Culture Aspect of Job Relevant in Budgeting

8/12/2019 Culture Aspect of Job Relevant in Budgeting

1/33

JOURNAL OF MANAGEMENT ACCOUNTING RESEARCH Volume Nineteen2007pp. 105136

Differences in the Role of Job-RelevantInformation in the Budget Participation-

Performance Relationship among U.S. andMexican Managers: A Question of

Culture or Communication

Maria A. Leach-Lopez Auburn University Montgomery

William W. Stammerjohan Louisiana Tech University

Frances M. McNair Mississippi State University

ABSTRACT: This study extends the stream of participative budgeting literature and,specically, the work of Frucot and Shearon (1991). This study employs an expandedversion of the path model introduced to this literature by Kren (1992) to examine andcompare the budget participation-performance relationship for U.S. and Mexican mid-level managers. The expanded path model allows the examination of both the directeffects of budget participation on performance and the indirect effects of budget par-ticipation on performance that run through job satisfaction and job-relevant information.

The primary ndings of this study are that while there are strong associationsbetween budget participation and performance for both U.S. managers working in theU.S. and Mexican managers working for U.S.-controlled maquiladoras in Mexico, thecausal mechanisms connecting budget participation to performance are quite differentbetween these two groups. The information-communication aspect of the budget

participation-performance relationship is much stronger among our Mexican managersand strongest among our Mexican managers who may face the greatest psychic dis-tance from their U.S. parent companies: those who are not bilingual, and/or those whoare supervised by U.S. nationals.

INTRODUCTION

T he relationship between budget participation and performance has a long history inthe managerial accounting literature (e.g., Argyris 1952; Brownell 1981, 1982a,1983; Frucot and Shearon 1991; Kren 1992; Chow et al. 1991; Chow et al. 1994;We thank M. Alicia Williamson for her unselsh assistance with translations. We thank Claire Stammerjohan andDeborah Seifert for their assistance and suggestions over the entire course of our research. We thank the participantsof the American Accounting Association 2005 Diversity Section Meeting and three anonymous referees for theirhelpful comments We also thank Joan Luft (editor) for her patience and guidance


2/33

106 Leach-Lopez, Stammerjohan, and McNair

Chow et al. 1996; Chow et al. 1999). Using meta-analysis, Greenberg et al. (1994) sum-marized the empirical literature up to that time and found that, on average, the positivecorrelations between participation and performance had been supported by both survey andexperimental studies.

We contribute to this literature by examining the relationship between budget partici-pation and performance among Mexican managers working for U.S.-controlled maqui-ladoras in Mexico. This setting is important because U.S. companies continue to moveoperations into Mexico (Lindquist 2001; The Economist 2000) and because U.S. multina-tionals export U.S.-based management control systems to manage their foreign operations(Harrison 1992; Harrison and McKinnon 1999). With the passage of the North AmericanFree Trade Agreement (NAFTA), U.S. companies have continued to expand the maquila-dora system, factories that import materials and/or parts and make goods for re-export. The Economist (2000) documents that maquila exports from Mexico have been growing at 20percent per year since the implementation of NAFTA, and Lindquist (2001) documents thatforeign companies spent $11.8 billion on maquila operations between 1994 and 1999. Basedon U.S. Department of State information, Greer and Stephens (2001) document that in 1999the U.S. consumed 88 percent of Mexicos exports and provided 78 percent of Mexicosimports.

While there is a large body of prior research on the relationship between budget par-ticipation and performance, and while the effects of management techniques employed inAsia have received considerable attention (e.g., Birnberg and Snodgrass 1988; Harrison1992; Ueno and Sekaran 1992; Harrison et al. 1994; Chow et al. 1991; Chow et al. 1994;Chow et al. 1996; Chow et al. 1999; Shields et al. 2000), Frucot and Shearon (1991) remainsthe lone study examining the effects of participative budgeting within Mexico. They theorizethat culture is an important variable in the budget participation-performance and budget

participation-job satisfaction relationships. Although Frucot and Shearons (1991) hypoth-eses are two-sided, their implied expectation is that the cultural differences between theU.S. and Mexico documented by Hofstede (1980) would negate the benets of participativebudgeting in Mexico. However, contrary to this expectation, Frucot and Shearon nd someevidence of positive relationships between budget participation and performance and be-tween budget participation and job satisfaction in Mexico.

The ndings of Frucot and Shearon (1991) coupled with the subsequent passage of NAFTA leave several unanswered questions: (1) Can parent companies expect budget par-ticipation to be associated with performance among their Mexican managers in U.S.-controlled maquiladoras? (2) Are there signicant cultural differences between Mexicanmanagers currently employed by U.S.-controlled maquiladoras and their U.S. counterparts?(3) Do these cultural, or other, differences lead to differences in the basic level of correlationbetween budget participation and performance and/or any differences in the causal mech-anisms connecting budget participation to performance between U.S. managers working inthe U.S. and Mexican managers working for U.S.-controlled maquiladoras? The purposeof our study is to address these questions.

We offer several renements to the existing literature by concentrating on the manage-ment environment faced by U.S.-controlled maquiladoras. Our renements include: (1) anupdated examination of the cultural measurement literature beyond the original work of Hofstede (1967, 1980, 1983a, 1983b); (2) direct measurement of Hofstedes cultural di-mensions over our samples; (3) including a U.S. sample to insure that our Mexican results

are not instrument-driven; (4) using mid-level managers working in manufacturing to focuson the maquiladora environment; (5) using managers from both the U.S.-Mexican borderand interior Mexico to control for potential regional differences; and (6) employing an


3/33


4/33


Our path model includes one independent variable, budget participation ( PART ), onedependent variable, performance ( PERF ), and two intervening variables, job satisfaction(SAT ) and job-relevant information ( JRI ). The six theorized relationships between thesevariables are described as Paths AF in Figure 1. Prior literature on each relationship is

summarized below. Path A: PART PERF

Shields and Shields (1998) cite three studies that found direct positive relationshipsbetween budget participation and performance (Milani 1975; Kenis 1979; Brownell andMcInnes 1986). In a path analysis study, Chenhall and Brownell (1988) failed to ndevidence of a direct effect. Frucot and Shearon (1991) nd mixed results on this relation-ship. We include Path A in our model to capture the direct effect and to provide compar-ability with prior research.

Path B: PART SAT

Shields and Shields (1998) cite Hofstede (1967) as a study that found a positive rela-tionship between budget participation and satisfaction. Chenhall and Brownell (1988) ndboth a signicant direct relationship between budget participation and satisfaction and asignicant indirect relationship between budget participation and satisfaction runningthrough role ambiguity. Frucot and Shearon (1991) nd a signicant positive relationshipbetween budget participation and satisfaction among Mexican managers working at thehighest levels of companies that are less than 100 percent foreign-owned. Lau and Tan(2003) nd a signicant path coefcient linking budget participation and job satisfactionamong Singaporean managers.

Path C: SAT PERF

To better understand the paths between participation and performance, we include thisrelationship in our model that is commonly assumed, but not tested, in the existing budgetparticipation-performance literature. 1

Path D: PART JRI Kren (1992) introduced a path model with job-relevant information as an intervening

variable between budget participation and performance. Kren (1992) theorized and foundevidence that increased budget participation leads to increased job-relevant information andthat increased job-relevant information leads to increased performance. Many other studieshave theorized and/ or found evidence of either a positive relationship between budgetparticipation and information or a negative relationship between participation and role am-biguity (e.g., Chenhall and Brownell 1988; Brownell and Dunk 1991; OConnor 1995;Hassel and Cunningham 1996; Magner et al. 1996; Covaleski et al. 2003; Lau and Tan2003).

Path E: JRI SAT Lau and Tan (2003) theorize and nd a positive relationship between job-relevant in-

formation and satisfaction. Insofar as job-relevant information reduces role ambiguity, the

1 See Luft and Shields (2003, Appendix A, 209211) for a comprehensive mapping of numerous budgetingrelationships modeled at the individual employee (manager) level in the extant literature. Note further that amongthe 42 studies mapped in Appendix A, 21 include performance as a dependent variable, 11 include satisfactionas a dependent variable, and that none model the relationship between satisfaction and performance modeledby our Path C.


5/33

Budget Participation-Performance among U.S. and Mexican Managers 109

FIGURE 1

The Path Model

Job Relevant Information JRI

Job SatisfactionSAT

Budget ParticipationPART

PerformancePERF

D

B

EF

A

C

negative relationship between role ambiguity and satisfaction theorized and found byChenhall and Brownells (1988) also supports our expectation of a positive relationshipbetween job-relevant information and satisfaction.

Path F: JRI PERFPath F represents the nal path in Krens (1992) model of an indirect relationship

between budget participation and performance running through job-relevant information.Our expectation of a positive relationship between job-relevant information and perform-ance is supported by Krens (1992) ndings and by the logic attributed to the psychology-based literature by Covaleski et al. (2003, 37): Participation in this sense can improveperformance by providing a forum for the superior to communicate information to subor-dinates that they can use to coordinate their efforts with others or choose actions withhigher returns.

