Indexation and Wage Change Settlements: Evidence from Spanish Manufacturing Firms
| DOI | http://doi.org/10.1111/1468-0084.00109 |
| Published date | 01 November 1998 |
| Date | 01 November 1998 |
| Author | Sergi Jimenez‐Martin |
OXFORD BULLETIN OF ECONOMICS AND STATISTICS, 60, 4 (1998)
0305-9049
INDEXATION AND WAGE CHANGE
SETTLEMENT: EVIDENCE FROM SPANISH
MANUFACTURING FIRMS
Sergi Jimenez-Martin*
I. INTRODUCTION
The goal of this paper is to analyse a joint model for wage changes and
cost of living allowance clauses (COLA) at the firm-level in Spain while
taking into account the institutional forces at play in the labour market.
We are interested in analysing the determinants and consequences of the
decision to index the settlement, the estimation of separate wage change
equations and the detection of spillover effects from sectoral to firm-level
bargaining. The consideration of separate wage equations may help
explain the weak relationship between wages and unemployment in the
Spanish economy at the aggregate level.1Of particular interest is the
possibility that non-indexed wage changes are sensitive to market condi-
tions while the indexed are not. As regards the spillover effect from
sectoral to firm-level bargaining, we want to assess the extent to which a
given firm-level wage change is related to what other bargaining units do
and unrelated to firm performance variables. In such circumstances, the
bargaining system prevents wages from reflecting productivity differences
between workers in different firms (see Blanchard et al., 1995).
The literature on wage indexation in the context of inflation uncer-
tainty has concluded that the firm as well as the union, seeking to
maximize a function of their real income, can benefit from indexing the
contract.2For instance, assuming that the implicit negotiation cost is
relatively lower than the potential gain of indexation and that the union
*I am very grateful fo J. Andr´es, J. J. Dolado, J. Garcia, P. Kujal, J. M. Labeaga, F.
Sancho, an anonymous referee and seminar participants at Barcelona (UPF and UAB),
London (UCL), Valencia, Aix-en-Provence (AEA meeting) and Warsaw (EALE meeting) for
very useful comments and suggestions. Financial support from DGICYT PB92-1036-V02-01,
DGES PB95-0980 and CIRIT EE:93-160 are acknowledged. All errors remaining are my own
responsibility.
1See Andr´es (1993) or Blanchard et al. (1995) for recent reviews of the macroeconomic
empirical evidence on the (weak) relationship between unemployment and wages in Spain.
2Shavell (1976), Gray (1978), Azariadis (1978), Blanchard (1979), Dazinger (1980, 1983),
Ehrenberg et al. (1983, 1984), Card (1986) and Gottfries (1992).
449
© Blackwell Publishers Ltd, 1998. Published by Blackwell Publishers, 108 Cowley Road, Oxford
OX4 1JF, UK & 350 Main Street, Malden, MA 02148, USA.
is relatively more risk averse than the firm,3Shavell (1976) showed that
there is at least one Pareto optimal contingent contract. In fact, the more
risk averse the union, the higher the degree of indexation degree required
and the lower is the expected real wage that it is willing to accept.
So far, the majority of applied work in this field, using micro data,
deals with North American labour markets.4Since earlier work (see, for
instance, Hamermesh, 1970, or Sparks and Wilton, 1971), great efforts
have been made to try and integrate the basic Phillips curve explanation
and the institutional features of the labour market. The availability of
contract data has facilitated this purpose. Using US contract data,
Vroman (1984) and Kaufman and Woglom (1984) analyse the role of
expectations in determining union wages, while Hendricks and Kahn
(1985) obtain an estimate of the price workers must pay to obtain wage
indexation. In turn, using contract data from Canada, Christofides (1990)
emphasizes the interaction between the degree of indexation (i.e., the
parameter which ties wage changes to inflation), contract duration and
non-contingent wage adjustments, while Prescott and Wilton (12992) deal
with an endogenous switching model of wage changes and indexation,
emphasizing the analysis of compensation inflation structure. In particu-
lar, they allow total inflation compensation to vary with the degree of
COLA protection. In addition, Christofides et al. (1980) or Prescott and
Wilton (1991) found important ‘wage spillover’ effects.
