Вы находитесь на странице: 1из 10

Journal of Occupational Psychology, 1985, 58,39-47.

Printed in Great Britain


1985 The British Psychological Society

The relationship between absenteeism


and production efficiency: An empirical
assessment
MICHAEL K. MOCH*
Michigan State University

AND DALE E. FITZGIBBONS


Illinois State University
Despite almost universal agreement that employee absenteeism leads to decreased
production efficiency, there is little documentation of a relationship between these
variables. Several authors have even suggested that absenteeism might have some
beneficial effects. The present study finds that absenteeism and department efficiency
are negatively associated only (1) when production processes are not highly automated, and (2) when the absences cannot be anticipated in advance. Despite these
limitations, however, the costs attributable to the impact of absenteeism on department efficiency are substantial. Programmes designed to decrease unanticipated
absenteeism, therefore, can result in considerable savings by increasing operating
efficiency where employees are directly involved in the production process.

There is a general consensus that absenteeism is costly to organizations (Steers & Rhodes,
1978). Macy & Mirvis (1976) identify several types of costs incurred through absenteeism.
For example, fringe benefits or salary paid to absent personnel, supervision time spent
finding replacements or training new personnel, and unabsorbed overheads are costs
which might be associated with employee absenteeism. These authors, like others
(Metzner & Mann, 1953; Morgan & Herman, 1976; Steers & Rhodes, 1978), also argue
that absenteeism hinders operating effectiveness and efficiency. Effectiveness obviously
would be constrained if absent personnel were not replaced. Replacing absent personnel,
however, may not eliminate the negative consequences of absenteeism. Effectiveness often
is associated with familiarity and practice, and replacements frequently are less familiar
and less well practised in the job than those they replace.
While the possibility generally is overlooked, absenteeism may have positive as well
as negative consequences for operating effectiveness and efficiency. Staw (1977), Staw &
Oldham (1978), Brandth et al. (1980) and Hammer et al. (1981), for example, argue that
absenteeism may actually be beneficial. Absenteeism may relieve dissatisfied employees of
job-related stress and thereby allow them to be more productive when they return to work.
Perhaps such a combination of positive and negative consequences is responsible for the
frequent failure to document a consistently negative association between absenteeism and
operating efficiency (e.g. Argyle et al., 1958; Seashore et al., 1960; Turner, 1960; Ronan,
i963). Because of these inconsistent findings and because contradictory hypotheses can
Requests for reprints should be addressed to Michael K. Moch, Michigan State University, Graduate
School of Business Administration, East Lansing, MI 48824, USA.
39

40

MICHAEL K. MOCH AND DALE E. FITZGIBBONS

be generated theoretically, more research is needed if we are to come to understand the


relationship between absenteeism and operating efficiency and effectiveness.
Research relating absenteeism and production efficiency and effectiveness for the
most part has been carried out at the individual level of analysis. From a departmental or
organizational perspective, however, we would expect the effects of absenteeism to vary
with a variety of more ' macro' variables. The effects, for example, are likely to vary
depending on the employee's job or location in the organization. Attendance by
employees responsible for ground maintenance, for example, is likely to be less important
than attendance by those more directly critical to the production function.
Two other contingencies are the objects of study here. First, automation may reduce
the critical functions performed by employees to those which can be carried out by anyone
with minimal ability and familiarity with the job. Automation, therefore, may reduce
or perhaps even eliminate losses attributable to replacing absent employees with less
well-trained and well-practised substitutes. Second, department supervisors or company
personnel directors may reduce the costs of absenteeism by anticipating and planning for
them. For example, vacations can be coordinated so that interdependent personnel will
be absent at the same time or to ensure that at least one person capable of performing a
critical function will be present at all times. Employers who require advance notification
for certain types of absences frequently do so precisely so that this sort of scheduling can
occur.
These two organization or department level contingencies may appear to be obvious,
but they have not been considered in studies relating absenteeism to productivity at the
individual level of analysis. Understanding this relationship is also particularly important
because the untested assumption that absenteeism generally hinders operating effectiveness frequently guides managerial decision making.

