unemployment be Reduced?           99
Japan, in which other bargainers follow a pattern settlement unless they face exceptional circumstances. Everyone recognizes the need for increasing flexibility in remuneration packages. But equally, it is important that most agreements stick within an accepted range of total remuneration and do not initiate a game of competitive leapfrogging.
However, this does presuppose a fairly high degree of social discipline. If this is not forthcoming, governments naturally consider direct intervention.
(ii) Conventional incomes policies
We then need to consider the case for some form of government wage controls, such as a maximum permitted percentage rate of growth of earnings. Incomes policies of this kind have been tried at many times arid places.
To control inflation, the Roman Emperor Diocletian issued a wage decree in ad 301 and those who breached it were sentenced to death. The policy was abandoned as a failure after 13 years.
In ad 1971 the US President Nixon introduced a three-month wage-price freeze, followed by two years of less rigid controls. The policy clearly restrained inflationary pressure while it lasted, but proved unsustainable under the pressure of shortages of labour and goods (Blinder 1979).
In Britain there was a statutory incomes policy in 1972-4 and a voluntary one (initially agreed with the Trades Union Congress) in 1975-9. Both of these were abandoned, mainly because of union opposition. However, the second of the policies was at first remarkably successful, and helped to reduce inflation from 27 to 12 per cent in two years with no increase in unemployment. After the policy was abandoned inflation rose again. Some people said this was due to a 'catching-up effect'. But the best econometric evidence does not support the view that in Britain reductions of inflation achieved during incomes policies are automatically undone once the policies end (Wadhwani 1985).
100                                  Chapter 12
In France an incomes policy was introduced in 1982 and inflation fell over four years from 12 to 3 per cent. The wage norms had statutory force _in the public sector, and the'employers' federation broadly followed the same norms.
Similarly, Belgium and Italy have, since 1982, had laws prescribing the maximum degree of wage indexation in between major renegotiations, which again implies a form of wage norm. Inflation has fallen.
Australia has a long-standing system of quasi-judicial determination of basic wage rates, above which 'over-award' payments can be negotiated. However, since 1983 the national government, in 'accord' with the union movement, has set the basic norm within which the system operates.
There are two main problems with fully centralized governmental incomes policies. First, they infringe the principle of free bargaining between workers and employers. Thus, many individual groups have a strong incentive to breach the norm. This is also the case, of course, where a norm has been bargained centrally between confederations of employers and unions. But individual groups are more inclined to accept a deal to which they are at least an indirect party. For this reason,/governmental incomes policies that have the support of the confederations of employers and unions are themselves more likely to last than those that are imposed] But history suggests that nearly all such policies are eventually breached. A permanent centralized incomes policy is probably infeasible.
The-second problem is that a centralized incomes policy is inherently inflexible} It is bound to impose rigidity on the structure of relative wages. But the reallocation of labour may be much easier if relative wages rise where labour is scarce and vice versa. Without this, structural unemployment is likely to become worse, unless major efforts are made, as in Sweden, to promote movement of labour between industries and regions. Incomes policies sometimes try to incorporate committee mechanisms for adjusting relativities, but these cannot work as effectively as the market.
The result is that incomes policies of this kind have
How Can Unemployment be Reduced?         101
always been short-lived. This does not mean they have always been useless. Indeed, a( temporary incomes policy is a much better way to disinflate than having a period of high unemployments And if unemployment is above the long-run NAIRU and there is hysteresis, a temporary incomes policy is an excellent way of helping unemployment to return to the NAIRU more quickly.
(Hi) Tax-based incomes policies
One would, however, like to achieve a permanent reduction in the NAIRU itself. If this is to be through incomes policy, it must be through some mechanism other than direct controls. This leads to the proposal for tax-based incomes policy. Under this there is a norm for the growth of nominal wages, but employers are free to pay more than the norm at the cost of a substantial financial penalty. Thus/if employers need to break the norm in order to recruit labour or avoid a strike, they will do so. But all bargainers will be subject to strong disincentives to excessive settlements! Let us see more clearly how this would work.