Formal Budget Participation-Performance HypothesesOur rst simple hypothesis, H1, is that there is a positive relationship between budget

participation and performance modeled by Path A when the other path coefcients arerestricted to zero. Our second simple hypothesis, H2, is that there is a positive directrelationship between budget participation and performance modeled by Path A when the

other paths are included in the model. The remaining formal budget participation-performance hypotheses are joint hypotheses of the relationships modeled by Paths BF.Support for H3H5 requires signicance of each path included in the respective hypothesis.Hypothesis 4 represents the joint hypothesis proposed by Kren (1992). Hypotheses 3 and5 include our addition of job satisfaction as a second intervening variable. The ve formalbudget participation-performance hypotheses stated in the alternative form are:

H1: Increased budget participation leads directly to increased performance.Path A: PART PERF (with path coefcients BF restricted to zero)

H2: Increased budget participation leads directly to increased performance.Path A: PART PERF (with path coefcients BF included in the model)


6/33


H3: Increased budget participation leads to increased job satisfaction and increased jobsatisfaction leads to increased performance.Path B and Path C: PART SAT PERF

H4: Increased budget participation leads to increased job-relevant information and in-creased job-relevant information leads to increased performance.Path D and Path F: PART JRI PERF

H5: Increased budget participation leads to increased job-relevant information, in-creased job-relevant information leads to increased job satisfaction, and increased job satisfaction leads to increased performance.Path D, Path E, and Path C: PART JRI SAT PERF

Cultural Differences between the U.S. and Mexico

Prior research summarized in the following subsection (Effects of Culture on the

Budget Participation-Performance Relationships) has found signicant (though inconsis-tent) associations between Hofstedes dimensions of national culture, budgetary participa-tion, and the effects of budgetary participation. Before we propose national-culture differ-ences in our budgetary participation-performance model, we briey introduce Hofstedesdimensions of national culture and then consider reasons why Hofstedes original measuresmight not discriminate well between present-day samples of U.S. and Mexican managers.

HofstedeHofstedes original measures of cultural factors are: Power Distance, Individualism

versus Collectivism, Uncertainty Avoidance, and Masculinity versus Femininity (Hofstede1983b, 78). 2 Sivakumar and Nakata (2001) nd that Hofstede (1980) was cited 1,101 timesbetween 19871997 in a variety of disciplines including accounting, marketing, and man-agement. As noted by Chenhall (2003, 153), Virtually all MCS [Management ControlSystem] contingency-based studies have used [Hofstedes] values to study the inuence of culture. Chenhall (2003) goes on to note that this science has not moved beyond Hofstedesoriginal measures and that most studies rely on his original characterization of nationalcultures and do not directly measure culture. 3 We believe it would be inappropriate to relyon Hofstedes original measurements of Mexican and U.S. cultural dimensions, because of possible measurement instability and selection bias in our sample, as described below.

Measurement Instability

While Hofstedes original measures have been widely used within management ac-counting research, there have been some recent attempts to re-measure Hofstedes dimen-sions. OConnor (1995) provides evidence of variability in Hofstedes power distance andindividualism measures between different samples and subsamples of Singaporean-Chinesemanagers. While Hofstede (2001) provides a variety of evidence that his cultural dimensionshave remained stable over time, Fernandez et al. (1997) provide evidence of temporal shiftsin cultural dimensions in the U.S. and Mexico.

2 We do not include Hofstedes fth cultural dimension, Long-Term, captured in his later studies because he didnot measure this dimension in Mexico (Hofstede 1994b).

3 While Baskerville (2003) questions whether nation can be equated with culture and whether Hofstedes dimen-sions are measures of culture or of other national tendencies, her arguments may not be directly relevant forour purposes. The fact is that Hofstede remains the most cited culture reference in the accounting literature andthat whether his measures are cultural dimensions or other measures of national tendencies may be of littleconsequence to those interested in predicting behavior and interactions in the workplace


7/33


Selection Bias and TrainingEven if Hofstedes measures are still representative of the Mexican and U.S. populations

as a whole, the Mexican managers working for U.S.-owned maquiladoras may be morelike their U.S. counterparts than a random sample of Mexican citizens. OConnor (1995)

provides evidence that organizational culture may move toward the national culture of theparent organization through the effects of selection, socialization, and/or training. Harrisonand McKinnon (1999) nd evidence that multinationals modify host-country culture asopposed to modifying their management control systems to t host cultures. Chenhall (2003,154) states that It is possible that strong organizational culture may dominate nationalculture in the work situation.

Cultural Difference HypothesisBecause it is not clear ex ante what, if any, signicant differences will exist on

Hofstedes dimensions between the Mexican and U.S. managers in our sample, we plan totest directly for such differences before proceeding to compare the budgetary-participation-performance relationships across Mexican and U.S. samples. We therefore state the follow-ing hypothesis:

H6: There are differences on Hofstedes national-culture dimensions between Mexicanmanagers working for U.S.-controlled maquiladoras in Mexico and U.S. managersworking for U.S. companies in the U.S.

Effects of Culture on the Budget Participation-Performance Relationships Relevant Literature

Researchers have proposed associations between budgetary participation and Hofstedes

dimensions of power distance, uncertainty avoidance, and collectivism, but evidence onthese associations has been mixed. Frucot and Shearon (1991, 85) expect the benets of budget participation on performance to be negated among Mexican managers because of the high power distance and high uncertainty avoidance documented by Hofstede (1980).Harrison and McKinnon (1999, 494496) indicate that high collectivism should be asso-ciated with a preference for participation. Sivakumar and Nakata (2001, 556) nd thatMexican managers are more apt to use participatory management techniques than theirAmerican counterparts [and that] collectivism is said to engender this effect.

Contrary to Harrison and McKinnon (1999) and Sivakumar and Nakata (2001), Tsui(2001) suggests that managers in collectivist cultures will desire less participation in thedecision making process. OConnor (1995, 392) nds that Singaporean managers workingfor foreign-owned subsidiaries exhibited less power distance and preferred a more consul-tative management style as compared to their counterparts working for locally owned com-panies. In their study of maquiladora automotive workers, Peterson et al. (2003) documentthat effective cross-cultural management of the Mexican workforce is much more compli-cated than would be expected based on commonly accepted cultural typologies.

Formal Cultural Effects HypothesesGiven the mixed expectations and results cited above, our two cultural effects hypoth-

eses are two-sided. Hypothesis 7 tests the question of whether there is a difference in thebasic budget participation-performance relationship between Mexican and U.S. managers.

Hypothesis 8 tests the more complex question of whether there are differences in the causalmechanisms, i.e., differences in the relationships modeled in H2H5 between U.S. man-agers and Mexican managers working for U S -controlled maquiladoras in Mexico


8/33


H7: There is a difference in the basic budget participation-performance relationship (ashypothesized in H1) between Mexican managers working for U.S.-controlled ma-quiladoras in Mexico and U.S. managers working for U.S. companies in the U.S.

H8: There are differences in the complex budget participation-performance relation-ships described in H2H5 between Mexican managers working for U.S.-controlledmaquiladoras in Mexico and U.S. managers working for U.S. companies in theU.S.

RESEARCH DESIGN

The sample for this study includes U.S. managers working in the U.S. and Mexicanmanagers working for U.S.-controlled maquiladoras in Mexico. The data were collected byhaving each mid-level manager complete a survey instrument that was written in eitherEnglish or Spanish. 4 Hypotheses 15 (budget participation to performance) and H7H8(cultural effects), are tested by estimating the models described below. Hypothesis 6 (dif-

ference in culture) is tested by analyzing the Hofstede scores over our samples.5

Path Model Variables

The level of budget participation ( PART ) is measured with a six-item scale developedby Milani (1975). This measure sums each item score measured with a seven-point Likert-type scale. This scale has been widely used in participative budget research (e.g., Brownell1982c, 1983; Chenhall and Brownell 1988; Mia 1989; Frucot and Shearon 1991; Nouri andParker 1998; Tsui 2001; Lau and Tan 2003; Hassel and Cunningham 2004). This scale hasconsistently produced reliable Cronbachs (Cronbach 1951) coefcients from 0.71(Chenhall and Brownell 1988) to 0.91 (Mia 1989).

Performance ( PERF ) is measured with an eight-dimension scale developed by Mahoneyet al. (1963). This scale measures eight performance dimensions: planning, investigating,coordinating, evaluating, supervising, stafng, negotiating, and representing. PERF is thesum of these eight individual measures. 6 The appropriateness of using self-reported mea-sures of performance and the reliability of the Mahoney et al. (1963) scale are well doc-umented (Heneman 1974). This scale has been widely used in participative budgeting re-search (e.g., Brownell 1982b, 1983; Brownell and Hirst 1986; Brownell and McInnes 1986;Frucot and Shearon 1991; Tsui 2001). Self-reported performance is also used in other recentstudies. Nouri and Parker (1998) employ a self-reported measure of performance adaptedfrom Govindarajan and Gupta (1985), and Shields et al. (2000) develop their own self-reported measure of performance. 7

The level of job satisfaction ( SAT ) is measured with the short-form of the MinnesotaSatisfaction Questionnaire (MSQ) (Weiss et al. 1967). The short-form version of the MSQhas been supported for its reliability and validity (Weiss et al. 1967), and has been used

4 Two procedures were performed to insure the equivalency of the English and Spanish versions of our instrument.First, two individuals who both were born in Mexico, residing in the U.S., and both highly procient in Englishand Spanish (including one of the authors) prepared translations of the English version into Spanish. The fewdifferences were then corrected as needed. Second, following the method suggested by Hui and Triandis (1985),three subjects answered both the English and Spanish versions. All three sets of answers were consistent acrossboth languages and indicated no need for further changes in the instrument.

