Professional Documents
Culture Documents
0022-2380 J3.50
LEX DONALDSON
Australian Graduate School oJ Management, University of New South Wales
ABSTRACT
Addrtss for reprints: Lex Donaldson, Aunralian Graduate School of Management, UniveraityofNew
South Wales, PO Box I, Kensington, NSW 2033, Australia.
LEX DONALDSON
Contingency • Structure .
1. Contingency dctcrminiBm
If
Contingency environment
illiberal
Choice by
dominant Fit Performance
coalition
Change in 4-
Slruclural adjusimenl
contingency
Environment
illiberality
Performance
and divisionalization). Similarly, the rationale for this match, the explanation
of why a particular situation requires a particular structure, is specific to
that pair of contingency-structure variables and is a theory such as is found
in Lawrence and Lorsch (1967) or Blau (1972). Thus SARFIT is a higher level
theory of change in the structure of organizations which subsumes lower
level theories about particular contingency-structure matches. Mismatch
produces a range of dysfunctional behaviours in the organizational system.
These include slow and poor quality decision-making, miscommunication,
demotivation and so on. These in turn lead to low economic performance
such as loss of opportunities to gain new sales, low profitability, low returns
to shareholders and the like. Low performance places pressures on the organiza-
tion and causes the rulers, the dominant coalition, to reorganize. Such critically
low performance occurs more readily in situations where business conditions
are sufficiently stringent to prevent the company having slack resources which
can be used to absorb suboptimal performance without the necessity for change
in structure. Child (1972) has termed this environmental illiberality, and
this includes factors such as the degree of competitiveness (Wilhamson, 1970).
Thus environmental illiberality plays the role of a moderator variable: low
performance leads to structural change only if the environment is illiberal.
When the structural adjustment has been made and match restored performance
rises.
The concept of match will now be examined by considering what constitutes
fu between corporate strategy and structure.
HYPOTHESES
Contingency Determinism
The idea of contingency determinism implies that change in structure in one
time period is the result of changes in strategy, the contingency variable, in
an earlier time period. Here this can be formalized to state:
(1) Increase in degree of diversification in the first period will lead to increases
in degree of decentralization of structure in the second period.
(2) Organizations which are in mismatch at the start of the period are more
likely to change their structure during that period than are those which
are in match at the start.
Further, this shift in structure does constitute adjustment, being towards fit.
Organizations which change only their structure in a decade are likely to move
from mismatch to match. More specifically:
Strategic Choice
Of organizations which move from mismatch to match in a decade, some adjust
only their structure, some only their contingency and some both jointly during
the decade. The hypothesis from the strategic choice model is that:
(4) Of the organizations which move from mismatch to match over a. decade,
as many adjust only their contingency as adjust only their structure.
Functional c (• i
Functional with
subsidiaries c c I i
Product c
divisional 1 c C
Holding i I i 1
company
Companies with sales deriving from one product are appropriately organized
functionally. Since up to 5 percent of the sales in a'single business' may come
from a secondary product line, which may be unrelated to the core, this minor
degree of diversity suggests that a 'functional with subsidiaries' structure also
would be appropriate for some companies classified as single business. The
larger proportionate size of secondary activities, up to 30 per cent of sales, in
a 'Dominant business' makes either a 'functional with subsidiaries' or 'product
divisional' form appropriate. The larger size of the second2iry activities in
the dominant as compared with the single category means that the chance
of their containing a line which is unrelated to the core is increased; hence
the appropriateness of a product divisional structure is increased. Again, a larger
size of secondary activities will tend to increase the number of subsidiaries
existing within a functional with subsidiaries structure. In this type of structure
STRATEGY AND STRUCTURAL ADJUSTMENT 9
(0 eg
E e •5 •g •g u E
E
mis
mis
Itch
GIUI
mis
«
E 1E 1 11 'E 1 E S
Ul<
e u B E o E O
E2 E o E 2 E o E2 s
.s o 2 2 g 2 o 2 2
b b. u. b.
i
be 3
2 -
S o w
CL.
