JOURNAL OF MANAGEMENT ACCOUNTING RESEARCH

Volume Twenty, Special Issue

2008
pp. 4759

Assessing JIT Performance:

An Econometric Approach
Adam S. Maiga
Florida International University
Fred A. Jacobs
Auburn University

ABSTRACT: This paper uses a sample of 131 just-in-time (JIT) firms and their matched
non-JIT firms obtained from Kinney and Wempe with 19771995 Compustat data to
assess whether the relationship between JIT adoption and firm performance is endog-
enous. Results indicate a significant positive association between JIT adoption and
firm performance and strongly indicate that the decision to adopt JIT is endogenous.
We also show that asset productivity, sales growth, and leverage, are important in
explaining the effect of JIT adoption on performance and that firm characteristics are
an important contributor to unobserved heterogeneity. Furthermore, the econometric
analyses in the form of both Wooldridge 2SLS and Heckman approaches suggest that
the underlying relationship between JIT adoption and performance is much stronger
after controlling for endogeneity and self-selection bias and that OLS estimates are
indeed biased.
Keywords: just-in-time; endogeneity; treatment effect; econometric analysis.

INTRODUCTION

I
nterest in Just-in-time (JIT) practices has been increasing for over two decades. How-
ever, the findings of empirical studies that examine the relationship between JIT adop-
tion and firm performance are mixed (Anyane-Ntow 1991; Balakrishnan et al. 1996;
Fullerton and McWatters 1999; Huson and Nanda 1995; Kinney and Wempe 2002; Mia
2000; Ockree 1993). For example, Balakrishnan et al. (1996), using a matched-pairs design,
investigate 46 firms that adopted JIT over the 19851989 period and found no significant
difference in the changes to return on assets (ROA) between the treatment group and the
control group. In fact, both groups reported a decline in their ROA, leading to the conclusion
that JIT implementation did not produce better financial performance. Kinney and Wempe
(2002) (hereafter K & W) compare the profitability of JIT and non-JIT adopters, and their
results indicate that JIT adoption positively affects adopters ROA. However, the results of
additional analyses suggest that relative ROA improvement is concentrated among the ear-
liest JIT adopters (which may have the most to gain from JIT adoption), and that by the

We are grateful to Professors Michael Kinney and William Wempe for graciously providing their sampling frames
employed in their 2002 study. This paper has benefited substantially from comments by the editor and two anon-
ymous reviewers. Please do not quote or forward without permission of the authors.

47
48 Maiga and Jacobs

fifth or sixth year following JIT adoption, adopters no longer exhibit superior performance
in ROA.
The above studies preliminarily utilized univariate tests to identify the relationship
between JIT adoption and firm performance. To this end, K & W (2002) suggest more
sophisticated modeling of the JIT choice and further study of the potential endogeneity in
the JIT adoption decision.1 The purpose of this paper is to improve the statistical analysis
through econometric analyses to assess whether the relationship between JIT adoption and
firm performance is endogenous. More specifically, our econometric approach takes into
account that the assignment to the subsample of firms that adopt JIT is not random and
thereby introduces a selection bias. Both issues of endogeneity and selectivity are dealt
with by employing the traditional approach suggested in Heckman (1979) as well as the
more recent control functions methods illustrated in Wooldridge (2002).2 This research is
an important contribution, in the presence of still growing interest in the effect of JIT
adoption on firm performance.
Results indicate a significant positive association between JIT adoption and firm per-
formance and confirm that the decision to adopt JIT is endogenous. We also show that firm
characteristics, such as asset productivity, sales growth, and leverage, are important in ex-
plaining the effect of JIT adoption on performance and that firm characteristics are an
important contributor to unobserved heterogeneity. Further, the econometric analyses in the
form of Wooldridge 2SLS and the Heckman approach suggest that the underlying relation-
ship between JIT adoption and performance is much stronger after controlling for endo-
geneity and self-selection bias and that OLS estimates are indeed biased.
In the next section we develop the literature review. The third section includes the
research methods, while the fourth section reports the study results. The final section pro-
vides a discussion and suggestions for future research.

