Description
Asset And Cost Retrenchment In Turnaround Strategies A Large-sample Study Of Corporate Responses To The Asian Crisis In Singapore
1
Asset and Cost Retrenchment in Turnaround Strategies –
A Large-Sample Study of Corporate Responses to the
Asian Crisis in Singapore
Alexander D. Falkenberg
University of St. Gallen
Institute for International Management – Asia Research Centre
Dufourstrasse 40a
CH-9000 St. Gallen, Switzerland
Wharton-Singapore Management University Research Centre
469 Bukit Timah Road, Eu Tong Sen Building
Singapore 259756, Singapore
[email protected]
Li-Choy Chong
University of St. Gallen
Dufourstrasse 40a
CH-9000 St. Gallen, Switzerland
[email protected]
Pascal P. Prinz
University of St. Gallen
Dufourstrasse 40a
CH-9000 St. Gallen, Switzerland
[email protected]
Acknowledgement
We thank Helena A. Glamheden (University of St. Gallen & Wharton-SMU Research
Center), Markus Fröhlich (University of St. Gallen) and Jerel Gareth Lee Chye Hock
(Singapore Management University) for support and helpful discussions. We
thankfully acknowledge the Wharton-Singapore Management University Research
Center, Singapore, for organizational support. We would also like to express deep
gratitude to the Swiss National Foundation for financially enabling the research
project.
2
Abstract
The role of retrenchment is the most heatedly and controversially debated element of
turnaround management for a Western environment and the debate becomes even
more ambiguous when an Asian perspective is taken.
Since preceding studies have mainly neglected the dualism of industry-contraction
based and firm-based decline situations, we take the Asian crisis as a prominent
example of a temporary, but stiff contraction across industries to inquire the role of
retrenchment strategies in response to such a crisis. We focus our analysis on
Singapore and utilize a triangulation of quantitative analyses of 134 companies listed
on the Singaporean stock exchange.
We conclude that both asset and cost retrenching companies face lower overall
performance levels throughout the crisis situation in Singapore than non-retrenching
firms, which indicates that the degree of distress and urgency in the turnaround
situation moderates the degree of asset and cost retrenchment.
We furthermore find differing effects on performance from asset and cost
retrenchment. Although the impact of asset retrenchment on company performance
in a crisis situation in Singapore is non-significant, the impact of cost retrenchment on
company performance in a crisis situation in Singapore is positive, however, limited.
Our findings suggest that the reduction of costs in a crisis situation might foster
company performance, whereas the reduction of assets has no performance impact.
3
Introduction
A long tradition of Western-based turnaround literature has developed a rather solid
body of knowledge of turnaround for a Western, mainly US-based environment.
Despite the large amount of academic work on turnaround, however, the role of
retrenchment is still a heatedly debated element of turnaround for a Western
environment. Retrenchment “entails deliberate reductions in costs, assets, products,
product lines and overhead” (Pearce and Robbins, 1993) and past studies on a
Western environment have produced differing results: Pearce and Robbins (1992;
1994) conclude that retrenchment is indispensable for a successful turnaround, while
Barker and Mone (1994) stress the importance of a strategic reorientation in
achieving turnaround.
The debate of the role of retrenchment becomes even more ambiguous when an
Asian perspective is taken: During the Asian crisis starting with the severe
devaluation of the Thai Baht on July 1
st
, 1997 it became apparent that trying to apply
turnaround concepts that proved successful in a Western environment to South-East
Asian companies generated dissatisfactory results and demonstrated the limited
applicability of Western methods.
Initial research on turnaround in South-East Asia has analytically reconfirmed the
limited applicability of Western findings in South-East Asia and has made first
inquiries into the turnaround process and turnaround strategies in South-East Asia
mainly based on qualitative data and focusing on Chinese-owned companies (Bruton,
Ahlstrom and Wan, 2001; 2003; Falkenberg, 2003; Falkenberg and Chong, 2004a,b).
Even the limited research done on South-East Asia already demonstrates
contradictious findings on the role of retrenchment. Whereas the general notion in
South-East Asia is that retrenchment is avoided and inappropriate for cultural
reasons (Bruton, Ahlstrom and Wan, 2001; Falkenberg, 2003), an initial analysis of
quantitative data on Chinese-owned businesses revealed the fact that where
retrenchment was implemented in turnaround situations, it resulted a positive effect
on performance (Bruton, Ahlstrom and Wan, 2003).
Given that only a few research studies on firm turnaround for Asia have been
conducted so far and that these studies already reveal equivocal findings of
retrenchment in firm turnaround, a large-sample study is conduced to contribute to
4
the clarification of the role of retrenchment. Such a large-sample study can validate
or reject previous findings on a quantitative basis and further develop the role of
retrenchment in turnaround strategies.
Acknowledging the differing structural and analytical situations of firm-based decline
and industry-contraction based decline, it is of particular interest to investigate
whether the role of retrenchment is different for the two logical situations as
suggested in preceding studies (Barker and Duhaime, 1997; Falkenberg and Chong
2004 a,b). Therefore, we take the most prominent example of an industry-contraction
based decline situation – a widespread economic crisis across all industries – as
reference point for our analysis. At the same time, we take a geographic focus rather
than the view on Chinese-owned businesses, which has been the focus in preceding
articles (Bruton, Ahlstrom and Wan, 2001, 2003). Taking a geographic focus on
Singapore allows us to assess the role of retrenchment for the specific legal, social
and (multi)cultural environment of Singapore.
Our contribution seeks to find, what the impact of both asset and cost retrenchment is
on company performance in a financial crisis situation in Singapore. Our investigation
of one of the most developed markets in South-East Asia during the Asian financial
crisis examines the influence of asset- and/or cost-retrenchment on performance and
consequently, the general significance of retrenchment as viable strategy in
improving performance in an Asian setting. The examined sample is not restricted to
firms that faced a severe decline in 1997 and the following years but studies whether
firms saw asset and/or cost retrenchment as a viable option for improving
performance being faced with a financial crisis situation.
In line with preceding studies on turnaround management, we take a strategic
management perspective on turnaround. Our perspective assumes that the
strategies implemented at the top of organizations translate into the observed
strategies. Our view, therefore, abstracts from the obstacles in organizational
processes, which relate observed and intended strategies and is thus not in the
tradition of an organizational behavior view.
Our article is organized in the following way: We review the literature on turnaround
management with a focus on retrenchment in the following section to develop
hypotheses on the role of retrenchment in turnaround endeavors. We base our
5
hypotheses on preceding studies investigating both the West and South-East Asia,
although the role of retrenchment in South-East Asia might be different from the
Western view, Western findings are the most accurate information available on the
role of retrenchment. Following the literature review, we present our methodology.
We triangulate our findings based on three methodologies: mean comparison (t-test),
regression and correlation analysis. Thereafter, we present and discuss the findings.
The last section provides conclusions of our research contribution to the ongoing
discussion of the role of retrenchment in turnaround strategies.
Literature Review
Multiple researchers in the field of turnaround management emphasize the
importance of asset- and/or cost retrenchment for achieving turnaround success (e.g.
Hofer, 1980; Bibeault, 1999; Robbins and Pearce, 1992).
Correctly identifying the cause of decline allows the turnaround response to be
tailored to the specific problem. An early perspective on turnaround strategies
dichotomized the cause of a firm’s decline into internal and external sources of
decline (Schendel and Patton, 1976; Schendel, Patton and Riggs, 1976).
Correspondingly, two types of turnaround strategies are distinguished: operating vs.
strategic turnaround strategies (Hofer, 1980). Strategic turnaround strategies should
be used to solve external problems, while operating strategies should be applied in
the case of internal problems. Operating strategies focus on improvement of firm
efficiency and therefore are closely related to retrenchment of non-performing assets
and overly high cost factors. Both types of turnaround responses, however, may
include asset- and cost-cutting elements, which are assumed to positively influence
performance if closely tied to the assessment of current operating and strategic
health of the firm (Hofer, 1980).
More recently, the cause of decline is dichotomized in firm-based and industry-
contraction based declines (Arogyaswamy, Barker and Yasai-Ardekani, 1995; Barker
and Duhaime, 1997, Falkenberg and Chong 2004a). Subsequently, the cause of
decline is found to decisively influence whether a company may achieve turnaround
through strategic renewal. If the decline is firm-based, a relatively high level of
strategic change is involved (Falkenberg and Chong 2004a). Conversely, an industry
contraction-based and cyclical decline involves relatively less strategic change and
may require capacity adjustments (Falkenberg and Chong, 2004a).
6
The need for retrenchment in the short-term oriented starting phase of the turnaround
process, i.e. the negotiation and decline-halting phase in Asian turnarounds
(Falkenberg and Chong, 2004b)
1
, is not dependent on the source of decline.
Retrenchment in the initial phase of the turnaround process varies depending on (a)
the severity of decline of the firm and (b) the availability of slack resources (Hofer,
1980; Hambrick, 1985; Arogyaswamy, Barker and Yasai-Ardekani, 1995, Falkenberg
and Chong, 2004a). If the decline is severe, however, the business of the troubled
firm is viable, but the survival of the firm is immediately endangered, a short-term
oriented retrenchment exercise is implemented independent of the long-term
recovery strategy. In line with the proposed relationship of financial pressure and
retrenchment for the initial phase of the turnaround process (Falkenberg and Chong,
2004b), we propose that the degree of retrenchment is related to the degree of
financial pressure in firms facing a crisis situation:
Hypothesis 1: Asset retrenching companies face lower overall performance levels
throughout the crisis situation in Singapore than non-asset retrenching firms.
Hypothesis 2: Cost retrenching firms face lower overall performance levels
throughout the crisis situation in Singapore than non-cost retrenching firms.
Retrenchment in the long-term oriented revitalization phase of the turnaround
process, however, is differently assessed in a Western environment by two opposing
groups of researchers: While one group of researchers focuses on the recovery
response, i.e. the change in company strategy (Barker and Duhaime, 1997; Barker
and Mone, 1994), another group of researchers finds retrenchment to be an integral
part of successful turnarounds (Robbins and Pearce, 1992; Pearce and Robbins,
1994). Latter conclude from their large-sample study that retrenchment is a critical
strategic element in attaining turnaround success (Robbins and Pearce, 1992). They
conclude that the degree of turnaround success is positively related to the level of
retrenchment and the association between the degree of asset retrenchment and
turnaround performance is more pronounced in severe turnaround situations.
The proposition of the relevance of retrenchment has raised methodological and
theoretical criticism by Barker and Mone (1994) denying a vital role of retrenchment
in facilitating recovery and proposing the role of strategic turnarounds and strategic
reorientation (Barker and Mone, 1994; Barker and Duhaime, 1997).
7
The impact of retrenchment on turnaround success in South-East Asia is even more
ambiguous. On the one hand there is empirical evidence for retrenchment to
positively influence turnaround success for Chinese-owned firms (Bruton, Ahlstom
and Wan, 2003) supporting the argument put forward by Robbins and Pearce (1992).
On the other hand, there is a notion of negative retrenchment effects in a South-East
Asian environment for cultural reasons (Bruton, Ahlstrom and Wan, 2001, Falkenberg,
2003) and the proposition of a limited ability to retrench due to the Asian operating
environment and already slim Asian organizations faced with comparably low human
resources costs (Bruton, Ahlstrom and Wan, 2001). Since the applicability of the
equivocal and controversial findings to a Singaporean environment is unknown, we
propose two competing hypothesis for asset and cost retrenchment respectively:
Hypothesis 3a: The impact of asset retrenchment on company performance in a
crisis situation in Singapore will be positive.
