OECD Economic Studies No . 21. Winter 1993 - PDF document

OECD Economic Studies No . 21. Winter 1993 CENTRALISATION WAGE BARGAINING II shape hypothesis and the effect competitive pressures basic theoretical to the basic model Ill Centralisation by IV empirical research Vl centralisation and Centralisation and parameters ~~ extended version this paper OECD Economics Department Working Papers. The author the Institute Economic Studies. Stockholm Jsrgen Elmeskov. Steinar Holden. Martin. Peter Skedinger. Peter Sturrn. Michael Wallerstein Domeij research assistance Helena Matheou secretarial assistance The wage countries exhibit great differences. One extreme States and Canada with firms. The Austria have traditionally represented the highly centralised cedures. Other Germany, France, Italy, Belgium, are between polar cases with wage setting mainly at the industry level. more decentralised bargaining countries during the last decade has reduced these differences but on the whole they a/., the macroeconomic various bargaining claimed that centralised low unemployment (early references are McCallum, 1985; Bean a/., 1986; Newell large amount few years, which has observation that both centralised and decentralised wagesystems seem to (e.g., 1987; Calmfors Driffill, One lesson appears that the wage formation may be far more complex acknowledged. The is to in the field and discuss the possibilities policy conclusions. The focus the theoretical the empirical the article follows. Section wage increases specific groups on the are internalised II restraining power decentralisation. Section 111 highlights different dimen IV bargaining. Section a critical assessment central- isation and wage behaviour. Finally, tentative policy conclusions are drawn I. INTERNALISATION EFFECTS AND CENTRALISATION aggregate real analysed within different theoretical frameworks. The simplest are union wagesetting and efficiencywage models (see Layard a/., 1991, for an exposition to be set unilaterally by unions, which trade the benefits from real- wage increase for employed union determined unilaterally by employers, who weigh disadvantages from the increase benefits in effort from reduced turnlabour. The most western European conditions appear to be a/., 1991), employers negotiate about revenues from production These latter models involve between the a higher wage for the employees the associated profit decrease for the employer. revenue sharing also depends upon the alternatives that two bargaining parties face in facing unions and is natural define the the literature is increases for one group have negative others that will be internalised under cooperative behaviour, create incentives for Such internalisation effects will work labour market, both other employees employers are to be adversely affected by wage least seven types negative wage externalities have been treated in the stems from fact that man.3 increase is mainly another man’s price increase” {Layard a/., 132). Every wage increase in the economy contributes general price level and therefore real disposable income and capital owners directly affected wage bargain 1987; Calmfors and Driffill, a/., 1991 ; Moene et a/., arises when wage increases in one part economy cause price rises for the products used material inputs other firms. consequence will lower output elsewhere and also lower employment material inputs (Wallerstein, 1990; a/., 1991). iii) fiscal externality imposed on the increases in sector causes unemployment there ciated cost for unemployment benefits have to be through higher or lower government expenditure in general. similar externality arises because output in wage increase there means tax base, paid for others (Blanchard Calmfors and wage increases unemployment externality The reason is that sector makes workers everywhere the econjobs (Hoel, 1991; Jackman, investment externality also arise under decentralised present employees given firm will quit before they can reap wages from present investment in union incentives current wage restraint promote such (Rmdseth, Hoel, 1991). the welfare individual workers depends negatively the economy will have that reduces the others (Oswald, 1984; Calmfors, 1993a; Udden-Jondal, 1993). vii) there may be the effort employees depends upon their relative which case one wage increase will reduce effort rise in may make difficult for keep labour (Hoel, a/., 1993; Rerdseth, The above externalities have been used explain why centralised bargainlower aggregate real wage hence, according to standard negatively demand schedule, higher employment. The simple idea is that interunion and interemployer cooperation imply that the one part the economy will ered. Thus marginal benefit wage increase is reduced andlor marginal cost consequence, the incentives for realwage restraint be strengthened bargaining is centralised. another advantage centralised bargaining may the aggregate economic development and the behaviour various wage setters. These aspects have recently been stressed Bhaskar (1990). his analysis wage setter strong incentives under decentralised conditions to match the envy or wage considerations. This makes (in Bhaskar’s terminology there a given natural rate employment but instead natural range, i.e., equilibrium employment Which equilibrium is realised will depend others. For instance, everyone anticipates wages to will indeed be that expectations are thus turn out rational.’ It follows that centralised bargaining may be one economy ends up in “good” instead “bad” equilibria. Somewhat paradoxically, adaptive expectations with respect US. facilitate coordination in the sense that they remove the ambiguity introduced a range of self-fulfilling rational-expectations equilibria. But such a lack of syn- react swiftly much discussed oil price the disinflation (e.g., Bruno and Sachs, 1985; Jackman, 1986; Layard eta/., a possiabove discussion that synchronisation deals with to time is not necessarily favourable in II. externality arguments a monotonic negative relationship the extent centralisation and the depicted by the 1. that the central- isation may also affect that competitive under decentralised These issues have been (1 987) Calmfors and Drif- fill (1988), and later on Rowthorn (1992), and Driffill der Ploeg basic theoretical hypothesis is that decen- systems are likely to produce realwage moderation and high this is explained by in the the restraint imposed by market forces. real wage and with intermediate the form industry level, because then both market forces and internalisation effects is then shaped relationand the real as illustrated by curve II a humpshaped relationship as follows (Calmfors and Driffill, eta/., 1993; Calmfors, 1993a, b, Consider the incentives raising the (the nominal consumer price index) in a bargaining model, where unions care both about the real and employment, and employers real value profits (nominal profits by the consumer across all firms producing similar products, the possibility rises on via an the relative will hold back the rise real product price). Since determined by the real given increase real consumption Figure 1. and the aggregate real wage II Nation a II: a closed the case of an open the case of when domestic and foreign are perfect This weakens the incentives for wage moderation on the union side. Similarly, the price increases reduce the profit employer side Compare then the industry bargaining with the two pletely decentralised at the level individual firm wage setting the national Assume furthermore that there is market, that i.e., no foreign the consumerprice one 1. firm is that it cannot raise its relative wage is increased the firm in the sector. Hence price offset the employment and profit decreases from wage increase bargaining. Exactly the instead consider completely centralised the national level. The simple is that, uniformly throughout no relative can change sectors are perfectly symmetrical). Therefore a given increase the same employment and profit effects under complete decentralisation. same way, the stated cases. It and hence employment higher industry bargaining. to the The conclusion that bargaining the individual firm national level gives wage outcome should, however, regarded only complete analysis, additional considerations need to be i) possibility that individual firms raise their relative prices under decentralised wage setting was ruled out above perfect competition product market. In the realistic case monopolistic competition, such a price increase will indeed a wage the individual tives for wage restraint under decentralised wage setting a/., the leftThe conclusion industry bargaining highest aggregate real wage however, still likely since interemployer coindustries will larger increases market power industries (and hence larger increase raise relative output prices uniform wage increases across co-operat- ing firms). The reason products within are closer substitutes than the aggregate outputs different industries eta/., The result that relative output price can change wages rise uniformly assumption that In an open economy with foreign there still exists wedge between real consumption the consumption basket Since the relative price between domestic foreign goods will rise consumption wages in all firms rise, losses from wage increases are dampened also under centralised baran open economy. The more open the economy the higher the rightthe humpshaped relationship in Figure compared to the lefthand end point (as indicated only externality is central- ised wage setting in an open economy must result in higher real wage determination can be decentralised perfectly competitive a/., 1991; Calmfors, 1993a, Driffill and van der there is monopolistic competition between firms (see i) above), it is less clear extremes produce wage restraint in an open economy. The conclusion that intermediate extent centralisation (wage setting industry level) may produce macroeconomic outcomes both very high and assumption that substantial amount competitive pressures for wage market power less, the international competition, foreign competitors has been demonstrated (1994), lower the competition there 111 has been 11). domestic and foreign perfect substitutes, it would domestic firms a small open to raise their relative they cooperate (Calmfors and Driffill, a/., In this relationship between centralisation and gate real into a line like IV in Figure 1 (Calmfors, 1993b). iv) It that the consumer price takes the other externalities II strong arguments result in lower the firm 1993a, b). The extent of internalisa- tion under industry to be too have a substantial effect on above has not considered received so for example, Gottfries Blanchard and Summers, this literature that temporary shocks reducof employment are likely wages permanently, unemployment risks insiders become smaller which weakens for wage moderation. often argued that these weaker under bargaining, because members and not disenfranchised the same extent as 1993). that this true, the incentives wage restraint ought to be decentralised bargaining after adverse labour- It is not evident though since it ought to be the same insiders that co central- decide union wage policy decen- tralised bargaining.3 vi, decentralisation may affect the employers and employees because the alternative a labourmarket conflict (the fallback positions are changed. the two the labour market. On hand, counions within individual employers one strike costs the extent that workers in these gain market during a conflict On the employer cooperation within industry may prevent the profit demand spillovers during labour 1993a). Layard et al. (1991) and Layard and Nickell (1 992) have reduce workemployer cothe national affects the made in, for instance, Sweden that the employer side on using economy- union strikes confined to key groups the union side (Elvander, 1988; Calmfors and Fors- lund, 1990). vii) issue in affect wage setting decen- between different production units does not apply in case (unless