METHOD

The research site


Data were gathered from a medium-sized assembly and packaging plant. The plant
produced over 30 different food products. Production was organized along conveyors
which carried materials from the initial assembly operation through to final packaging
and movement to the distribution centre. The assembly and packaging areas constituted
distinct departments. Approximately 130 people were employed in the assembly department. Four hundred and fifty employees worked in the packaging area. Maintenance of
the conveyors and a variety of assembly and packaging equipment was the responsibility
of approximately 90 maintenance personnel, primarily millwrights and electricians, in
the maintenance department. Department supervisors allocated personnel to different
conveyors on an as-needed basis. Seniority and job experience determined which
employees were assigned to specific locations and to specific tasks.
The assembly and the packaging operations were both highly mechanized. Input
material was transported to the assembly area through pheumatic tubes. When it arrived,
it was pressed into form by large rollers and laid automatically on the moving conveyor.
In the packaging department, wrapping and boxing was done mechanically. The only time
employees touched any of the product was when they put it together in fixed amounts
prior to wrapping and packaging.
While all production was highly mechanized, there was variation in the degree to
which the production processes were automated. Some products were produced in batches
and the production process was organized in a series of discrete steps. Other products were
run more continuously and the production process was organized more as a continuous

ABSENTEEISM AND PRODUCTION EFFICIENCY

41

Table 1. Absenteeism levels by department and by absence


type (n = 103 weeks)

Packaging department (/J = 4 5 0 )


Sickness absence
Other unanticipated absence
Vacation absence
Assennbly department (n = 130)
Sickness absence
Other unanticipated absence
Vacation absence
iVIaintenance department (/7 = 9O)
Sickness absence
Other unanticipated absence
Vacation absence

X days absent
per week per
100 employees

SD

39-24
284
30-76

9-88
1-12
11-59

17-38
0-92
30-23

5-16
0-85
13-40

14-56
4-44
30-00

6-88
3-33
18-54

flow (Woodward, 1965). In addition, only one product was run on a conveyor at any one
time. Since there were considerably fewer than 30 conveyors, changeovers in which a
conveyor was stopped and refitted to carry a different product were frequent. The
mechanization, automation, and frequent changeovers combined to make the maintenance function as critical as the assembly and packaging functions in the overall production
process.
Measures
Absenteeism. Absence data were gathered from company records. Data were gathered on
the number of days each employee in each of the three departments was absent during
the previous two years. These data were coded to reflect whether the employee was on
vacation or absent for other reasons subsequently provided by the absent employee. The
company distinguished between unanticipated absences due to sickness and those due to
other reasons. These were combined, and the indicator of absenteeism was then taken to
be the number of absent employee days, expressed as a percentage of the total number of
possible absent employee days.
Average weekly days absence per 100 employees for each department and absence
category are presented in Table 1. Correlations among measures of absenteeism are
presented in Table 2. The high correlations among vacation absenteeism across departments reflect management efl"orts to synchronize them. Since we anticipate no efl'ect of
vacation absences, these high correlations do not present problems of multicolinearity.
The other correlations in Table 2 are relatively low. Correlations between sickness absence
and other unanticipated absence present no problem, since these were combined to provide an overall index of unanticipated absenteeism. While there are small but statistically
significant correlations between unanticipated absences in the packaging department and
sickness-related absences in both the assembly and maintenance departments, there are no
significant correlations between unanticipated absenteeism in the assembly department
and unanticipated absenteeism in the maintenance department.
Vacation absences were anticipated, and it was therefore expected that operations
would not be noticeably less efl"ective or eflicient during weeks when relatively many
employees were on vacation. Production schedules could be adjusted depending on the
number of employees available. In addition, demand for the product was seasonal, and
vacation time was coordinated to match the period of low product demand.