Hypothesis 3b: The impact of asset retrenchment on company performance in a
crisis situation in Singapore will be negative or non-present.
Hypothesis 4a: The impact of cost retrenchment on company performance in a
crisis situation in Singapore will be positive.
Hypothesis 4b: The impact of cost retrenchment on company performance in a
crisis situation in Singapore will be negative or non-present.
Methodology
The structure of our hypothesis indicates that we are seeking to identify the
relationship of two continuous independent variables, asset and cost retrenchment,
on one dependent continuous variable, company performance. We triangulate our
findings based on a data sample of listed Singaporean companies.
Operationalization of Independent and Dependent Variables
The independent variables for our analysis are asset and cost retrenchment. We
composed asset and cost retrenchment based on balance sheet items to construct a
measurement of changes in assets and costs of the Singaporean listed firms
appropriately.
8
Past large-sample studies commonly employ financial ratios to examine efficiency
gains interpreted as the result of retrenchment. Such measures comprise for example
of cost of goods sold/sales (Schendel and Patton, 1976), receivables/sales, R&D
expenditure/sales and increased sales per employee (Hambrick and Schecter, 1983).
Increases in the ratios were interpreted as efficiency gains based on asset and cost
retrenchment. Such study design has raised criticism since all of these ratios include
sales in the denominator and are therefore not only dependent on the numerator but
also on sales. Therefore, improvements in the ratios can both be a result of the
decrease of the numerator or increases in sales, which makes an interpretation of
such ratios in light of retrenchment very difficult. It can thus not be determined
unambiguously whether efficiency gains result from sales gains, retrenchment or both
(Arogyaswamy, Barker and Yasai-Ardekani, 1995; Barker and Duhaime, 1997).
In order to avoid the dependency on sales, we model retrenchment independent of
sales in absolute numbers. Consequently, we define assets and asset retrenchment
as follows:
Assets Cash & Due from Banks + Receivables + Inventory +
Property, Plants, and Equipment (PPE)
Asset Retrenchment - ? (Cash & Due from Banks + Receivables +
Inventory + Property, Plants, and Equipment (PPE))
= Assets
t
– Assets
t-1
Assets are measured as the sum of cash & due from banks, receivables, inventory,
and property, plants and equipment (PPE). The variable asset retrenchment is
defined as the (negative) delta of the sum of cash & due from banks, receivables,
inventory, and property, plants and equipment (PPE).
Similarly, costs are measured as the sum of operating expense and total interest
expense. The variable cost retrenchment is defined as the (negative) delta of the sum
of operating expense and total interest expense.
Costs Operating Expense + Interest Expense Total
Cost Retrenchment - ? (Operating Expense + Interest Expense Total)
= Costs
t
– Costs
t-1
The dependent variable is company performance. Most past studies use return on
investment (ROI) or a similar indicator as performance measure (e.g. Robbins and
9
Pearce, 1992). Following these studies, we employ return on total assets (ROA) as a
measure of performance during the period of investigation.
Financial measures like ROI, ROA, and return on sales (ROS) are widespread and
frequently used indicators, which deliver reliable information about actual
performance (Bernard, 2000).
Thus, performance and performance change are measured as follows:
Performance 100 * ROA
? Performance Performance
t
– Performance
t-1
Data Sample
The data source from which the dependent and the two independent variables were
drawn is Standard & Poor’s COMPUSTAT Global data base. COMPUSTAT reports
418 companies for Singapore. Due to missing data, only 134 companies could be
included in the sample. Data was initially collected for the years 1995 to 2002. The
period of time that was later investigated was restricted to the years 1996 to 1999 to
avoid a blurring of the effects of the Asian financial crisis with other factors (e.g.
downturn of world economy after the collapse of the Internet bubble in 2000). Pre-
crisis year, year of crisis, and 2 years of potential recovery were therefore a plausible
period of observation.
Assuming that the Asian financial crisis led to industry contraction for nearly all
companies, we examined the financial situation of the companies before 1997. We
evaluated an indicator of potential bankruptcy to confirm our assumption: A widely-
used predictor of corporate bankruptcy in turnaround studies is Altman’s Z-score,
(Altman 1968, 1971). Altman uses five financial ratios
2
and combines this set in a
discriminant analysis approach. A Z-score of more than 2.99 indicates a financially
healthy firm; below 1.8 suggests immediate danger, while a Z-score of 1.8 to 2.99 is
a warning sign of financial distress.
3
Our analysis of the 134 companies in our sample revealed that the number of firms
with Altman’s Z of a score lower than 2.99 increases dramatically after the Asian
crisis hit in 1997 reaching a peak of 112 distressed firms in 1998 (Table 1). Since it is
unknown to what degree the health of the firms with an Altman’s Z higher than 2.99 is
dependent on the implementation of retrenchment strategies, we analyzed both
groups of firms with respect to the performance impact of retrenchment.
10
Table 1: Number of Companies and Altman Z-score
Z-score ? 2.99 Z-score < 2.99
1995 64 70
1996 53 81
1997 28 106
1998 22 112
1999 38 96
We classify firms as asset/cost retrenching firms if assets/costs decreased by more
than 5% from 1996 to 1999. If a firm’s assets/costs diminished by less than 5% or did
even increase between 1996 and 1999, the firm was classified as non-asset
respectively non-cost retrenching firm. Table 2 shows our sample along the resulting
four categories: (1) firms that were found to retrench both in assets and costs, (2)
firms that did retrench in assets but not in costs, (3) firms that did retrench in costs
but not in assets and (4) firms that did not retrench at all.
Table 2: Classification of Asset and Cost Retrenching Firms
Cost retrenching firms Cost non-retrenching firms
Asset retrenching firms 60 20
Asset non-retrenching firms 21 33
Only 33 companies of the total number of 134 did not retrench at all. The majority of
firms retrenched both assets and costs, and about the same number of firms did
utilize either cost or asset retrenchment.
In order to assess the structure of the industries of the firms in our sample, we
utilized the categories used by the Singapore Stock Exchange (Singapore Exchange,
2004) (Table 3 and Figure 1). More than one third of the companies in the sample
operated in a manufacturing industry. The second most important sector was
commerce with 16%. The share of multi-industry, hotels & restaurants, properties,
construction, transport/ storage/ communication (TSC), and services was
approximately equal, ranging from 6% to 10%. Just one company in the sample
operated in the finance sector, which percentage share in the sample is therefore
less than 1%. No companies came from the loans & debentures, electricity/ gas/
water, CLOB (Central Limit Order Book) international, agriculture and
mining/quarrying sector.
11
In order to obtain deeper
insights into the composition
of our sample with relation to
the total structure of the
listed Singaporean firms, it
was necessary to investigate
how many of all companies
of a sector (listed at
Singapore Stock Exchange)
were part of our sample. The
percentage share of the
companies in a specific
industry in relation to the total number of companies listed in this sector was
computed and is presented in Figure 2.
The analysis revealed that the multi-industry and hotels & restaurants sector are
overrepresented in our sample, while the services and particularly the finance sector
are underrepresented. 67% of all companies in the multi-industry sector respectively
60% in the hotels & restaurants sector were part of our sample. Conversely, only
8.7% of all services companies and a meager 2.4% of all finance companies were in
our sample. The other sectors namely commerce (22%), manufacturing (23.7%),
properties (26.5%), construction (22%), and TSC (27.8%) were represented in a
balanced way in the sample. The non-inclusion of sectors like agriculture or
mining/quarrying is less crucial in view of the small number of companies listed in
those sectors except for loans & debentures, a sector with 75 companies totally.
Table 3: Industry Classification
Multi-Industry 14
Commerce 21
Finance 1
Manufacturing 50
Hotels & Restaurants 12
Properties 9
Construction 9
Transport, Storage, Communication 10
Services 8
Figure 1: Percentage of Industries in Sample
Multi-Industry
10%
Commerce
16%
Finance
1%
Manufacturing
37%
Hotels &
Restaurants
9%
Properties
7%
Construction
7%
TSC
7%
Services
6%
12
Findings
Hypotheses 1 & 2:
Analysis t-tests
The first and second
hypotheses state
that asset
retrenching companies as well as cost retrenching firms face lower overall
performance levels throughout the crisis situation in Singapore as compared to non-
asset and non-cost retrenching firms.
Accordingly, the mean ROA of asset cost retrenching firms respectively should be
found to be significantly lower than that of non-retrenching companies. To examine
this relationship, mean ROA of asset and non-asset retrenching firms was compared
in pair wise t-tests over the period of observation between 1996 and 1999. The same
was done for cost and non-cost retrenching firms (see Table 4 and 5).
Table 4: Mean Performance Comparison between Asset Retrenching and Non Asset Retrenching
Companies
Mean
performance
Asset retrenching
companies
Non Asset retrenching
companies t-value p-value
1996 2.40675867 7.27142956 -3.9856918
***
0.000
1997 0.4355338 6.12316862 -3.52055495
***
0.000
1998 -2.17733058 4.70841509 -4.00419641
***
0.000
1999 -1.11679691 5.30690223 -3.66902157
***
0.000
avg ROA 96-99 -0.11295875 5.85247888
No. of companies 80 54
*** significant at the 1% level, ** significant at the 5% level, * significant at the 10% level
Definitions: Asset retrenching firms reduced assets by more than 5% between 1996 and 1999; Non asset retrenching firms
reduced by less or increased assets between 1996 and 1999.
Table 5: Mean Performance Comparison between Cost Retrenching and Non Cost Retrenching
Companies
Mean
performance
Cost retrenching
companies
Non Cost retrenching
companies t-value p-value
1996 3.03769277 6.39895804 -2.66996136
***
0.004
1997 1.01411379 5.3462376 -2.64480988
***
0.004
1998 -0.96778836 3.08915686 -1.93836691
*
0.026
1999 0.47563784 3.02442901 -1.12081277 0.131
avg ROA 96-99 0.88991401 4.46469538
No. of companies 81 53
Figure 2: Firms in Sample as Percentage of Total
Number of Companies
0,0%
15,0%
30,0%
45,0%
60,0%
75,0%
M
u
l
t
i
-
I
n
d
u
s
t
r
y
C
o
m
m
e
r
c
e
F
i
n
a
n
c
e
M
a
n
u
f
a
c
t
u
r
i
n
g
H
o
t
e
l
s
&
R
e
s
t
a
u
r
a
n
t
s
P
r
o
p
e
r
t
i
e
s
C
o
n
s
t
r
u
c
t
i
o
n
T
S
C
S
e
r
v
i
c
e
s
Average: 28.9 %
13
*** significant at the 1% level, ** significant at the 5% level, * significant at the 10% level
Definitions: Cost retrenching firms reduced costs by more than 5% between 1996 and 1999; Non cost retrenching firms
reduced by less or increased costs between 1996 and 1999.
The magnitude of the results for mean performance in each group seems plausible.
Both the maximum and the minimum mean ROAs are realistic, as well as the
average ROAs between 1996 and 1999. Moreover, a characteristic pattern in the way
the mean ROAs develop over time is identified (see Figure 3).
Indisputably, 1996 represents the year with the highest mean ROA among all groups.
With the beginning of the Asian financial crisis in mid-1997 a clear-cut downward
trend starts, leading to a minimum in mean ROAs in 1998. An upswing begins in
1999 with the subsiding of negative effects of the crisis. The results are therefore
compatible with the chronological sequence of events during the Asian financial crisis.
