Worker turnover and job reallocation: the role of fixed-term contracts.

I. Introduction

The increasing interest in the dynamics of labour markets since the 1970s has generated much work on the computation of worker and job turnover. Such work has concluded that the amount of job reallocation (the sum of job creation and destruction) is large in OECD countries. Moreover, new evidence from databases incorporating information on the number of hirings and separations at firm or establishment-level suggests that workers mostly enter jobs that existed before their appointment and that will continue to exist after they quit or are fired. In other words, the rotation component of worker flows accounts for a large proportion of total worker turnover.

Are those findings in accordance with homogeneous firm behaviour (the representative firm assumption in labour demand models)? The answer to this question requires the analysis of the likely relationship between net employment changes and hirings/separations (i.e. whether employment growth (decline) and hirings (separations) are interchangeable terms for the same phenomenon), and the simultaneity of hirings and separations at firm-level. Empirical work on these issues is limited. Only Hamermesh et al. (1994) and Lane et al. (1996) have provided some evidence. Their results suggest that those terms are not interchangeable and that there are simultaneous worker flows.

The contribution of this paper is to address all the above-mentioned issues for the Spanish case while taking into account the fact that employment relationships between firms and workers can be of two types: permanent contracts and fixed-term contracts. In most OECD countries, the use of the latter has increased substantially during the 1980s and 1990s (see Bentolila and Dolado, 1994).(1) The distinction between the type of contracts highlights the difference in separation costs. In Spain, firms need collective layoffs to be approved before-hand by the corresponding labour authority, and they pay severance payments to workers with permanent contracts if workers are laid off. For temporary workers, firms know exactly the amount to be paid as separation costs (close to zero), but those costs are subject to bargaining in the case of permanent workers.

The data used in this paper is a sample of large plants (more than 500 employees) for the period 1993-4 from a database, the Survey of Economic Situation (Encuesta de Coyuntura Laboral, ECL), launched by the Spanish Ministry of Labour and Social Affairs in 1990. This is the first time that dataset has been used. Its main advantage is that it collects information at the establishment level for the non-agricultural economy, providing quarterly information on the total number of workers arriving at and leaving establishments by type of contract. However, given the specific group of establishments included in the sample, the results should be interpreted with care as they cannot be extended to the whole Spanish economy.(2)

The structure of the paper is as follows. Section 2 provides the definitions used throughout the paper, surveys the relevant literature, and describes the data in more detail. The empirical analysis proceeds in the next two sections, where the role of fixed-term contracts is addressed in detail. In Section 3 the relationship between net employment changes and hirings/separations is examined using simple cross-tabulations of the data. Next, in Section 4 the extent of worker turnover and the relative importance of its components is documented. Section 5 concludes.

2. Definitions, previous literature, and data 2.1 Definitions

Total worker turnover or worker reallocation (WR) is defined as the formation and dissolution of employee-employer job matches, thus consisting of total hirings (H) and separations (S) in the economy: WR H + S. Worker reallocation can be divided into two components. The first element arises as a consequence of firms creating and destroying job positions (gross job reallocation or job turnover, JR), leading to changes in the level of employment. The second component of turnover occurs independently of job flows, with no effect on the level of firms' employment positions. This sort of worker flows is called rotation (RR). This is a result of workers moving between job positions and may be either worker-initiated (due to personal factors, low job satisfaction, higher promotion expectations in alternative firms) or firm-initiated (as firms replace workers without changing employment positions, perhaps because job matches are poor).

How are those measures of turnover estimated? And how are they related to net employment changes? First, the net employment change ([Delta]E) in an establishment is measured as the difference between total hirings and total separations. It may be decomposed using the flows of workers as follows

[Delta]E [equivalent to] H- S [equivalent to] H- (Q+L+ T+R) (1)

where total separations consist of quits (Q), layoffs (L), transfers of workers to other plants belonging to the same firm (T), and retirements (R).

Firms may use two types of contract when hiring employees: permanent and fixed-term contracts. With this distinction, the previous identity can be rewritten in the following way

[Delta]E [equivalent to] H - S [equivalent to] ([H.sub.p] + [H.sub.F] - (Q + L + C + T + R) (2)

where total hires are now comprised of permanent hires ([H.sub.P]) and temporary hires ([H.sub.F]). In addition, termination of fixed-term contracts (C) is a new source of separations.

Net employment change may also be computed using gross flows based on positions, not people. Thus, the net employment change in an establishment is the difference between the number of jobs in periods t and t + 1. It also may be defined as the difference between job creation and job destruction in that interval of time:

[Delta]E [equivalent to] [J.sub.t+1] - [J.sub.t] [equivalent to] JC - JD.

Moreover, the sum of job creation and job destruction gives the gross job reallocation or job turnover: JR = JC + JD. It measures the reallocation of employment positions across and within establishments. In fact, as our dataset does not contain information on positions, job reallocation is proxied by plant-level net employment changes (or disaggregate job flows). This measure has been used in recent empirical work as a measure of gross job flows (see, for instance, Davis and Haltiwanger, 1990). Gross job reallocation is calculated by summing employment gains at expanding plants and employment losses at shrinking plants:

JR[prime] = JPOS + JNEG = [Delta][E.sup.+] + [Delta][E.sup.-].

See the Appendix for a description of …