5 Each manager provided demographic information and answered questions used to measure each of our scalevariables and to calculate the Hofstede scores. The questionnaire is presented in the Appendix.

6 The Mahoney et al. (1963) scale also includes a single overall measure of performance that is commonly usedas a validity check. PERF was successfully validated in the current study by measuring the correlation betweenthe single overall measure ( PER9 ) and PERF (Pearson correlation 0.807, p 0.0001).

7 See Bommer et al (1995) for a complete discussion of the implications of self-reported performance measures


9/33


extensively in both applied psychology (e.g., Pulakos and Schmitt 1983; Butler 1983) andmanagerial accounting research (e.g., Brownell 1981, 1982b, 1982c; Harrison 1992, 1993;Lau and Tan 2003). 8

The level of job-relevant information ( JRI ) is measured with a scale developed by Kren

(1992). The objective of this variable is to measure the managers perception of the avail-ability of information for effective job-related decisions. JRI consists of the sum of threequestions answered on a scale of 1 to 5, with anchors of strongly disagree and stronglyagree. Kren (1992) documented the validity of his scale that continues to be used in themanagement accounting literature (e.g., Lau and Tan 2003). 9

Cultural Variables

Our primary indicator of culture is whether the subject was a U.S. manager workingin the U.S. or a Mexican manager working for a U.S.-controlled maquiladora in Mexico.In our supplemental analyses and robustness checks we further differentiate culture intowhether the Mexican manager was working on the U.S.-Mexican border or in interiorMexico. We calculate Hofstede cultural dimension scores for power distance ( PDI ), uncer-tainty avoidance ( UAI ), individualism ( IDV ), and masculinity ( MAS ) over our U.S., Mexi-can, Interior, and Border samples using Hofstedes VSM94 (Hofstede 1994a). 10

Other Variables

In our supplemental analyses and robustness checks, we include whether the Mexicanmanager was bilingual or non-bilingual and whether the manager was supervised by asupervisor from his/her home country. We also capture several variables that we use in ourrobustness tests that include sex, age, education, years with the company, years in currentposition, and the number of employees, number of managers, and number of hourly em-ployees supervised.

Basic Model

Our test of H1 is the simple correlation between budget participation ( PART ) and per-formance ( PERF ) represented by Model (1): 11

8 We use the modied rating categories advocated by Weiss et al. (1967). The modied ratings anchor on notsatised and extremely satised. This modication overcomes the ceiling effect of response means locatedclose to the maximum possible score when the categories are anchored on very dissatised and very sat-ised, and centered on neither satised nor dissatised.

9 For ease of interpretation, we measure PART , PERF , SAT , and JRI so that higher scores are associated with

higher levels, i.e., higher scores indicate higher levels of budget participation, performance, job satisfaction, and job-relevant information.10 The Hofstede cultural dimension scores for the current samples are calculated by applying the following formulas

to the average responses to selected VSM94 questions for each sample (Hofstede 1994a):

PDI 35( Q3) 35( Q6) 25( Q14) 20( Q17) 20

UAI 25( Q13) 20( Q16) 50(Q18) 15( Q19) 120

IDV 50( Q1) 30( Q2) 20( Q4) 25(Q8) 130

MAS 60( Q5) 20( Q7) 20( Q15) 70( Q20) 100

Please see the Appendix for the VSM94 questions used in these calculations.11 All of our models are estimated over standardized versions of our primary variables ( PART , PERF , SAT , and

JRI ) i e there are no intercept terms in our basic models


10/33


PERF PART e . (1)i 1 i i

An interaction term is added to Model (1) to test H7, the incremental effect of the managerbeing Mexican. The expanded model is represented by Model (2):

PERF MX PART PART * MX e (2)i 0 i 1 i 2 i i i

where:

PERF performance measured using the Mahoney et al. (1963) eight-item scale;PART budget participation measured as the sum of the six-item scale developed by

Milani (1975); and MX 1 if the manager is Mexican, and 0 if the manager is U.S.

Path Model

While multiple regression is an appropriate tool for measuring the incremental effectsof two or more independent variables on a single dependent variable, multiple regressiondoes not capture the subtleties of a phenomenon where the correlation between an explan-atory variable and the dependent variable is not direct, but ows through one or moreintervening variables. Luft and Shields (2003, 191) provide a clear explanation of why anadditive model (multiple regression) would be inappropriate for our analysis.

Our path model captures both the direct and indirect effects of budget participation onperformance by including budget participation as the sole independent variable, perform-ance as the sole dependent variable, and satisfaction and job-relevant information as inter-vening variables. 12 The path coefcients are estimated by solving the set of simultaneousequations represented by Model (3):

PERF A*PART C *SAT F * JRI ei i i i 1 i

SAT B*PART E * JRI ei i i 2 i

JRI D*PART e (3)i i 3 i

where:


Milani (1975);

12 Other studies have also found signicant intervening variables between participative standard setting and jobperformance (e.g., Nouri and Parker 1998; Shields et al. 2000). Nouri and Parker (1998) found that the pathsbetween budget participation and performance that ran through budget adequacy and organizational commitmentwere both signicant. Shields et al. (2000) found that the paths between participative standard setting andperformance that ow through standard tightness and the level of standard-based incentives are only signicantwhen the level of job-related stress was inserted between standard tightness and job performance, and betweenthe level of standard-based incentives and job performance. In the interest of a parsimonious model, we restrictour intervening variables to the two most common variables found in the extant literature, job-relevant infor-mation and job satisfaction, as documented by Shields and Shields (1998).

While our path model falls into the causes and effects of budgeting at the individual level described bythe map in Luft and Shields (2003, Appendix A, 209), our path model is not fully described by any of thespecic maps in their appendix. While some of the studies reviewed by Luft and Shields (2003) included job-relevant information as an intervening variable job satisfaction was only included as a dependent variable


11/33


SAT job satisfaction measured using a modied form of the Minnesota SatisfactionQuestionnaire (Weiss et al. 1967); and

JRI job-relevant information measured as the sum of the three-item scale developedby Kren (1992).

The estimated path coefcients AF are used to test H2H5. Differences in the esti-mated path coefcients between the U.S. and Mexican samples are used to test H8.

SAMPLE, SUMMARY STATISTICS, AND TESTSOF CULTURAL DIFFERENCES

SampleAll of the subjects in our study are mid-level managers working for U.S. manufacturing

companies located in the U.S., or U.S.-controlled maquiladoras located in Mexico. Eachmanager completed a survey instrument written in either English or Spanish. The U.S.subjects work for 15 manufacturing companies located in the southeastern U.S. TheU.S. responses were obtained by gaining permission from the HR manager and then sendingthe survey instruments to the HR manager by email for distribution. The U.S. responseswere then collected via email.

The border subjects are Mexican managers working for 20 maquiladoras located in, ornear, Nuevo Laredo. The interior subjects are Mexican managers working for 29 maqui-ladoras located in, or near, Puebla, a metropolitan city located about 250 miles south of Mexico City. The Mexican companies were obtained from lists of 39 active U.S.-controlledmaquiladoras in the Nuevo Laredo region and 49 active U.S.-controlled maquiladoras inthe Puebla region. Both lists were provided by Mexicos Department of Commerce. Theresearch instruments for both Mexican samples were hand-delivered to the maquiladoras

and the completed instruments were hand collected a few days later.13

Demographic StatisticsTable 1 provides the ranges and sample means for demographic variables. The nal

sample includes 143 managers, 45 working in the U.S., 58 working on the U.S.-Mexicanborder, and 40 working in interior Mexico. The number of usable responses available forthe calculation of variable means and for drawing statistical comparisons between the sam-ples are reported in the column labeled n.

The U.S. versus Mexico, and Border versus Interior, sample means that differ at p 0.05 are reported in bold. The subsamples differed demographically in several respects.

(Robustness tests reported later in the paper indicate that these demographic differences donot explain our participation-performance effects.) While the U.S. sample has a signicantlylarger portion of male managers than does the Mexican sample, 89 percent versus 70percent, p 0.0068, the proportion of males in the Border and Interior samples is notstatistically different, 71 percent versus 69 percent, p 0.8793. Age is reported in ranges,

20 1, 2024 2, 2529 3, 3034 4, 3539 5, 4049 6, 5059 7, and 59 8. The mean age for U.S. subjects (5.84) indicates that the average U.S. subject

was in the range of late-thirties to late-forties and the mean age for Mexican subjects (5.05)indicates that the average Mexican manager was in the range of mid-thirties to early-forties,p 0.0093. The average number of years of school was not statistically different acrossany of the samples, p 0.2426.