sector production is opened in some cases others). One way reasoning focuses instead on fiscal as a factor. Attempts create incentives for wage moderation the public through cash limits spending may under centralised bargaining, because large can then general cutservices that are regarded as politically intolerable. One would expect it to fiscal discipline the parties wage agreements cannot expect to central government cash the central bargaining, since it then in a significant electorate (Calmfors 1985, 1988). likely relationship the aggregate real - - - - Firm-level IndUStN National bargaining bargaining 169 bargaining conclusion is that should expect complete centralisation (interunion and interemployer cothe national level) produce more realwage restraint than complete decentralisation (non-coopera- therefore expect relationship like Moreover, it appears important distinguish between different sectors tradeable sectors with strong foreign competidifferences in terms wage outcomes Industry bargaining is wages than firmlevel bargaining in the private nonThis presupposes, however, that there is a reasonable domestic competitors. When this case, policies strengthening competition decentralised bargaining deliver wage restraint. Especially under industry bargaining, increased international integration markets for goods and services powerful tool increasing pressures for wage m~deration.~ However, this will succeed only union and interwage bargaining within sectors (for instance, integrated Europe) is a~oided.~ Centralisation and decentralisation results above are clearcut, the ambiguous. The reason is actual wage bargaining systems seldom their theoretical counterparts. There is neither complete centralisation with nationwide determination nor complete decentralisation with independent bargaining the level individual firms. does not therefore the extremes. Instead actual various hybrid forms. the literature exaggerates the actual amount centralisation in, for the Nordic countries (e.g., Bruno and Sachs, Newell and Symons, 1987; Layard, 1990; Jackman, 1990; Layard et a/., these economies there has never been complete centralisation sense that been determined in same bargain. Instead, the traditional systems are better wages have been determined units (Elvander, 1988; Calmfors and Nymoen, 1990). Complete centralisation does not appear alternative even and homogeneous countries like the Nordic ones Netherlands; it the question larger economies holding large coalitions together and handling all the information necessary (Freeman, a/., likely to make important difference for wage moderation whether the effects wage increases are internalised completely partly.6 few, very large competing wageearner organisations may also reinforce union concerns over relative wages as more decentralised that the incentives for ened (Calmfors, 1986; Uddenequally important consideration concerns actual degree in decentralised systems. substantial amount emerge also when wage formally decentralised. Since wage part the economy have strong influence there will this case be strong incentives informal consultations between bargaining units on each tendencies have been to be the strongest side, where there usually exist strong personal ties between different firms, may be profitable them not to jeopardise their longrun relationships through “irresponsible” setting behaviour (Soskice, 1990). however, expect similar mechanisms to for unions well. They, too, have strong incentives maintain stable longrun relationships, for achieve common political aims and to be able against employers centralised levels also when bargaining is decentralised, including working time, bargaining procedures rules for settling disputes about the interpretation et a/., 1983; Bratt, 1986; Windmuller et a/., been argued that such informal interand interemployer cooperation is characteristic feature both the German systems (in the former case across industry employer associations across industry in the latter across individual across local unions the firm level for example, a/., 1987, or some coboth sides labour market always likely to emerge is the demand for insurance against labour market conemployer associations union confederations national levels in many countries is to provide such insurance through the buildcentral conflict funds (Soskice, 1990). conflicts risks appears historically have been one the driving forces the emergence more centralised labour market organisations the first place Such an insurance system does create tradeoff problems under independent decentralised bargaining. On the hand, compensation conflicts must be high enough to provide the individual firm (union) with sufficient bargaining strength relative other side. On the other hand, high compensation levels create problems moral hazard, the incentives for agents to avoid labour market conflicts are influence from the insurer (employer associations and industry unionshnion local wage natural way Many formally decentralised wagesetting systems pattern bargaining, i.e., stable pattern according (or firms) act as wage leaders, setting the pace for wage increases in economy (Flanagan a/., 1983; Windmuller et a/., In systems where the actual bargaining takes place at the industry level, the metal often fulfil this role (e.g., in Germany, Australia, Netherlands, New Zealand, Such bargaining be regarded as a informal coemployer association and the union in the wageleading sector not interests but also negotiate “on behalf” the economy (Soskice, 1990). Alternatively, keysector bargaining this type another form intermediate centralisation, where the wage employer association take their recognise that wage they set will affect wage decisions implications for their members’ welfare (Calmfors, promote wage wage increase in leading sector tends increases have negative externalities, is then and employees restrain wages cooperative situation with independent bargaining, because their wage increases will trigger wage increases for others that reduce the own elf are.