I
STRATEGY AND STRUCTURAL ADJUSTMENT
UJO, 00 — * • —
E E Sb E
u E o C g
o be o fd 4.
So bo
I -
bo -C
u
^
c — c a
-E
n em u
_E
bo n
u
•E
a 0 -a (J
*N
M
"S c(g s >^ >,, a
1*1
M '3 u
o
,OU1
nali
nali
OUI
div ersi fica
u
"3
ure
ure
E
g£ 1 E u u
_c
ate]
div ersi
lEUI
inly
lUl UIO,
onl
u 2 LI u
g
; |
g E 0 O E Id
itch
Itch
div
UIO
div
to:
O
i: ° o
structi
strateg
ems
tch
tch
o be "o aa o c 'o ,^
be 'o 0 o JZ -E
tch
be
c
> •id
c 1 S -^ u u
1
tag.
rce tag.
Kill'
EUI
flllt
rce tag
rce tag
bo *> E E n n
rce tag
a c u c E E E E E U
u 4J
a. a. a. a.
o p o o o o o
in
12 LEX DONALDSON
Despite these anomalies and small-scale variations across the national studies,
there is a high degree of consistency in the findings cross-nationally and
temporally. The results are based on combining data from the five countries
for 14 main analyses. Disaggregating for each country and the two decades,
there are (14 x 5 x 2 = ) 140 analyses. The pattern revealed in the aggregate
holds within each country and decade in 137 of the disaggregated statistics
(table I). In each of the three exceptions, the result arises from only a single
organization and hence reflects small numbers. Thus the aggregate patterns
reflect relationships which hold generally across the five countries and the two
decades. Thus combining data is legitimate, and gives the advantage of larger
numbers for statistical analysis, which yield more robust fmdings.
For the contingency determinism proposition, the requirement for testing
is data on the same organization over two successive time periods, e.g. 1949-59
and 1959-69. This information is available only for the USA (Rumelt, 1974).
Accordingly, the number of organizations used in these analyses will be smaller
and restricted in national origin. Similarly, for investigating performance, data
is available only for the USA, so these analyses are likewise restricted.
In the analysis of fit and performance, performance is measured for a time
period after the period over which fit is measured. In this way the effect of
fit on performance is studied. The causal inference is clear in that fit occurs
before performance, and performance is in the period immediately subsequent to
the fit period. In the performance analysis, an organization is categorized as
being in fit if it is in that state both in 1959 and 1969, and similarly for misfit.
This provides an estimate of chronic fit and misfit in order to maximize the
treatment effect of organization design so that any such effect may be revealed.
The period chosen for the measurement of performance was 1969-71.
Perfom:iance is indexed here by five measures of the rate of growth in nnancial
variables over the years 1969-71. The five variables were: growth in sales,
growth in profit, growth in the ratio of profit to sales, growth in the ratio of profit
to invested capital, and growth in earnings per share. In all cases the growth
measure was calculated by taking the incremental growth over the two years
and dividing by the initial level. For example, sales growth was calculated as:
RESULTS
Contingency Determinism
For the USA only, hypothesis 1 was tested by examining those companies
which diversified or not during the period 1949-59 and seeing whether, or not,
each of those organizations changed its structure towards greater decentralization
in the following 10 years (1959-69). Of the non-diversifying companies
(n = 132), the percentage which changed their structures was 34. The percentage
which diversified and changed structure was 25 (n = 48). Thus the proportion
of structure changes was less among those which diversified than is the
proportion of those which had not diversified. Thus hypothesis 1, based on
contingency determinism, was not supported. These results hold also when
controlling for the possibility of shorter time-lags and under variations in method
(Donaldson, 19B4). Contingency determinism does not predict structural
change.
