The earliest empirical HRM–performance research was conducted by Arthur (1994),
who used a sample of steel mini-mills operating under two different HR systems (control
vs. commitment). Arthur (1994) found that commitment HR systems, in contrast to control
systems, resulted in higher productivity, lower scrap rates and reduced turnover. In a
seminal study using data from 1000 firms, Huselid (1995) examined the relationship between HPWS and organizational performance and found that the relationship between
HPWS and corporate financial performance was mediated by turnover and productivity.
HPWS reduced turnover and increased productivity, thus in turn having a positive effect
on corporate financial performance. An equally important study was conducted by
MacDuffie (1995), who used a survey of 62 international automotive assembly plants to
test whether HR bundles affected productivity and quality. HR bundles were defined as
interrelated and internally consistent HR practices. MacDuffie (1995) found support for
the effect of HR bundles on plant productivity and quality. These pioneering studies were
then followed by others using different industries and regions (Delaney and Huselid 1996;
Becker and Huselid 1998; Ichniowski and Shaw 1999; Bae and Lawler 2000; Evans and
Davis 2005; Chand and Katou 2007; Chang and Chen 2011; Lertxundi and Landeta 2011;
Zhang and Jia 2012).