See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/236004003

Level of Analysis in Export Performance Research

Article  in  International Marketing Review · February 2012

DOI: 10.1108/02651331211201561

CITATIONS READS

57 1,202

3 authors:

João S Oliveira John William Cadogan

Loughborough University Loughborough University
22 PUBLICATIONS   93 CITATIONS    139 PUBLICATIONS   5,286 CITATIONS   

SEE PROFILE SEE PROFILE

Anne L. Souchon
Loughborough University
48 PUBLICATIONS   1,231 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Frontline employee behaviours View project

Export Mode Portfolio- Transaction Cost Economics and Real Options Perspectives View project

All content following this page was uploaded by John William Cadogan on 29 May 2014.

The user has requested enhancement of the downloaded file.

Level of analysis in export performance research

João S. Oliveira
PhD Candidate
School of Business and Economics
Loughborough University
Loughborough, UK
Email: [email protected]

John W. Cadogan
Professor of Marketing
School of Business and Economics
Loughborough University
Loughborough, UK
Email: [email protected]

Anne Souchon
Professor of International Marketing
School of Business and Economics
Loughborough University
Loughborough, UK
Email: [email protected]

Forthcoming:

International Marketing Review, 2012, Volume 29, Issue 1.

1
Level of analysis in export performance research

Introduction

Many researchers are interested in developing and testing models of the

determinants of export performance. In marketing and international business,

researchers interested in models of this kind tend to analyze the antecedents of

export performance either from the export function level of analysis or from the

export venture level of analysis. Export function level of analysis studies focus on the

overall export performance achieved by the exporting entity (e.g., Aulakh et al.,

2000; Brouthers and Xu, 2002; Cadogan et al., 2009; Ito and Pucik, 1993). Venture

level studies of export performance (e.g., Cavusgil and Zou, 1994; Katsikeas et al.,

2006; Lages et al., 2008; Morgan et al., 2004) look at the performance of an export

venture within the firm, where an export venture is defined as a single product or

product line exported by a company to a particular foreign market (Cavusgil and Zou,

1994; Morgan et al., 2004).

Of course, the export function and the export venture levels of analyses are related

since firms’ export functions can be thought of as portfolios composed of those firms’

multiple export ventures (Katsikeas et al., 2000; Madsen, 1998). Thus, one can

observe the presence of hierarchical ordering in export performance, with firms’

export functions being the higher level units and firms’ export ventures being the

lower level units, nested within the export functions.

2
As Sousa et al. (2008) recognize in their literature review of the determinants of

export performance, there is an increase in the number of studies adopting the

export venture as the unit of analysis. Two observations are worrying, however. First,

there seems to be an assumption made by some authors and journal reviewers that

performance assessment at the export venture level is the only correct level to adopt

when studying export performance: this is an incorrect assumption. Second, there is

a lack of recognition that in some instances, export performance assessment at the

export venture level is inappropriate: indeed, studies that measure export

performance at the export venture level when measurement should take place at the

export function level may be presenting invalid recommendations for managerial

practice and theory advancement.

In this article, we examine the issue of level of analysis in models where export

performance is a dependent variable, pinpointing the potential pitfalls of measuring

export performance at the incorrect level of analysis, and providing for researchers

recommendations on the appropriate level of analysis to be used depending on the

research objectives. In the next section, we begin by elaborating on the definition of

export performance, paying particular attention to issues that are pertinent to the

levels of analysis debate.

Export Performance

As Katsikeas et al. (2000) highlight, there are multiple ways of conceptualizing export

success. One approach is to view export performance in terms of export

effectiveness, export efficiency, and export adaptiveness. Export effectiveness is

3
concerned with the degree to which exporters achieve their exporting goals and

objectives. For example, a firm might evaluate itself according to whether export

profit or export sales revenue objectives are achieved, or whether a certain

penetration level is achieved or exceeded. Export efficiency refers to the comparison

of outputs (e.g., revenues) to inputs (e.g., costs), and so includes traditional profit

ratios, such as return on investment (ROI).

Interestingly, the export effectiveness and efficiency constructs are not mutually

exclusive. Indeed export effectiveness and efficiency measures may be identical: a

firm might have, as a performance objective, the goal of making a certain return on

investment for every dollar invested in their export operations. As such, an efficiency

measure of performance (ROI) might also act as an effectiveness measure of

performance.