LITERATURE REVIEW
The direct costs associated with excess inventories include storage costs, cost of capital,
and the cost of obsolescence. In addition, the presence of excessive inventory levels may
also reflect poorly on the ability and competence of the firms management team and on
the effectiveness of the firms supply chain processes. Thus, carrying excess inventories
may signal a lack of coordination and collaboration among supply chain partners as well
as a lack of flexibility and agility to adjust to sudden demand shifts. This can result in
increasing uncertainty about future earnings and their growth prospects, thereby negatively
affecting the reputation of the firm and its market value (Singhal 2005).
Hendricks and Singhal (2003) investigated the stock market reaction to the public an-
nouncement by a firm that they are experiencing supply chain problems causing production
or shipping delays. Using a sample of 519 announcements made during 19892000, they
found that the supply chain problems significantly decrease shareholder value. They find
market prices react when problems in normal inventory control are large enough to require
announcement in the Wall Street Journal and the Dow Jones News Service. However, their
study neither indicates whether inefficient inventory control practices that may be reflected

1
Endogeneity is caused whenever a predictor is also a choice variable that is correlated with the random error in
the structural model. This misspecification causes the parameter estimates to be inconsistent, which renders the
interpretation of the model and hypothesis tests problematic (Ittner and Larcker 2001).
2
We used 131 sample firms of the 201 provided by K & W (2002) to investigate the effect of JIT adoption on
firm performance through econometric analyses. Complete data for only 131 firms were available through year
5 to compute Tobins Q. For consistency, the 131 firms were used to compute ROA.

Journal of Management Accounting Research, Special Issue 2008

Assessing JIT Performance: An Econometric Approach 49

in the financial statements, but not otherwise publicly announced, are deemed important to
the equity market; nor does it provide information about the importance of the market
response to inventory holdings.
Tobin (1969) argues that inventory reductions due to JIT implementation will lower
inventory carrying costs that, in turn, will improve firm financial position. He further argues
that firms highly valued by the equity market will have a high Tobins Q. Therefore, if a
JIT firm is highly valued because it leads to lower cost and higher returns, then a JIT firm
will have a particularly high Tobins Q.
If efficiencies result, the market for equity securities might view this investment as
having a positive net present value and adjust the price of the firms securities upward
(Howton et al. 2000). This is consistent with McConnell and Muscarella (1985) who find
that the announcements of capital expenditure are met with positive stock price reactions
in the market. This is also consistent with Inman and Mehra (1993) who find evidence to
suggest that JIT adoption is correlated with the future financial success of the firm. Thus,
if JIT adoption is viewed as beneficial capital expenditure, it will result in increased market
prices.
However, prior studies indicate that JIT may not be appropriate in some situations and,
indeed, JIT has not been successful for all firms (Golhar and Deshpande 1993; Inman and
Brandon 1992; Milligan 1999). Also, Balakrishnan et al. (1996) suggest that JIT adoption
does not automatically increase profit because the benefits of JIT adoption may be offset
by its many direct and indirect costs. For example, capital expenditures associated with JIT
adoption increase the asset base and the associated depreciation depresses short-term profit.
However, Sterman et al. (1997) suggest that, although in the short run, innovation adoption,
such as TQM or JIT can create operational and financial stresses, in the long run it can
increase productivity and lower costs.
In summary, although the empirical research results on the effect of JIT adoption on
firm performance have been inconclusive, none of the prior studies explicitly attempted to
control for endogeneity, which is the main issue investigated in this paper. The following
section outlines the research methods which are used in the empirical analysis.

RESEARCH METHODS
Performance Measures
We use two widely used measures of firm performance: Tobins Q as a measure of
market valuation of the firm (Woidtke 2001) and ROA as a measure of operating perform-
ance (Balakrishnan et al. 1996; Barber and Lyon 1996; Barua et al. 1995; Bharadwaj 2000;
Hitt and Brynjolfsson 1996; K & W 2002; Weill 1992). Market measures, such as Tobins
Q, reflect a great many macro- and micro-economic considerations in addition to the ac-
counting information provided to the market, including the accounting information neces-
sary to calculate ROA. Using both market and accounting measures of firm performance,
we can expect more valid research findings.
Tobins Q is used as the first measure of firm performance in this study. Although
various methods have been proposed for calculating Tobins Q, according to Chung and
Pruitt (1994), different approaches tend to yield similar Q values. In this study, to calculate
Tobins Q we use the following formula proposed by Chung and Pruitt (1994) and used in
prior studies (Lee and Tompkins 1999; Mitra and Rashid 2002)3:

3
According to Chung and Pruitt (1994), approximate q in Equation (1) explains 96.6 percent of the variability of
Tobins Q obtained by using the Lindenberg and Ross (1981) method.

Journal of Management Accounting Research, Special Issue 2008

50 Maiga and Jacobs

Tobins Q (MVE PS DEBT)/TA (1)

where:

MVE product of a firms share price and the number of common stock shares
outstanding;
PS liquidating value of the firms outstanding preferred stock;
DEBT book value of the firms short-term liabilities net of its short-term assets, plus
the book value of the firms long-term debt; and
TA book value of the total assets of the firm.