^ the resulting outcome will be as favourable for employment co-opera- tion, with noncooperative wage setting (Calmfors, 1987; Forslund, 1 990).8 addition, keysector bargainprovide an efficient means ordinating expectations about Ill. DIFFERENT DIMENSIONS DECENTRALISATION The discussion bargaining institutions usually focused whether wage setting should occur at the level i.e., have labelled the extent other dimensions decentralisation according profession (trade) along regional lines and unionised sector will treated briefly. Decentralisation according Consider first unions organise different professions. most obvious independent decentralised bargaining such unions course, the United Kingdom, several unions different professions that all bargain independently at the same work place. Similar conditions and New Zealand. But also in Nordic countries, whitecollar workers bargained independently. In continental Europe mon that unions organise both blueThe humpshaped relationship between vertical centralisation discussed in Section II, the assumption industry level substantial increase the market power wage setters at the the firm, because negotiations then close substitutes. is unlikely that similar argument would in the case horizontal centralisation across professions, which not to be easily substitutable for each other are the different firms within one draws diagram with the extent horizontal centralisation and real the axes one should therefore expect monotonically negative relationship between the two variables. different professions comple- 172 The extent horizontal centralisation and the across professions that a pay for one reduces employment for others versa.9 Considerations with the relative bargaining strength and unions give individual groups can inflict heavy production losses the relative bargaining power the union side becomes unions negotiate separately instead (Horn and 1993a).1° The outcome lower profits. There are hence likely to operation in wage bargaining between different professions, for instance, eta/., horizontal cooperation are obtained independently vertical nation, the industry bargaining occurs. is interesting to register the recent tendency in the United Kingdom different unions bargain jointly with the employer a/., as well the tendency adopt a single status for Sweden provides another especially the employer side appears to aim a change bargaining system that all categories employees should be encompassed same collective agreement (Elvander, Note also that the above analysis applies operation across professions. Another issue is how setting is affected several unions organise the employees. This often the case France, Italy, the Netherlands, along political and confessional a/., 1983; Bratt, 1986; Windmuller et a/., 173 Decentralised bargaining organise workers who raising effects discussed losing employment opportunities to other unions ought wage restraint under these circumstances, just competition between firms (cf. Section 11). This effect competing unions try attract members through their ability wages. It though, how important these considerations practice, since bargaining cooperation between different unions frequent in countries mentioned above and the same collective agreement often encompasses the members all unions (Flanagan 1983; Windmuller discussed dimension is the This is surprising since bargaining the regional level occurs sectors in, for example, industries in, e.g., operation in setting between different industries region is likely to contribute substantially wage restraint, because several externalities that discussed in Section internalised already level. One reason is that the labour market employers in region ought to have strong interest in preventing their competilabour spilling over into higher Unions should also have incentives provide alternative employment opportunioffs. Moreover, regional and municipal taxes play important role in many countries, which reduce the regional increases in region also have negative effects the regional price level, especially in service sector, where the proportion tradeables” across regions is high. In addition, within wage increases may be important because most natural wage with those living in the area (see, e.g., Finally cooperation within regions is likely to entail smaller competitive pressures for industry bargaining, since firms with firms These considerations suggest that intraregional cooperation across industries in to result in lower real wages and higher employment than intraindustry cooperation across One might expect negative relationship between geographical centralisation and aggregate real shaped one, intermediate level centralisation is regional one, in Figure conclusion illustrated in the figure the major part externalities takes place already when going from independent at the firm to cooperation within the region, further gains in terms wage moderation moving to centralisation national level be limited. one cannot even out the possibilregional cooperation across industries might produce more real- restraint than bargaining the national level (curve II in Figure in the regional and the aggregate real - - - bargaining bargaining bargaining case there would exist an optimal the region from the point aggregate wage one hand, small enough that there sufficient interregional competition; the other'hand, must be large enough key externalities from decentralised regime to be regional bargaining internalisation effects weaker. But the internalisation unemployment and competilabour externalities could still the mobility the region within rather between industries. Similarly, wage to be important within between sectors (Nilsson, union density can as a decentralisation but will have different effects reduced cooperation between is important unionised firms and union coverage, i.e., the share economy that collective agreements (Layard et a/., The differences between these respects differences