Atypical work: a threat to labour productivity growth? Some evidence from Italy

Abstract

Various theories suggest the existence of a negative relationship between the use of atypical employment contracts and productivity growth, arguing that firms’ utilisation of atypical contracts may reduce the incentive to innovate and internal training, inducing firms to follow a ‘low-road’ to competitiveness, based upon cost-cutting strategies.

This paper aims to provide new evidence on the occurrence of these effects in the Italian economy, where changes in labour legislation from the mid-Nineties onwards, associated with an ‘institutional’ wage moderation period, have brought about a significant process of job creation, but also an appreciable slowdown in labour productivity.

This issue is investigated using a microeconomic approach, taking a rich source of microdata for firms and estimating a dynamic model for labour productivity on a pseudo-panel of firms for the period 2003-2008.

The results support the hypothesis of a negative impact of external labour flexibility on labour productivity growth at firm level, such effect proving stronger for small and medium than for large enterprises and of varying magnitude for the different atypical contracts.

Keywords:

• Temporary work
• labour productivity
• panel data

JEL Classifications:

• J24
• D24

Acknowledgements

We thank the staff at Laboratorio Adele of ISTAT for their help with the datasets used in this work. Data processing and model estimation were carried out at the regional data centre of Laboratorio Adele in compliance with confidentiality regulations. ISTAT does not bear any responsibility for the views expressed here. We also benefited from the comments of participants in the IZA session of the European Meeting of the International Microsimulation Association. The usual disclaimers apply.

Notes

1. Employment protection legislation is made up of a series of rules that regulate dismissal and which derive from both legislation and collective bargaining. Among other things, EPL covers the monetary compensation necessary to put an end to the working relationship (severance pay), the notification procedures to be respected to fire a worker, and the definition of ‘just causes’ for dismissal.

2. According to legislative decree 368/2001, from both the economic and legal points of view fixed-term employees must be treated in the same way as those with open-ended contracts.

3. The international comparison is based on Eurostat data, which also include among atypicals those with a causa mista contract.

4. Rlil is a survey carried out by ISFOL. The Rlil sample for 2010 contains information on 24,459 private firms in non-agricultural sectors; these firms are mostly small-sized (almost 98.6% of the sample have fewer than 50 employees).

5. Only firms employing at least one worker are considered in the analysis.

6. See Duranti (2009) for a logit estimation of the firm characteristics influencing the probability of using different types of atypical work.

7. Small and medium firms are defined as enterprises with fewer than 100 workers, while large firms have at least 100 workers.

8. These studies are part of the wider stream of literature on the effects of Employment Protection Legislation (EPL) and the consequent effect on productivity of workforce turnover (see Auer, Berg, and Coulibaly 2005; Autor, Kerr, and Kugler 2007; Bassanini, Nunziata, and Venn 2009; and Blakemore and Hoffman 1989).

9. Following one line of thought (Lucidi 2008; Lucidi and Kleinknecht 2010; Vergeer and Kleinknecht 2007), such a corporate strategy may be favoured by a modest growth in real wages. This is what happened in Italy from the 1990s onwards because of the combined effect of the new system of wage bargaining introduced in the early 1990s and of the lower wages usually paid to atypical workers (for some empirical evidence on the lower wages of temporary workers, see Bentolila and Dolado 1994; Rossetti and Tanda 2007; and Picchio 2006).

10. The authors specify that the use of temporary workers has a positive effect on ‘imitative’ (or ‘new to the firm’) products, but not on ‘new to the market’ products.

11. These data are collected within the context of Council Regulation 58/97 on structural business statistics. According to this regulation, the SBS surveys must be fully representative at the local level and for certain classes of firm size (typically 1–9 workers, 10–19, 20–49 and 50+). The SBS cover the business economy, which includes industry, construction and services, but do not cover agriculture, forestry and fishing, or public administration or (to a large extent) non-market services, such as education and health.