We do not calculate the replacement value of tangible assets as proposed by Lindenberg

and Ross (1981) because past research has found little qualitative difference between this
measure and the version used in this analysis (Chung and Pruitt 1994).4 Also, the main
advantage of the method of measuring Tobins Q used in this paper is that it requires less
work to calculate than the traditional method of Lindenberg and Ross (1981), while pos-
sessing a high degree of observed consistency between the resultant measures (Chung and
Pruitt 1994; Lee and Tompkins 1999; Perfect and Wiles 1994). The second measure, ROA,
is computed as income before extraordinary items (available for common stockholders),
divided by the average of the beginning and ending total assets.

Econometric Analyses
A standard way to estimate the effect of a treatment on outcomes with cross-section
data is to control for observable differences between treated and non-treated firms (i.e., JIT
and non-JIT), using Ordinary Least-Squares (OLS). Assuming that firm performance
(Tobins Q or ROA) can be explained by a linear model of the form:

Performanceit 0 1Adoptit 2 ASSETPRODit 3SALESGROWTHit

4SIZEit 5INNOVATEit 6LEVERAGEit it (2)

where Performanceit is Tobins Q or ROA. Adoptit is a dummy treatment variable that is

set equal to 1 for firms that adopt JIT, and 0 otherwise. The remaining variables are control
variables. Asset productivity (ASSETPROD) is return on cash-adjusted assets measures, the
return on book value of assets adjusted for cash balances, and short-term investments such
as marketable securities.5 Sales growth (SALESGROWTH) is the average of the change in
sales from the previous year divided by sales in the previous year. SIZE is measured
by log of firm total assets. INNOVATE is proxied by R & D spending scaled by sales.
LEVERAGE is total debt over total assets. In this basic case, 1 is the parameter of interest:
the effect of the treatment (Adoptit) on the outcome (Performance).

4
Lindenberg and Ross (1981) calculate Q via the following formula: (PREFST VCOMS LTDEBT STDEBT
ADJ) / (TOTASST BKCAP NETCAP) where PREFST is defined as the liquidating value of a firms preferred
stock, VCOMS is the price of the firms common stock multiplied by the number of shares outstanding at the
close of the year (December 31), LTDEBT is the value of the firms long-term debt adjusted for its age structure,
STDEBT is the book value of the firms current liabilities, ADJ is the value of the firms net short-term assets,
TOTASST is the book value of the firms total assets, BKCAP is the book value of the firms net capital stock,
and NETCAP is the firms inflation-adjusted net capital stock.
5
By removing cash, this study focuses on the productivity of physical assets such as equipment and machinery.
The assets held during the year in which the income is generated are deemed most relevant.

Journal of Management Accounting Research, Special Issue 2008

Assessing JIT Performance: An Econometric Approach 51

A potential selection problem in estimating Equation (2) may arise because the sample
of firms involved in JIT adoption is not random. That is, it is possible that firm performance
and the decision to adopt JIT are jointly determined (Wooldridge 2002). For simplicity,
suppose that each firms decision on whether to adopt JIT can be modeled according to the
following probit model:

Adoptit 0 1LEVERAGEit 2INNOVATEit 3EARNBONUSit

4SIZEit 5EMPLPRODit 6UNIONit it (3)

where:

LEVERAGE total debt over total assets;

INNOVATE proxied by R & D spending scaled by sales;
EARNBONUS 1 if the firm has an earnings-based bonus plan, and 0 otherwise;
SIZE measured by the natural log of firm total assets;
EMPLPROD sales per employee; and
UNION 1 if the financial press reports union negotiations for the firm, and 0
otherwise.

Additionally, industry dummy variables are also included in the regression.