in bargaining structure the unionised sector Union membership a given firm determines how large force can strike, and hence also the damage that the union can inflict the employer in the case conflict. Therefore, decrease in union member- 175 weakens the relative bargaining strength and increase employment union coverage is not complete there a nonthe unionised one can assume the sector are to equalise supply and demand (Minford, 1983; Oswald, 1986). run analysis, can be the supply unionised jobs labour demand from to the cussed by a/. (1 991).11 In the paid union jobs employment for rise. The wage in economy is likely to The effects wage increase the nonregarded as inferior to are lower paid) labour force participation that the the nonbecomes smaller than the employment decrease The net theoretically ambiguous, although one former effect to a/., Layard and Changes in union coverage may have different effects depending sector. The above occurs decrease in union coverage also means is reduced (Holden and Raaum, union sector due to be larger decentralised than bargaining.l2 go some towards explaining with centralised bargaining in economies with decentralised bargaining and low union United States) have performed well the past (Layard, 1990; Layard and Nickell, 1992). MULTILEVEL BARGAINING The above analysis has implicitly assumed that not take into that the centralised e.g., Nordic countries have multilevel bargaining, and/ have regularly been followed by local about their implementation. These subsequent negotiations have sistently resulted driff, i.e., wage increases higher levels central wage agreements and drift in the counties has long been neglected research models, which analyse wage local bar (e.g., Holden, 1988; 1990a, b; Holmlund and Skedinger, 1990; Calmfors, 1993a) have been developed. basic conclusion that one should wage drift the subsequent local bargaining takes place under typically been the in Finland, Norway the contracts higher bargaining levels have been concluded). The reason bargaining positions the parties local level then become metrical. On the one employers are obliged the centrally on the inflict damage employer through informal methods the local parties slow actions individually providing less the asymmetrical bargaining positions, the employees extra wage increases in local negotiations. Wage setters higher levels increases there subsequent wage drift, desire is to be achieved (Calmfors Radseth and Holden, 1990). low rates productivity growth this real-wage the bargaining the central Whether these can be made large enough for the wage target not depend upon price inflation. envisage central even seems suggest that there may lower floor central money per cent per is true, hard to explain such from traditional assumptions about rational does represent “stylised fact” possible hypothesis is therefore that high inflation may be prerequisite for multilevel bargaining system deliver realwage restraint (Holden, 1992; Calmfors, 1993a). On the bargainers may strive for realwage restraint because various internalisation effects discussed above. On the be unable unless inflation high enough to wage increases consistent with realmoderation when there is local wage This may explain the coincidence high inflation and realwage moderation which characterised the Nordic econo1980s (Calmfors and Nymoen, 1993a). appears that the rate relationship between the centralisation and The implicit assumption is then that centralised wage setting involves bargaining levels, whereas decentral- ised bargaining not. At low inflation, the humpshaped relationship from Figure 2 ought is in II or 111 where the ranking between level bargaining wage restraint has been and C’ are above is below). the curves have been drawn, decentralised setting thus produces wage restraint than centralised bargaining does not, however, produce much realwage moderation centralised bargaining with inflation, but diagram this realwage outcome feasible alternative in Figure 5. The extent of centralisation and the aggregate real wage under multi-level bargaining and low inflation 111 Real I wage ‘.- / bargaining bargaining bargaining intermediate inflation an economy possible conclusion that the the extent formal centralisation when there low inflation. This put, for instance, Japanese, German Austrian bargaining systems another perspective. plausible hypothesis is low inflation reasonable employment growth these countries that bargaining takes place only one level enterprise one Japan, and industry one Austria and Germany though there is strong interunion and interemployer colevel organisations influence relative an important factor influencing multilevel bargaining systems. The difficulties mean that the higher central ambiaffecting the more “nominal room” is needed central level drift may local level in order has been imposed on the wage structure. This has been highlighted as a serious problem in Sweden (Calmfors, Hibbs and presented empirical evidence the hypothesis increased central push wage equalisation raised both central money wage increases in The Swedish experience can be compared Austria and where central attempts to even out differentials have been much weaker (Flanagan et a/., In the system, industry negotia- 178 workers: the minimum wage increases agreed however, act guidelines also for the actual wage increases, but unions seem to interfere the distribution within firms the total increases there conform the norm. This may necessary prerequisite combining bargaining industry (or central) level with EMPIRICAL RESEARCH discussion about the centralisation wage bargaining and macroecoperformance has provoked substantial amount empirical research during the last The focus has been rival hypotheses whether there is monotonic relationship between centralisation and aggregate real wages (macroeconomic performance) decentralised systems better than intermediate ones (the humpshape hypothstudies can i) that have centralisation to various measures macroeconomic