The results of the
t-tests for asset
and non-asset
retrenching firms
are unambiguous:
The differences of
mean performance
between asset and
non-asset
retrenching firms
are highly significant for all years. Non-asset retrenching firms achieve an average
ROA of 5.85 between 1996 and 1999, while asset retrenching firms in contrast are
confronted with a negative average ROA of -0.11. Hypothesis 1, therefore, receives
strong support.
The results for cost and non-cost retrenching firms are not as clear as those for asset
retrenchment, but similar. The t values for 1996 and 1997 are -2.67 and -2.65 hence
significant at the 1% level. For 1998 the t value of -1.94 is almost significant at the
5% level. The only year with a clearly non-significant difference of mean ROA
between cost retrenching and non-cost retrenching firms is 1999 with a t value of -
1.12. Non-cost retrenching firms experienced an average ROA of 4.47 between 1996
and 1999, while cost retrenching firms were challenged by a low average ROA of
Figure 3: Time Pattern of Development in Mean ROAs
-4
-2
0
2
4
6
8
1996 1997 1998 1999
Years
M
e
a
n
R
O
A
Asset retrenching
companies
Non asset
retrenching
companies
Cost retrenching
companies
Non cost
retrenching
companies
14
0.89. These findings represent support for Hypothesis 2 in spite of the non-significant
t value for 1999.
Analyses of Hypotheses 3 & 4: Regression model
The previous analysis showed differences between companies that retrenched
assets and/or costs during the whole period. In a next step, we examine in more
detail the relationship between ROA and assets and costs at a particular point in time
using a linear regression model:
ROA
i,t
= alpha + beta1 * asset
i,t
+ beta2 * cost
i,t
+ epsilon
i,t
(1)
where t denotes time and i is an indicator of the company.
Three regressions were conducted. Firstly, equation (1) was estimated by ordinary
least-squares (OLS) with robust standard errors. Secondly, equation (1) is estimated
as a random effects specification, where epsilon
i,t
consists of a company specific
time-invariant error term and a time-varying error term. Third, the regression equation
is estimated in first differences:
?ROA
i,t
= alpha + beta1 * ?asset
i,t
+ beta2 * ?cost
i,t
+ ?epsilon
i,t
(2)
which allows to eliminate a time-invariant error term.
All regressions were performed with EViews 4.1 for the years 1996 to 1999. The
results for regression equation (2) are presented in table 6.
The method of OLS was used to produce estimates that were the best linear
unbiased estimates under classical assumptions (von Auer, 2003; Gujarati, 2003).
The model yielded an R-square of 0.001, which indicates a very weak fit between the
model and the data. The F value for the overall model was 0.222, which has a
significance level of 0.801. Both independent variables were non-significant with a t
statistic of 1.139 (p-value: 0.255) for asset retrenchment and a t value of -1.209 (p-
value: 0.227) for cost retrenchment. The result did not change when the regression
was repeated with only one independent variable, i.e. a regression of performance on
asset retrenchment and a regression of performance on cost retrenchment
individually.
Table 6: Regression of ?ROA on ?Asset and ?Cost
15
Dependent Variable: ?ROA
Method: Least Squares
Sample: 135 536
Included observations: 399
Excluded observations: 3
Newey-West HAC Standard Errors & Covariance (lag truncation=5)
Variable Coefficient Std. Error t-Statistic Prob.
C -0.966518 0.445953 -2.167309 0.0308
?ASSET 0.001003 0.000881 1.138972 0.2554
?COST -0.001212 0.001002 -1.209146 0.2273
R-squared 0.001121 Mean dependent var -0.977738
Adjusted R-squared -0.003924 S.D. dependent var 9.508195
S.E. of regression 9.526831 Akaike info criterion 7.353592
Sum squared resid 35941.16 Schwarz criterion 7.383584
Log likelihood -1464.042 F-statistic 0.222232
Durbin-Watson stat 2.023429 Prob(F-statistic) 0.800829
?ROA = ROA
t
- ROA
t-1
, ?asset = asset
t
- asset
t-1
, ?cost = cost
t
- cost
t-1
The finding of non-significant effects may be due to the large loss of information by
taking first differences. Estimates of the regression equation (1) in levels are given in
table 7.
Table 7: Regression of ROA on Asset and Cost
Dependent Variable: ROA
Method: Least Squares
Sample: 135 670
Included observations: 534
Excluded observations: 2
Newey-West HAC Standard Errors & Covariance (lag truncation=5)
Variable Coefficient Std. Error t-Statistic Prob.
C 1.792712 0.494422 3.625871 0.0003
ASSET -0.000305 0.000166 -1.841946 0.0660
COST 0.002323 0.000674 3.448921 0.0006
R-squared 0.010268 Mean dependent var 2.305773
Adjusted R-squared 0.006540 S.D. dependent var 10.38383
S.E. of regression 10.34981 Akaike info criterion 7.517416
Sum squared resid 56879.99 Schwarz criterion 7.541463
Log likelihood -2004.150 F-statistic 2.754500
Durbin-Watson stat 1.996541 Prob(F-statistic) 0.064550
ROA = ROA
t
, asset = asset
t
, cost = cost
t
The model yielded an R-square of 0.01, indicating again that only a small part of
movements in ROA may be explained by the two independent variables asset and
cost. The F value for the overall model was 2.755 which correspond to a significance
level of 0.065. The F value is therefore marginally significant at about the 10% level.
In contrast to the first regression, we can reject the null hypothesis that the two
regressors have no impact, notwithstanding the fact that the R-square is only 0.010.
16
The independent variable asset was marginally significant with a t value of -1.842 (p-
value: 0.066) at the 10% level. Cost was significant with a t value of 3.449 (p-value:
0.0006) at the 1% level.
The significant regressor cost tends to positively influence ROA, contrary to the
impact stated in Hypothesis 4b and supporting Hypothesis 4a.
How strong is this positive influence of cost on performance? The 81 cost retrenching
companies of our sample did retrench on average by -26.39 between 1996 and 1999.
Using the coefficient of cost of 0.002323 from our model, an average performance
(i.e. ROA) improvement of 0.0613 percentage points results. In contrast, the 53 non-
cost retrenching companies faced on average increasing costs of 60.12 between
1996 and 1999. Again using the results of our regression, an average performance
deterioration of 0.1397 percentage points results.
The independent variable asset has a negative sign, which opposes Hypothesis 3a.
The 80 asset retrenching companies of our sample retrenched on average by -46.74
between 1996 and 1999, while the 54 non-asset retrenching companies increased
their assets by 65.56. From both the small coefficient of asset of -0.000305 and the
fact that the variable is only marginally significant, it may be inferred that the impact
of asset reductions on performance is very small, if at all existing. Therefore,
Hypothesis 3b receives some support.
In the next step the regression was repeated twice, in each case just with one
independent variable as regressor (see Table 8 and 9).
In the regression of performance on asset, the independent variable asset is again
marginally significant with a t value of 1.761 (p-value: 0.079) at the 10% level. The F
value for the model was 1.176 which had a significance level of 0.279. The coefficient
of asset is comparable in magnitude to the result of the preceding regression with two
independent variables, but the sign has changed from negative to positive. This
indicates a correlation between asset retrenchment and cost retrenchment.
Table 8: Regression of ROA on Asset
Dependent Variable: ROA
Method: Least Squares
Sample: 135 670
Included observations: 534
Excluded observations: 2
Newey-West HAC Standard Errors & Covariance (lag truncation=5)
17
Variable Coefficient Std. Error t-Statistic Prob.
C 2.157021 0.460547 4.683605 0.0000
ASSET 0.000353 0.000201 1.760566 0.0789
R-squared 0.002206 Mean dependent var 2.305773
Adjusted R-squared 0.000331 S.D. dependent var 10.38383
S.E. of regression 10.38211 Akaike info criterion 7.521783
Sum squared resid 57343.30 Schwarz criterion 7.537815
Log likelihood -2006.316 F-statistic 1.176427
Durbin-Watson stat 1.990281 Prob(F-statistic) 0.278576
ROA = ROA
t
, asset = asset
t
Table 9: Regression of ROA on Cost
Dependent Variable: ROA
Method: Least Squares
Sample: 135 670
Included observations: 534
Excluded observations: 2
Newey-West HAC Standard Errors & Covariance (lag truncation=5)
Variable Coefficient Std. Error t-Statistic Prob.
C 1.809094 0.492258 3.675095 0.0003
COST 0.001799 0.000483 3.721827 0.0002
R-squared 0.009425 Mean dependent var 2.305773
Adjusted R-squared 0.007563 S.D. dependent var 10.38383
S.E. of regression 10.34448 Akaike info criterion 7.514522
Sum squared resid 56928.43 Schwarz criterion 7.530553
Log likelihood -2004.377 F-statistic 5.062017
Durbin-Watson stat 1.996262 Prob(F-statistic) 0.024864
ROA = ROA
t
, cost = cost
t
In the regression of performance on cost, the independent variable cost is significant
with a t value of 3.722 (p-value: 0.0002) at the 1% level. The F value of the model
was 5.062 corresponding to a significance level of 0.025. The null hypothesis is
therefore rejected.
A regression of asset on cost was performed to examine the correlation between
asset and cost retrenchment (see Table 10).
Cost as independent variable is highly significant with a t value of 6.373 (p-value: 0).
The F value of the model is 505.395 (p-value: 0). Consequently, the null hypothesis
that the regressor has no impact on the regressand is rejected.
The estimated coefficient of cost may be interpreted as follows:
) cost var(
) cost asset, cov(
= ?
The coefficient of cost corresponds to the covariance between asset and cost divided
by the variance of cost.
18
As a next step, the correlation coefficient between asset and cost retrenchment may
be computed. The correlation coefficient r is defined as follows:
) cost var( * ) asset var(
) cost asset, cov(
= r
Table 10: Regression of Asset on Cost
Dependent Variable: ASSET
Method: Least Squares
Sample: 135 670
Included observations: 534
Excluded observations: 2
Newey-West HAC Standard Errors & Covariance (lag truncation=5)
Variable Coefficient Std. Error t-Statistic Prob.
C -53.69611 40.50148 -1.325781 0.1855
COST 1.718811 0.269706 6.372916 0.0000
R-squared 0.487177 Mean dependent var 420.8188
Adjusted R-squared 0.486213 S.D. dependent var 1379.868
S.E. of regression 989.0750 Akaike info criterion 16.63516
Sum squared resid 5.20E+08 Schwarz criterion 16.65119
Log likelihood -4439.587 F-statistic 505.3945
Durbin-Watson stat 2.032769 Prob(F-statistic) 0.000000
asset = asset
t
, cost = cost
t
The correlation coefficient r corresponds to the covariance between asset and cost
divided by the square of the product of the variances of asset and cost. A r-value of
0.69809 is the result using our values for the covariance and the two variances.
Consequently, a significant, positive correlation between asset and cost retrenchment
exists, which means that asset and cost retrenchment are likely to appear either
together or not at all. In sum, it may be concluded that performing just a regression of
performance on asset is misleading due to correlation between the two independent
variables. Asset reflects in this case effects of the non-included variable cost, which
is the variable that actually explains impact on performance.