14 LEX DONALDSON
Strategic Choice
What ofthe idea that there are two routes back towards fit and that organiza-
tions may adjust contingency to structure as well as structure to contingency
(hypothesis 4)? Of those moving from mismatch to match, during either decade,
202 changed their structure only and 13 changed their strategy only (there were
61 organizations which changed both). Thus the ratio of structure-only adjusters
to contingency-only adjusters was 15 to 1. The proportions of structure-only
adjusters and contingency-only adjusters are 73 per cent and 4.7 per cent,
respectively, which are significanUy different from each other (at J(J<.0002, one
tail). The 61 cases where both strategy and structure were changed during the
same decade to regain fit, indicate some latitude for choice, in that by altering
the contingency a structure may have become fitting which would not othewise
have done so. Yet, equally, the move to fit in these cases did require a change
in structure, so in a more general way does involve structural adjustment. These
61 firms constitute perhaps an intermediary position between purely structural
and purely contingency routes to match. They cannot be used to support (he
thesis that re-engineering the contingencies allows structural change to be
avoided. Thus contingency adjustment, in the absence of structural adjustment,
was not a frequently occurring mode of attaining fit. Hypothesis 4 was not
supported.
There is evidence here of a process whereby organizations diversify and thus
move out of fit, then adjust their structure and attain a new match.
Table II. Results of multiple regressions of pcrfonnancc on fit controllin;; for ejects of
strategy, industry and diversifying {n-93)
Notes: For each performance indicator, results are from regressions in which strategy (diversification
in both 1959 and 1969) and industry (science based versus non-science based) controlled
by inclusion. Diversifying has been contrc^ed by excluding any corporations which diversified
between 1959 and I%9,
significant at (p<i. 1) as was growth in profit on invested capital (at p<,06). Thus
there is evidence of an association between fit and the ability to generate profit
from sales and capital subsequently. Hypothesis 5 was confirmed. Fit causes
higher performance and misfit causes lower performance. This supports those
who have argued that fit affects performance (Chiindler, 1962; Child, 1972;
Donaldson, 1982a; Galbraith and Nathanson, 1978; Schreyogg, 1980).
Environmental IlUberality
Theorists have suggested that environment adversity may hasten structural
adjustment, wheras exogenously created slack may allow mismatch to continue
unchecked (Chandler, 1962; Child, 1972; Williamson, 1970) (hypothesis 6).
There is support for this here. Among those organizations which are diversified,
begin the decade in misfit and do not diversify further in the decade, a com-
parison was made between those that retain an inappropriate functional
structure with those which adjust their structure by divisionalizing. By
examining financial performance at the start of the decade, in 1959, some
assessment may be made of whether these firms were operating in muni-
ficient or in competitive environments. In this way the proposition may be tested
that necessary structural adjustment is more likely to occur in competitive than
in munificent environments. This was done by regressing whether or not the
corporation divisionalized between 1959 and 1969 on financial performance
18 LEX DONALDSON
in 1959. Five financial indicators were measured (sales, profit, profit on sales,
profit on invested capital and earnings per share). The analysis was restricted
to corporations which did not diversify further during the decade (n = 26).
The results of the five multiple regressions are given in table III, The negative
associations for sales, profit and earnings per share in 1959 with divisionalizing
were (at/K0.06 or less) significant and conformed to theoretical expectations.
Profit on sales and profit on invested capital in 1959 were positively associated
with divisionalizing, but the associations were not statistically significant. The
correlation between low earnings per share and divisionalizing (r= -0.54) is
particularly striking. These results suggest that it is the cumulative effect of
low sales and profit performance flowing through to low earnings which is
decisive for delivering the shock to the managerial system. They imply that
it is shareholder dislike of low earnings which may be the key process in
activating the superordinate formal authority of the board of directors to
intervene in the executive management, This helps to explain why the lags
and structural adjustment to strategy can be so long. Thus hypothesis 6 receives
support.
