An analysis of the length of labour and financial contracts: a study for Spain

Abstract

This study investigates the connection between the duration of financial contracts and that of labour contracts. Workers with long–term contracts have incentives to invest in training. This makes them attractive to the entrepreneur. Furthermore, this behaviour will be reinforced if financial contracts are long–term, because it reduces the probability of an early liquidation as well as the dismissal of trained workers. As a conclusion, significant increases in the length of financing contracts should be accompanied by corresponding increases in the length of labour contracts. Support for this theoretical contention is found by testing it on a dataset composed of Spanish manufacturing firms for the period 1991–2000.

Acknowledgements

I deeply appreciate the comments of Ignacio Hernando from the Bank of Spain as well as those of participants in the EALE conference, Paris (2002). All errors are my own responsibility.

Notes

This model is formally developed in Hernando and Tribó (2003).

An alternative approach would have been to introduce the possibility to extend short-term workers’ contracts if their efforts, e, are high enough. The problem is that this arrangement is contingent on e, which is not verifiable and only produces some tangible results at the end of the second period.

Questions of renegotiation can be ruled out at this point.

Topel (1986) describes a long-term contract as a bond between the worker and the firm. This gives log-term workers incentives to implement specific human capital investment.

The firm is competing in a perfect competitive market and it is a contract taker.

By construction, half of the observations have a zero value for these variables.

Although dependent variables are clearly truncated, for completeness, regression estimations have been conducted controlling for fixed effects (available upon request). The results found are qualitatively the same as those obtained from Tobit estimations. This is not surprising as Hausman Test on regressions of RDEF reveals the non-existence of the endogeneity problem for the unobservable heterogeneity. See footnote 11

This technique has been employed to tackle the potential endogeneity problem between variations of a firm's labour ratio and workers’ productivity. Dependent variables have been advanced by one period.

One should say that these figures are higher than the figures shown in studies like Hernando and Tribó (2003), because the database used is more balanced in terms of size of firms because it includes a higher proportion of small firms. Smaller firms imply, according to Table 2, a higher temporal labour ratio.

This is true in the Tobit estimations but not strictly in the logit ones. In these latter estimations increases in the length of a firm's financial structure generate incentives to hire significantly more workers with any contract length. However, in the next section, it will be seen that when controlling for additional endogeneity problems, increases in long-term debt have a superior impact on increases in the length of labour contracts compared with the impact generated by increases in short-term debt.

When one estimates RDELR with a fixed-effect regression (available upon request), one also finds a positive and significant coefficient of RDEFR (0.019***), but not for RIFR. Similarly, in the estimation of RILR, the coefficients of RIFR (0.036***) and of RDEFR (0.023***) are both significant and positive. Hausman tests also reveal the non-existence of fixed-effects on RDEFR estimation. That is why we rely on Tobit estimations as dependent variables RDEFR and RIFR are totally truncated.

There is also another indirect interpretation. Those firms that have been able to reduce their financing costs (good firms) are potential subjects of a takeover. Thus, they can hire long-term workers as a ‘poison pill’ to avoid to be taken over (Pagano and Volpin, 2002).

Note that it has been found in the descriptive analysis that in the recessive year of 1992, there was a significant decrease in the temporal labour ratio. This is in accordance with our theory.

This result also works when different variables like the log of sales are used.

There is a relationship between a firm's productivity with union power. The greater the negotiated power of the unions, the lower firm's productivity. They are also less likely to negotiate long-term contracts; preferring short-term labour contracts as a way to ‘hold up’ workers with continuous negotiations. This allows the unions to maintain their bargaining power (Murphy, 1992). Thus, a direct connection is found between a firm's productivity and a bias towards long-term labour contracting.

It may be argued that the connection between financial and labour contract length is spurious in a sense that an increase in a firm's growth expectations would make long-term financial and labour contracts more attractive. By incorporating the R&D variable as a proxy of a firm's growth expectations, the robustness of the main result in the paper has been proved. However, an additional component potentially correlated with independent variables may exist in the error term. To treat this issue, one needs to address endogeneiry issues more systematically. This is made in the following section.

Workers under 30 years of age, workers over 45 years of age, the long-term unemployed, women under-represented in their occupations and disabled workers.