19
Figure 4: Correlogram Asset-Cost Retrenchment
(1996-1999, with trend line)
-5000
0
5000
10000
15000
20000
0 1000 2000 3000 4000 5000
Cost retrenchment (absolute values)
A
s
s
e
t
r
e
t
r
e
n
c
h
m
e
n
t
(
a
b
s
o
l
u
t
e
v
a
l
u
e
s
)
In the third step of the analysis a random effects model (REM) was estimated, which
delivers a more efficient estimation under the assumption of strict exogeneity. A
random effects model assumes that the intercept ?
1i
is the sum of a common
constant ?
1
and a time-invariant firm-specific random variable ?
i
which is uncorrelated
with the residual u
it
(Gujarati, 2003) (see Table 11).
The random effects model yieldes an R-square of 0.0264, which is the highest –
though still low – R-square that resulted from our regressions. Both independent
variables’ coefficients are approximately of the same magnitude like those in the
second regression. Moreover, they have the same signs, again opposing Hypotheses
3a and 4b and supporting 3b and 4a. The independent variable asset is now non-
significant with a t value of -0.704 (p-value: 0.482). Cost, which was significant at the
1% level in the second regression is now significant only at the 5% level with a t
value of 2.133 (p-value: 0.033).
Since the magnitude of the coefficient of cost has not considerably changed
compared to the second regression with absolute values, the scale of the influence of
cost retrenchment on performance is practically the same. For cost retrenching
companies an average performance improvement of 0.0624 (earlier: 0.0613)
percentage points results, while for non-cost retrenching companies an average
performance deterioration of 0.1421 (earlier: 0.1397) percentage points results.
20
Table 11: Random Effects Model (REM)
Dependent Variable: ROA?
Method: GLS (Variance Components)
Sample: 1 134
Included observations: 134
Number of cross-sections used: 4
Total panel (unbalanced) observations: 534
Variable Coefficient Std. Error t-Statistic Prob.
C 1.786600 1.211543 1.474649 0.1409
ASSET? -0.000317 0.000450 -0.703979 0.4818
COST? 0.002363 0.001108 2.132470 0.0334
Random Effects
_1996--C 1.801159
_1997--C 0.354394
_1998--C -1.444388
_1999--C -0.727371
GLS Transformed
Regression
R-squared 0.026422 Mean dependent var 2.305773
Adjusted R-squared 0.022755 S.D. dependent var 10.38383
S.E. of regression 10.26501 Sum squared resid 55951.65
Durbin-Watson stat 2.033263
Unweighted Statistics
including Random
Effects
R-squared 0.028668 Mean dependent var 2.305773
Adjusted R-squared 0.025009 S.D. dependent var 10.38383
S.E. of regression 10.25316 Sum squared resid 55822.56
Durbin-Watson stat 2.037965
In sum, we conclude that cost is a significant explanatory variable for performance,
while asset and with that asset retrenchment has no impact on performance.
Consequently, Hypotheses 3a and 4b are rejected, while Hypotheses 3b and 4a
received support. However, the result of a positive impact of cost retrenchment on
performance has to be qualified. Despite the statistical significance of cost, its actual
positive influence on performance is limited.
Discussion
We find a significant difference in the performance levels between retrenching and
non-retrenching firms. The t-tests that were performed to examine the hypothesized
relationship fully supported Hypotheses 1 and 2. Only for 1999 an non-significant
difference in ROA between cost retrenching and non-cost retrenching companies
resulted.
4
The lower levels of performance of retrenching firms indicate the higher financial
pressure on such firms induced by the Asian crisis. A higher degree of pressure
21
makes firms more inclined to take sharp measures including retrenchment to achieve
relief from the crisis situation.
Preceding literature suggests that the urgency in the turnaround situation and the
pressure on the distressed firm determines the degree of retrenchment (Falkenberg
and Chong, 2004b). Since cultural difference between the West and South-East Asia
suggest a resistance to retrenchment in an Asian setting due to loyalty between
employees and management and the life-long employment principle, in tendency
South-East Asian firms that retrench will face severe turnaround situations and even
more severe situations as compared to Western firms. On the other hand, however,
Singapore is a place where retrenchment has started to become a practical business
policy independent of public resistance.
Our findings suggest that the degree of distress and urgency in the turnaround
situation moderates the degree of asset and cost retrenchment
The utility of retrenchment for firms has been discussed controversially in preceding
literature. Robbins and Pearce (1992) see retrenchment as critical strategic element
in attaining turnaround and argue that retrenchment not only improves the liquidity
situation of a firm, but also the efficiency of a company. A fundamental proposition
underlying their research methodology was that a retrenchment response is present
in a substantial number of firms in the sample. This assumption was validated with
retrenchment put into effect among 15 of the 32 firms in the sample.
Similarly, retrenchment among our sample firms was widespread with just 33 of 134
companies not retrenching in response to the crisis. But in contrast to Robbins and
Pearce (1992), our quantitative analyses did not find strong evidence for the utility of
retrenchment. Our findings suggest that asset retrenchment has no impact on
performance at all and the influence of cost retrenchment is positive, however, rather
weak.
Retrenching companies face significantly lower performance levels than non-
retrenching companies throughout the crisis situation in Singapore.
Asset retrenchment has proven to be rather useless in improving performance during
the crisis situation. Cost retrenchment has a significantly positive influence on
performance, but the magnitude of its impact is limited.
22
Generally, retrenchment seems to have a limited efficiency-increasing impact on
companies. Given the limited effect of retrenchment, the substantial number of
companies in the sample which where found to retrench indicates an over
expectation of the effect of retrenchment and a misperception of its positive impacts.
Seemingly, the majority of companies saw retrenchment as an important element of
their decline-stemming strategies to improve their performance during the financial
crisis. Despite the high number of retrenching firms, the positive effects on
performance on retrenchment are limited.
Conclusion
Our review of Western turnaround literature together with turnaround literature
focusing particularly on turnaround management in South-East Asia led us to
propose that retrenching companies face lower overall performance levels throughout
the crisis situation in Singapore as compared to non-retrenching firms and to put
forward competing hypotheses concerning the utility of asset and cost retrenchment
in improving performance in a crisis situation. Our findings, based on quantitative
analyses, confirm that retrenching companies face lower performance levels than
non-retrenching firms. Moreover, our findings suggest asset and cost retrenchment to
have limited positive impact on company performance. The impact of retrenchment
as measure to increase the overall efficiency of the assets of a firm is overestimated.
The empirical investigation of the relationships between both performance and
probability to retrench and retrenchment and performance remains a fruitful field of
future research. Forthcoming research could control for company size, severity of
decline and industry in their model to create more homogenous samples.
23
Bibliography
Altman, E. I. (1968). Financial Ratios, Discriminant Analysis and the Prediction of
Corporate Bankruptcy. Journal of Finance, 23 (4), 589-609.
Altman, E. I. (1971). Corporate Bankruptcy in America. Heath-Lexington: Lexington,
MA.
Arogyaswamy, K., Barker, V. L. & Yasai-Ardekani, M. (1995). Firm Turnarounds: An
Integrative Two-Stage Model. Journal of Management Studies, 32 (4), 493-525.
Backman, M. (1999). Asian Eclipse: Exposing the Dark Side of Business in Asia.
John Wiley: Singapore.
Barker, V. L. & Duhaime, I. M. (1997). Strategic Change in the Turnaround Process:
Theory and Empirical Evidence. Strategic Management Journal, 18 (1), 13-38.
Barker, V. L. & Mone, M. A. (1994). Retrenchment: Cause of Turnaround or
Consequence of Decline? Strategic Management Journal, 15 (5), 395-405.
Bernard, R. (2000). Social Research Methods – Qualitative and Quantitative
Approaches. Thousand Oaks: Sage Publications.
Bibeault, D. B. (1999). Corporate Turnaround: How Managers Turn Losers Into
Winners. McGraw-Hill: New York.
Bruton, G. D., Ahlstrom, D. & Wan J. C. C. (2001). Turnaround Success of Large and
Midsize Chinese Owned Firms: Evidence from Hong Kong and Thailand.
Journal of World Business, 36 (2), 146-165.
Bruton, G. D., Ahlstrom, D. & Wan J. C. C. (2003). Turnaround in East Asian Firms:
Evidence from Ethnic Overseas Chinese Communities. Strategic Management
Journal, 24, 519-540.
Falkenberg, A. D. (2003). Turnaround Management in South-East Asia – Research
Challenges. Working Paper 215, Indian Institute of Management Bangalore:
Bangalore.
Falkenberg, A. D. & Chong L.-C. (2004a). Turnaround Strategies in South-East Asia.
24
th
Strategic Management Society Annual International Conference ‘Strategic
Balance: Driving Innovation and Maintaining Performance’, San Juan, Puerto
Rico, October 2004.
Falkenberg, A. D. & Chong, L.-C. (2004b). Turnaround Process in South-East Asia.
64
th
Academy of Management Conference ‘Creating Actionable Knowledge’,
New Orleans, USA, August 2004.
Gujarati, D. N. (2003). Basic Econometrics (4
th
ed.). McGraw-Hill: New York.
24
Hambrick, D. C. (1985). Turnaround Strategies. In Guth, W. D. (Ed.), Handbook of
Business Strategy (pp. 10.1-10.32). WG&L: Boston, New York.
Hambrick, D. C. & Schecter S. M. (1983). Turnaround Strategies for Mature
Industrial-Product Business Units. Academy of Management Journal, 26 (2),
231-248.
Hofer, C. W. (1980). Turnaround Strategies. Journal of Business Strategy, 1, 19-31.
Pearce, J. A. & Robbins, D. K. (1993). Toward Improved Theory and Research on
Business Turnaround. Journal of Management, 19 (3), 613-636.
Pearce, J. A. & Robbins, D. K. (1994). Retrenchment Remains the Foundation of
Business Turnaround. Strategic Management Journal, 15 (5), 407-417.
Robbins, D. K. & Pearce, J. A. (1992). Turnaround: Retrenchment and Recovery.
Strategic Management Journal, 13 (4), 287-309.
Schendel, D. & Patton, R. (1976). Corporate stagnation and turnaround. Journal of
Economics and Business, 28, 236-241.
Schendel, D., Patton, R. & Riggs, J. (1976). Corporate turnaround strategies: A study
of profit decline and recovery. Journal of General Management, 3, 3-11.
Singapore Exchange. (2004). Singapore Exchange (SGX) Homepage. May 10, 2004,
www.singaporeexchange.com
Von Auer, L. (2003). Ökonometrie: Eine Einführung (2. Aufl.). Springer: Berlin,
Heidelberg.
1
The negotiation and decline-halting phase in South-East Asia is the second phase in the South-East
Asian turnaround process and actions might have been taken in the recognition phase already. The
strategies in this phase are comparable to the decline-stemming strategies in a Western setting
(Arogyaswamy, Barker, Yasai-Ardekani, 1995).
2
(1) Working capital/Total assets, (2) Retained Earnings/Total assets, (3) Earnings before interest and
taxes/Total assets, (4) Market value equity/Book value of total debt, (5) Sales/Total assets)
3
Altman’s Z score has proven to be quite correct in predicting insolvency for the next 2 years.
However, for longer periods of time accuracy of prediction diminishes considerably (Altman, 1968).