13 The company/maquiladora response rates are 55.6 percent (15/27), 51.3 percent (20/39), and 59.2 percent (29/ 49) for the U S Mexican border and Mexican interior samples respectively


12/33


TABLE 1Demographic Variable Ranges and Sample Means

Variable Description Range n ALL U.S. Mexico Border Interior

n 143 45 98 58 40Sex 141 76% 89% 70% 71% 69%Age, reported in ranges 18 133 5.27 5.84 5.05 4.95 5.21Number of years of school 620 135 15.70 16.00 15.56 15.67 15.41Percentage of bilingual managers 140 64% 24% 83% 88% 76%Years with company 640 130 9.41 13.23 7.77 6.50 10.00Years in current position 040 128 4.91 4.26 5.20 3.73 7.70Number supervised 0472 128 26.84 14.59 31.82 37.30 22.65Number of managers supervised 050 127 6.57 6.08 6.77 7.37 5.73Number of hourly employees supervised 0450 127 21.09 10.22 25.57 29.93 18.03

U.S. versus Mexico sample means, and Border versus Interior sample means, that differ at p 0.05 are reported

in bold.Sex is coded: male 1 and female 0.Age is reported in ranges: 20 1, 2024 2, 2529 3, 3034 4, 3539 5, 4049 6, 5059 7,and 60 8.

The Mexican sample had a much larger percentage of bilingual managers than the U.S.sample, 84 percent and 25 percent, respectively, p 0.0001. Although the U.S. managershad worked for their respective companies longer, 13.23 years versus 7.77 years, p

0.0093, they had not been in their respective current positions for signicantly longerperiods, 4.26 years versus 5.20 years, p 0.4076. The interior managers had been in theircurrent positions for more years than their border counterparts, 7.70 years versus 3.73 years,p 0.0169.

Path Variable MeansTable 2 reports the sample means and the differences in sample means for the four

path model variables, budget participation ( PART ), performance ( PERF ), job satisfaction(SAT ), and job-relevant information ( JRI ). These results are reported in Panels A, B, C, andD, respectively. The sample means are reported for the 45 U.S. managers, the 98 Mexicanmanagers, the 58 Mexican managers working on the U.S.-Mexican border, and the 40Mexican managers working in interior Mexico. Separate Cronbach s are presented for

the U.S. and Mexican samples to provide measures of reliability. All of the scales appearreliable at conventional levels, Cronbachs 0.80. The differences between the samplemeans and the corresponding t-statistics and p-values are presented in the right columns.Sample means that differ at p 0.05 are presented in bold.

The Mexican managers reported higher average performance scores, PERF , p 0.0295and higher average levels of job-relevant information, JRI , p 0.0002. The mean valuesfor the four path variables are not signicantly different between the border and interiorsamples, p 0.5063.

Path Variable Simple Correlations

Table 3 reports the simple correlations between the four path model variables and sevenof the demographic variables. Panel A reports results over the U.S. sample and Panel B


13/33


TABLE 2Path Model Variable Summary Statistics

Sample n Mean Std. Dev. Cronbachs Difference t-statistic p-value

Panel A: Budget Participation ( PART )U.S. 45 25.600 7.460 0.88Mexico 98 26.929 9.971 0.89 1.329 0.89 0.3778Border 58 26.448 10.231Interior 40 27.625 9.067 1.177 0.58 0.5646

Panel B: Performance ( PERF )

U.S. 45 50.267 8.010 0.80Mexico 98 53.898 11.241 0.89 3.631 2.20 0.0295Border 58 53.431 11.041Interior 40 54.575 11.633 1.144 0.49 0.6265

Panel C: Job Satisfaction ( SAT )

U.S. 45 69.933 11.042 0.93Mexico 98 73.010 12.403 0.91 3.077 1.49 0.1402Border 58 73.552 13.207Interior 40 72.225 11.251 1.327 0.53 0.5946

Panel D: Job Relevant Information ( JRI )

U.S. 45 15.222 3.111 0.86Mexico 98 17.469 3.253 0.85 2.247 3.95 0.0002Border 58 17.655 3.138

Interior 40 17.200 3.436 0.455 0.67 0.5063Sample mean differences with p-values 0.05 are reported in bold.PART budget participation measured as the sum of the six-item scale developed by Milani (1975);PERF performance measured using the Mahoney et al. (1963) eight-item scale;

SAT job satisfaction measured using a modied form of the Minnesota Satisfaction Questionnaire (Weiss etal. 1967); and

JRI job-relevant information measured as the sum of the three-item scale developed by Kren (1992).

reports results over the Mexican sample. Simple correlations that are signicant at p 0.05are reported in bold. 14

Culture DifferencesWe test for differences on Hofstedes national-culture dimensions (H6) as a preliminary

to testing the individual paths in the proposed model (H1H5). Signicant national-culturedifferences between the Mexican and U.S. samples will suggest caution in interpreting theresults of the full-sample path model and will prompt estimation of separate path modelsfor the Mexican and U.S. samples.

14 Simple correlations for the three variables measuring the number of employees, number of managers, and numberof hourly employees supervised are not reported because these variables are only signicantly correlated withone another.


14/33


TABLE 3Path Model and Demographic Variable Correlation Tables

PART PERF SAT JRI SEX AGE SCH BLD BLR FYR JYR

Panel A: U.S. SamplePART 1PERF 0.385 1SAT 0.222 0.331 1 JRI 0.237 0.156 0.086 1SEX 0.050 0.082 0.107 0.135 1 AGE 0.015 0.029 0.167 0.218 0.214 1SCH 0.068 0.140 0.393 0.091 0.100 0.250 1 BLD 0.010 0.196 0.146 0.360 0.041 0.096 0.034 1 BLR 0.010 0.196 0.146 0.360 0.041 0.096 0.034 1.000 1

FYR 0.238 0.233 0.111 0.318 0.073 0.609 0.379 0.082 0.082 1 JYR 0.247 0.225 0.247 0.573 0.081 0.404 0.021 0.099 0.099 0.491 1

Panel B: Mexican Sample

PART 1PERF 0.310 1SAT 0.485 0.278 1 JRI 0.420 0.534 0.458 1SEX 0.177 0.093 0.064 0.081 1 AGE 0.352 0.281 0.231 0.088 0.351 1

SCH 0.044 0.204 0.081 0.020 0.084 0.007 1 BLD 0.150 0.090 0.035 0.151 0.264 0.071 0.118 1 BLR 0.038 0.047 0.006 0.066 0.231 0.097 0.112 0.707 1FYR 0.280 0.187 0.096 0.023 0.292 0.511 0.011 0.001 0.066 1 JYR 0.244 0.083 0.023 0.063 0.134 0.436 0.098 0.164 0.157 0.770 1

Signicant simple correlations with two-tailed p-values 0.05 are reported in bold.PART budget participation measured as the sum of the six-item scale developed by Milani (1975);PERF performance measured using the Mahoney et al. (1963) eight-item scale;

SAT job satisfaction measured using a modied form of the Minnesota Satisfaction Questionnaire (Weiss etal. 1967);

JRI job-relevant information measured as the sum of the three-item scale developed by Kren (1992);

SEX

sex: male

1 and female

0; AGE age reported in ranges: 20 1, 2024 2, 2529 3, 3034 4, 3539 5, 4049 6, 5059 7, and 60 8;

SCH years of school; BLD bilingual indicator equal to 1 if bilingual, and 0 otherwise; BLR bilingual rating where 3 very well, 2 somewhat well, 1 poorly, and 0 non-bilingual;FYR years with the company; and JYR years in current position.

Hofstede National-Culture Dimension Tests (H6)

We present the cultural dimension scores calculated using Hofstedes VSM94 formu-lation (Hofstede 1994a) for our U.S. and Mexican samples in the rst two rows of Table4 We present the cultural dimension scores using the VSM94 formulation or our Mexican


15/33


TABLE 4Tests of Cultural Differences (H6) over Hofstedes Cultural Dimensions

PDI UAI IDV MAS

Current U.S. sample 31 33 95 70Current Mexico sample 30 61 62 84

Abs(difference) 1 28 33 14Current Border sample 29 48 75 83Current Interior sample 31 77 48 85

Abs(difference) 2 29 27 2Hofstedes U.S. sample 40 46 91 62Hofstedes Mexico sample 81 82 30 69

Abs(difference) 41 36 61 7

Material differences are reported in bold.

Chow et al. (1999, 447) reference Hofstede (1980, 1991) in suggesting that absolute differences in countryscores that differ by 20 or more are material. A difference of 20 represents about one standard deviation in thedistribution of by country scores reported in Hofstede (1983a).The Hofstede cultural dimension scores for the current samples are calculated by applying the followingformulas to the average responses to selected VSM94 questions for each sample (Hofstede 1994a):

Power distance: PDI 35( Q3) 35( Q6) 25( Q14) 20( Q17) 20

Uncertainty avoidance: UAI 25( Q13) 20( Q16) 50(Q18) 15( Q19) 120

Individualism: IDV 50( Q1) 30( Q2) 20(Q4) 25( Q8) 130

Masculinity: MAS 60( Q5) 20(Q7) 20( Q15) 70( Q20) 100

Please see the Appendix for the VSM94 questions used in these calculations.