12. The sample of SMEs varies over time and it includes about 50–60,000 firms. The average population of large enterprises is about 10,000 units.

13. The data do not allow information on contract types to be crossed with the type of employee qualification. The education levels and skills of the workforce are not available.

14. In Italy, the phenomenon of labour hoarding has been favoured by the availability of a widely-used short-time-working arrangement, called Cassa Integrazione Guadagni (CIG), which allows firms to adjust the hours worked while preserving their workforces.

15. A higher disaggregation by type of working relationship is deemed unnecessary, given the low number of companies using some labour contracts, such as apprenticeships. On average, in our sample data, the hours worked by apprentices represent 6% and 1% of the total hours for SMEs and LEs respectively.

16. Bruno, Caroleo, and Dessy (2014) find that job satisfaction for young Italian temporary workers is higher for temporary employees with levels comparable to those of permanent employees, while external collaborators are the least satisfied. Job satisfaction can affect the work effort and therefore productivity. Lucidi and Raitano (2009) provide evidence of the existence of a wage gap between different types of atypical contracts, which penalises external temps compared with fixed-term employees.

17. With this distinction, differences in the design of the underlying surveys – respectively a sample ‘rotating’ survey for SMEs and a census survey for LEs – are considered as well as the different selection of variables collected with the two questionnaires.

18. Some examples of applied studies based on firm pseudo-panels are given in Dwenger, Rattenhuber, and Steiner (2011) and Caponera, Lugaresi, and Riti (2008). Boeri and Garibaldi (2007) use a pseudo-panel in the first part of their empirical work.

19. If firms are grouped into cohorts according to three-digit NACE and geographical macro-areas of their headquarters, the loss of heterogeneity is too significant. Therefore, we rely on the option of allowing a higher number of cohorts.

20. The yearly numbers of firms and cohorts after each step of the above procedure are reported in Table A.2 of the Technical Appendix.

21. All monetary variables are deflated by the appropriate price deflators.

22. As the apprenticeship contract requires a compulsory training period inside the firm, correlation between training expenses and the use of apprenticeship contracts has been tested. The correlation index for SMEs is –0.0045 and for large enterprises is –0.0074, although neither value is statistically significant.

23. The use of pseudo-panel data does not allow the perpetual inventory approach to be adopted to provide a measure of capital stock.

24. According to Lucidi (2012), the inclusion of lagged productivity growth allows for variations in the utilisation of productive capacity over the period, so that a firm which had an abnormally low productivity growth at the beginning of the period for transitory reasons and then returns to its ‘normal’ level is not considered a fast-growing firm.

25. The Hansen test is weakened by the inclusion of an excessively high number of instruments compared with the number of observations.

26. On the contrary, since the model is estimated on first differences, the equation will show first-order serial correlation.

27. All the estimates are carried out for the total economy and according to business sector (manufacturing and service sectors). The complete sets of results are available from the authors upon request.

28. These results appear different from those obtained in Lucidi's OLS estimate, which point to a positive relationship between contract workers and productivity growth and to a negative relationship for fixed-term and on-the-job training employees (Lucidi 2012). The difference in the results may be explained in part by the different methodology used in the estimation but, above all, by the fact that we use a wider definition of atypical workers (divided into only two broad categories) and we measure their shares by relying on hours actually worked instead of workers. According to our data, shares of external staff out of total workers as opposed to hours worked overestimate the contribution of these atypical workers, in particular for SMEs, while this effect is more limited in the case of apprentices and fixed-term employees.

29. More than a third of our panel of service firms belong to trade, hotel or restaurant activities, which are characterised by a high volatility and seasonality of demand.

30. Sectors with a high innovative content or based on high-level professions, such as information and communication services, professional, scientific and technical activities and health, represent around 30% of the total service panel.

31. This question was the title of an international workshop held in 2011, contributions to which are summarised by Jahn, Riphahn, and Schnabel (2012).