Efficiency in the Mondragon Cooperatives: Evidence from an Econometric case study

1. Efficiency in the Mondragon Cooperatives: Evidence from an Econometric case study Saioa Arando (MIK, S.Coop. & MU-Enpresagintza) Monica Gago (MIK, S.Coop. & MU-Enpresagintza) Derek C. Jones (Hamilton College) Takao Kato (Colgate University)

2. The case: EROSKI • Data • Insider econometric evidence • Conclusions

3. Eroski Chain • One of the largest and rapidly growing members of Mondragon Group • Core businesses=supermarkets (705) and hypermarkets (109)=the focus of our investigation.

4. Group Structure

5. Distribution Area Retail Agro-food Eroski Supermarkets Hypermarkets Dapargel Forum Sport Eroski Travel

6. Retail & Allied Group, Sales History, 1988-2008

7. Eroski Chain • Total employment ~ 50, 600 • Eroski the third largest retail chain in Spain. • Eroski is among the ten best spanish brands (Branding Global y Brand Finance).

10. RQ1: Is the legal structure important to explain firm productivity? • H1: Cooperatives are more productive than others.

11. Opportunity High Performance Work System Ability/skill Incentive Goal Alignment Job Security • for teamwork; • to produce and share valuable local knowledge; • to respond to local shocks quickly; • to accumulate firm-specific human capital;

13. RQ2: Which legal structure is nearest to the HPWS? • H2: Cooperatives are more likely to perform as a HPWS

14. Key performance & financial panel data for: • 435 supermarkets (142 coop, 26 Gespa, 267 conventional) and 80 hypermarkets (25 Coop, 55 Gespa). • Monthly data (feb-06/may-08) • 10.000 observations for supermarkets and 2.150 observations for hypermarkets.

15. First difference model (2)

16. Δ indicates the first difference between month t and t-1; Qit = output (real sales) in store i in month t; Lit = employment (measured by the number of full-time equivalent workers) in store i in month t; COOPi = 1 if store i is a coop store, 0 otherwise; GESPAi = 1 if store i is a GESPA store, 0 otherwise.

18. In addition to labor (L), store space often considered crucial capital input (K) in retail service production. For all Eroski stores during the time period under study, however, month to month variations of store space are zero and hence in our first-difference model, lnKit = 0.

19. Control variables • A store located in a rapidly growing market with rising population and average household income will naturally grow its sales faster. • To control for such differences in each store’s market condition, • MARKETit where MARKETit = monthly market index in month t for the area which store i serves.

20. Due to the standard lifecycle model of retail stores, younger stores tend to grow faster than older stores. To control for such a lifecycle effect, we also include • YEAROPENEDi = the year store i was opened.

21. constant (to capture an Eroski-wide time trend which is common to all Eroski stores regardless of its ownership types), monthly dummy variables (to capture seasonality of retail sales), and year dummy variables (to control for year time effects)

23. The first-difference model adopted for two reasons. • Field research at Eroski  sales growth a primary business goal, • First-difference models control for all time-invariant unobserved heterogeneity of stores that affects the level of sales.

25. 1. The extent of “Opportunity” measured by: • INVOLVEi = proportion of scheduled work hours spent on joint labor-management meetings (monthly average of store i during the time period under study).

27. The strength of “Incentive” gauged by: • STAKEi = average stake of employee owners (monthly average of store i during the time period under study). • MEMBERi= proportion of workers who are COOP or GESPA members (monthly average of store i during the time period under study).

29. 3. The extent of “skill/ability” measured by: • TRAININGi = proportion of scheduled hours spent on training in general (crude).

31. New model: (3)

34. We also estimated a fully nested version of Eq. (3) with all four HPWP variables considered simultaneously. • The results turned out to be quite robust to the use of such a fully nested specification although the estimates are slightly less precise due to multicollinearily as expected.

35. RQ1: Is the legal structure important to explain firm productivity? H1: Cooperatives are more productive than others. Hypermarket stores with cooperative ownership grow sales significantly faster than do Gespa stores. City supermarket: coop ownership stores are more productive than conventionally owned stores. However for Center supermarkets we find that conventional owned stores grow as fast as both coops and Gespa.

36. RQ2: Which legal structure is nearest to the HPWS? H2: Cooperatives are more likely to perform as a HPWS Consistence with those who argue for the existence of powerful incentive mechanisms for coop members who work under institutional arrangements that differ from those facing workers in other firms: a large financial stake in the firm; substantial employee involvement; unusual job security; and working in firms with earnings differences that are substantially more compressed.

37. Thank you!