Border and Mexican Interior samples in the next two rows. For comparison with the com-monly cited ndings of Hofstedes 19671973 worldwide surveys of IBM employees, wepresent the U.S. and Mexican results reported in Hofstede (1983a) in the last two rows.Material differences are reported in bold. 15

The cultural differences between U.S. managers working for U.S. companies and Mex-

ican managers working for U.S.-controlled maquiladoras hypothesized by H6 is supportedby material differences in uncertainty avoidance and individualism. While the difference inpower distance between the U.S. and Mexico reported by Hofstede and relied upon byFrucot and Shearon (1991) is not apparent in our sample, our results are consistent withHofstedes original ndings over the other three dimensions.

Discussion of National-Culture ResultsPower distance. While the failure of our Mexican sample to reproduce the high levels

of power distance reported by Hofstede (1983a) may seem surprising, on further review

15 Chow et al. (1999, 447) reference Hofstede (1980, 1991) in suggesting that absolute differences in countryscores that differ by 20 or more are material. A difference of 20 represents about one standard deviation in thedistribution of by-country scores reported in Hofstede (1983a)


16/33


our ndings may represent several subtle factors that were not commonly recognized at thetime of Frucot and Shearon (1991). Hofstedes between-country comparisons are based onsamples of IBM marketing-plus-service organization personnel employed in a variety of occupations (Hofstede 2001). While Hofstede (2001, 88) nds that uncertainty avoidance

and individualism are unaffected by occupation, he nds that power distance is much lowerfor managers and that power distance is negatively correlated with years of education.Therefore, the fact that our Mexican subjects are managers and more highly educated thanmany of Hofstedes subjects may indicate that our Mexican power distance scores arereasonable even if the general Mexican population remains high on this dimension. Thisalso means that expecting an aversion to participation based on high power distance maybe unreasonable.

Uncertainty avoidance. While Offermann and Hellmann (1997) nd that leaders inhigh uncertainty avoidance countries maintain more control, delegate less, and are lessapproachable, Newman and Nollen (1996, 773) nd that performance is higher in highuncertainty avoidance countries when direction and policies are clearer. Hofstede reinforcesthe idea that high uncertainty avoidance cultures require more structure (Hofstede 2001,442) and that group decisions are preferred in these cultures (Hofstede 2001, 160). If participation plays a role in making the companys direction, policies, and structure moreclear, then it is quite plausible that participation is positively linked to performance amongour high uncertainty avoidance Mexican managers.

Individualism-collectivism. As discussed earlier, it is not clear whether collectivist(low individualism) cultures prefer more or less participation (Harrison and McKinnon1999; Sivakumar and Nakata 2001; Tsui 2001). Harrison and McKinnon (1999) make astrong argument that collectivism is directed at different collective groups in different lowindividualism societies, e.g., organization, family, community. Hofstede (2001, 239) makes

the point that in collectivist cultures a relationship of trust must be established beforeboth parties become part of the same in-group. In this sense, if budget participation facil-itates the process of the Mexican manager adopting the company as part of his/her in-group, then a positive relationship between participation and performance is again quiteplausible.

BUDGET PARTICIPATION-PERFORMANCE RESULTSThe simple correlations between budget participation and performance and the esti-

mated path coefcients are presented in Table 5 for our U.S. and Mexican samples. Thepath model results for the U.S. and Mexican samples are also reported graphically in Figure2, Panels A and B. The path model coefcients are estimated by solving the system of equations in Model (3) using the Covariance Analysis of Linear Structural Equations(CALIS) procedure, provided by SAS Institute Inc. (1990, 292365).

The simple correlations provide our test of H1, the basic participation-performancerelationship, and the difference in the simple correlations between the U.S. and Mexicansamples provide our test of H7, difference in the basic participation-performance relation-ship. The path coefcients provide our tests of H2H5, the complex participation-performance relationships. The differences in path coefcients between the U.S. and Mex-ican samples provide our tests of H8, differences in the complex participation-performancerelationships. The simple correlations and path coefcient p-values are based on one-sided

tests. The joint hypotheses (H3H5) p-values are based on the joint probabilities of theindividual path coefcients. The simple correlations and path coefcients in Table 5 and


17/33


TABLE 5Tests of the Direct and Indirect Budget Participation to Performance Path Analysis Results

Tests of Hypotheses: H1, H2, H3, H4, H5, H7, and H8


Job SatisfactionSAT


PerformancePERF

D

A

B

E

C

F

Path Description

SampleU.S.

n 45Mexicann 98

DifferencesH7 and H8

PART PERF corr. 0.3852 0.3101 0.0751H1: simple correlation p-value 0.0001 0.0045 0.7722A PART PERF coef. 0.3142 0.0991 0.2151H2: direct effect p-value 0.0158 0.1657 0.2768B PART SAT coef. 0.2137 0.3635 0.1498

p-value 0.0829 0.0001 0.6517C SAT PERF coef. 0.2560 0.0053 0.2507

p-value 0.0348 0.4757 0.0282D PART JRI coef. 0.2370 0.4254 0.1884

p-value 0.0570 0.0001 0.7882E JRI SAT coef. 0.0351 0.3032 0.2681

p-value 0.4088 0.0008 0.0020F JRI PERF coef. 0.0599 0.4899 0.4300

p-value 0.3330 0.0001 0.0001H3: PART SAT PERF joint prob 0.1147 0.4798H4: PART JRI PERF joint prob 0.3710 0.0001H5: PART JRI SAT PERF joint prob 0.4618 0.4801

The correlations and path coefcients with p-values

0.05 are reported in bold.The correlation and path coefcient p-values are based on one-sided tests. The joint hypotheses p-values arebased on joint probabilities.The simple Pearson correlations between budget participation, PART , and performance, PERF , tests hypothesisH1 and are presented in the rst row labeled PART PERF .The signicance tests for H3H5 are based on the joint probabilities that the individual path coefcients linkingparticipation with performance in that hypothesis. The four path model-based hypotheses linking budgetparticipation, PART , to performance, PERF are:

H2: the direct path following A, PART PERF ;H3: the indirect path running through B and C, PART SAT PERF ;H4: the indirect path running through D and F, PART JRI PERF ; andH5: the indirect path running through D, E, and C, PART JRI SAT PERF .

The signicance test of H7, the difference in the direct connection s between budget participation, PART , andperformance, PERF , between the U.S. and Mexico is tested by the signicance in the difference in simplecorrelations reported in the differences column for simple correlations rows.

(continued on next page )


18/33


TABLE 5 (continued)

The signicance tests of H8, the differences in the causal paths connecting budget participation, PART , andperformance, PERF , between the U.S. and Mexico is tested by the signicance in the difference in pathcoefcients reported in the differences column for the respective coefcients.

PART

budget participation measured as the sum of the six-item scale developed by Milani (1975);PERF performance measured using the Mahoney et al. (1963) eight-item scale;SAT job satisfaction measured using a modied form of the Minnesota Satisfaction Questionnaire (Weiss et

al. 1967); and JRI job-relevant information measured as the sum of the three-item scale developed by Kren (1992).

Figure 2, Panels A and B with p-values 0.05 are reported in bold. Hypotheses 35 jointp-values are reported at the bottom of Table 5. 16

Hypotheses 1 and 7: Simple Correlation Tests

With the other path coefcients restricted to zero, the simple correlations betweenbudget participation and performance provide consistent support for H1, the positiveparticipation-performance relationship, across both the U.S. and Mexican samples, p

0.0045. These results provide no support for H7, differences in the basic budgetparticipation-performance relationship between U.S. and Mexican managers, p 0.7722based on a two-sided test.

Hypotheses 2, 3, 4, 5, and 8: Path Model TestsWhen the other path coefcients are unrestricted, the analysis provides a different

picture. While the U.S. sample supports H2, a direct relationship between participation andperformance with a signicant Path A coefcient, p 0.0158, the Mexican sample doesnot, p 0.1657. While correlations between budget participation and performance areapparent in both samples, the paths between them are quite different. The Mexican resultsclearly support H4 with signicant path coefcients between budget participation and job-relevant information, Path D, and between job-relevant information and performance, PathF, joint-p 0.0001. Hypothesis 3, the indirect relationship between participation and per-formance running through job satisfaction, and H5, the indirect relationship between par-ticipation and performance running through job-relevant information and job satisfaction,are not supported by the Mexican sample and none of the joint hypotheses (H3H5) aresupported by the U.S. sample.

Hypothesis 8, differences in the complex relationships described in H2H5 betweenthe U.S. and Mexican samples are supported. Based on two-sided tests there are signicantdifferences in three of the individual path coefcients: Path C, SAT PERF , p 0.0282;Path E, JRI SAT , p 0.0020; and Path F, JRI PERF , p 0.0001. While Path D,PART JRI , is at least marginally supported in both cultures, p 0.0570, the nonsignif-icant Path F coefcient among our U.S. managers, p 0.3330, fails to complete the con-nection between participation and performance running through job-relevant informationthat is observed among our Mexican managers.

16 The simple correlations and path A coefcient differ because the alternative path coefcients are restricted tozero with the simple correlations. Path A represents the direct effect when alternative paths are included in themodel.