We include LEVERAGE in Equation (3) because a firms capital structure may be
inversely related to its value (Smith and Watts 1992). The variable, INNOVATE, is included
to reflect the possibility that innovative firms may be more likely to adopt new technologies.
We include EARNBONUS to control for the possibility that earnings-based bonus plans are
associated with managers JIT adoption decisions. That is, it is possible that firms in which
managers are compensated through earnings-based bonus plans are more likely to adopt
JIT (K & W 2004). We include SIZE because firms with greater resources may be more
inclined to adopt new technologies. EMPLPROD is included because JIT may require a
motivated and productive work force. UNION is included because, as suggested by Inman
and Mehra (1989) labor unions may resist JIT adoption. Dummy variables for two-digit
SIC codes are used to control for industry effects. The individual error terms, it and uit,
are assumed to have a bivariate normal distribution where it N(0, ), uit N(0,1) and
corr(it, uit) .
If 0, then Adoptit is endogenous in Equation (2), and OLS estimates would be
biased. The bias of such an exogenous treatment approach is referred to as one of self-
selection or selection into treatment (Wooldridge 2002). A solution to this problem is a
treatment effects model that can provide consistent estimates of the associated parameters
(Greene 1997; Wooldridge 2002).6
It is well known that endogeneity can induce bias in the regression coefficients. To
address this issue we follow two strategies: a Wooldridge Two-Stage Least-Squares (2SLS)
approach and a Heckman approach. Following the procedure 18.1 outlined in Wooldridge
(2002, 621625), a probit Equation (3) is estimated for the variable capturing the treatment
effects, that is the Adoptit variable, and the fitted values from this estimation are included
as instrumental variables in the Wooldridge 2SLS estimation of Equation (2). In the
Heckman (1979) approach the inverse Mills ratio (IMR) from Equation (3) is used as an

6
A treatment effect is inferred if there is a statistically significant regression coefficient relating the dummy variable
to the dependent variable. Cook and Campbell (1979, 298).

Journal of Management Accounting Research, Special Issue 2008

52 Maiga and Jacobs

additional regressor in Equation (2). A t-test for the significance of the IMRs coefficient
can be considered as a test on the presence of selection bias. The Wooldridge 2SLS ap-
proach does not assume a normal distribution for the error terms, which makes it less
restrictive than the Heckmans two-stage procedure.7

RESULTS
Descriptive Statistics
Table 1, Panel A presents the sample selection, while Table 1, Panel B summarizes the
temporal distribution of the 131 treatment firms used in the study. Table 1, Panel C provides
the two-digit industry distribution for the treatment and matched control firms. As in K &
W (2002), firms are mainly represented in industries 35 and 36 (Industrial Equipment and
Commercial Machinery [including computers] and Electronic Equipment, respectively) with
48.8 percent of all firm-pairs in our sample compared to 43 percent in K & W (2002). We
also use K & Ws (2002) matching design on inventory valuation method on the treatment
and control firms.
We align the annual data of each treatment firm and its matched non-adopter in event
time based on the adoption year (year 0). We define the pre- (post) adoption period to be
five years preceding (following) adoption. Similarly, for each firm, we assess a pre- (post)
adoption variable as its mean or median value for the five-year pre- (post) adoption period.
Table 2 provides descriptive statistics of firm attributes in the pre-adoption period for
both the treatment and the control samples in terms of mean, median, and significance. As
reported in the first two columns, the mean or median assets, investment, and equity dif-
ferences are large and statistically significant, indicating that JIT adopters are, on average,
larger than their non-adopting matched firms, supporting K & W (2002). The mean (me-
dian) sales growth and return on assets differences, however, are not statistically significant.
Table 3 reports the results of the probit model of the selection equation where the
independent variable is firm likelihood to adopt JIT. Due to the non-linearity of the probit
equation, we report the marginal effect at the mean level of the independent variables.8 We
find that innovative firms are more likely to engage in JIT adoption. The results of the
probit model also indicate that the larger firms and firms with high earnings-based bonus
plans are more likely to adopt JIT.
Our multivariate tests of the hypothesis that firm performance as measured by Tobins
Q and ROA respectively, is related to JIT adoption are conducted via OLS, Wooldridge
2SLS, and Heckman (1979) two-step procedure. Table 4 reports the regression results.

OLS Model
The model (Table 4) shows that JIT adoption is significantly and positively related to
both Tobins Q and ROA. The coefficients of the control variables show, in most cases, the
expected signs. On one hand, asset productivity and sales growth are significantly positively
related to both Tobins Q and ROA, while leverage is significantly negatively related to
Tobins Q and ROA. On the other hand, size and innovation are not significant. However,
as discussed earlier, the OLS results of Equation (2) may bias the treatment effect of JIT
performance.

7
Wooldridge (2002) argues that, for the purpose of estimating treatment effects, selection models need not be
correctly specified. What matters is finding estimates of IMR and fitted values orthogonal to independent variables
in the performance equation.
8
Thus, marginal effects have been estimated so as to measure the magnitude of the effect that the explanatory
variables exercise upon the probability of firms adopting JIT.