If the free market generates excessive wage pressure, the obvious solution is to_tax^excessive jwages, This is generally the most efficient way to deal with market failure, unless direct controls have some particular advantage. One approach is through a tax on excessive wage growth; ^ another is through a progressive tax on wage levels. For p the sake of clarity, we shall discuss them in reverse order, starting with a tax on the level of wages.
Suppose that the tax is paid by firms. If a firm pays its workers a gross real wage Wi, it also has to pay the Exchequer a net tax per worker of tWi-S, where / is the tax rate and S a positive per worker subsidy. Hence the firm faces an ex ante schedule of labour cost per worker (C,) equal to
C/= Wi{\ + t)-S.
We assume that the scheme is self-financing, so that ex post in the representative firm C,- = W{.
102                                  Chapter 12
How does this reduce wage pressure and thus unemployment? The basic mechanism is that, when workers gain an extra £1 of wages, it costs the firm an extra £(1 + ř)>Thus, the firm is more willing to resist any claim, while the workers may be more anxious about making the claihi because of its greater employment effect. As on p. 39, the bargained wage Wi is that which maximizes ß\og(Wi- A)Si + logu/. Differentiating this expression with respect to Wi, the firm chooses the wage so that
ß        ß dSi ad    Ni öd _ Wi - A     Si od dWi ~ Hi dWi ~~   '
where by the envelope theorem a unit rise in labour cost (Q) reduces profit by Nt so that MildQ = -Nt. í     Since the tax sets dQIdWi = 1 + t, and ex post it is self-financing with Q - Wu the mark-up of the wage over outside opportunities (Equation 8) is now given by
Wj-A _           1-ctK
Wi     - (1 + t)(esN + cLKlft
The higher the tax rate, the less will wages tend to leapfrog each other. Thus unemployment will be lower. To be precise, since Wi= W= We, equation (8') now becomes
_________________1 - CtK
"   ~ (1 + t)(eSN + cuc/flqfl - BIW)
| so that unemployment falls as the tax rate rises. A similar
í result holds in the case of efficiency wages.
Needless to say, it makes no difference whether the tax is levied on firms or workers.18 But it must be progressive so that, when wages rise, labour cost rises faster than wages do; i.e., a part of wage cost must be tax-exempt, through a positive S. A proportional tax at rate t whose proceeds were given to the Martians would have no effect.
Of course, any tax introduces some distortions, even while it offsets others. A tax on weekly earnings could have
How Can Unemployment be Reduced?          103
severe effects on work incentives, so the tax should be levied on hourly earnings to make it as near an ideal tax as possible.
An alternative, and more understandable, policy is to tax the growth in hourly earnings. The upshot again is lower wage pressure and lower unemployment/But the tax bites less hard, because raising wages this year rather than next / costs you taxes this year but saves you taxes next yearľ) ' Thus, to achieve a given reduction in wage pressure, the tax rate has to be l/(r - n) times what is needed with a wage level tax, where r is the real discount rate and n the permitted (tax-free) growth rate of real wages.
According to many people, a tax-based incomes policy is very difficult to administer. This is not true. Provided it is part of the law of the land and the definition of earnings is as for the income tax (or the...social security tax), it can be readily collected from firms at the same time as they pay ;, j . the withholding income tax (or the social security tax). , ' There are, as with any tax, some obvious ways of trying to dodge the tax. Most of these can be dealt with. Even so, any tax has some distorting effects and so does TIP. But on balance we believe thatí! if the political will were there to implement it, in most countries', it would not only decrease        / /
unemployment but would raise social welfare. )
We should stress that the aim of all incomes policies is not to reduce real take-home pay but only to reduce wage pressure and thus the NAIRU. Indeed, since higher output yields higher tax returns, it will normally be possible to cut tax rates when employment increases.
Marginal employment subsidies
Incomes policy works by reducing the target real wage at given unemployment. An alternative way to reduce unemployment is to raise the feasible real wage in a way that does not lead to equal changes in the target real wage. A good way to do this is by a marginal .employment subsidy.
104                                  Chapter 12
Suppose that we subsidize at a rate s all employment above some fixed proportion of last period's employment. If the scheme is self-financing, it can be paid for by a tax on the rest of last year's employment. If the firm is monop-olistically competitive, it sets prices as a mark-up on marginal cost. Thus the price equation becomes
p-(we-s) = ßo-ßiu.