19/33


FIGURE 2Estimated Path Coefcients and p-values

Panel A: Path Model with U.S. Coefcients and p-values


Job SatisfactionSAT

Budget Participation

PART

Performance

PERF

D

A

B

E

C

F

0.3142(0.0157)

0.2370(0.0570)

0.2137(0.0829)

0.0351(0.4088)

0.2560(0.0348)

0.0599(0.3330)

Panel B: Path Model with Mexican Coefcients and p-values


Job SatisfactionSAT


PerformancePERF

D

A

B

E

C

F

0.0991(0.1657)

0.4254(


20/33


our U.S. and Mexican results can be explained by differences in these two cultural dimen-sions. We reject the cultural difference explanation because the regional results are similarto our tabulated results with both regions supporting H4, the indirect relationship betweenparticipation and performance running through job-relevant information, with joint-p

0.0118.Communication

To further explore the communication aspect, we turn to a broader concept of nationaldifferences that has origins in the marketing and management literature, psychic distance(e.g., Dow 2000; Child et al. 2002). While the measurement of psychic distance is not asprecisely dened as Hofstedes measurements, and while psychic distance has only veryrecently found its way into the accounting literature (e.g., Hassel and Cunningham 2004),psychic distance has implications for our study because it is comprised of a broad set of concepts that inhibit communication including language, education, business practices, cul-ture, religion, political system, industrial development, and geographic distance (OGradyand Lane 1996; Dow 2000; Child et al. 2002; Hassel and Cunningham 2004).

The strongest implication of our results is that the information-communication aspectof budget participation leads to a stronger connection with performance among our Mexicanmanagers. While the connection between budget participation and job-relevant informationis at least marginally signicant in the U.S. and Mexican samples, Path D, PART JRI ,p 0.0570, the connection between job-relevant information and performance is muchstronger among our Mexican managers, difference in Path F, JRI PERF , p 0.0001.

We conduct three additional tests to determine whether the information-communicationaspect of our Mexican ndings can be explained by the level of difculty our Mexicanmanagers may have in communicating with their U.S.-controlled companies. We explore

bilingual versus non-bilingual, bilingual ability, and supervisor nationality. We conductthese tests over a reduced form of Model (3) consistent with Kren (1992) that has JRI asthe only intervening variable between participation ( PART ) and performance ( PERF ). Thisreduced model restricts the Path B, PART SAT , Path C, SAT PERF , and Path E, JRI SAT , coefcients to zero. 17

Bilingual AbilityHypothesis 4, the indirect relationship between budget participation and performance

running through job-relevant information, is supported by estimations over both the bilin-gual and non-bilingual groups, joint-p 0.0349. Furthermore, while there is no signicantdifference in the Path D, PART JRI , coefcients, p 0.5757, between the bilingual andnon-bilingual samples; the Path F, JRI PERF , coefcient is signicantly larger for thenon-bilingual Mexican managers in a two-sided test, p 0.0361. Hypothesis 4 is alsomarginally supported for every level of bilingual ability other than very well, joint-p

0.0601. While there is at least a marginally signicant correlation between budget par-ticipation and job-relevant information, Path D, PART JRI , across all four levels of bilingual ability p 0.0554, and while there is a signicant correlation between job-relevantinformation and performance, Path F, JRI PERF , over the rst three levels of bilingual

17 Given the relative small sample sizes of several of the Mexican groups used for these tests, bilingual, 79; non-bilingual, 16; bilingual ability: poor, 30; somewhat well, 32; very well, 17; supervised by a U.S. national, 28;and supervised by a Mexican national, 63; estimations of the full Model (3) is impractical due to insufcientdegrees of freedom. The reduced model appears reasonable because it produces results highly consistent withthe full model when estimated over the full U.S. and Mexican samples.


21/33


22/33


ROBUSTNESS TESTSWe estimate an expanded version of Model (2), Model (2a), that includes demographic

variables to test whether the lack of support for H7, difference in the basic participation-performance relationship, is an artifact of differences in the demographics between the U.S.

and Mexican samples:

PERF MX PART PART * MX X e (2a)i 0 i 1 i 2 i i k ki i

where:


Milani (1975); MX 1 if the manager is Mexican, and 0 otherwise; and

X k the k th control variable.

The results of these tests are similar to the primary results reported above. There isconsistent support for H1, the basic participation-performance relationship, and a consistentlack of support for H7, differences in the basic budget participation-performance relation-ship between U.S. and Mexican managers, when the demographic variables listed in Table1 are added individually or in groups as control variables, 1 , p 0.0063, and 2 , p

0.5955.Finally, we estimate Model (3) with the samples dichotomized on sex, median age,

median time with the rm, and median number of employees supervised, to test whetherour tabulated path results dichotomized on culture are driven by differences in demographicfactors. None of these tests provide any evidence that our tabulated results are driven bydemographic differences between our U.S. and Mexican samples.

CONCLUSIONS, LIMITATIONS, AND FUTURE RESEARCHThe primary ndings of this study are that while there are strong associations between

budget participation and performance for both U.S. managers working in the U.S. andMexican managers working for U.S.-controlled maquiladoras in Mexico, the causal mech-anisms connecting budget participation with performance are quite different. Theinformation-communication aspect of budget participation appears much more importantamong the Mexican managers working for U.S.-controlled maquiladoras in Mexico thanfor their U.S. counterparts working in the U.S. Our Mexican ndings suggest that theimportance of the information-communication aspect in Mexico may be related to the psy-chic distance, the level of difculty our Mexican managers face in communicating withtheir U.S. parent companies. These conclusions are supported by our primary nding of asignicant connection between budget participation and performance running through job-relevant information that is absent among our U.S. managers and further supported byadditional ndings within our ex post analyses: (1) the job-relevant information to perform-ance coefcient is larger among our non-bilingual Mexican managers; (2) the joint pathbetween budget participation and performance running through job-relevant information is

at least marginally signicant for every level of bilingual ability among our Mexican man-agers with the exception of those who should have the least difculty in communicating


23/33


with their U.S. owners, those with the highest levels of bilingual ability; and (3) the job-relevant information to performance link is signicantly stronger among those Mexicanmanagers supervised by U.S. nationals than by Mexican nationals.

Our study suffers from three common limitations found in all cross-sectional survey

research: (1) the lack of temporal precedence between the independent and dependent var-iables; (2) any limitations imbedded in the scales used to measure our variables; and, (3)the generalizability of our samples. While the lack of temporal precedence between budgetparticipation, job-relevant information, satisfaction, and performance precludes formal testsof causality, and while we cannot say that better performers are not provided more job-relevant information and allowed greater participation, the primary implications of our nd-ings are largely unaffected by the direction of these relationships. Our nding that job-relevant information plays a bigger role among our Mexican managers offers insights toU.S.-controlled maquiladoras whether the process starts with participation or performance.While our scales may contain their own limitations, we employ scales that have been welltested within the literature and scales that produce Cronbachs s that are well withinacceptable ranges. Finally, while our southeastern U.S. sample may not fully represent theentire U.S. and our two Mexican samples may not fully represent all of Mexico, our studyis the rst study we are aware of that applies a single research instrument to managersfrom these three locations. The fact that our U.S. results appear consistent with the existingliterature and the fact that our Mexican results are consistent across two far-separatedregions of Mexico offers support that our results are not driven by regional anomalies withineither the U.S. or Mexico.

We also acknowledge the potential limitation of using self-reported measures of budgetparticipation and performance. Locke et al. (1997) dene these single-source measures asthe percept-percept method. While the ndings of both Greenberg et al. (1994) and Locke

et al. (1997) suggest that the single-source method may produce higher correlations thanmulti-source methods, both Locke et al. and Greenberg et al. nd that the positive corre-lations persist in the multi-source studies. Self-reported levels of participation may be morerelevant than external measures of participation because it should be the subjects perceptionof budget participation that inuences behavior. While external measures of performancehave some documented benets, self-reported measures of performance remain a commonpractice in the literature (e.g., Nouri and Parker 1998; Shields et al. 2000).

Even with these limitations, the ndings of this study have important implications forU.S. companies employing U.S. management techniques in their foreign operations. Ourndings may go beyond budget participation and suggest that management tools that pro-vide additional job-relevant information to foreign managers may help improve perform-ance. While Frucot and Shearon (1991) suggest that U.S. companies employing Mexicanmanagers should exercise caution in implementing participative budgeting techniques, ourresults suggest that participative budgeting may be very effective in the current maquiladoraenvironment.

The ndings of this study also suggest several avenues for future research. While theinformation provided by this study has implications for the U.S. companies operating inMexico, similar research may be warranted across a variety of cultures and locations. Fur-ther research is needed to determine whether the ndings of this study are driven by theuniqueness of the Mexican managers in our study, managers who have self-selected towork for U.S.-controlled maquiladoras, or whether the information-communication aspect

of budget participation may prove to be useful in other cross-cultural settings where psychicdistance, communication difculty, is high.


24/33


APPENDIXSURVEY QUESTIONNAIRE

Introduction Letter Information and Consent

The following questionnaire is designed to help the researcher understand differentaspects related to your job and about yourself. Its main purpose is to determine youropinions and perceptions.