42

MICHAEL K. MOCH AND DALE E. FITZGIBBONS

Table 2. Pearson product moment correlations among measures of absenteeism by department and by reason for absence (n = 103 weeks)
Packaging department
1. Sickness absence
2. Other unanticipated
absence
3. Vacation absence
Assembiy department
4. Sickness absence
5. Other unanticipated
absence
6. Vacation absence
Maintenance department
7. Sickness absence
8. Other unanticipated
absence
9. Vacation absence

023*
003 - 0 0 8
031*
002
-001
0 34*
008
-007
1

016*
-005
-001
021*
-003

-on
2

000
-008
077*

002
002

-004

-026*

012

-009

-019*

0 17* - 0 1 0
083*

-001
-010

012
071*

-0-12
-OW

017*

-on

The other absences were not anticipated. When employees in the assembly and
packaging departments unexpectedly failed to report for work, they were replaced from
a pool of people with less experience and seniority which was maintained by each department for this purpose. The same departmental pool supplied replacement personnel for
each ofthe conveyors. The number of employees in this pool was decreased or increased as
needed by drawing upon or adding to a list of employees on temporary layoff. Since many
of the jobs in each ofthe three departments required familiarity with machinery and with
different products, however, it was generally believed that there were costs associated with
employing relatively inexperienced people. It was felt that substitutes could not perform at
the same level as regular employees and that unanticipated absences therefore adversely
affected operating effectiveness and efficiency. This was particularly true for the maintenance department which had no replacement pool from which to draw. Jobs which
otherwise would have been done by a maintenance employee who failed to report
therefore had to be reassigned or postponed.
Automation. To assess the role of automation in determining the consequences of absenteeism, production data were gathered on two specific productsthe product produced by
the most automated process and a comparable product which required much more direct
employee involvement. During the production process for the automated product,
employees monitored equipment as the material passed from assembly through packaging
on its way to the distribution area. The less automated product was produced in batches.
Planning, scheduling, starting, and stopping production of this product required direct
employee involvement. In addition, the wrapping equipment frequently fouled, requiring
direct employee intervention. Employees were also involved directly in maintaining a
queue of packaged products as packages were fed into the final package sealing process.
Employees were also more directly involved in the production process for the less
automated product, because it was run on a conveyor which produced several other
products as well. The average changeover time charged to this product was 3-35 hours per
week during the period under study. The comparable figure for the more automated
product was 0 2 hours per week during the same period. While the less automated product

ABSENTEEISM AND PRODUCTION EFFICIENCY

43

required skilled intervention in assembly, packaging, and maintenance, therefore, the


more automated product moved along continuously and required very little direct
employee involvement. The monitoring functions required for this product were very
straightforward and could be performed equally well by veteran employees or by newcomers. It was expected, therefore, that absenteeism would have less effect on production
efficiency for this product relative to the less automated product.
Operating effectiveness or efficiency. The company kept records both of the number of
pounds of each product produced and of the pounds of each product rejected as waste
for each week of the two-year observation period. In addition, the number of labour
hours used in both the packaging and assembly departments was recorded and saved. It
was decided to measure efficiency via waste per unit resource expended. Because waste is
directly related to lost production in this technology this also constitutes a measure of
eftectiveness.
Rather than combine assembly and packaging hours assigned to each of the two
products, it was felt that greater precision could be gained by concentrating on the largest
department, the packaging department, while controlling for labour hours assigned in
the assembly department. Accordingly, the measures of efficiency used in the study are
the number of pounds of product produced and wasted per labour hour assigned in the
packaging department.
ANALYSIS: COMPARING ALTERNATIVE TIME SERIES REGRESSION MODELS

Analysis compared the relative predictive utility of regressions which generate


separate coefficients for high versus low absenteeism weeks with regressions which did not
allow for separate estimates. If the former explained significantly greater variance in
pounds produced or wasted, it would constitute evidence that the type of absenteeism
under investigation had an effect. All regressions included a control for labour hours
assigned in the assembly department, and, because of the high correlations among
measures of anticipated (vacation) absences, separate regressions were run for anticipated
and unanticipated absences in each department.
The baseline model. The regression equation which did not allow for separate coefficients
for high versus low absenteeism weeks took the following form:
Y=aXi-\-fiX2-\-C,

(1)

where Y= pounds product produced or wasted;


Xi - labour hours assigned in the packaging department;
X2 = labour hours assigned in the assembly department;
C = a constant*
Once the coefficients generated by applying equation (1) had been estimated and a
value for variance explained, /?^(1), had been obtained, the data were run again, on an
alternative model.
The alternative model. The regression equation which allowed for differential effects of
labour hours for high versus low absenteeism weeks took the following form:

y=a,;sr,,+a,x,,+)?A',+c,

(2)

*It is conceivable that competitors of the plant could identify it as the subject of this research. They also
might identify the products under study. If this were to happen, competitors would obtain valuable information concerning operating efliciency. Inclusion of the constant term precludes this possibility, and for
this reason the values of this term will not be reported here.