This paper, in line with recent work for Canada, considers the interplay
between the indexing decision and wage changes while taking into
account the inflation compensation structure and the potential spillover
effects. In addition, we develop the following issues. First, we investigate
whether some union groups behave strategically when negotiating indexa-
tion.5This possibility, which may help explain the evidence from the
Spanish data that half of settlements are not indexed, is especially
relevant in the Continental-Europe Collective Bargaining (CB) context,
where works councils replace unions at firm-level negotiations. Several
union groups can form a works council. We consider two types of groups:
strong and weak. Strong groups are linked to nationwide unions, which
3The fact that there are less opportunities for diversifying human capital against risk than
for diversifying other kinds of capital, might be argued in favour of such an assumption (see
Danziger, 1980).
4Sparks and Wilton (1971), Ehrenberg et al. (1983) and Card (1986) are good examples
using industry level data. Hamermesh (1970) and Kaufman and Woglom (1984) used a long
time series of a few firms. Christofides et al. (1980b), Vroman (1984), Hendricks and Kahn
(1985), Christofides (1990) and Prescott and Wilton (1991, 1992) are also good examples
which employ firm level data.
5In general, rejection of indexation clauses could be either due to firms being relatively
more risk averse than unions and/or by the presence of relatively high transaction costs or by
alternative sources of price uncertainty (see Card, 1986). An interesting exception is Gottfries
(1992) who show, extending the Baily (1974) and Azariadis (1975) standard labour contract
model, an equilibrium in which, provided all other contracts are not contingent, there is no
indexation.
© Blackwell Publishers 1998
450 BULLETIN
carry out negotiations at national or industry levels. Thus, they benefit
from aggregate wage setting. Alternatively, weak groups are either linked
to regional (or professional) unions or are formed by independent
workers. They do not participate in aggregate wage setting and hence
receive little credit from it. Consequently, as noted by Ehrenberg et al.
(1983), they may prefer not to ask for indexation but to achieve relatively
higher non-contingent wage changes, in order to gain credit from the
workers.
Secondly, we evaluate the differential between ex-ante indexed and
non-indexed wage change settlements. In our context, this differential
evaluates the ex-ante cost (or premium) that workers ought to pay to
obtain the COLA clause. However, the former measure does not evaluate
the expected cost of a COLA (i.e., the price of indexation as defined, for
instance, in Hendricks and Kahn, 1985), since the expected contingent
compensation is not taken into account. An evaluation of the expected
cost of COLA is obtained by comparing the realized average contingent
compensation with the former ex-ante cost.
Thirdly, we solve a problem with the degree of indexation, a key part of
contingent contracts that is unobservable ex-ante but which becomes
observable ex-post for triggered contracts. When a contract triggers, we
observe ex-ante and total wage adjustments, and the inflation rate. In
most cases, we can deduce the inflation threshold (i.e., the inflation rate
at which the contract triggers). For non-triggered contracts, Prescott and
Wilton (1991, 1992) used the realized degree of indexation. This can have
strong consequences since it introduces measurement error into the
model. To avoid this problem we instead use predictions for the degree of
indexation. These predictions are obtained from a joint model of the
probability that a clause triggers and the degree of indexation (see
Appendix B for details). Ehrenberg et al. (1983), Card (1986) or Christo-
fides (1990) provide guidelines to specify such a model.
Finally, we carefully examine the role of the government’s inflation
target and the information ‘flow’. The inflation target is important since
agents use it to recommend a wage change and, in most cases, it deter-
mines the inflation threshold. Likewise, the flow of information, which
captures any spillover effect, is important because agents in any particular
negotiation unit use this information to see how the alternative wage is
progressing (see Christofides et al., 1980b, or Prescott and Wilton, 1991).
As a novel feature, we consider different measures of the flow of informa-
tion for indexed and non-indexed wage change equations.
To carry out the analysis we use data from the ‘Estadistica de Conve-
nios Colectivos’ (ECC), which has information on all the agreements
settled in Spain between 1981 and 1991. The analysis is restricted to
manufacturing firms that are observable for at least four consecutive
years. Note that we use annual observations (and no contract observa-
tions as is common in, for instance, North American data) because multi-
© Blackwell Publishers 1998
451INDEXATION AND WAGE CHANGE SETTLEMENTS
To continue reading
Request your trial