Your responses are anonymous and will be kept in strict condence. The results willbe presented as a summarized total so that individual responses cannot be identied. Yourconsent to participate is given by lling out this questionnaire.

This study has been developed for research purposes. Participation in this study isvoluntary and you may withdraw at any time.

Thank you for your participating in this survey. Please answer each section fully andgive your opinion as honestly as possible.

Budget Participation Questions (PART)Please respond by circling a number from 1 to 7 on the scale for each of the following

items (1 Strongly Disagree, 7 Strongly Agree).

StronglyDisagree

StronglyAgree

B1. I am involved in setting allportions of my budget. 1 2 3 4 5 6 7

B2. The reasoning provided bymy supervisor when budget

revisions are made is verysound and/or logical. 1 2 3 4 5 6 7B3. I very frequently state my

requests, opinions, and/orsuggestions about the budgetto my supervisor withoutbeing asked. 1 2 3 4 5 6 7

B4. I have a high amount of inuence on the nal budget. 1 2 3 4 5 6 7

B5. My contribution to the

budget is very important. 1 2 3 4 5 6 7B6. When the budget is being set,

my supervisor seeks myrequests, opinions, and/orsuggestions very frequently. 1 2 3 4 5 6 7

Job Performance Questions (PERF)For each area of activity, please rate your own recent performance using the following

scale:

Below Average

Performance

Average

Performance

Above Average

Performance1 2 3 4 5 6 7 8 9


25/33


PERFORMANCENumber from

1 to 9

P1. PlanningDetermining goals, policies, and courses of action; work scheduling, budgeting,setting up procedures, programming.

P2. InvestigatingCollecting and preparing information for records, reports andaccounts; measuring output; inventorying, job analysis.

P3. CoordinatingExchanging information with people in other organizational unitsin order to relate and adjust programs; advising other departments,liaison with other managers.

P4. EvaluatingAssessment and appraisal of proposals or of reported or observedperformance; employee appraisals, judging output records, judgingnancial reports; product inspection.

P5. SupervisingDirecting, leading and developing your subordinates; counseling,training, and explaining work rules to subordinates; assigning work and handling complaints.

P6. StafngMaintaining the work force of your unit; recruiting, interviewing,and selecting new employees; placing, promoting, and transferring

employees.P7. Negotiating

Purchasing, selling, or contracting for goods or services,contacting suppliers, dealing with sales representatives; collectivebargaining.

P8. RepresentingAttending conventions, consultation with other rms, business clubmeetings, public speeches, community drives; advancing thegeneral interests of your organization.

P9. Overall Performance

Includes all of the areas listed above.

Minnesota Satisfaction Questionnaire (SAT)

If you feel that your job gives you a lot more than you expected , check the boxunder ES (Extremely Satised);

If you feel that your job gives you more than you expected , check the box underVS (Very Satised);

If you feel that your job gives you what you expected , check the box under S(Satised);

If you feel that your job gives you less than you expected , check the box under

SS (Somewhat Satised); If you feel that your job gives you much less than you expected , check the boxunder NS (Not Satised)


26/33


On my present job, this is how I feel about:ES VS S SS NS

S1. Being able to keep busy all the time. . . . . . . . . . . . . . . . . . . . . . .S2. The chance to work alone on the job . . . . . . . . . . . . . . . . . . . . . .

S3. The chance to do different things from time to time . . . . . .S4. The chance to be somebody in the community. . . . . . . . .S5. The way my boss handles his workers.. . . . . . . . . . . . . . . . . . . .S6. The competence of my supervisor in making decisions. . .S7. Being able to do things that dont go against my

consc ience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .S8. The way my job provides for steady employment . . . . . . . .S9. The chance to do things for other people.. . . . . . . . . . . . . . . . .

S10. The chance to tell people what to do . . . . . . . . . . . . . . . . . . . . . .S11. The chance to do something that makes use of myabi l i t i e s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

S12. The way company policies are put into practice ..........S13. My pay and the amount of work I do . . . . . . . . . . . . . . . . . . . . .S14. The chances for advancement on this job.. . . . . . . . . . . . . . . . .S15. The freedom to use my own judgment . . . . . . . . . . . . . . . . . . . .S16. The chance to try my own methods of doing the job . . . . .S17. The working conditions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

S18. The way my co-workers get along with each other .......S19. The praise I get for doing a good job......................S20. The feeling of accomplishment I get from the job ........

Copyright 1977, Vocational Psychology Research, University of Minnesota. Reproduced bypermission.

Job-Relevant Information Questions (JRI)Please respond by circling a number from 1 to 7 on the scale for each of the following

items (1 Strongly Disagree, 7 Strongly Agree).

StronglyDisagree

StronglyAgree

RI1. I am always clear about what isnecessary to perform well on my job. 1 2 3 4 5 6 7

RI2. I have adequate information tomake optimal decisions toaccomplish my performanceobjectives. 1 2 3 4 5 6 7

RI3. I am able to obtain the strategic

information necessary to evaluateimportant decision alternatives. 1 2 3 4 5 6 7


27/33


Hofstede VSM94 Cultural Values QuestionnairePlease think of an ideal job, disregarding your present job. In choosing an ideal job,

how important would it be to you to ... (please circle one answer in each line across):1 of utmost importance

2

very important3 of moderate importance4 of little importance5 of very little importance

Q1. have sufcient time for your personal life 1 2 3 4 5Q2. have good physical working conditions (good ventilation

and lighting, adequate work space, etc.) 1 2 3 4 5Q3. have a good working relationship with your direct superior 1 2 3 4 5Q4. have security of employment 1 2 3 4 5Q5. work with people who cooperate well with one another 1 2 3 4 5Q6. be consulted by your direct superior in his/her decisions 1 2 3 4 5Q7. have an opportunity for advancement to higher level jobs 1 2 3 4 5Q8. have an element of variety and adventure in the job 1 2 3 4 5

In your private life, how important is each of the following to you? (please circle oneanswer in each line across):

1 of utmost importance2 very important3 of moderate importance4 of little importance5 of very little importance

Q9. Personal steadiness and stability 1 2 3 4 5Q10. Thrift 1 2 3 4 5Q11. Persistence (perseverance) 1 2 3 4 5Q12. Respect for tradition 1 2 3 4 5Q13. How often do you feel nervous or tense at work?

1. never

2. seldom3. sometimes4. usually5. always

Q14. How frequently, in your experience, are subordinates afraid to express disagreementwith their superiors?

1. very seldom2. seldom3. sometimes4. frequently5. very frequently

To what extent do you agree or disagree with each of the following statements? (pleasecircle one answer in each line across):


28/33


1 strongly agree2 agree3 undecided4 disagree

5

strongly disagreeQ15. Most people can be trusted 1 2 3 4 5Q16. One can be a good manager without having precise

answers to most questions that subordinates may raiseabout their work 1 2 3 4 5

Q17. An organization structure in which certain subordinateshave two bosses should be avoided at all cost 1 2 3 4 5

Q18. Competition between employees usually does more harmthan good 1 2 3 4 5

Q19. A companys organizations rules should not be brokennot even when the employee thinks it is in the companysbest interest 1 2 3 4 5

Q20. When people have failed in life, it is often their own fault 1 2 3 4 5

Hofstedes VSM94 is copyrighted by Geert Hofstede and freely available for research pur-poses (Hofstede 1994a).

Demographic Data

Job Title:

How long have you worked for this rm? Years: Months:

How long have you held your current position? Years: Months:

Please check one: male female

How many people do you supervise?

Of those you supervise, how many are salaried and/or managers?

how many hourly?

What is your nationality?

What was your nationality at birth (if different)?

Circle the number of years of schooling you completed:4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20Grade School High School College Graduate or

Professional

Did you complete your highest level of school education in the United States?

If not in the United States, where?


29/33


How old are you:1. Under 202. 20243. 2529

4. 30345. 35396. 40497. 50598. 60 or over

Is your supervisor of: U.S. descent? Mexican descent? Other?

Do you speak Spanish (English)? If you answered yes, please rate your ability:

I speak Spanish (English): Very well Somewhat well Poorly

REFERENCES

Argyris, C. 1952. The Impact of Budgets on People . New York, NY: The Controllership Foundation.Baskerville, R. F. 2003. Hofstede never studied culture. Accounting, Organizations and Society 28:

114.Birnberg, J. G., and C. Snodgrass. 1988. Culture and control: A eld study. Accounting, Organizations

and Society 13: 447464.Bommer, W. H., J. L. Johnson, G. A. Rich, P. M. Podsakoff, and S. B. Mackenzie. 1995. On the

interchangeability of objective and subjective measures of employee performance: A meta-analysis. Personnel Psychology 48: 587605.Brownell, P. 1981. Participation in budgeting, locus of control and organizational effectiveness. The

Accounting Review 56 (October): 844860.. 1982a. Participation in the budgeting processWhen it works and when it doesnt. Journal

of Accounting Literature (Spring): 124153.. 1982b. A eld study examination of budgetary participation and locus of control. The Ac-

counting Review 57 (October): 766777.. 1982c. The role of accounting data in performance evaluation, budgetary participation and

organizational effectiveness. Journal of Accounting Research (Spring): 1227.. 1983. Leadership style, budgetary participation and managerial behavior. Accounting, Or-

ganizations and Society 8 (4): 307321., and M. K. Hirst. 1986. Reliance on accounting information, budgetary participation, and task

uncertainty: Tests of a three-way interaction. Journal of Accounting Research (Autumn): 241249.