44

MICHAEL K. MOCH AND DALE E. FITZGIBBONS

where Y= pounds product produced or wasted;


A'i^ = labour hours assigned in packaging for weeks experiencing greater than the
median percentage employee days absence (low absenteeism weeks were dummy
coded zero on this variable);
^"1^ = labour hours assigned in packaging for weeks experiencing median percentage employee days absence or less (high absenteeism weeks were dummy coded
zero on this variable);
X2 = labour hours assigned in the assembly department
C = a constant (see footnote).
Regression coefficients a, and Oj in equation (2) provided independent estimates of
the number of pounds produced or wasted per labour hour under conditions of high (a,)
versus low (Oj) absenteeism. Separate regressions were run for each department for both
types of absenteeism (anticipated and unanticipated), for two products (more vs. less
automated), and for each of two dependent variables: pounds product produced and
pounds product wasted.
Comparing the predictive utility of the competing models. The only difference between the
baseline model and the alternative model is whether they allow for the possibility that the
impact of labour hours on pounds produced or wasted could be significantly different
during weeks of high vs. low absenteeism. The significance of the difference in the
explanatory power ofthe baseline model, R^{1), and that ofthe alternative model, /?^(2),
therefore, is also a test of the significance of the difference of the impact of labour hours
on pounds produced or wasted under high vs. low absenteeism weeks. Initial regressions
based upon equation (1) revealed substantial amounts of autocolinearity. Values of the
Durbin-Watson d statistic tended to be close to 1 0. Accordingly, the Cochrane-Orcutt
technique was used to transform the measures to reduce correlation among first-order
residuals (Johnston, 1963, pp. 192ff). Producton decisions were made on a weekly basis.
Not infrequently, product not produced in one week had to be postponed to the next
week. Adjustments for first-order residuals, therefore, were appropriate. There was no
reason to expect higher-order autocorrelation and the analyses, therefore, were limited to
first-order ad.justments.
RESULTS

Regression coefiScients reflecting pounds wasted per packaging department labour


hour under conditions of high versus low absenteeism for reasons which could be
anticipated and for reasons which could not be anticipated are presented in Tables 3 and
4. Table 3 presents the data for the less automated product. Coefficients generated from
data on the more automated product are presented in Table 4. All regressions explained a
statistically significant proportion ofthe variance in pounds wasted (39-51 per cent).
It is clear from Tables 3 and 4 that absenteeism is associated with department
efficiency. However, this relationship occurs only when the absences are not anticipated
and the production process was not completely automated. The difference in pounds
wasted per labour hour under conditions of high versus low unanticipated absenteeism in
the packaging department approaches but does not attain generally accepted levels of
statistical significance {P<005); however, the trend is evident in all three departments,
and statistically significant {P<005) differences in efficiency by level of absenteeism were
observed in the assembly and maintenance departments.
Tables 3 and 4 generally show no relationship between absenteeism and packaging
pounds wasted per labour hour when absences were anticipated. This relationship also
fails to appear for the highly automated product. This is particularly interesting, given the
fact that distinction between absence types is seldom made in the literature. The data.

ABSENTEEISM AND PRODUCTION EFFICIENCY

45

Table 3, Pounds waste produced per labour hour (regression coefficients) by absenteeism
level and department for anticipated and unanticipated absence (less automated product.
n= 101 weeks)

Packaging
absenteeism

Pounds waste produced per labour hour


Assembly
Maintenance
absenteeism
absenteeism

Reasons for

absenteeism

Low

Anticipated
Unanticipated

1 70
1 81
081 250

High

Low

High

221
1 58
- 0 3 1 *' 2-12

Low

High

1 78
1 40
047 280

'Difference between coefficients statistically significant at P<0-10,


"Difference between coefficients statistically significant at / ' < 0 05,
""Difference between coefficientsstatistically significant at P < 0 01,