According to Katsikeas et al. (2000), export adaptiveness refers to the firm’s capacity

to respond to changes occurring in the market environment. Unfortunately,

adaptiveness is a more awkward inclusion as a dimension of success since its

definition shares conceptual overlaps with market orientation (Kohli and Jaworski,

1990). Market orientation is likely a potential cause of export sales performance

(Cadogan et al., 2009), but there are few who would support the argument that

market orientation is, or falls within the domain of, firm performance. Furthermore,

there is evidence to suggest that the links between market orientation and sales

performance and between market orientation and profit performance are non-linear

(see Narver and Slater, 1990; Cadogan et al. 2009). If true, this is a problem for the

face validity of the “adaptiveness is export success” notion, because it implies that

4
increases in adaptiveness may sometimes harm sales and profits. Of course,

adaptiveness may play a role in shaping a firm’s success, but adaptiveness as a

construct is rarely (if ever) going to be a variable that exporters themselves would

consider as a performance outcome, and for the reasons outlined above, we exclude

adaptiveness as a necessary or defining aspect of export performance.

Another approach to the classification of export success is to categorize

performance in terms of non-economic and economic outcomes (Katsikeas et al.,

2000). Non-economic outcomes are potentially of interest to exporting researchers,

since they may play key mediating roles in shaping economic outcomes. For

instance, a firm’s degree of internationalization is partially a function of the number of

country markets it operates in: firms operating in more country markets may find they

are less vulnerable to fluctuations in demand happening within individual markets,

and so may be better able to ride out market shocks. Simultaneously, such firms

may be better positioned to capitalize on growth opportunities arising in markets in

which they already operate. Accordingly, it would not be unusual for exporters to

have as a formal strategic objective the expansion of their export operations into new

markets, and as a result, market expansion goals might legitimately be translated

into non-economic effectiveness measures of export success. From this discussion,

one can clearly see that non-economic factors may become important export

performance indicators.

However, the export performance indicators of greatest interest are the economic

measures: these contain sales-related dimensions, such as sales growth and

revenues relative to competitors (market share), and profit-related dimensions, such

5
as profit margins, ratios, and growth trends. Ultimately, all profit-oriented businesses

are seeking success in terms of economic performance dimensions and, justifiably,

economic export performance indicators receive most research attention. In terms of

modeling determinants of export success, economic performance indicators are the

ultimate endogenous variables, with non-economic indicators playing important but

subservient mediating roles. Economic performance indicators can be of the

effectiveness kind or the efficiency kind, or both simultaneously, since the

operational distinction between effectiveness and efficiency is often vague.

With these observations in mind, we now turn our attention to the issue of the level at

which theory is developed and tested.

Levels issues

Different research questions are posed at different levels within the firm, and as a

result, require data for theory testing that originate and refer to variables that exist at

different levels in the firm. When analyzing export performance determinants, the

researcher may be interested in studying one or more types of broad research

questions. The first type of research question involves explaining variations in levels

of export performance across businesses (i.e., inter-firm export performance

variation). In short, the researcher wants to know why some firms outperform others

across some aspect of their export operations, and as a result, the export

performance assessment has an external orientation, focusing on how the firm

performs relative to other firms. This type of research study has immediate and

obvious value, since it results in the identification of factors that improve the export

6
performance of companies. The second type of research question that researchers

may wish to address involves identifying the determinants of variance in the

performance of individual export ventures within the firm (i.e., intra-firm export

venture performance variation). Here, the researcher has an internal orientation,

since they are seeking to understand why some export ventures within a firm

outperform other export ventures in the same firm. In our experience, it is not

unusual for small to medium-sized exporting firms to have dozens of export

ventures, even hundreds, and so identifying factors that cause poor or exceptional

performance within the population of firms’ export ventures can have merit,

potentially leading to the identification of factors that enhance the overall

performance of the export function.

Importantly, a fundamental principle of theory testing is that in order for research

findings to be valid, there needs to be a match between the level at which the theory

is constructed and the level at which it is tested (Klein et al. 1994; Mossholder and

Bedeian 1983; Rousseau, 1985). The implications for export performance research

are straightforward and clear. If one is interested in testing a theory about the

determinants of export performance, then the level at which the theory is developed

should govern the level at which variables are measured. What this means in

practice is laid down in Figure 1, which provides a road map in terms of the level of

the theory being developed (i.e., the focus of the research question) and the kinds of

data that are required in order to provide valid answers for these theories. We

consider the options presented in Figure 1 now1.

Figure 1 about here

7
First type of research question: determinants of the performance of firms’ entire

export functions

When the purpose of the study is the analysis of how the overall degree of export

performance achieved by the export function varies across firms, export performance

needs to be assessed at the level of the export function. It may be possible to do this

in two ways. First, one can collect data that directly provides information on the

performance of the entire export function (first column, Figure 1). For instance,

economic data in the form of total sales revenues across all the firms’ export markets

can be collated, or total profits accruing from export markets can be estimated.