4
There are indications that year-specific influences of 1999 led to this result. T-tests that were done
on a trial basis for 2000, 2001 and 2002 showed t-values of more than 3 – indicating again highly
significant differences in ROA between cost retrenching and non cost retrenching companies for the
following years.
doc_446053614.pdf
Asset And Cost Retrenchment In Turnaround Strategies A Large-sample Study Of Corporate Responses To The Asian Crisis In Singapore
1
Asset and Cost Retrenchment in Turnaround Strategies –
A Large-Sample Study of Corporate Responses to the
Asian Crisis in Singapore
Alexander D. Falkenberg
University of St. Gallen
Institute for International Management – Asia Research Centre
Dufourstrasse 40a
CH-9000 St. Gallen, Switzerland
Wharton-Singapore Management University Research Centre
469 Bukit Timah Road, Eu Tong Sen Building
Singapore 259756, Singapore
[email protected]
Li-Choy Chong
University of St. Gallen
Dufourstrasse 40a
CH-9000 St. Gallen, Switzerland
[email protected]
Pascal P. Prinz
University of St. Gallen
Dufourstrasse 40a
CH-9000 St. Gallen, Switzerland
[email protected]
Acknowledgement
We thank Helena A. Glamheden (University of St. Gallen & Wharton-SMU Research
Center), Markus Fröhlich (University of St. Gallen) and Jerel Gareth Lee Chye Hock
(Singapore Management University) for support and helpful discussions. We
thankfully acknowledge the Wharton-Singapore Management University Research
Center, Singapore, for organizational support. We would also like to express deep
gratitude to the Swiss National Foundation for financially enabling the research
project.
2
Abstract
The role of retrenchment is the most heatedly and controversially debated element of
turnaround management for a Western environment and the debate becomes even
more ambiguous when an Asian perspective is taken.
Since preceding studies have mainly neglected the dualism of industry-contraction
based and firm-based decline situations, we take the Asian crisis as a prominent
example of a temporary, but stiff contraction across industries to inquire the role of
retrenchment strategies in response to such a crisis. We focus our analysis on
Singapore and utilize a triangulation of quantitative analyses of 134 companies listed
on the Singaporean stock exchange.
We conclude that both asset and cost retrenching companies face lower overall
performance levels throughout the crisis situation in Singapore than non-retrenching
firms, which indicates that the degree of distress and urgency in the turnaround
situation moderates the degree of asset and cost retrenchment.
We furthermore find differing effects on performance from asset and cost
retrenchment. Although the impact of asset retrenchment on company performance
in a crisis situation in Singapore is non-significant, the impact of cost retrenchment on
company performance in a crisis situation in Singapore is positive, however, limited.
Our findings suggest that the reduction of costs in a crisis situation might foster
company performance, whereas the reduction of assets has no performance impact.
3
Introduction
A long tradition of Western-based turnaround literature has developed a rather solid
body of knowledge of turnaround for a Western, mainly US-based environment.
Despite the large amount of academic work on turnaround, however, the role of
retrenchment is still a heatedly debated element of turnaround for a Western
environment. Retrenchment “entails deliberate reductions in costs, assets, products,
product lines and overhead” (Pearce and Robbins, 1993) and past studies on a
Western environment have produced differing results: Pearce and Robbins (1992;
1994) conclude that retrenchment is indispensable for a successful turnaround, while
Barker and Mone (1994) stress the importance of a strategic reorientation in
achieving turnaround.
The debate of the role of retrenchment becomes even more ambiguous when an
Asian perspective is taken: During the Asian crisis starting with the severe
devaluation of the Thai Baht on July 1
st
, 1997 it became apparent that trying to apply
turnaround concepts that proved successful in a Western environment to South-East
Asian companies generated dissatisfactory results and demonstrated the limited
applicability of Western methods.
Initial research on turnaround in South-East Asia has analytically reconfirmed the
limited applicability of Western findings in South-East Asia and has made first
inquiries into the turnaround process and turnaround strategies in South-East Asia
mainly based on qualitative data and focusing on Chinese-owned companies (Bruton,
Ahlstrom and Wan, 2001; 2003; Falkenberg, 2003; Falkenberg and Chong, 2004a,b).
Even the limited research done on South-East Asia already demonstrates
contradictious findings on the role of retrenchment. Whereas the general notion in
South-East Asia is that retrenchment is avoided and inappropriate for cultural
reasons (Bruton, Ahlstrom and Wan, 2001; Falkenberg, 2003), an initial analysis of
quantitative data on Chinese-owned businesses revealed the fact that where
retrenchment was implemented in turnaround situations, it resulted a positive effect
on performance (Bruton, Ahlstrom and Wan, 2003).
Given that only a few research studies on firm turnaround for Asia have been
conducted so far and that these studies already reveal equivocal findings of
retrenchment in firm turnaround, a large-sample study is conduced to contribute to
4
the clarification of the role of retrenchment. Such a large-sample study can validate
or reject previous findings on a quantitative basis and further develop the role of
retrenchment in turnaround strategies.
Acknowledging the differing structural and analytical situations of firm-based decline
and industry-contraction based decline, it is of particular interest to investigate
whether the role of retrenchment is different for the two logical situations as
suggested in preceding studies (Barker and Duhaime, 1997; Falkenberg and Chong
2004 a,b). Therefore, we take the most prominent example of an industry-contraction
based decline situation – a widespread economic crisis across all industries – as
reference point for our analysis. At the same time, we take a geographic focus rather
than the view on Chinese-owned businesses, which has been the focus in preceding
articles (Bruton, Ahlstrom and Wan, 2001, 2003). Taking a geographic focus on
Singapore allows us to assess the role of retrenchment for the specific legal, social
and (multi)cultural environment of Singapore.
Our contribution seeks to find, what the impact of both asset and cost retrenchment is
on company performance in a financial crisis situation in Singapore. Our investigation
of one of the most developed markets in South-East Asia during the Asian financial
crisis examines the influence of asset- and/or cost-retrenchment on performance and
consequently, the general significance of retrenchment as viable strategy in
improving performance in an Asian setting. The examined sample is not restricted to
firms that faced a severe decline in 1997 and the following years but studies whether
firms saw asset and/or cost retrenchment as a viable option for improving
performance being faced with a financial crisis situation.
In line with preceding studies on turnaround management, we take a strategic
management perspective on turnaround. Our perspective assumes that the
strategies implemented at the top of organizations translate into the observed
strategies. Our view, therefore, abstracts from the obstacles in organizational
processes, which relate observed and intended strategies and is thus not in the
tradition of an organizational behavior view.
Our article is organized in the following way: We review the literature on turnaround
management with a focus on retrenchment in the following section to develop
hypotheses on the role of retrenchment in turnaround endeavors. We base our
5
hypotheses on preceding studies investigating both the West and South-East Asia,
although the role of retrenchment in South-East Asia might be different from the
Western view, Western findings are the most accurate information available on the
role of retrenchment. Following the literature review, we present our methodology.
We triangulate our findings based on three methodologies: mean comparison (t-test),
regression and correlation analysis. Thereafter, we present and discuss the findings.
The last section provides conclusions of our research contribution to the ongoing
discussion of the role of retrenchment in turnaround strategies.
Literature Review
Multiple researchers in the field of turnaround management emphasize the
importance of asset- and/or cost retrenchment for achieving turnaround success (e.g.
Hofer, 1980; Bibeault, 1999; Robbins and Pearce, 1992).
Correctly identifying the cause of decline allows the turnaround response to be
tailored to the specific problem. An early perspective on turnaround strategies
dichotomized the cause of a firm’s decline into internal and external sources of
decline (Schendel and Patton, 1976; Schendel, Patton and Riggs, 1976).
Correspondingly, two types of turnaround strategies are distinguished: operating vs.
strategic turnaround strategies (Hofer, 1980). Strategic turnaround strategies should
be used to solve external problems, while operating strategies should be applied in
the case of internal problems. Operating strategies focus on improvement of firm
efficiency and therefore are closely related to retrenchment of non-performing assets
and overly high cost factors. Both types of turnaround responses, however, may
include asset- and cost-cutting elements, which are assumed to positively influence
performance if closely tied to the assessment of current operating and strategic
health of the firm (Hofer, 1980).
More recently, the cause of decline is dichotomized in firm-based and industry-
contraction based declines (Arogyaswamy, Barker and Yasai-Ardekani, 1995; Barker
and Duhaime, 1997, Falkenberg and Chong 2004a). Subsequently, the cause of
decline is found to decisively influence whether a company may achieve turnaround
through strategic renewal. If the decline is firm-based, a relatively high level of
strategic change is involved (Falkenberg and Chong 2004a). Conversely, an industry
contraction-based and cyclical decline involves relatively less strategic change and
may require capacity adjustments (Falkenberg and Chong, 2004a).
6
The need for retrenchment in the short-term oriented starting phase of the turnaround
process, i.e. the negotiation and decline-halting phase in Asian turnarounds
(Falkenberg and Chong, 2004b)
1
, is not dependent on the source of decline.
Retrenchment in the initial phase of the turnaround process varies depending on (a)
the severity of decline of the firm and (b) the availability of slack resources (Hofer,
1980; Hambrick, 1985; Arogyaswamy, Barker and Yasai-Ardekani, 1995, Falkenberg
and Chong, 2004a). If the decline is severe, however, the business of the troubled
firm is viable, but the survival of the firm is immediately endangered, a short-term
oriented retrenchment exercise is implemented independent of the long-term
recovery strategy. In line with the proposed relationship of financial pressure and
retrenchment for the initial phase of the turnaround process (Falkenberg and Chong,
2004b), we propose that the degree of retrenchment is related to the degree of
financial pressure in firms facing a crisis situation:
Hypothesis 1: Asset retrenching companies face lower overall performance levels
throughout the crisis situation in Singapore than non-asset retrenching firms.
Hypothesis 2: Cost retrenching firms face lower overall performance levels
throughout the crisis situation in Singapore than non-cost retrenching firms.
Retrenchment in the long-term oriented revitalization phase of the turnaround
process, however, is differently assessed in a Western environment by two opposing
groups of researchers: While one group of researchers focuses on the recovery
response, i.e. the change in company strategy (Barker and Duhaime, 1997; Barker
and Mone, 1994), another group of researchers finds retrenchment to be an integral
part of successful turnarounds (Robbins and Pearce, 1992; Pearce and Robbins,
1994). Latter conclude from their large-sample study that retrenchment is a critical
strategic element in attaining turnaround success (Robbins and Pearce, 1992). They
conclude that the degree of turnaround success is positively related to the level of
retrenchment and the association between the degree of asset retrenchment and
turnaround performance is more pronounced in severe turnaround situations.
The proposition of the relevance of retrenchment has raised methodological and
theoretical criticism by Barker and Mone (1994) denying a vital role of retrenchment
in facilitating recovery and proposing the role of strategic turnarounds and strategic
reorientation (Barker and Mone, 1994; Barker and Duhaime, 1997).
7
The impact of retrenchment on turnaround success in South-East Asia is even more
ambiguous. On the one hand there is empirical evidence for retrenchment to
positively influence turnaround success for Chinese-owned firms (Bruton, Ahlstom
and Wan, 2003) supporting the argument put forward by Robbins and Pearce (1992).
On the other hand, there is a notion of negative retrenchment effects in a South-East
Asian environment for cultural reasons (Bruton, Ahlstrom and Wan, 2001, Falkenberg,
2003) and the proposition of a limited ability to retrench due to the Asian operating
environment and already slim Asian organizations faced with comparably low human
resources costs (Bruton, Ahlstrom and Wan, 2001). Since the applicability of the
equivocal and controversial findings to a Singaporean environment is unknown, we
propose two competing hypothesis for asset and cost retrenchment respectively:
Hypothesis 3a: The impact of asset retrenchment on company performance in a
crisis situation in Singapore will be positive.