Thejeasible_ real wage is increased and unemployment falls. This is an attractive way of reducing inflationary pressure.
Clearly, we do not want this process to reduce post-tax profits, but post-tax profits can be restored by reductions in the profit tax financed by proportional taxes on workers. The latter, as we have seen, would not affect unemployment.
Another way to reduce the profit mark-up is by increased product-market competition (e.g. via the 1992 programme in Europe). Under wage-bargaining (though not efficiency wages), this will reduce unemployment.
We turn now to policies that are much less likely to have this effect.
Profit-sharing
There has been much recent excitement over profit-sharing, generated by the work of Martih Weitzman. Social reformers have, of course, advocated profit-sharing for many years as a way of improving productivity—and there is good evidence to support their case. But the extra productivity would not of itself .increase^ernployment. That would
require^šomě additional mechanism.                         ........
In his original book, Weitzman (1984) proposed such a mechanism in the context of a labour market which in equilibrium is market-clearing. He argued that under the wage system firms equate the real wage to the marginal revenue product of labour^ In the short period the real wage is fixed,; so that any fall in marginal revenue product will
How Can Unemployment be Reduced?         105
reduce employment. Under profit-sharing, by contrast, competition for labour ensures that in general equilibrium the marginal revenue product equals the total remuneration of labour (i.e. the base wage plus the profit share). Hence the marginal revenue product exceeds the base wage. But, once the base wage has been set,(ex post firms seek to employ labour up to where the marginal revenue product equals the base wage. So there is permanent excess demand for labour./1 Ä fall in labour demand (marginal revenue product) will not cause a fall in empioýměnť:::-měrelý in profits. Weitzman claimed that this explained the Japanese miracle.
But there are problems with the theory and with the Japanese evidence. The theory assumed that, after the package of base wage and profit share had been determined, workers would stand idly by while the firm tried to employ people, thus eroding the profit share of the existing workers. It seems unlikely that workers would react in this way, rather than trying to bargain also about employment. Second, the theory assumed long-run market-clearing in the labour market. In many countries this may not be the right model, and it is easy to f show that in both an efficiency wage model and our bargaining model profit-sharing would have no effect on the NAIRU.
So what about Japan? Why exactly is unemployment in Japan so low and so stable? It is not because of any of the mechanisms Weitzman describes, as the following facts make clear.
1.  Output is not stable. It fluctuates (about its trend) more than in most countries. It responds to monetary shocks exactly as elsewhere.
2.  Nominal prices are affected by cost factors and not simply by demand.
3.  Excess demand for labour, as reported by firms, is rather lower than in other countries.
4.  It does not appear that employment is determined in the short run by base wages.
106                                  Chapter 12
Having said all this, the basic fact remains that employment in Japan is stable compared with elsewhere. What happens is roughly as follows. Only 40 per cent of Japanese workers are in the organized sector (where bonuses are paid); another 30 per cent are employees in the small-firm sector, and 30 per cent are family workers. When output fluctuates, employment in the formal sector fluctuates quite a lot.' But employment in small firms varies much less. This ^ is quite simply because.thelElöalMlity_ofjay per^worker is f so high in the market-clearing small-firm sector, while it is much less high in large bonus-paying firms. Thus, Japan's stable employment record is due mainly to the wage flexi-bility in the small-firm sector.
This flexibility has the result that in Japan the total labour input (hours x employment, HN) fluctuates less than in other countries. On top of this, the Japanese value their human capital highly, so they use hours per worker (H) as a shock-absorber more than most other countries, further dampening fluctuations in employment (JV). In addition, the labour force (L) shrinks in recession, as 'secondary' female workers shrink back home. This makes unemployment (L-N) even more stable than employment (compared with other countries).
So what does the Japanese evidence tell us about profit-sharing? Since the intermediate predictions of Weitzman's theory are not borne out, one can say either that his theory is wrong or that Japan is not a case of profit-sharing. There is a lot to the latter view. While some 25 per cent of remuneration is in bonuses, much of this is indeed a fixed element. Thus, we must probably conclude that Japan provides little evidence either for or against profit-sharing.