performance; ii) have concentrated various parameters estimated aggregate equations. The has been country differences. macroeconomic performance first studies in field (Tarantelli, Cameron, 1984; Bruno 1985) pointed to positive association between macroeconomic performusually measured low values for some adding up the degree designed to capture aspects believed to be conducive to real- wage restraint. Centralisation bargaining was such crucial factor general attitude consensus between labour ment involvement negotiations were others. (1988), the concept corporatism. They substituted earlier corporatism indicators. Their finding shaped association between hand the the other other macroeconomic misery indicators (unemployment plus inflation, unemployment plus current account deficit in the post oil- shock period 1974results were confirmed drawn by Freeman (1988) using instead the effective degree work has provided diverse results. Rowthorn (1 shaped association countries between centralisation and unemployment in the eighties but seventies. Soskice (1990), argued for monotonic negative relationship, mainly highly centralised rather ones as in the CalmforsDriffill analysis (see Section 1I.B). The empirical work surveyed far must fairly unsophisticamainly focused simple correlations. few recent studies however, estimated unemployment equations in which crosscountry differences are explained by a number (e.g., unemployment benefits, expenditures active labourcoverage, the early eighties). Layard eta/. and Layard and Nickel monotonic unemployreducing effect employer coweaker such effect from union cooperation. This finding employer coordination received further support from regressions Nickell (1992), which pooled crossand time 993), however, found data are consistent also with the humpshape hypothesis. Other work has instead studied the relationship between centralisation harder to judge context, since wages on growth is complex to However, Heitger both centralised decentralised economies mies with intermediate centralisation internalisation benefits centralisation and relativewage flexibility under decentralisation. Similar 993b). conclusion is that the are conducive wage restraint should also accept meaimplying labouraugmenting productivity growth.13 Grier (1993), the results instead concluded monotonic negative relationship between centralisation and more decentralisation always implies Centralisation and parameters in quoted above investigating the effects centralisa- tion wage bargaining in form context. direct method examine how the structural parameters aggregate wage price equations are affected. The first attempts in this direction McCallum (1983, 1986) 985). They estimated crosscountry Phillipscurve equations relating inflation its change), closely associated with aka, capacity utilisation and corporatism. The latter variable found to contribute monotonically inflation. However, these are subject the imprecision the corporatism Indeed, Calmfors showed that the results were robust to substituting centralisation indicator the corporatist Later studies have the responsiveness real wages unemployment, which both theoretical arguments and empirical observations to be equilibrium unemployment, eta/., variable that has been emphasised is the sensitivity real product and relative import prices between the real product and the real consumption a/. found that corporatism real- wage responsiveness unemployment and the wedge. and Driffill instead using their centralisation indicator. They also wage equations for studies but could relationship in only for the shaped relationhowever, even weaker. the relationship centralisation and the wage responto unemployment have come results. The results in both and Alogoskoufis Manning (1988) were latter study, other evidence suggests not reflect how the attempts moderate wages centralisation (instead the parameter capturing the desire for eta/. Heylen (1993) have related the structural characteristics including centralisation. The former study found evidence favour of a association, whereas latter were hump- Finally, there wage equations. Alogoskoufis that price higher the centralisa- tion. Layard et a/. (1991) found reduce hysteresis i.e., increases following labour market, but unionindicate that firmconditions that only affect for wage is bargaining. the empirical on the impact centralisation on ing and a mixed picture. It is which significance setting has been captured through only one most studies and Layard and Nickell, has, howhighlighted the different aspects: the firm-indus- dimension, centralisation professions and across bargaining levels, union versus employer cooperation, coinformation as internal- isation union coverage and union the national level will tive centralisation depending upon sug- 181 gests the need distinguish different aspects further empirical Second, since bargaining systems in individual countries are stable over time, studies variations in the degree of centralisation must large build country differences. This centralisation is less restricted the number OECD countries. additional information can be added to the extent that the gradual developmore decentralisation in countries such Kingdom, Denmark, Finland, and Sweden exploited. The upshot is the degrees freedom when studying several institutional variables (including the labour market political conditions) become very few. This problem one tries to a proper analysis the various dimensions One must therefore regard the empirical research this area with a fair is always better knowledge there is in more than individual country many interpretations. Therefore, may have rely more the theoretical generalisations can be the relationships between centralisation, aggregate wage formation macroeconomic performance has discussed a number theoretical mechanisms. The extent likely to have different effects depending upon whether it refers to sectors, professions, regions union- isation. should not expected from empirical studies