Journal of Management Accounting Research, Special Issue 2008

Assessing JIT Performance: An Econometric Approach 53

TABLE 1
Sample Selection, Screening and Distribution

Panel A: Sample Selection and Screening

Sample Size
Sample Provided by Kinney and Wempe (2002) 201
Firms Missing Data Necessary to Calculate Tobins q in Each Year from 5 70
through 5 Relative to JIT Adoption Year
Final Sample (Matched Pairs) 131

Panel B: Distribution of JIT Adoption Years

Year Number of Firms Percentage
1983 4 3.053
1984 8 6.107
1985 9 6.870
1986 11 8.397
1987 8 6.107
1988 19 14.504
1989 17 12.977
1990 21 16.031
1991 15 11.450
1992 11 8.397
1993 8 6.107
131 100.000

Panel C: Distribution of Two-Digit Industry Classifications

Two-Digit Industry
Industry Code Description Number of Firms Percentage
25 Furniture 6 4.580
27 Printing, Publishing 2 1.527
28 Chemicals 3 2.290
30 Rubber, Plastics 4 3.053
33 Primary Metals 8 6.107
34 Fabricated Metals 9 6.870
35 Industrial Equipment 29 22.137
36 Electronic Equipment 35 26.718
37 Motor Vehicles 10 7.634
38 Instrumentation 18 13.740
39 Other Manufacturing 2 1.527
50 Wholesale Durables 1 0.763
53 Department Stores 1 0.763
59 Miscellaneous Retail 1 0.763
73 Packaged Software 2 1.527
131 100.000

Journal of Management Accounting Research, Special Issue 2008

54 Maiga and Jacobs

TABLE 2
Distribution of Mean and Median Sample Data and Comparison of Treatment and
Control Firms in the Pre-Adoption Period
(n 131 pairs)

Std. Wilcoxon
Mean Median Dev. t-statistic z-statistic
($ mil ($ mil ($ mil (difference (difference
Firm Attributes or ratio) or ratio) or ratio) in means) in medians)
Average Assets (millions)
Treatment 1817.190 304.724 5044.310 3.856 8.055
Control 903.240 153.753 2509.323 (0.000) (0.000)
Average Investment (millions)
Treatment 381.900 77.920 980.463 3.863 8.023
Control 200.140 37.626 538.120 (0.000) (0.000)
Average Equity (millions)
Treatment 866.810 136.580 2758.613 4.230 7.732
Control 349.550 69.143 915.922 (0.000) (0.000)
Inventory Turnover
Treatment 4.141 3.398 3.011 1.489 1.505
Control 4.443 3.689 3.759 (0.000) (0.066)
Current Ratio
Treatment 2.482 2.251 0.991 3.377 0.876
Control 2.809 2.294 2.072 (0.000) (0.038)
Sales Growth
Treatment 0.119 0.091 0.238 1.139 0.236
Control 0.139 0.089 0.360 (0.255) (0.407)
ROA
Treatment 0.051 0.062 0.070 1.709 2.658
Control 0.048 0.058 0.073 (0.103) (0.139)

Wooldridge 2SLS
As indicated above, a probit probabilistic choice model is used for this first stage and
the fitted probability values are estimated. Lee and Trost (1978), Barnow et al. (1981), and
Khanna and Damon (1999) show that the fitted probability value from the probit model
serves as a good instrument in the presence of selection bias. We then use the omitted
variable version of the Hausman test to examine whether selection bias is present (Kennedy
1998, 150151). If the coefficient for the fitted probability value is not significant in the
two-stage model, then we can rule out selection bias. In addition, in the absence of selection
bias, the coefficient estimates with simple OLS are unbiased.
The Wooldridge 2SLS model (Table 4) yields a significant positive relationship between
JIT adoption and outcome (Tobins Q and ROA). The coefficients of the control variables
show that asset productivity, sales growth, and leverage are significantly related to both
Tobins Q and ROA, while size and innovation are not significant. Finally, the Hausman F-
Test statistic is statistically significant at the p 0.05 level for both ROA and Tobins Q.
Therefore, we reject the null hypothesis that JIT adoption is exogenous in both equations.

Heckman Approach
Following Heckmans two-step procedure, we estimate probit model (3) to get the
coefficients. Then, these are used to estimate the inverse of Mills ratio which represents
the correction for self-selection. After incorporating in the second-stage model (Equation