, and M. McInnes. 1986. Budgetary participation, motivation, and managerial performance.The Accounting Review 61 (October): 587600.

, and A. S. Dunk. 1991. Task uncertainty and its interaction with budgetary participation andbudget emphasis: Some methodological issues and empirical investigation. Accounting, Orga-nizations and Society 16: 693703.

Butler, J. K. 1983. Value importance as a moderator of the value fulllmentjob satisfaction relation-ship: Group differences. Journal of Applied Psychology (August): 420428.

Chenhall, R. H., and P. Brownell. 1988. The effect of participative budgeting on job satisfaction and

performance: Role ambiguity as an intervening variable. Accounting, Organizations and Society13 (3): 225233.


30/33


. 2003. Management control systems design within its organizational context: Findings fromcontingency-based research and directions for the future. Accounting, Organizations and Society28: 127168.

Child, J., S. H. Ng, and C. Wong. 2002. Psychic distance and internationalization. InternationalStudies of Management and Organization 32 (Spring): 3656.

Chow, C. W., M. D. Shields, and Y. K. Chan. 1991. The effects of management controls and nationalculture on manufacturing performance: An experimental investigation. Accounting, Organiza-tions and Society 16 (3): 209226

, Y. Kato, and M. D. Shields. 1994. National culture and the preference for managementcontrols: An exploratory study of the rm-labor market interface. Accounting, Organizationsand Society 19 (4,5): 381400.

, , and K. A. Merchant. 1996. The use of organizational controls and their effect ondata manipulation and management myopia: A Japan versus U.S. comparison. Accounting, Or-ganizations and Society 21 (2,3): 175192.

, M. D. Shields, and A. Wu. 1999. The importance of national culture in the design of andpreference for management controls for multi-national operations. Accounting, Organizations

and Society 24: 441461.Covaleski, M. A., J. H. Evans III, J. L. Luft, and M. D. Shields. 2003. Budgeting research: Three

theoretical perspectives and criteria for selective integration. Journal of Management Accounting Research 15: 349.

Cronbach, L. J. 1951. Coefcient alpha and internal structure of tests. Psychometrika (September):297334.

Dow, D. 2000. A note on psychological distance and export market selection. Journal of International Marketing 8: 5164.

Economist, The . 2000. If not for NAFTA, when? 357 (Special Section, October 28): 710.Fernandez, D. R., D. S. Carlson, L. P. Stepina, J. D. Nicholson. 1997. Hofstedes country classication

25 years later. The Journal of Social Psychology 137: 4354.Frucot, V., and W. T. Shearon. 1991. Budgetary participation, locus of control, and Mexican mana-

gerial performance and job satisfaction. The Accounting Review 66 (January): 8098.Govindarajan, V., and A. K. Gupta. 1985. Linking control systems to business unit strategy: Impact

on performance. Accounting, Organizations and Society 10: 5166.Greenberg, P. S., R. H. Greenberg, and H. Nouri. 1994. Participative budgeting: A meta-analytic

examination of methodological moderators. Journal of Accounting Literature 13: 117141.Greer, C. R., and G. K. Stephens. 2001. Escalation of commitment: A comparison of differences

between Mexican and U.S. decision-makers. Journal of Management 27: 5178.Harrison, G. L. 1992. The cross-cultural generalizability of the relation between participation, budget

emphasis and job related attitudes. Accounting, Organizations and Society 17 (1): 115.. 1993. Reliance on accounting performance measures in superior evaluative styleThe inu-

ence of national culture and personality. Accounting, Organizations and Society 18 (4): 319

339., L. McKinnon, S. Panchapakesan, and M. Leung. 1994. The inuence of culture on organi-

zational design and planning and control in Australia and the United States compared withSingapore and Hong Kong. Journal of International Financial Management and Accounting 5:242261.

, and J. L. McKinnon. 1999. Cross-cultural research in management control systems design:A review of the current state. Accounting, Organizations and Society 24 (5/6): 483506.

Hassel, L. G., and G. M. Cunningham. 1996. Budget effectiveness in multinational corporations: Anempirical test of the use of budget controls moderated by two dimensions of budgetary partic-ipation under high and low environmental dynamism. Management International Review 36(Third Quarter): 245266.

, and . 2004. Psychic distance and budget control of foreign subsidiaries. Journal of International Accounting Research 3: 7993.


31/33


Heneman, H. G. III. 1974. Comparisons of self and superior ratings of managerial performance. Journal of Applied Psychology 59 (5): 2953.

Hofstede, G. H. 1967. The Game of Budget Control . Assen, The Netherlands: Van Gorum.. 1980. Cultures Consequences: International Differences in Work-Related Values. Beverly

Hills, CA: Sage Publications.. 1983a. Dimensions of national cultures in fty countries and three regions. In Expectations

in Cross-Cultural Psychology , edited by J. B. Deregowski, D. Dzuirawiec, and R. C. Annis.Brighton, U.K.: Swets and Zeitlinger.

. 1983b. The cultural relativity of organizational practices and theories. Journal of International Business Studies 14 (Fall): 7589.

. 1991. Cultures and Organizations: Software of the Mind . London, U.K.: McGraw-Hill.. 1994a. VSM94: Value survey module 1994. Available at: http:/ / feweb.uvt.nl/ center/ hofstede/

english.html.. 1994b. Management scientists are human. Management Science 40 (January): 413.. 2001. Cultures Consequences: Comparing Values, Behaviors, Institutions, and Organizations

Across Nations . 2nd edition. Thousand Oaks, CA: Sage Publications, Inc.

Hui, C. H., and H. C. Triandis. 1985. Measurement in cross-cultural psychology: A review andcomparison of strategies. Journal of Cross-Cultural Psychology 16 (June): 131152.

Kenis, I. 1979. Effects of budgetary goal characteristics on managerial attitudes and performance. The Accounting Review 54 (October): 707721.

Kren, L. 1992. Budgetary participation and managerial performance: The impact of information andenvironmental volatility. The Accounting Review 67 (July): 511526.

Lau, C. M., and S. L. C. Tan. 2003. The effects of participation and job-relevant information on therelationship between evaluative style and job satisfaction. Review of Quantitative Finance and Accounting 17: 1734.

Lindquist, D. 2001. Rules change for maquiladoras. Industry Week/IW 250 (1): 2325.Locke, E. A., M. Alavi, and J. A. Wagner III. 1997. Participation in decision making: An informative

exchange perspective. Research in Personnel and Human Resource Management 15: 293331.Luft, J., and M. D. Shields. 2003. Mapping management accounting: Graphics and guidelines for

theory-consistent empirical research. Accounting, Organizations and Society 28: 169249.Magner, N., R. B. Welker, and T. L. Campbell. 1996. Testing a model of cognitive budgetary partic-

ipation process in a latent variable structural equations format. Accounting and Business Re-search 27: 4150.

Mahoney, T. A., T. H. Jerdee, and S. J. Carroll. 1963. Development of Managerial Performance: A Research Approach . Cincinnati, OH: South-Western Publishing, Co.

Mia, L. 1989. The impact of participation in budgeting and job difculty on managerial performanceand work motivation: A research note. Accounting, Organizations and Society 14 (4): 347357.

Milani, K. 1975. The relationship of participation in budget-setting to industrial supervisor perform-ance and attitudes: A eld study. The Accounting Review 50 (April): 274284.

Newman, K. L., and S. D. Nollen. 1996. Culture and consequence: The t between managementpractices and national culture. Journal of International Business Studies 27 (4): 753779.

Nouri, H., and R. J. Parker. 1998. The relationship between budget participation and job performance:The roles of budget adequacy and organizational commitment. Accounting, Organizations and Society 23: 467483.

OConnor, N. G. 1995. The inuence of organizational culture on the usefulness of budget partici-pation by Singaporean-Chinese managers. Accounting, Organizations and Society 20: 383403.

Offermann, L. R., and P. S. Hellmann. 1997. Cultures consequences for leadership behavior: Nationalvalues in action. Journal of Cross-Cultural Psychology 28 (May): 341342.

OGrady, S., and H. W. Lane. 1996. The psychic distance paradox. Journal of International BusinessStudies 27 (Second Quarter): 309333.

Peterson, D. K., G. M. Puia, and F. R. Suess. 2003. Yo tengo la camiseta (I have the shirt on): Anexploration of job satisfaction and commitment among workers in Mexico. Journal of Lead-ership and Organizational Studies 10 (Fall): 7388.
http://feweb.uvt.nl/center/hofstede/english.htmlhttp://feweb.uvt.nl/center/hofstede/english.htmlhttp://feweb.uvt.nl/center/hofstede/english.htmlhttp://feweb.uvt.nl/center/hofstede/englis

Culture Aspect of Job Relevant in Budgeting

Documents