Hypothesis 3b: The impact of asset retrenchment on company performance in a
crisis situation in Singapore will be negative or non-present.
Hypothesis 4a: The impact of cost retrenchment on company performance in a
crisis situation in Singapore will be positive.
Hypothesis 4b: The impact of cost retrenchment on company performance in a
crisis situation in Singapore will be negative or non-present.
Methodology
The structure of our hypothesis indicates that we are seeking to identify the
relationship of two continuous independent variables, asset and cost retrenchment,
on one dependent continuous variable, company performance. We triangulate our
findings based on a data sample of listed Singaporean companies.
Operationalization of Independent and Dependent Variables
The independent variables for our analysis are asset and cost retrenchment. We
composed asset and cost retrenchment based on balance sheet items to construct a
measurement of changes in assets and costs of the Singaporean listed firms
appropriately.
8
Past large-sample studies commonly employ financial ratios to examine efficiency
gains interpreted as the result of retrenchment. Such measures comprise for example
of cost of goods sold/sales (Schendel and Patton, 1976), receivables/sales, R&D
expenditure/sales and increased sales per employee (Hambrick and Schecter, 1983).
Increases in the ratios were interpreted as efficiency gains based on asset and cost
retrenchment. Such study design has raised criticism since all of these ratios include
sales in the denominator and are therefore not only dependent on the numerator but
also on sales. Therefore, improvements in the ratios can both be a result of the
decrease of the numerator or increases in sales, which makes an interpretation of
such ratios in light of retrenchment very difficult. It can thus not be determined
unambiguously whether efficiency gains result from sales gains, retrenchment or both
(Arogyaswamy, Barker and Yasai-Ardekani, 1995; Barker and Duhaime, 1997).
In order to avoid the dependency on sales, we model retrenchment independent of
sales in absolute numbers. Consequently, we define assets and asset retrenchment
as follows:
Assets Cash & Due from Banks + Receivables + Inventory +
Property, Plants, and Equipment (PPE)
Asset Retrenchment - ? (Cash & Due from Banks + Receivables +
Inventory + Property, Plants, and Equipment (PPE))
= Assets
t
– Assets
t-1
Assets are measured as the sum of cash & due from banks, receivables, inventory,
and property, plants and equipment (PPE). The variable asset retrenchment is
defined as the (negative) delta of the sum of cash & due from banks, receivables,
inventory, and property, plants and equipment (PPE).
Similarly, costs are measured as the sum of operating expense and total interest
expense. The variable cost retrenchment is defined as the (negative) delta of the sum
of operating expense and total interest expense.
Costs Operating Expense + Interest Expense Total
Cost Retrenchment - ? (Operating Expense + Interest Expense Total)
= Costs
t
– Costs
t-1
The dependent variable is company performance. Most past studies use return on
investment (ROI) or a similar indicator as performance measure (e.g. Robbins and
9
Pearce, 1992). Following these studies, we employ return on total assets (ROA) as a
measure of performance during the period of investigation.
Financial measures like ROI, ROA, and return on sales (ROS) are widespread and
frequently used indicators, which deliver reliable information about actual
performance (Bernard, 2000).
Thus, performance and performance change are measured as follows:
Performance 100 * ROA
? Performance Performance
t
– Performance
t-1
Data Sample
The data source from which the dependent and the two independent variables were
drawn is Standard & Poor’s COMPUSTAT Global data base. COMPUSTAT reports
418 companies for Singapore. Due to missing data, only 134 companies could be
included in the sample. Data was initially collected for the years 1995 to 2002. The
period of time that was later investigated was restricted to the years 1996 to 1999 to
avoid a blurring of the effects of the Asian financial crisis with other factors (e.g.
downturn of world economy after the collapse of the Internet bubble in 2000). Pre-
crisis year, year of crisis, and 2 years of potential recovery were therefore a plausible
period of observation.
Assuming that the Asian financial crisis led to industry contraction for nearly all
companies, we examined the financial situation of the companies before 1997. We
evaluated an indicator of potential bankruptcy to confirm our assumption: A widely-
used predictor of corporate bankruptcy in turnaround studies is Altman’s Z-score,
(Altman 1968, 1971). Altman uses five financial ratios
2
and combines this set in a
discriminant analysis approach. A Z-score of more than 2.99 indicates a financially
healthy firm; below 1.8 suggests immediate danger, while a Z-score of 1.8 to 2.99 is
a warning sign of financial distress.
3
Our analysis of the 134 companies in our sample revealed that the number of firms
with Altman’s Z of a score lower than 2.99 increases dramatically after the Asian
crisis hit in 1997 reaching a peak of 112 distressed firms in 1998 (Table 1). Since it is
unknown to what degree the health of the firms with an Altman’s Z higher than 2.99 is
dependent on the implementation of retrenchment strategies, we analyzed both
groups of firms with respect to the performance impact of retrenchment.
10
Table 1: Number of Companies and Altman Z-score
Z-score ? 2.99 Z-score < 2.99
1995 64 70
1996 53 81
1997 28 106
1998 22 112
1999 38 96
We classify firms as asset/cost retrenching firms if assets/costs decreased by more
than 5% from 1996 to 1999. If a firm’s assets/costs diminished by less than 5% or did
even increase between 1996 and 1999, the firm was classified as non-asset
respectively non-cost retrenching firm. Table 2 shows our sample along the resulting
four categories: (1) firms that were found to retrench both in assets and costs, (2)
firms that did retrench in assets but not in costs, (3) firms that did retrench in costs
but not in assets and (4) firms that did not retrench at all.
Table 2: Classification of Asset and Cost Retrenching Firms
Cost retrenching firms Cost non-retrenching firms
Asset retrenching firms 60 20
Asset non-retrenching firms 21 33
Only 33 companies of the total number of 134 did not retrench at all. The majority of
firms retrenched both assets and costs, and about the same number of firms did
utilize either cost or asset retrenchment.
In order to assess the structure of the industries of the firms in our sample, we
utilized the categories used by the Singapore Stock Exchange (Singapore Exchange,
2004) (Table 3 and Figure 1). More than one third of the companies in the sample
operated in a manufacturing industry. The second most important sector was
commerce with 16%. The share of multi-industry, hotels & restaurants, properties,
construction, transport/ storage/ communication (TSC), and services was
approximately equal, ranging from 6% to 10%. Just one company in the sample
operated in the finance sector, which percentage share in the sample is therefore
less than 1%. No companies came from the loans & debentures, electricity/ gas/
water, CLOB (Central Limit Order Book) international, agriculture and
mining/quarrying sector.
11
In order to obtain deeper
insights into the composition
of our sample with relation to
the total structure of the
listed Singaporean firms, it
was necessary to investigate
how many of all companies
of a sector (listed at
Singapore Stock Exchange)
were part of our sample. The
percentage share of the
companies in a specific
industry in relation to the total number of companies listed in this sector was
computed and is presented in Figure 2.
The analysis revealed that the multi-industry and hotels & restaurants sector are
overrepresented in our sample, while the services and particularly the finance sector
are underrepresented. 67% of all companies in the multi-industry sector respectively
60% in the hotels & restaurants sector were part of our sample. Conversely, only
8.7% of all services companies and a meager 2.4% of all finance companies were in
our sample. The other sectors namely commerce (22%), manufacturing (23.7%),
properties (26.5%), construction (22%), and TSC (27.8%) were represented in a
balanced way in the sample. The non-inclusion of sectors like agriculture or
mining/quarrying is less crucial in view of the small number of companies listed in
those sectors except for loans & debentures, a sector with 75 companies totally.
Table 3: Industry Classification
Multi-Industry 14
Commerce 21
Finance 1
Manufacturing 50
Hotels & Restaurants 12
Properties 9
Construction 9
Transport, Storage, Communication 10
Services 8
Figure 1: Percentage of Industries in Sample
Multi-Industry
10%
Commerce
16%
Finance
1%
Manufacturing
37%
Hotels &
Restaurants
9%
Properties
7%
Construction
7%
TSC
7%
Services
6%
12
Findings
Hypotheses 1 & 2:
Analysis t-tests
The first and second
hypotheses state
that asset
retrenching companies as well as cost retrenching firms face lower overall
performance levels throughout the crisis situation in Singapore as compared to non-
asset and non-cost retrenching firms.
Accordingly, the mean ROA of asset cost retrenching firms respectively should be
found to be significantly lower than that of non-retrenching companies. To examine
this relationship, mean ROA of asset and non-asset retrenching firms was compared
in pair wise t-tests over the period of observation between 1996 and 1999. The same
was done for cost and non-cost retrenching firms (see Table 4 and 5).
Table 4: Mean Performance Comparison between Asset Retrenching and Non Asset Retrenching
Companies
Mean
performance
Asset retrenching
companies
Non Asset retrenching
companies t-value p-value
1996 2.40675867 7.27142956 -3.9856918
***
0.000
1997 0.4355338 6.12316862 -3.52055495
***
0.000
1998 -2.17733058 4.70841509 -4.00419641
***
0.000
1999 -1.11679691 5.30690223 -3.66902157
***
0.000
avg ROA 96-99 -0.11295875 5.85247888
No. of companies 80 54
*** significant at the 1% level, ** significant at the 5% level, * significant at the 10% level
Definitions: Asset retrenching firms reduced assets by more than 5% between 1996 and 1999; Non asset retrenching firms
reduced by less or increased assets between 1996 and 1999.
Table 5: Mean Performance Comparison between Cost Retrenching and Non Cost Retrenching
Companies
Mean
performance
Cost retrenching
companies
Non Cost retrenching
companies t-value p-value
1996 3.03769277 6.39895804 -2.66996136
***
0.004
1997 1.01411379 5.3462376 -2.64480988
***
0.004
1998 -0.96778836 3.08915686 -1.93836691
*
0.026
1999 0.47563784 3.02442901 -1.12081277 0.131
avg ROA 96-99 0.88991401 4.46469538
No. of companies 81 53
Figure 2: Firms in Sample as Percentage of Total
Number of Companies
0,0%
15,0%
30,0%
45,0%
60,0%
75,0%
M
u
l
t
i
-
I
n
d
u
s
t
r
y
C
o
m
m
e
r
c
e
F
i
n
a
n
c
e
M
a
n
u
f
a
c
t
u
r
i
n
g
H
o
t
e
l
s
&
R
e
s
t
a
u
r
a
n
t
s
P
r
o
p
e
r
t
i
e
s
C
o
n
s
t
r
u
c
t
i
o
n
T
S
C
S
e
r
v
i
c
e
s
Average: 28.9 %
13
*** significant at the 1% level, ** significant at the 5% level, * significant at the 10% level
Definitions: Cost retrenching firms reduced costs by more than 5% between 1996 and 1999; Non cost retrenching firms
reduced by less or increased costs between 1996 and 1999.
The magnitude of the results for mean performance in each group seems plausible.
Both the maximum and the minimum mean ROAs are realistic, as well as the
average ROAs between 1996 and 1999. Moreover, a characteristic pattern in the way
the mean ROAs develop over time is identified (see Figure 3).
Indisputably, 1996 represents the year with the highest mean ROA among all groups.
With the beginning of the Asian financial crisis in mid-1997 a clear-cut downward
trend starts, leading to a minimum in mean ROAs in 1998. An upswing begins in
1999 with the subsiding of negative effects of the crisis. The results are therefore
compatible with the chronological sequence of events during the Asian financial crisis.