Journal of Management Accounting Research, Special Issue 2008

Assessing JIT Performance: An Econometric Approach 55

TABLE 3
Probit Estimatesa

Variables Coefficient Marginal Effect

Intercept 0.609*
(4.501)
LEVERAGE 0.016 0.007
(1.095)
INNOVATE 0.897* 0.581
(3.497)
EARNBONUS 0.018* 0.007
(8.906)
SIZE 0.001* 2.93 105
(7.712)
EMPLPROD 0.423 0.166
(0.925)
UNION 0.007 0.004
(1.021)
Industry Dummy Variables yes
Pseudo R2 0.165
2 39.272*
[Prob. 2] 0.000
* Indicates significance at the 1 percent level.
a
This table reports coefficients, corresponding z-statistics in parentheses, and marginal effects for a probit model
using JIT adoption dummy variable (ADOPT ) as dependent variable. The estimates of this model are used to
calculate the fitted probabilities for the Wooldridge 2SLS and inverse Mills ratio for the Heckman approach in
Table 4. Industry dummy variables are also included in the regression.
Variable Definitions:
LEVERAGE total debt over total assets;
INNOVATE proxied by R & D spending scaled by sales;
EARNBONUS 1 if the firm has an earnings-based bonus plan, 0 otherwise;
SIZE log of firm total assets;
EMPLPROD sales per employee; and
UNION 1 if the financial press reports union negotiations for the firm, 0 otherwise.

(2)), we estimate the model with the estimated error term as an additional explanatory
variable.
Results from the Heckman approach (Table 4) show a significant positive coefficient
for the inverse of Mills ratio, the correction for self-selection, indicating that characteristics
that make firms choose to adopt JIT are positively correlated with performance, and that
the relationship between JIT adoption and performance is stronger than was initially re-
vealed in the OLS model. This implies that without control for self-selection the impact of
JIT adoption on firms Tobins Q or ROA would be understated. The coefficients of asset
productivity, sales growth, and leverage are significantly related to both Tobins Q and ROA,
while the coefficients of size and innovation are not significant. In summary, the Heckman
approach indicates that JIT adoption is endogenous.
The nonsignificant impact of size on performance is consistent with prior studies
(Schmalensee 1989; Robins and Wiersema 1995; Capon et al. 1990; OLeary-Kelly and
Flores 2002). Similarly, the nonsignificant effect of innovation on performance is consistent
with Heunks (1998) who gives empirical evidence for a profit-reducing effect of innovation.

Journal of Management Accounting Research, Special Issue 2008

56 Maiga and Jacobs

TABLE 4
Effect of JIT Adoption on Performancea

OLS Wooldridge 2SLS Heckman Approach

Tobins Q ROA Tobins Q ROA Tobins Q ROA
Intercept 0.593*** 0.534*** 0.664*** 0.610*** 0.653*** 0.591***
(4.112) (3.570) (3.129) (2.811) (3.709) (3.917)
Adopt 0.042*** 0.373*** 0.148*** 0.431*** 0.381*** 0.395***
(5.343) (3.874) (5.014) (4.908) (5.296) (4.837)
ASSETPROD 0.016*** 0.313*** 0.021*** 0.613*** 0.024*** 0.712***
(8.273) (4.994) (8.392) (3.417) (8.761) (5.631)
SALESGROWTH 0.017** 0.132 0.151*** 0.138*** 0.131*** 0.217***
(2.181) (2.254) (4.167) (3.653) (4.566) (3.991)
SIZE 0.003 0.001 0.011 0.017 0.004 0.024
(1.162) (1.386) (0.212) (1.209) (0.917) (0.813)
INNOVATE 0.012 0.003 0.001 0.010 0.001 0.011
(1.452) (1.287) (0.241) (1.471) (0.371) (1.501)
LEVERAGE 0.012*** 0.010** 0.033** 0.031* 0.036** 0.043***
(2.993) (2.397) (2.451) (1.991) (2.482) (3.493)
Industry Dummy yes yes yes yes yes yes
Variables
Hausman F-Test 46.217 38.171
[p-value] [0.000] [0.000]
0.132*** 0.164***
(t-value) (4.239) (5.109)
R2 0.059 0.031 0.134 0.092
*, **, *** Indicate significance at the 10 percent, 5 percent, and 1 percent levels, respectively.
a
This table reports coefficients for regressions of the performance measures (Tobins Q and ROA) on the JIT
adoption dummy ADOPT using the OLS, Wooldridge 2SLS, and the Heckman approach. To determine whether
the relation between JIT adoption and performance (Tobins Q and ROA) is endogenous, we use the Hausman
test statistic in the Wooldridge 2SLS model, and the significance of the inverse of Mills ratio (IMR) coefficient
in the Heckman approach. Dependent variable is Tobins Q or ROA. Industry dummy variables are also
included in the regression.
Variable Definitions:
Adopt an indicator variable that takes the value of 1 if the firm adopts JIT and 0 otherwise;
ASSETPROD return on cash-adjusted assets, measures the return on book value of assets adjusted for cash
balances, and short-term investments;
SALESGROWTH the average of the change in sales from the previous year divided by sales in the previous
year;
SIZE log of firm total assets;
INNOVATE proxied by R & D spending scaled by sales; and
LEVERAGE total debt over total assets.