Table 4, Pounds waste produced per labour hour (regression coefficients) by absenteeism
level and department for anticipated and unanticipated absence- (more automated product,
n = 103 weeks)

Packaging
absenteeism

Pounds waste produced per labour hour


Assembly
Maintenance
absenteeism
absenteeism

Reasons for
absenteeism

Low

High

Low

High

Low

High

Anticipated
Unanticipated

773
7 94

7-13
738

829
756

* 6 46
763

793
727

725

"Difference between coefficients statistically significant at P < 0 05,

however, also show that in the automated condition when absenteeism can be anticipated,
pounds wasted per labour hour tend to be greater under conditions of low as opposed to
high absenteeism. For the assembly department, this difference is statistically significant.
It is possible that when supervision can anticipate absences in assembly, they lower their
production quotas, reduce stress on available employees, and thereby reduce waste per
labour hour in packaging. There is no apparent reason why this does not occur in the case
of the maintenance department, however, and this finding would have to be more fully
explained by further research.
There were no statistically significant differences in pounds produced per labour hour
under conditions of high as opposed to low absenteeism. Labour hours explained between
86 and 97 per cent of the variance in pounds produced. This was possibly due to the fact
that the number of labour hours allocated to produce the prespecified number of pounds
was established by policy. Despite this, however, some interesting trends were evident in
the data.
The number of pounds produced per labour hour was generally constant for the more
automated product. It ranged from 124 pounds per labour hour to 126 pounds per labour
hour, with labour hours always explaining 97 per cent ofthe variance in pounds produced.
There was more variability in pounds produced per labour hour for the less automated
product. Labour hours explained a constant 86 per cent of the variance in the number of

46

MICHAEL K. MOCH AND DALE E. FITZGIBBONS

pounds of this product produced. However, under conditions of high absenteeism for
unanticipated reasons in the packaging department, an average of 295 pounds per labour
hour were produced. During weeks when the absenteeism was relatively low, 324 pounds
per labour hour were produced. This relationship was reversed for anticipated absences.
During weeks when absenteeism was high in packaging, 321 pounds per labour hour were
produced. When the absenteeism was relatively low, packaging produced 291 pounds per
labour hour. These differences were not statistically significant but they do suggest that
unanticipated absenteeism may reduce efficiency measured in pounds produced per labour
hour. More interestingly perhaps, they suggest that when management can anticipate
absences, adjustments can be made which may actually increase efficiency.
SUMMARY AND DISCUSSION

The data do not support the contention that absenteeism adversely affects operating
efficiency under all conditions. This common assumption does not hold when absences
can be anticipated and therefore planned for in advance or when production procedures
are highly automated. In the plant studied here, when absences were not anticipated and
production was not highly automated, absenteeism in the assembly and maintenance
departments was associated with lower levels of efficiency, though absenteeism in the
packaging department was not associated with lower efficiency at a statistically significant level. Packagers in the plant studied here were relatively low paid and unskilled,
while assemblers and maintenance personnel required more training and skills to operate
equipment and respond to variability in the product. It is possible that the relatively
unskilled nature of the work made packagers more interchangeable and thereby reduced
much ofthe potentially adverse effect of absence on operating efficiency. Without further
information, this/)05/ hoc interpretation must be regarded as speculative.
The impact of absenteeism among assembly and maintenance personnel was clearly
visible in the data. When the product was not completely automated and when absenteeism could not be anticipated, absenteeism in the assembly and maintenance departments was associated with significantly more pounds wasted per labour hour. Moreover,
since the levels of unanticipated employee days absence in these departments were not
correlated, their effects can be taken to be independent.
The differences in pounds wasted per labour hour under conditions of high vs. low
absenteeism in the assembly and maintenance departments may seem to be rather small,
ranging from 2 33 to 2-43 pounds. These differences, however, represent substantial
dollar losses when projected on an annual basis. An average of 462 packaging hours
were allocated to the relatively unautomated product each week. This means that
2 43 X 462= 1123 more pounds of this product were wasted during high as opposed to
low weeks of unanticipated absenteeism in the assembly department. This figure for
maintenance is 2 33x462=1076 pounds wasted. Together, 1123+1076 = 2199 pounds
were wasted, on the average, during high as opposed to low unanticipated absenteeism
weeks in these two departments. By definition (median split) half of the weeks reflected
high absenteeism. Unanticipated absenteeism, therefore, may be held responsible for
2199 X 50-5= 111 050 pounds wasted during the course ofthe 101-week study.
This product retails for approximately S0.85 for a half-pound container. If 5 per cent
of this cost represents retail mark-up and 25 per cent represents transportation and the
cost of packaging materials, then each pound wasted represents (0-85 x 2) 0 70 = $ 1 19 lost
income per pound. This totals 111 050xSI 19 = $132 150 lost during the course of the
two-year study. This represents $66 075 annually lost due to the effect of unanticipated
absences for only one of the over 30 products produced by the plant being studied. Only
one of these products was completely automated. The effect of absenteeism in the
company labour pool which packaged all of these products therefore is likely to have