in the field. This to arrive unambiguous conclusions. Keeping these caveats mind, the preceding analysis i) is unrealistic to expect universally optimal setinstitutions to for all countries. Because varying historical traditions structural characteristics different economies, different wagegood macroeconomic performance in different places. This there are good reasons why both centralised (codecentralised (competitive) solutions In countries the United States with strong traditions wage bargaining and union density, cooperative solutions such practised in Germany Scandinavian countries just not relevant. In the latter countries truly decentralised solutions might equally irrelesetting systems change only slowly over time, any change must necessity build existing institutions and ii) The outcome industry level, such occurs in many western European economies, is likely to depend the degree foreign competition that domestic producers are exposed to. is high, forces restraining to be much weakened the level the is an increased international integration (and wage setting within industries between producers different countries). In nonindustry bargaining may, however, have strong adverse effects incentives for If little scope for interindustry cosetting, decentralisation firms in these sectors iii) can take many forms. Barthe national level, such have occurred Nordic countries, A drawback effect implies multilevel bargaining, could make combine low inflation and realwage moderation. This problem avoided with operation (including the possibility bargaining), between level bargaining informal cois not likely through synchronisation in time. Horizontal cowage setting different occupational groups when there are high unionisation appears likely advantageous from the point aggregate realwage restraint. This applies independently bargaining takes place industry, region little interest have been devoted as a suitable level possible conclusion negative exterincreases could be internalised already with intrathere seem to sometimes legislated many western European countries extend the collective sector to firms there. effect this means competition from nonrestraint and adverse effects probable consequences. The above discussion has focused the relationship between bargaining institutions and formation. However, complete evaluation also take into account the extent centralisation affects relative There has been much less research on this there is decentralised wage setting is associated with dispersion and wage flexibility (Abraham Agell Freeman, 1988; Ramaswamy This raises the possibility optimal bargaining institutions off between aggregatewage considerations. In arguments for decentralised bargaining are strengthened. One could also hypothesise that informal methods employer cocombining centralisation benefits with respect aggregate wages with reasonable degree wage flexibility. Since regional labourmarket imbalances is considered an imporpolicy aim most countries, similar argument might apply to cobargaining within regions. It wage determination, provide more various bargaining institutions. striking conclusion my survey is perhaps that links between centralisation macroeconomic performance appear complex that scientific consensus on how best to organise 1 a4 NOTES 1. for multiple equilibria that preferences are asymmetrical in the sense that the other sectors greater than the perceived benefit from being this introduces the possibility of “corner the expected See Calmfors (1993~) for a diagrammati- Note that this statement applies only relation between conditions with given degree increased openness on realwage levels given degrees cussed below (see also note practice, only employed members local unions union officials that take decisions on therefore not should be of unemployed outsiders under The same conclusion applies level bargaining are few domestic It is analyse the consequences increased international integration and van der Ploeg (1993) find a lowering in this reduces real increases employment same time real consumption wsges increase. This is the combined effect a reduced real- increases from given increases of the real consumption effect from the increase relative price domestic output consumption increases share of imports). This analysis does not, howtake into the full general (Le., the proceeds from are used). realise internalisation benefits international cooperation in centralisation both across borders and sectors, as discussed by In a simulation a model economy will raise more than a from two pattern bargaining can be analysed as called Stackelberg equiliboptimises against the reaction reply) functhe other agents. The bargaining case diseach agent as given, the solution is given by the “intersection” the various reaction conclusion differs from that Wallerstein (1990), case where employees leads for other groups as there is externality. This would seem less the one discussed the text. eg., Ekberg (1984) and white that production are HicksAllen substitutes, i.e., the outputconstant cross elasticities are positive 1993). This the possibility workers may be gross output effect dominates the 10. occur, different types labour must sufficiently complementary the sense that the the other. Complementarity this sense does two factors production are Hicks (Layard and Walters, 1978). See also note 11. The dynamics complex, since the short entails shifts labour demand and labour supply from the nonsector (most workers will remain in a firm even collective agreements are the long when individual are mobile between firms, the sectors should relative expected Theoretically, one a decrease union coverage might increase the centralised bargaining (though the averbecause the nonlower than the been pointed out that cooperation between game situation may be sustained by the threat beha- viour one union will lead from the others, causing reducing the potential utility other unions can a noncoverage may cause (Holden and Raaum, The argument is