CONCLUSION
The purpose of this paper is to improve the statistical analysis through econometric
analyses to assess whether the relationship between JIT adoption and firm performance is
endogenous. Thus, the contribution of this paper is to carry out econometric analyses to
assess whether the relationship between JIT adoption and firm performance is endogenous.
Our econometric analysis indicates strong and significant positive association between JIT
adoption and firm performance and represents strong evidence that the decision to adopt
JIT is endogenous. We also show that characteristics, such as asset productivity, sales
growth, and leverage, are important in explaining the effect of JIT adoption on performance.

Journal of Management Accounting Research, Special Issue 2008

Assessing JIT Performance: An Econometric Approach 57

Interestingly, the econometric analyses in the form of Wooldridge 2SLS and Heckman
approaches suggest that the underlying relationship between JIT adoption and performance
is much stronger after controlling for endogeneity and self-selection bias and that OLS
estimates are indeed biased.
This paper has some limitations. First, estimates of the relationship between JIT adop-
tion and performance are based on data drawn from the 1980s and 1990s when JIT adoption
was being widely introduced across a broad spectrum of firms. Presumably larger and more
innovative firms would have embraced this new technology. Further investigation of the
relationship is called for using a sample of current or late adopters of JIT. Second, this
study has used Wooldridge 2SLS and Heckman approaches to assess the impact of JIT
adoption on firm performance. Further studies may use alternative estimation techniques
(three-stage least-squares or maximum-likelihood methods) to assess this relationship. Also,
as more companies globalize their operations, it would be valuable to use an international
sample of firms to determine to what extent the firm performance effect of JIT adoption
can be generalized.

REFERENCES
Anyane-Ntow, K. 1991. Just-in-time manufacturing systems and inventory reported in financial state-
ments: A cross-national comparison of manufacturing firms. The International Journal of
Accounting 20: 277285.
Balakrishnan, R., T. J. Linsmeier, and M. Venkatachalam. 1996. Financial benefits from JIT adoption:
Effects of customer concentration and cost structure. The Accounting Review 71: 183205.
Barber, B. M., and J. D. Lyon. 1996. Detecting abnormal operating performance: The empirical power
and specification of test statistics. The Journal of Financial Economics 41: 359399.
Barnow, B. S., G. G. Cain, and A. S. Goldberger. 1981. Issues in the analysis of selectivity bias. In
Evaluation Studies Review Annual, edited by W. Stromdorfer, and G. Farkas. Beverly Hills,
CA: Sage Publications.
Barua, A., C. H. Kriebel, and T. Mukhopadhyay. 1995. Information technologies and business value:
An analytic and empirical investigation. Information Systems Research 6 (1): 323.
Bharadwaj, A. S. 2000. A resource-based perspective on information technology capability and firm
performance: An empirical investigation. MIS Quarterly 24 (1): 169196.
Capon, N., J. U. Farly, and S. M. Hoenig. 1990. A meta-analysis of financial performance. Manage-
ment Science 36 (10): 11431159.
Chung, K. H., and S. W. Pruitt. 1994. A simple approximation of Tobins Q. Financial Management
23 (3): 7074.
Cook, T. D., and D. T. Campbell. 1979. Quasi-Experimentation, Design and Analysis Issues for Field
Settings. Boston, MA: Houghton Mifflin.
Fullerton, R. R., and C. S. McWatters. 1999. An empirical investigation of the relationship between
JIT and financial performance. Working paper, McGill University, Montreal, Quebec.
Golhar, D. Y., and S. P. Deshpande. 1993. An empirical investigation of HRM practices in JIT firms.
Production and Inventory Management Journal 34 (4): 2832.
Greene, W. 1997. Econometric Analysis. 3rd edition. Englewood Cliffs, NJ: Prentice Hall.
Heckman, J. J. 1979. Sample selection bias as a specification error. Econometrica 47 (1): 153162.
Hendricks, K. B., and V. R. Singhal. 2003. The effect of supply chain glitches on shareholder wealth.
Journal of Operations Management 21 (5): 501533.
Heunks, F. J. 1998. Innovation, creativity and success. Small Business Economics 10: 263272.
Hitt, L. M., and E. Brynjolfsson. 1996. Productivity, business profitability, and consumer surplus:
Three different measures of information technology value. MIS Quarterly 20 (2):121142.
Howton, S. D., E. J. Higgins, and T. B. Biggart. 2000. The information content of just-in-time inven-
tory system adoption announcement. Journal of Business Finance & Accounting 27 (5/ 6): 711.