ABSENTEEISM AND PRODUCTION EFFICIENCY

47

been felt by all of the other products. The costs of unanticipated absenteeism for the
organization, therefore, are doubtless much greater than the $66 075 lost annually on the
single product studied here. Steers & Rhodes (1978) estimated the annual nationwide cost
of absenteeism as high as $26-4 billion. The data presented here suggest that this estimate
may not be unrealistically high.
Data reported here support but modify the commonly held assumption that absenteeism reduces operating efficiency. It appears to be possible to 'people proof production
through automation and thereby to reduce and even eliminate the eflfects of absenteeism
on production efficiency. It also appears that it is equally effective and less costly to
develop ways to better anticipate absenteeism as to reduce absenteeism itself. Organizational interventions, therefore, might usefully be directed toward establishing the levels
of trust required for employees to seek advance approval for planned days absence for
reasons which have not generally been considered legitimate.
The present study does not consider costs of absenteeism other than production
efficiency. Such costs doubtless exist. The study, however, does indicate that programmes
designed to reduce or anticipate absenteeism can yield substantial financial benefits if
properly directed to certain types of absence, in settings which depend heavily on human
resources.
REFERENCES
ARGYLE et al. (1958). Supervisory methods related to productivity, absenteeism and labour turnover.
Human Relations. 11, 23-40.
BRANDTH, B., RANGNES, K . & AARVIK, K . (1980). Absenteeism as an indicator of work climate problems.
Research Report No. STF82A79009. Trondheim, Norway.
HAMMER, T . H . , LANDAU, J. C. & STERN, R. (1981). Absenteeism when workers have a voice: The case of
employee ownership. Journal of Applied Psychology, 66,561-573.
JOHNSTON, J. (1963). Econometric Methods. New York: McGraw-Hill.
MACY, B. A . & MiRVis, P. H. (1976). Measuring quality of work and organizational effectiveness in
behavioral-economic terms. Administrative Science Quarterty, 21, 212-226.
METZNER, H . P. & MANN, F . (1953). Employee attitudes and absences. Personnel Psychotogy. 6,467-485.
MORGAN, L. G . & HERMAN, J. B. (1978). Perceived consequences of absenteeism. Journal of Applied
Psychology, 61, 738-742.
RONAN, W. W. (1963). A factor analysis of eleven job performance measures. Personnel Psychology, 16,
255-267.
SEASHORE, S., E., INDIK, B. P. & GEORGOPOULOS, B. S. (I960). Relationships among criteria of job
performance. Journal of Applied Psychology, 44, 195-202.
STAW, B. M . (1977). Motivation in organization: Toward a synthesis and redirection. In B. M. Staw & G.
R. Salanck (eds). New Direction in Organizational Behavior. Chicago: St Clair Press.
STAW, B. M . & OLDHAM, G . R. (1978). Reconsidering our dependent variables: A critique and empirical
study. Academy of Management Journal, 21, 539-559.
STEERS, R. M . & RHODES, S. R . (1978). Major influences on employee attendance: A process model. Journal
of Applied Psychology, 63,391-407
TURNER (1960). Dimensions of foreman performance: A factor analysis. Journal of Applied Psychology, 44,
216-223.
WOODWARD, J. (1965). Industrial Organization: Theory and Practice. London: Oxford University Press.
Received 10 January 1983: revised version received 18 August 1984

Вам также может понравиться