Even so, we would support profit-sharing as a device to improve productivity and industrial relations. As a device to reduce unemployment, it is no straightforward panacea.
How Can Unemployment be Reduced?          107
Early retirement and work-sharing
Two policies that are very popular would be clearly counter-productive.i The first is the policy of reducing the labour force by early retirement. As we have shown, it is the unemployment rate, that„gguiĽbxate_s_tiie labour market. If the size of the labour force is reduced, the equilibrium unemployment rate is unaffected. Employment has to fall to eliminate the wage pressure that would otherwise emerge, as the supply of labour becomes more scarce relative to the demand. Thus, early retirement does not make jobs available for people..who. .would otherwise be unemployed: it just reduces employment.
This is what reasonable theory says, and it is "confirmed by the evidence. In time-series regressions wage behaviour is affected not only (positively) by employment but also (negatively) by the size of the labour force—and the absolute elasticities are of roughly equal size. Moreover, if one compares countries, it is striking that early retirement has expanded most in countries with the greatest increase in unemployment. In Japaá, where retirement behaviour is unchanged, unemployment has not risen at all. This suggests that early retirement is not an effective means of reducing unemployment» It is an excellent way of making a country poorer.)
The other policy with the same effect is work-sharing. The idea here is to redistribute the available work to more people. Butoncel again, the availablejjvorkis nqt_a given— that is the 'lump-of-output fallacy'. The equilibrium unemployment rate is independent of hours of work. Thus, if hours a^ejreduc^.MLd_emplo.yineiitjrises_for_ a while, wage pressure will soon increase and jth^amount of work available will have to be reduced. Employment will revert to its former level.
We can understand why this happens by taking our wage-setting models, inserting hours, and making Wi represent  the   hourly  wage.   The  conclusion  from  theory  is
108                                   Chapter 12
confirmed by time-series regressions, which show that hours do not affect the relation between wage pressure and the unemployment rate) Again, the countries that have reduced hours most have been those where unemployment has grown most. In Japan and the USA, with fairly steady unemployment, hours have fallen little. Thus, cuts in hours provide a poor antidote to~uhémployment. But they cer--1 tainly provide a lower standard of living.
Employment protection legislation
Another policy of importance relates to the laws of employment protection. In most European countries the law requires that, when a worker is laid off, he be given advance notice, severance pay (redundancy payments), and a satisfacto^reasojj (äš opposed to 'unfair dismissal').
Laws of this kind must reduce the rate of flow into unemployment (S/N), and this effect tends to reduce unemployment. But such laws also discourage hiring, since firms are less willing to hire workers* whom they cannot later dismiss without incurring costs. Thus, the outflow rate from unemployment (H/U) is also reduced. In equilibrium the outflow from unemployment has to equal the inflow (H = S). Thus
U_S/N N    H/U'
Unemployment is reduced by employment protection only if the inflow rate (S/N) falls more than the outflow rate (H/U). Studies on this matter yield ambiguous results.
On balance, employment protection laws are probably bad for employment, since they strengthen insider power and encourage the payment of efficiency wages to motivate workers who cannot be threatened with dismissal. But there are equity arguments in their favour, and the evidence on adverse employment effects is not strong enough to warrant ä total abandonment of the practice.
How Can Unemployment be Reduced?          109
Demand management
On the supply side, we have seen that there exist policies which would really help (policies towards the unemployed, towards 5'wage"" cTěteřmination, and? marginal employment subsidies)—and some others which would probably not. What about the demand side?        ,
This is not mainly a book about the demand side of the economy, or about 'stabilization policy'. We would make only two comments.
First, whenjiysteresis is strong, it is very important to avoid big rises in unemployment. If inflation is too high, it is better to eliminate it by small amounts of extra unemployment over a longish time period than by anything approaching 'cold turkey' (see Annex 5). Had this been understood in 1980, some of the disaster of European unemployment could have been avoided.
Second, once inflation is at an acceptable level, it is normally desirable to avoid disturbances to nominal demand, by holding the growth of nominal demand stable. But šHoůíd^nflationary supply shocks happen, the case for some accommodation ť through faster nominal demand growth is stronger the higher the degree of hysteresis. Stabilization policy should be highly sensitive to the supply Z mechanisms of the economy.