Journal of Management Accounting Research, Special Issue 2008

58 Maiga and Jacobs

Huson, M., and D. Nanda. 1995. The impact of just-in-time manufacturing on firm performance in
the U.S. Journal of Operations Management 9 (1): 514.
Inman, R. A., and S. Mehra. 1989. Potential union conflict in JIT implementation. Production and
Inventory Management Journal 30 (4): 1922.
, and L. D. Brandon. 1992. An undesirable effect of JIT. Production and Inventory Manage-
ment Journal: 5558.
, and S. Mehra. 1993. Financial justification of JIT implementation. International Journal of
Operations & Production Management 13 (4): 3239.
Ittner, C., and D. Larcker. 2001. Assessing empirical research in managerial accounting: A value-
based management perspective. Journal of Accounting and Economics 32: 349410.
Kennedy, P. 1998. A Guide to Econometrics. 4th edition. Cambridge, MA: The MIT Press.
Khanna, M., and L. A. Damon. 1999. EPAs voluntary 33 / 50 program: Impact on toxic releases and
economic performance of firms. Journal of Environmental Economics and Management 37: 1
25.
Kinney, M., and W. Wempe. 2002. Further evidence on the extent and origins of JITs profitability
effects. The Accounting Review 77: 203225.
, and . 2004. JIT adoption: The effects of LIFO reserves and financial reporting and
tax incentives. Contemporary Accounting Research 21 (3): 603638.
Lee, D. E., and J. G. Tompkins. 1999. A modified version of the Lewellen and Badrinath measure of
Tobins Q. Financial Management 28: 2031.
Lee, L., and R. P. Trost. 1978. Estimation of some limited dependent variable models with application
to housing demand. Journal of Econometrics 8: 357382.
Lindenberg, E., and S. Ross. 1981. Tobins Q ratio and industrial organization. Journal of Business
54: 132.
McConnell, J., and C. Muscarella. 1985. Corporate capital expenditure decisions and the market value
of the firm. Journal of Financial Economics 14 (3): 399422.
Mia, L. 2000. Just-in-time manufacturing, management accounting systems and profitability. Account-
ing and Business Review 30 (2): 137151.
Milligan, B. 1999. Whats it going to take to make it work? Purchasing 127: 4044.
Mitra, D., and M. Rashid. 2002. Dividend omission announcement effects, the firms information
environment, earning volatility and Q. Journal of American Academy of Business 2 (1): 3341.
OLeary-Kelly, S. W., and B. E. Flores. 2002. The integration of manufacturing and marketing / sales
decisions: Impact on organizational performance. Journal of Operations Management 20: 221
240.
Ockree, K. A. 1993. A just-in-time production philosophy: Empirical analyses and field study. Un-
published doctoral dissertation, University of Kansas.
Perfect, S., and K. Wiles. 1994. Alternative construction of Tobins Q: An empirical comparison.
Journal of Empirical Finance 1: 313341.
Robins, J., and M. F. Wiersema. 1995. A resource-based approach to the multibusiness firm: Empirical
analysis of the portfolio interrelationships and corporate financial performance. Strategic Man-
agement Journal 16: 277299.
Schmalensee, R. 1989. Inter-industry studies of structure and performance. In Handbook of Industrial
Organization, edited by R. Schmalensee, and R. Willig., Vol. 2. 9511011. Amsterdam, North
Holland.
Singhal, V. R. 2005. Excess inventory and long-term stock price performance. Working paper.
Smith, C. W., Jr., and R. L. Watts. 1992. The investment opportunity set and corporate financing,
dividend, and compensation policies. The Journal of Financial Economics 32: 263292.
Sterman, J. D., N. P. Repenning, and F. Kofman. 1997. Unanticipated side effects of successful quality
programs: Exploring a paradox of organizational improvement. Management Science 43 (4):
503521.
Tobin, J. 1969. A general equilibrium approach to monetary theory. Journal of Money, Credit and
Banking 1: 1529.

Journal of Management Accounting Research, Special Issue 2008

Assessing JIT Performance: An Econometric Approach 59

Weill, P. 1992. The relationship between investment in information technology and firm performance:
A study of the valve manufacturing sector. Information Systems Research 3 (4): 307333.
Woidtke, T. 2001. Agents Watching Agents? Evidence from Pension Fund Ownership and Firm Value.
College Station, TX: Mimeo, Texas A&M University.
Wooldridge, J. 2002. Econometric Analysis of Cross Section and Panel Data. London, U.K.: MIT
Press.

Journal of Management Accounting Research, Special Issue 2008