Both structural mismatch and environmental illiberality cause low perfor-
mance and thus jointly trigger structural adjustment. Thus the SARFIT model
is supported that posits diversification"* misfit"* lower performance-* structtiral
adjustment-* fit-* higher performance.
DISCUSSION
Notes: ' Corporations which diversified between 1959 and 1969 have been excluded,
Divisionalizing is defmed as moving from a functional to a multidivisional structure
between 1959 and 1969.
STRATEGY AND STRUCTURAL ADJUSTMENT 19
here for the idea that structurzil change is am automatic response to a change
in the contingency variable. Contingency fit offers an explanation of these
organizational changes. Changes in the contingency variable lead to changes
in structure, but the actual process of occurrence is via the cycle specified by
the SARFIT model that changes in contingency produce mismatch and lower
performance which lead eventually to structural adjustment. Thus the present
enquiry vindicates those who have distinguished the long-run contingency fit
from the shorter-run contingency determinism models (Child, 1972; Donaldson,
1982a; Schreyogg, 1980). And it attests to the former as a more accurate
depiction of the actual causal processes. The elongated causal chain of misfit-*
poor performance-^ adjustment-* fit articulated by Child (1972) in the strategic
choice thesis is to a large degree corroborated. However, the path of adjustment
is less open than stated there.
The adjustment is overwhelmingly of structure to contingency, rather than
of contingency to structure. Few organizations adjust contingencies to structure.
Thus, as a generalization, the statement that organizations adjust structure
to contingencies is true sufficiently often for this to be a useful summary of
the main thrust. Contingencies were shown to be relatively fixed here, so that
once a company had diversified it needed to adjust its structure.
Thus the present enquiry rejects both contingency determinism and strategic
choice, and supports the structural adjustment to regain fit (SARFIT) model. The
prime mover is an initial change in the contingency variable which disequilibriates
the organization moving it out of fit into mismatch, and then the structure is
adjusted to attain a new fit (figure 5). Structural change is a response to misfit
which in turn is driven by contingency change. However, not every change in the
contingency is predicated to produce organizational change. The key is shifts
which create lack of fit. Contingency change causes departures from equilibrium,
and this provides the pressure for structural accommodation. Thus equilibrium
analysis is vindicated in the study of organizational structure. A continuing role
for choice over matters of organization design may be present in that the dominant
coalition may still exercise a degree of selectivity in courses of action, most notably
in affecting the rate at which structurtd adjustment is made. A summary overview
of the differences between the three formulations of contingency theory (con-
tingency determinism, SARFIT and strategy choice) is provided in table IV.
By demonstrating that shifts in contingency lead to structural changes, the
model is broadly congruent with arguments which emphasize the inexorable
pressures of the situation (Lawrence and Lorsch, 1967; Schreyogg, 1982;
Woodward, 1965). The emphasis on contingencies as imperatives towards
structural accommodation is revealed as sound if understood as referring to long-
run tendencies in organizations. The model proposed here offers a coherent inter-
pretation of the traditional juxtaposition of arguments by contingency researchers
(Lawrence and Lorsch, 1967; Woodward, 1965) that managers need consciously
to attend to mismatches and adjust structure to contingency, and that structural
change is a response to a change in contingency (Schreyogg, 1980, 1982).
20 LEX DONALDSON
Diversifying Divisionalization
Environmental
illiberatitv
Performance
AspKi Thtoty
Noter ' Dominant coalition is the group with power over the organization (see Child, 1972).
position taken here is also broadly in accord with Aldrich (1979) and Hage
(1980) who both argue that the situation constrains the discretion of senior organiza-
tional management, and that consequently there is less freedom to choose struc-
ture than Child (1972) implies.
STRATEGY AND STRUCTURAL ADJUSTMENT 21
CONCLUSIONS
REFERENCES
CHANNON, D . F . (1978). The Service Industries: Strategy, Structure and Financial Performance,
London and Basingstoke: Macmillan.
