The Relation between Total Employment and Cooperative Employment: A Convergence and Causality Analysis

Abstract

The article analyses the evolution of the employment created by cooperative societies and its relationship with total employment in Spain during 1994–2007. We focus, first, on the comparison of the convergence processes of cooperative and total employments and second, on their relationship, in the Granger sense. For this, we use the institutional approach, which assumes the existence of a different degree of relationship between the cooperative societies and their environment according to the number of these societies in a region. Results show that the evolutions of cooperative and total employment are opposite and that total employment influences cooperative employment at the aggregate level.

Résumé

cet article analyse l’évolution de l'emploi créé par des sociétés coopératives, et son rapport avec la situation de l'emploi en Espagne, au cours de la période 1994–2007. Nous nous concentrons premièrement sur la comparaison entre les procédés de convergence dans les coopératives et l'emploi dans son ensemble, deuxièmement sur leur rapport du point de vue de Granger. Pour ceci, nous utilisons une approche institutionnelle, qui présuppose l'existence de rapports à un niveau différent entre les sociétés coopératives et leur environnement en fonction du nombre de ces sociétés dans une région. Les résultats indiquent que l’évolution des coopératives est opposée à celle de l'emploi dans son ensemble, et que l'emploi dans son ensemble influe sur l'emploi dans les coopératives au niveau agrégé.

Extracto

El artículo analiza la evolución del empleo creado por sociedades cooperativas y su relación con el empleo total en España durante el período 1994–2007. En primer lugar, nos centramos en la comparación de los procesos de convergencia de los empleos cooperativo y total, y, en segundo lugar, en su relación, en el sentido de Granger. Para ello, utilizamos el planteamiento institucional, que supone la existencia de un grado diferente de relación entre las sociedades cooperativas y su entorno, conforme al número de dichas sociedades en una región. Los resultados muestran que las evoluciones del empelo cooperativo y el total son opuestas, y que el empleo total influye sobre el empleo cooperativo en el nivel agregado.

摘要: 本论文分析了1994–2007

年期间西班牙合作社所引起的就业演变及其与总就业的关系。我们首先重点讨论合作就业与总就业的趋同过程的对比 , 然后从格兰杰意义上讨论它们的关系。为此我 们采用机构研究法, 假定根据合作社在一个区域中的数量, 合作社与其环境之间存在不同程度的关系。结果显示 , 合作就业与总就业的演变是对立的 , 而总就业会在 总体层面上影响合作就业。

Keywords:

• Cooperatives
• employment
• institutional approach

JEL classification:

• J21
• J54
• P13
• R11

Authors are grateful the financial support to the European Social Fund and to Aragonese Regional Government for this research.

Authors are grateful the financial support to the European Social Fund and to Aragonese Regional Government for this research.

Notes

1. ‘The Union shall take measures to ensure coordination of the employment policies of the Member States’ (Treaty of Lisbon, 2007: 50).

2. This assumption is also behind the Article 129.2 of the Spanish Constitution. It points out that public authorities shall efficiently promote the various forms of participation in the enterprise and shall encourage cooperative societies by means of appropriate legislation. The autonomous communities, as the delegated administrative structure of the Spanish State, are obliged to support and promote them.

3. For an explanation about the formation of these clusters and the common elements between them, see Dow (2003: 230).

4. Regarding the level of literature development, most of the papers we analyse in next paragraphs are related more to worker cooperatives, as the importance of employment is mainly observed in worker cooperatives. Nevertheless, due to the limitations of the database, the next empirical analysis focuses on all types of cooperative, and at the end of the section, we justify the extension of this approach to agricultural and banking cooperatives.

5. The next literature review discusses the issue with the relation between the business cycle and cooperative development. However, we apply a broad vision of this relationship without focusing specifically on the Ward's (1958) paper and his contribution to this matter. It has a lot of assumptions (revenue per member maximization as the objective of cooperatives, the ignorance about the effect that variations in employment have on workers’ welfare, the assumption of the absence of a labour market for cooperative members, etc. –for a wider list, see Burdin & Dean, 2009) which would need a vast explanation. This is not the point of the research of this paper, which focuses on the whole cooperative sector (see note 4) and is done from an empirical point of view.

6. This methodology has mainly been applied not only to the study of the relationship between income and GDP development (Quah, 1993) but also has been extended to other areas such as productivity (Byrne et al., 2009), effects of urban agglomerations (Bosker, 2007), inequalities amongst regions (Beenstock & Felsenstein, 2008; Tselios, 2009) and even employment and unemployment (O'Donoghuea & Townshend, 2005; Desmet & Fafchamps, 2006; Bayer & Jüßen, 2007; Tyrowicz & Wojcik, 2009).

7. β-Convergence appears when β > 0, but it can actually be greater than one (1 > β > 0). If β > 1, the partial correlation between one variable and its previous-period value would be negative and the series would oscillate, potentially from positive to negative values and back, making little economic sense (Young et al., 2008).

8. The stationary states in β-convergence can also be analysed with conditional β-convergence (Barro & Sala-i-Martin, 1992). This analysis is commented on in the next endnotes.

9. We want to thank the anonymous referee for the proposal to use the LR method for choosing the optimal number of lags. Numbers of lags for each estimation is equal to those obtained with the modified Akaike Information Criterion (AIC; Freeman, 1983), except for Group 1, where the number of lags with AIC would be 2 (4), instead of 1 (3) lags with LR method (see Table 5).

10. We applied a time series regression with ARIMA noise, missing values and outliers (TRAMO) and signal extraction in ARIMA time series (SEATS) seasonal adjustment (Gomez & Maravall, 1996) on the quarterly series in order to avoid biases introduced by outliers.

11. The differences between the groups are significant according to the analysis of variance analysis performed.

12. Group 1: Andalusia, Valencian Community, Extremadura, Murcia and Basque Country.Group 2: Aragon, Castilla-La Mancha, Castilla y León, Catalonia, Navarra and La Rioja.Group 3: Asturias, Balearic Islands, Canary Islands, Cantabria, Galicia and Madrid.

13. The existence of different steady states is also demonstrated through the study of conditional β-convergence, as the anonymous reviewer proposed. The calculation of the β-convergence is completed by the introduction of variables such as unemployment (measured as the average variation during the period), population growth (average on the period), the amount of bank deposits by region, human capital (average years of schooling in the period), health (measured as year life expectancy in 2007) and public spending (2000–2007 average percentage of GDP spending on each region). The result shows that there would be a slight conditional β-convergence (.0005*) in cooperative employment (in total employment, the importance of the β-convergence process would increase slightly to .0474***). However, this result does not much change the study's conclusions: Reducing by half the number of years needed to affect that difference between regions (Equation 3) tends to infinity (more than 1,000 years).