“…These may further improve our test procedure, but go beyond the scope of this note. 7 Examples include Black and Lynch (2001) with a number of observations of about 1,000, Galia and Legros (2004) with about 1,800, Laursen and Foss (2003) with about 1,900, Belderbos et al (2006) with about 2,000, Bresnahan et al (2002) with about 2,200, Catozzella and Vivarelli (2007) with about 3,000, Mohnen and Röller (2005) with about 5,500 and Cozzarin and Percival (2006) with about 5,900 observations. 8 For comparison, we executed similar Monte Carlo simulations with correlation coefficients set at 0.8, -0.8 and 0, respectively and without systematic correlation between the practices.…”