CHENHALL, R . H . (1979). 'Some elements of organizational control in Australian
divisionalized firms'. Australian Journal of Management, supplement to 4, 1, 1-36.
CHILD, J. (1972). "Organizational structure, environment and performance: the role
of strategic choice'. Sociology, 6, 1, 1-22.
C H I L D , J . (1984). Organization: A Guide to Problems and Practice {2nd Ed.). London: Harper
and Row.
DEWAR, R . and HAGE, J. (1978). 'Size, technology, complexity and structural differen-
tiation: toward a theoretical synthesis'. AdministrativeScierue Quarterly, 23, 1, 111-36.
DILL, W . R . (1958). 'Environment as an influence on managerial autonomy'.
Administrative Scierue Quarterly, 2 , 3 , 4 0 9 - 4 3 .
DONALDSON, L . (1979a). T h e transition from functional system to multidivisional
structure: a critical analysis'. In McLennan, R., Smith, D., Inkson, K. and
Marsh, N . (Eds.), People in Organizations: Studies in Australia and New Zealand, Depart-
ment of Business Administration, Victoria University of WeJlington, Wellington,
New Zealand.
DONALDSON, L . (1979b). 'Regaining control at Nipont'. Journal of General Management,
4, 14-30.
DONALDSON, L. (1982a). 'Comments on "Contingency and choice in orgeinization theory"'.
Organization Studies, 3 , 1, 6 5 - 7 2 .
DONALDSON, L . (1982b). 'Divisionalization and diversification: a longitudinal stud/.
Academy of Management Journal, 2 5 , 4, 909-14.
DONALDSON, L . (1982C). 'Divisionalization and size: a theoretical and empirical critique'.
Organization Studies, 3 , 4, 321-37.
DONALDSON, L . (1983). 'Organizational growth, the M-form and performance: a critique
of the thesis of Williamson'. Working Paper 83-041, Australian Graduate School
of Management.
DONALDSON, L. (1984). 'Explaining structural change in organizations: contingency deter-
minism or contingency-fit'. Australian Journal of Management, 9, 2, 15-24.
DONALDSON, L . (1985a). In Defence of Organization Theory: A Reply to the Critics, Cambridge:
Cambridge University Press.
DONALDSON, L . (1985b). 'Organization design and the tife-cyclcs of products'. Jouma/
of Management Studies, 2 2 , 1, 25-37.
DONALDSON, L . (1986a). 'The interaction of size and diversification as a determinant
of divisionalization: Grinyer revisited'. Organization Studies, 7, 4, 369-81.
DONALDSON, L . (1986b). 'Divisionedization and size: a reply to Grinyef. Australian Journal
of Management, H , 2 (in press).
DYAS, G . P . a n d THANHEESER, H . T . (1976), TTu Ernerging European Enterprise - Strategy
and Structure in French and German Industry. London: Macmillan.
EcELHOFF, W. G. (1982). 'Strategy and structure in multinational corporations: an
information-processing approach'. Administrative Scierue Quarterly, 27, 3, 435-58.
GALBRAITH, J . R . and NATHANSON, D . A. (1978). Strategy Implementation: The Role of
Structure and Process. St. Paul Minnesota: West.
GRINYER, P. H. (1982). 'Discussion note: divisionalization and size: a rejoinder'.
Organization Studies, 3 , 4, 339-50.
GRINYER, P. H. and YASAI-ARDEKANI, M . (1981). 'Strategy, structure, size and
bureaucracy'. Academy of Management Journal, 24, 3, 471-86.
GRINYER, P. H., YASAI-ARDEKANI, M . and AL-BAZZAZ, S. (1980). 'Strategy and structure,
the environment and financial performance in 48 United Kingdom companies'.
Academy of Management Journal, 2 3 , 193-220.
H A G E , J . (1980). Theories of Organizations: Form, Process and Transformation. New York:
24 LEX DONALDSON