The results of the
t-tests for asset
and non-asset
retrenching firms
are unambiguous:
The differences of
mean performance
between asset and
non-asset
retrenching firms
are highly significant for all years. Non-asset retrenching firms achieve an average
ROA of 5.85 between 1996 and 1999, while asset retrenching firms in contrast are
confronted with a negative average ROA of -0.11. Hypothesis 1, therefore, receives
strong support.
The results for cost and non-cost retrenching firms are not as clear as those for asset
retrenchment, but similar. The t values for 1996 and 1997 are -2.67 and -2.65 hence
significant at the 1% level. For 1998 the t value of -1.94 is almost significant at the
5% level. The only year with a clearly non-significant difference of mean ROA
between cost retrenching and non-cost retrenching firms is 1999 with a t value of -
1.12. Non-cost retrenching firms experienced an average ROA of 4.47 between 1996
and 1999, while cost retrenching firms were challenged by a low average ROA of
Figure 3: Time Pattern of Development in Mean ROAs
-4
-2
0
2
4
6
8
1996 1997 1998 1999
Years
M
e
a
n
R
O
A
Asset retrenching
companies
Non asset
retrenching
companies
Cost retrenching
companies
Non cost
retrenching
companies
14
0.89. These findings represent support for Hypothesis 2 in spite of the non-significant
t value for 1999.
Analyses of Hypotheses 3 & 4: Regression model
The previous analysis showed differences between companies that retrenched
assets and/or costs during the whole period. In a next step, we examine in more
detail the relationship between ROA and assets and costs at a particular point in time
using a linear regression model:
ROA
i,t
= alpha + beta1 * asset
i,t
+ beta2 * cost
i,t
+ epsilon
i,t
(1)
where t denotes time and i is an indicator of the company.
Three regressions were conducted. Firstly, equation (1) was estimated by ordinary
least-squares (OLS) with robust standard errors. Secondly, equation (1) is estimated
as a random effects specification, where epsilon
i,t
consists of a company specific
time-invariant error term and a time-varying error term. Third, the regression equation
is estimated in first differences:
?ROA
i,t
= alpha + beta1 * ?asset
i,t
+ beta2 * ?cost
i,t
+ ?epsilon
i,t
(2)
which allows to eliminate a time-invariant error term.
All regressions were performed with EViews 4.1 for the years 1996 to 1999. The
results for regression equation (2) are presented in table 6.
The method of OLS was used to produce estimates that were the best linear
unbiased estimates under classical assumptions (von Auer, 2003; Gujarati, 2003).
The model yielded an R-square of 0.001, which indicates a very weak fit between the
model and the data. The F value for the overall model was 0.222, which has a
significance level of 0.801. Both independent variables were non-significant with a t
statistic of 1.139 (p-value: 0.255) for asset retrenchment and a t value of -1.209 (p-
value: 0.227) for cost retrenchment. The result did not change when the regression
was repeated with only one independent variable, i.e. a regression of performance on
asset retrenchment and a regression of performance on cost retrenchment
individually.
Table 6: Regression of ?ROA on ?Asset and ?Cost
15
Dependent Variable: ?ROA
Method: Least Squares
Sample: 135 536
Included observations: 399
Excluded observations: 3
Newey-West HAC Standard Errors & Covariance (lag truncation=5)
Variable Coefficient Std. Error t-Statistic Prob.
C -0.966518 0.445953 -2.167309 0.0308
?ASSET 0.001003 0.000881 1.138972 0.2554
?COST -0.001212 0.001002 -1.209146 0.2273
R-squared 0.001121 Mean dependent var -0.977738
Adjusted R-squared -0.003924 S.D. dependent var 9.508195
S.E. of regression 9.526831 Akaike info criterion 7.353592
Sum squared resid 35941.16 Schwarz criterion 7.383584
Log likelihood -1464.042 F-statistic 0.222232
Durbin-Watson stat 2.023429 Prob(F-statistic) 0.800829
?ROA = ROA
t
- ROA
t-1
, ?asset = asset
t
- asset
t-1
, ?cost = cost
t
- cost
t-1
The finding of non-significant effects may be due to the large loss of information by
taking first differences. Estimates of the regression equation (1) in levels are given in
table 7.
Table 7: Regression of ROA on Asset and Cost
Dependent Variable: ROA
Method: Least Squares
Sample: 135 670
Included observations: 534
Excluded observations: 2
Newey-West HAC Standard Errors & Covariance (lag truncation=5)
Variable Coefficient Std. Error t-Statistic Prob.
C 1.792712 0.494422 3.625871 0.0003
ASSET -0.000305 0.000166 -1.841946 0.0660
COST 0.002323 0.000674 3.448921 0.0006
R-squared 0.010268 Mean dependent var 2.305773
Adjusted R-squared 0.006540 S.D. dependent var 10.38383
S.E. of regression 10.34981 Akaike info criterion 7.517416
Sum squared resid 56879.99 Schwarz criterion 7.541463
Log likelihood -2004.150 F-statistic 2.754500
Durbin-Watson stat 1.996541 Prob(F-statistic) 0.064550
ROA = ROA
t
, asset = asset
t
, cost = cost
t
The model yielded an R-square of 0.01, indicating again that only a small part of
movements in ROA may be explained by the two independent variables asset and
cost. The F value for the overall model was 2.755 which correspond to a significance
level of 0.065. The F value is therefore marginally significant at about the 10% level.
In contrast to the first regression, we can reject the null hypothesis that the two
regressors have no impact, notwithstanding the fact that the R-square is only 0.010.
16
The independent variable asset was marginally significant with a t value of -1.842 (p-
value: 0.066) at the 10% level. Cost was significant with a t value of 3.449 (p-value:
0.0006) at the 1% level.
The significant regressor cost tends to positively influence ROA, contrary to the
impact stated in Hypothesis 4b and supporting Hypothesis 4a.
How strong is this positive influence of cost on performance? The 81 cost retrenching
companies of our sample did retrench on average by -26.39 between 1996 and 1999.
Using the coefficient of cost of 0.002323 from our model, an average performance
(i.e. ROA) improvement of 0.0613 percentage points results. In contrast, the 53 non-
cost retrenching companies faced on average increasing costs of 60.12 between
1996 and 1999. Again using the results of our regression, an average performance
deterioration of 0.1397 percentage points results.
The independent variable asset has a negative sign, which opposes Hypothesis 3a.
The 80 asset retrenching companies of our sample retrenched on average by -46.74
between 1996 and 1999, while the 54 non-asset retrenching companies increased
their assets by 65.56. From both the small coefficient of asset of -0.000305 and the
fact that the variable is only marginally significant, it may be inferred that the impact
of asset reductions on performance is very small, if at all existing. Therefore,
Hypothesis 3b receives some support.
In the next step the regression was repeated twice, in each case just with one
independent variable as regressor (see Table 8 and 9).
In the regression of performance on asset, the independent variable asset is again
marginally significant with a t value of 1.761 (p-value: 0.079) at the 10% level. The F
value for the model was 1.176 which had a significance level of 0.279. The coefficient
of asset is comparable in magnitude to the result of the preceding regression with two
independent variables, but the sign has changed from negative to positive. This
indicates a correlation between asset retrenchment and cost retrenchment.
Table 8: Regression of ROA on Asset
Dependent Variable: ROA
Method: Least Squares
Sample: 135 670
Included observations: 534
Excluded observations: 2
Newey-West HAC Standard Errors & Covariance (lag truncation=5)
17
Variable Coefficient Std. Error t-Statistic Prob.
C 2.157021 0.460547 4.683605 0.0000
ASSET 0.000353 0.000201 1.760566 0.0789
R-squared 0.002206 Mean dependent var 2.305773
Adjusted R-squared 0.000331 S.D. dependent var 10.38383
S.E. of regression 10.38211 Akaike info criterion 7.521783
Sum squared resid 57343.30 Schwarz criterion 7.537815
Log likelihood -2006.316 F-statistic 1.176427
Durbin-Watson stat 1.990281 Prob(F-statistic) 0.278576
ROA = ROA
t
, asset = asset
t
Table 9: Regression of ROA on Cost
Dependent Variable: ROA
Method: Least Squares
Sample: 135 670
Included observations: 534
Excluded observations: 2
Newey-West HAC Standard Errors & Covariance (lag truncation=5)
Variable Coefficient Std. Error t-Statistic Prob.
C 1.809094 0.492258 3.675095 0.0003
COST 0.001799 0.000483 3.721827 0.0002
R-squared 0.009425 Mean dependent var 2.305773
Adjusted R-squared 0.007563 S.D. dependent var 10.38383
S.E. of regression 10.34448 Akaike info criterion 7.514522
Sum squared resid 56928.43 Schwarz criterion 7.530553
Log likelihood -2004.377 F-statistic 5.062017
Durbin-Watson stat 1.996262 Prob(F-statistic) 0.024864
ROA = ROA
t
, cost = cost
t
In the regression of performance on cost, the independent variable cost is significant
with a t value of 3.722 (p-value: 0.0002) at the 1% level. The F value of the model
was 5.062 corresponding to a significance level of 0.025. The null hypothesis is
therefore rejected.
A regression of asset on cost was performed to examine the correlation between
asset and cost retrenchment (see Table 10).
Cost as independent variable is highly significant with a t value of 6.373 (p-value: 0).
The F value of the model is 505.395 (p-value: 0). Consequently, the null hypothesis
that the regressor has no impact on the regressand is rejected.
The estimated coefficient of cost may be interpreted as follows:
) cost var(
) cost asset, cov(
= ?
The coefficient of cost corresponds to the covariance between asset and cost divided
by the variance of cost.
18
As a next step, the correlation coefficient between asset and cost retrenchment may
be computed. The correlation coefficient r is defined as follows:
) cost var( * ) asset var(
) cost asset, cov(
= r
Table 10: Regression of Asset on Cost
Dependent Variable: ASSET
Method: Least Squares
Sample: 135 670
Included observations: 534
Excluded observations: 2
Newey-West HAC Standard Errors & Covariance (lag truncation=5)
Variable Coefficient Std. Error t-Statistic Prob.
C -53.69611 40.50148 -1.325781 0.1855
COST 1.718811 0.269706 6.372916 0.0000
R-squared 0.487177 Mean dependent var 420.8188
Adjusted R-squared 0.486213 S.D. dependent var 1379.868
S.E. of regression 989.0750 Akaike info criterion 16.63516
Sum squared resid 5.20E+08 Schwarz criterion 16.65119
Log likelihood -4439.587 F-statistic 505.3945
Durbin-Watson stat 2.032769 Prob(F-statistic) 0.000000
asset = asset
t
, cost = cost
t
The correlation coefficient r corresponds to the covariance between asset and cost
divided by the square of the product of the variances of asset and cost. A r-value of
0.69809 is the result using our values for the covariance and the two variances.
Consequently, a significant, positive correlation between asset and cost retrenchment
exists, which means that asset and cost retrenchment are likely to appear either
together or not at all. In sum, it may be concluded that performing just a regression of
performance on asset is misleading due to correlation between the two independent
variables. Asset reflects in this case effects of the non-included variable cost, which
is the variable that actually explains impact on performance.