that workers perceive elasticity between the demand labour and wage setting in the individual market shares the expense adverse employment progress, they be more industry bargaining. There will also strong such incentives cen- bargaining, because the full benefits and higher and Houseman, inequality in for the Comparative Labour Markets Conference on structures”, Cambridge, July 2324, 1992. Agell, “Union egalitarianism Economica, 59, 1992. “On “Wage relativities and the natural Vol. 100, 1990. Bean, C., Layard, R. and Nickell, a multi Economica, 1986, Vol. 53, No. 2. Blanchard, 0. and Summers, L., “Hysteresis the European S. Macroeconomic Annual, “Fiscal increasing returns, hysteresis, real Relations in 78 Stockholm, 1986. The Economics Worldwide Stagflation, Lonebildning under 14 till Lhngtidsutredningen, Finansdeparte- Stockholm, 1992. “Centralized versus wage setting: och Ionerna”, Debatt, “Comment to and equality Swedish labour kets”, in Rivlin, The Swedish Brookings Institu“Kan decentraliserad lonebildning fungera?” in Herin, J, (ed.), Lonebildning i forandring, SAF, Stockholm, 1992. Calmfors, L., “Lessons from the macroeconomic experience of Sweden”, European Jour- nal of Political Economy, 1993a, Hoel: bargaining economic performance”, Behaviour, Pay Bargaining, and Economic Performance, Oxford University 1993b. 187 “The extent and macroeconomic Economics Department Working Papers, 1993c. and Driffill, “Bargaining structure, corporatism, and macroeconomic Economic Policy, L. and Forslund, formation in formation and Macroeconomic Policy Nordic Countries, versity Press, 1990. Calmfors, L. and Nymoen, R., “Real employment policies Holmlund, B., Lindbeck, A,, och samhallsekonomisk stabilisering, Svenska I., Jonung, L., Myhrman, and Soderstrom, Hur ska valfarden Konjunkturrddets rapport, SNS forlag, corporatism, labour quiescense the representation interest in capitalist societies”. In Capifalism, Clarendon, Oxford, “Wage bargaining employment and economic gration”, Forthcoming unit bargaining in oligopolistic industries”, Labour Econom S., “Enterprise bargaining, union structure Economic Record, 1993a. bargaining systems productivity growth OECD’, Report pre- pared for the Office of EPAC, 1993b. “Monopoly unions and the liberalisation The Economic “Oversikt av forskning kring arbetskrafisefterfrAgan”, in Arbetsmarknadspolitik under omprovning, SOU 1984:31, Stockholm,l984. Elvander, N., Den svenska modellen - loneforhandlingar och inkomstpolitik 1982-86, Pub- lica, Stockholm, 1988. Lokal lonemarknad i Sverige holm, 1991. Scandinavian Journal “Efficiency and Swedish labour markets”, in A,, (eds.), The Swedish Economy, Brookings Institution, 1987. Flanagan, R., Soskice, D. and Ulman, Unionism, Economic Stabilization lncome European Experiences, “Labour market “Goverments, unions and economic growth”, Mimeo, Center the Study Public Choice, George Manson University, 1993. and Horn, and the Vol. oligopolistic approach”, Economica, Vol. 51, 1984. 188 in the run”, in Heitger, B., “Corporatism, technological gaps and growth in OECD countries”, Weltwirt- “Labour market structures, labour market policy and wage Hibbs, D.A. “Wage compression, and wage inflation”, Working Labour Economics, Vol. 7, 1993. Paper No. 1991. Hoel, M., “Efficiency wages and local versus central wage bargaining”. Economic Leffers, 30, 1989. Hoel, policy: the importance of labour and the degree Economica, Vol. 58, 1991. Scandinavian Journal Vol. 90, 1988. bargaining: evidence from Norway”. Economica, Vol. 56, 1990a. a bargaining Formation and Macroeconomic Policy in the SNS and Oxford Uni 1990b. policy in an economy with local bargaining”. Mimeo, Raaum, O., 43, 1992. and Skedinger, evidence from in Calmfors, Wage Formation and Macroeconomic the Nordic Countries, University Press, 1990. effects in determinalion: from five Economic Review, ‘Horn, H. A,, policy in a unionized economy with nonCalmfors, L. and Horn, “Wage formation in the countries viewed spective”, in Formation Macroeconomic Policy Oxford University Press, bargaining and income policy: possible eastern Europe”, Paper No. Economic Performance, 1990. and Nickell, Discussion Paper for Economic A,, Publication, 1978. R. and Nickell, Oxford University Press, 1991. Outsider Theory and Unemploy McCallum, social consensus Economic Journal, Press, 1988. an open economy”, and Wallerstein, destruction and the collective bargaining”, University and Depatt- Angeles, 1992. Hoel, M., “Bargaining structure and economic perform- ance”. In Flanagan, R.J., Moene, K.O. and Wallerstein, M., (eds.), Trade Union Beha- viour, Pay Bargaining, and Economic Performance, and Oxford Newell, and Symons, Economic Review, Vol. 31, 1987. Lokal och loneinflation, Employment Outlook, Paris, 1985. Economic Outlook, Transition. Structural Adjustment in OECD Paris, 1989. Employment Outlook, Paris, 1991. an economy with many trade unions”, Papers, unions: an introductory survey”, in Calmfors, Unions, Wage Formation and Macroeconomic “Centralized bargaining, efficiency wages and and Faculty OR., in Denmark”, Oxford Bulletin Statis. tics, Vol. 55, 1993. Rowthorn, R.E., wage dispersion”, Economic Jour A,, “Centralized unions and macroeconomic policy: a Sosialerkonomisk institutt, University of Oslo, 1985. Rerdseth, A,, local versus central bargaining”, Mimeo, Economics, University Rerdseth, A. and Holden, S., “Wage Norway”, in Policy in the Nordic Countries, and Oxford Uni 1990. determination: the Vol. 6, No. 4. Skogh, G., “Employers’ associations in Sweden”, in Windmuller, J. and Gladstone, A,, Employers’ Associations lndustrial A Comparative Tarantelli, E., “The regulation of inflation in western countries and the degree of neocorporatism“, Economia, Vol. unionized economy, Ph.D. dissertation, Stockholm School of Economics, 1993. 190 Windmuller, J., ef a/., Collecfive Bargaining in Industrial “Arbetsloshet, arbetsmarknadspolitik och Poli- International Labour Office, Geneva, 1987. tik mot arbetslosbet, Betankande av 1993:43, Stockholm, 1993. 191