THE INCIDENCE OF PAYMENT BY RESULTS SYSTEMS: SOME RESULTS FROM FURTHER ECONOMETRIC ANALYSIS
| Author | R. F. Wynn,K. Holden |
| Date | 01 November 1972 |
| DOI | http://doi.org/10.1111/j.1467-8543.1972.tb00597.x |
| Published date | 01 November 1972 |
RESEARCH
NOTES
THE INCIDENCE
OF
PAYMENT
BY
RESULTS SYSTEMS:
SOME RESULTS FROM FURTHER ECONOMETRIC ANALYSIS
K.
HOLDEN
and
R.
F.
WYNN*
IN
a
recent article in this Journal, Bean and Garsidel discussed some of the
factors affecting the incidence
of
payment by results
(P.B.R.)
systems in British
manufacturing industry. In particular, they defined the variables
&
=
percentage of wage earners within
P.B.R.
schemes in minimum list
XI,
=
average number
of
operatives employed per establishment
in
mini-
X,,
=
wages and salaries of operatives as
a
percentage of the value of net
heading industry
i,
mum list heading industry
i,
output in minimum list heading industry
i,
and estimated the regression equation
r,
=
a
+
BX,,
4-
7x2,
+
4,
(1)
where
a,
/3
and
y
are unknown parameters and
ul
is
a
random disturbance, for
the
108
observations published at the level of minimum list headings. Their
result of
Ra
=
0.282
and
F
=
22.0
indicates that while the overall
fit
is
signifi-
cant
at
the
1
per cent level
of
confidence
a
Iarge part
of
the variation in the
incidence of
P.B.R.
schemes remains unexplained.a This note was stimulated
by two issues arising out
of
these regression results: (i)
a
technical point relating
to the further analysis
of
the residuals,
e,,
which separate observations on
r
from the predicted values of
Y,
PI,
which lie
on
the fitted regression line
and (ii) the bearing that this further analysis has
on
the discovery
of
other
variables which are associated with
T.
THE
ORDERING
OF
CROSS-SECTION
DATA
AND
THE
DETECTION
OF
SPECIFICATION
ERRORS
Classical least squares regression and correlation analysis methods are
valid subject to
a
number
of
assumptions including the assumption that the
disturbances are unrelated; violation
of
this particular condition will result in
*
Lecturers in Economics, Univemity of Ziverpool
l
R. Bean and
D.
A.
Garside, ‘Pa ment by Results Systems: Some Indicators
of
Incidence
and Relevance
to
Capital Intensive &rations’,
British
JOM~Q~
of
Industrial
Relations,
Vol.
IX,
a
The measure
Ia,
known
as
the coefficient
of
determination adjusted
for
degrees
of
freedom,
is
an estimate of the proportion
of
the variation found
in
Y
that
is
associated
with the independent
variables. See
R.
J. and
T.
H.
Wonnacott,
Econometrics,
John Wiley, 1970,
pp.
116-20
440
NO.
2,J~ly 1971,
pp.
182-97
To continue reading
Request your trial