19
Figure 4: Correlogram Asset-Cost Retrenchment
(1996-1999, with trend line)
-5000
0
5000
10000
15000
20000
0 1000 2000 3000 4000 5000
Cost retrenchment (absolute values)
A
s
s
e
t
r
e
t
r
e
n
c
h
m
e
n
t
(
a
b
s
o
l
u
t
e
v
a
l
u
e
s
)
In the third step of the analysis a random effects model (REM) was estimated, which
delivers a more efficient estimation under the assumption of strict exogeneity. A
random effects model assumes that the intercept ?
1i
is the sum of a common
constant ?
1
and a time-invariant firm-specific random variable ?
i
which is uncorrelated
with the residual u
it
(Gujarati, 2003) (see Table 11).
The random effects model yieldes an R-square of 0.0264, which is the highest –
though still low – R-square that resulted from our regressions. Both independent
variables’ coefficients are approximately of the same magnitude like those in the
second regression. Moreover, they have the same signs, again opposing Hypotheses
3a and 4b and supporting 3b and 4a. The independent variable asset is now non-
significant with a t value of -0.704 (p-value: 0.482). Cost, which was significant at the
1% level in the second regression is now significant only at the 5% level with a t
value of 2.133 (p-value: 0.033).
Since the magnitude of the coefficient of cost has not considerably changed
compared to the second regression with absolute values, the scale of the influence of
cost retrenchment on performance is practically the same. For cost retrenching
companies an average performance improvement of 0.0624 (earlier: 0.0613)
percentage points results, while for non-cost retrenching companies an average
performance deterioration of 0.1421 (earlier: 0.1397) percentage points results.
20
Table 11: Random Effects Model (REM)
Dependent Variable: ROA?
Method: GLS (Variance Components)
Sample: 1 134
Included observations: 134
Number of cross-sections used: 4
Total panel (unbalanced) observations: 534
Variable Coefficient Std. Error t-Statistic Prob.
C 1.786600 1.211543 1.474649 0.1409
ASSET? -0.000317 0.000450 -0.703979 0.4818
COST? 0.002363 0.001108 2.132470 0.0334
Random Effects
_1996--C 1.801159
_1997--C 0.354394
_1998--C -1.444388
_1999--C -0.727371
GLS Transformed
Regression
R-squared 0.026422 Mean dependent var 2.305773
Adjusted R-squared 0.022755 S.D. dependent var 10.38383
S.E. of regression 10.26501 Sum squared resid 55951.65
Durbin-Watson stat 2.033263
Unweighted Statistics
including Random
Effects
R-squared 0.028668 Mean dependent var 2.305773
Adjusted R-squared 0.025009 S.D. dependent var 10.38383
S.E. of regression 10.25316 Sum squared resid 55822.56
Durbin-Watson stat 2.037965
In sum, we conclude that cost is a significant explanatory variable for performance,
while asset and with that asset retrenchment has no impact on performance.
Consequently, Hypotheses 3a and 4b are rejected, while Hypotheses 3b and 4a
received support. However, the result of a positive impact of cost retrenchment on
performance has to be qualified. Despite the statistical significance of cost, its actual
positive influence on performance is limited.
Discussion
We find a significant difference in the performance levels between retrenching and
non-retrenching firms. The t-tests that were performed to examine the hypothesized
relationship fully supported Hypotheses 1 and 2. Only for 1999 an non-significant
difference in ROA between cost retrenching and non-cost retrenching companies
resulted.
4
The lower levels of performance of retrenching firms indicate the higher financial
pressure on such firms induced by the Asian crisis. A higher degree of pressure
21
makes firms more inclined to take sharp measures including retrenchment to achieve
relief from the crisis situation.
Preceding literature suggests that the urgency in the turnaround situation and the
pressure on the distressed firm determines the degree of retrenchment (Falkenberg
and Chong, 2004b). Since cultural difference between the West and South-East Asia
suggest a resistance to retrenchment in an Asian setting due to loyalty between
employees and management and the life-long employment principle, in tendency
South-East Asian firms that retrench will face severe turnaround situations and even
more severe situations as compared to Western firms. On the other hand, however,
Singapore is a place where retrenchment has started to become a practical business
policy independent of public resistance.
Our findings suggest that the degree of distress and urgency in the turnaround
situation moderates the degree of asset and cost retrenchment
The utility of retrenchment for firms has been discussed controversially in preceding
literature. Robbins and Pearce (1992) see retrenchment as critical strategic element
in attaining turnaround and argue that retrenchment not only improves the liquidity
situation of a firm, but also the efficiency of a company. A fundamental proposition
underlying their research methodology was that a retrenchment response is present
in a substantial number of firms in the sample. This assumption was validated with
retrenchment put into effect among 15 of the 32 firms in the sample.
Similarly, retrenchment among our sample firms was widespread with just 33 of 134
companies not retrenching in response to the crisis. But in contrast to Robbins and
Pearce (1992), our quantitative analyses did not find strong evidence for the utility of
retrenchment. Our findings suggest that asset retrenchment has no impact on
performance at all and the influence of cost retrenchment is positive, however, rather
weak.
Retrenching companies face significantly lower performance levels than non-
retrenching companies throughout the crisis situation in Singapore.
Asset retrenchment has proven to be rather useless in improving performance during
the crisis situation. Cost retrenchment has a significantly positive influence on
performance, but the magnitude of its impact is limited.
22
Generally, retrenchment seems to have a limited efficiency-increasing impact on
companies. Given the limited effect of retrenchment, the substantial number of
companies in the sample which where found to retrench indicates an over
expectation of the effect of retrenchment and a misperception of its positive impacts.
Seemingly, the majority of companies saw retrenchment as an important element of
their decline-stemming strategies to improve their performance during the financial
crisis. Despite the high number of retrenching firms, the positive effects on
performance on retrenchment are limited.
Conclusion
Our review of Western turnaround literature together with turnaround literature
focusing particularly on turnaround management in South-East Asia led us to
propose that retrenching companies face lower overall performance levels throughout
the crisis situation in Singapore as compared to non-retrenching firms and to put
forward competing hypotheses concerning the utility of asset and cost retrenchment
in improving performance in a crisis situation. Our findings, based on quantitative
analyses, confirm that retrenching companies face lower performance levels than
non-retrenching firms. Moreover, our findings suggest asset and cost retrenchment to
have limited positive impact on company performance. The impact of retrenchment
as measure to increase the overall efficiency of the assets of a firm is overestimated.
The empirical investigation of the relationships between both performance and
probability to retrench and retrenchment and performance remains a fruitful field of
future research. Forthcoming research could control for company size, severity of
decline and industry in their model to create more homogenous samples.
23
Bibliography
Altman, E. I. (1968). Financial Ratios, Discriminant Analysis and the Prediction of
Corporate Bankruptcy. Journal of Finance, 23 (4), 589-609.
Altman, E. I. (1971). Corporate Bankruptcy in America. Heath-Lexington: Lexington,
MA.
Arogyaswamy, K., Barker, V. L. & Yasai-Ardekani, M. (1995). Firm Turnarounds: An
Integrative Two-Stage Model. Journal of Management Studies, 32 (4), 493-525.
Backman, M. (1999). Asian Eclipse: Exposing the Dark Side of Business in Asia.
John Wiley: Singapore.
Barker, V. L. & Duhaime, I. M. (1997). Strategic Change in the Turnaround Process:
Theory and Empirical Evidence. Strategic Management Journal, 18 (1), 13-38.
Barker, V. L. & Mone, M. A. (1994). Retrenchment: Cause of Turnaround or
Consequence of Decline? Strategic Management Journal, 15 (5), 395-405.
Bernard, R. (2000). Social Research Methods – Qualitative and Quantitative
Approaches. Thousand Oaks: Sage Publications.
Bibeault, D. B. (1999). Corporate Turnaround: How Managers Turn Losers Into
Winners. McGraw-Hill: New York.
Bruton, G. D., Ahlstrom, D. & Wan J. C. C. (2001). Turnaround Success of Large and
Midsize Chinese Owned Firms: Evidence from Hong Kong and Thailand.
Journal of World Business, 36 (2), 146-165.
Bruton, G. D., Ahlstrom, D. & Wan J. C. C. (2003). Turnaround in East Asian Firms:
Evidence from Ethnic Overseas Chinese Communities. Strategic Management
Journal, 24, 519-540.
Falkenberg, A. D. (2003). Turnaround Management in South-East Asia – Research
Challenges. Working Paper 215, Indian Institute of Management Bangalore:
Bangalore.
Falkenberg, A. D. & Chong L.-C. (2004a). Turnaround Strategies in South-East Asia.
24
th
Strategic Management Society Annual International Conference ‘Strategic
Balance: Driving Innovation and Maintaining Performance’, San Juan, Puerto
Rico, October 2004.
Falkenberg, A. D. & Chong, L.-C. (2004b). Turnaround Process in South-East Asia.
64
th
Academy of Management Conference ‘Creating Actionable Knowledge’,
New Orleans, USA, August 2004.
Gujarati, D. N. (2003). Basic Econometrics (4
th
ed.). McGraw-Hill: New York.
24
Hambrick, D. C. (1985). Turnaround Strategies. In Guth, W. D. (Ed.), Handbook of
Business Strategy (pp. 10.1-10.32). WG&L: Boston, New York.
Hambrick, D. C. & Schecter S. M. (1983). Turnaround Strategies for Mature
Industrial-Product Business Units. Academy of Management Journal, 26 (2),
231-248.
Hofer, C. W. (1980). Turnaround Strategies. Journal of Business Strategy, 1, 19-31.
Pearce, J. A. & Robbins, D. K. (1993). Toward Improved Theory and Research on
Business Turnaround. Journal of Management, 19 (3), 613-636.
Pearce, J. A. & Robbins, D. K. (1994). Retrenchment Remains the Foundation of
Business Turnaround. Strategic Management Journal, 15 (5), 407-417.
Robbins, D. K. & Pearce, J. A. (1992). Turnaround: Retrenchment and Recovery.
Strategic Management Journal, 13 (4), 287-309.
Schendel, D. & Patton, R. (1976). Corporate stagnation and turnaround. Journal of
Economics and Business, 28, 236-241.
Schendel, D., Patton, R. & Riggs, J. (1976). Corporate turnaround strategies: A study
of profit decline and recovery. Journal of General Management, 3, 3-11.
Singapore Exchange. (2004). Singapore Exchange (SGX) Homepage. May 10, 2004,
www.singaporeexchange.com
Von Auer, L. (2003). Ökonometrie: Eine Einführung (2. Aufl.). Springer: Berlin,
Heidelberg.
1
The negotiation and decline-halting phase in South-East Asia is the second phase in the South-East
Asian turnaround process and actions might have been taken in the recognition phase already. The
strategies in this phase are comparable to the decline-stemming strategies in a Western setting
(Arogyaswamy, Barker, Yasai-Ardekani, 1995).
2
(1) Working capital/Total assets, (2) Retained Earnings/Total assets, (3) Earnings before interest and
taxes/Total assets, (4) Market value equity/Book value of total debt, (5) Sales/Total assets)
3
Altman’s Z score has proven to be quite correct in predicting insolvency for the next 2 years.
However, for longer periods of time accuracy of prediction diminishes considerably (Altman, 1968).
4
There are indications that year-specific influences of 1999 led to this result. T-tests that were done
on a trial basis for 2000, 2001 and 2002 showed t-values of more than 3 – indicating again highly
significant differences in ROA between cost retrenching and non cost retrenching companies for the
following years.
doc_446053614.pdf