Researchers are often interested in knowing how environmental or firm level factors

can shape the success of businesses’ export operations, and so in many situations,

the theory being presented is inherently at the level of the export function, and the

appropriate level of measurement for export performance is at the export function

level too. Nevertheless, researchers interested in theory that is developed at the

export function level may not, for one reason or another, have data that pertains to

the performance of the export function

The columns in Figure 1 show some additional sources of data. First, researchers

may have collected data at the level of a cohort of export ventures. For instance, the

performance data collected may pertain only to the firm’s export ventures within a

defined geographic region. Sales performance in this instance might be concerned

with assessing “total export sales to geographic area X”, since the cohort of ventures

8
contains all export ventures in the geographic area in question. Unless the ventures

within the cohort are representative of the entire set of export ventures making up

the totality of the firms’ export operations, the cohort based approach to export

performance assessment will provide biased information about the determinants of

export function performance.

The same can be said if the performance data are collected from a single export

venture. Here, the problem revolves around the question of whether the single

venture assessed is representative of the entire set of a firm’s export ventures,

especially as the number and diversity of firms’ export operations increases. An

additional problem arises when one considers the fact that in practice, it is very

difficult to accurately measure export performance at the single venture level. That

is, while it may be possible to assess some kinds of economic performance (e.g.,

sales) at the venture level, the allocation of costs, and hence the determination of

accurate profit figures, is harder. As a result, export profit assessment at the venture

level is likely to be highly subjective, and perhaps even unreliable, because the

managers who provide the information simply do not know how profitable any single

venture may be.

Unfortunately, it is fairly common for researchers to sample only one export venture

(i.e., one lower level unit) per firm in their studies and use the results obtained in that

single export venture to make inferences regarding the firm’s overall export function

(the higher level unit) performance. This type of extrapolation corresponds to what is

often referred to in organizational research as an upward cross level inference (e.g.,

Mossholder and Bedeian, 1983). This is an attempt to infer relationships at a higher

9
level (in this case, the firm’s overall export function) based on results found at a

lower level (in this case, at the single export venture level). Such inferences may not

be valid because relationships found in a single export venture of the firm do not

necessarily hold for all the other export ventures in the firm and hence may not hold

for the export function of the firm as a whole. For example, as defended by Douglas

and Wind (1987), it is not realistic to expect that the same marketing strategy will

lead to the same performance outcomes in all the firm’s export ventures or, in other

words, it is unrealistic to assume that the relationship between venture level

marketing strategy and venture level performance will be constant across all the

firm’s export ventures. Since such homogeneity is a pre-requisite necessary for one

to be able to extrapolate the results found at the single venture level to the export

function level without incurring an upward cross level bias, the researcher should not

collect performance data at the single venture level for testing theory developed at

the export function level.

Exploring this argument further, a research study at the single export venture level

might indicate that adaptation of the marketing mix enhances the profit of the

venture. However, even assuming that the finding has internal validity for the export

venture in question (and this is questionable, given the difficulty managers have in

assessing the profitability of an individual venture), the generalizability of the finding

to firms’ remaining export ventures is not a given. Specifically, if the finding is

generalizable, then the implication would be that all firms should adapt the marketing

mix in every export venture, regardless of the number of ventures that they have. Yet

the latter conclusion is by no means certain, simply because the costs of adapting

the marketing mix in every product market combination may escalate quickly,

10
resulting in reduced profits for a large proportion of ventures, and hence lower profits

for the export function as a whole. Consequently, the dangers of generalizing from

the single export venture level to the export function level stem from the fact that

findings may be contradictory across the two levels. Single export venture level

findings, if replicated across all firms’ export ventures and aggregated (i.e., to reflect

the function level) may indicate very different outcomes from any one individual

venture level study. In this fictitious example, the results of the export venture study

would provide invalid information and using the findings from a single export venture

to guide managerial practice would be harmful to the firm’s success.

Moving to the final column in Figure 1, a researcher may collect performance data

from multiple export ventures within the firm in order to test theory developed at the

export function level. Data of this kind might be appropriate if one is confident that

the sample of ventures provides representative information, or that it sheds light, on

what is happening across the firm’s totality of export ventures. Obtaining a

representative sample of ventures may not be difficult in the case of firms that

operate in a small number of relatively homogenous ventures, in which case the

number of ventures in the sample is close to the total number of ventures which

constitute firms’ export functions. Nonetheless, the task of obtaining representative

samples may become more difficult as the number and/or the diversity of ventures in

which firms operate increases – simply because data from larger samples may be

required. Therefore, in circumstances where firms operate in many and/or diverse

ventures, collecting information on samples of ventures and using the results to

make inferences at the export function level may not be practical if one wishes to

avoid biased data.

11
Second type of research question: determinants of performance of a cohort of

ventures within firms

A researcher may also be interested in analyzing how a particular cohort of export

ventures (i.e., a lower level unit) within firms’ export functions (i.e., a higher level

unit) varies across firms. This type of research objective suits situations where the

research is driven by theory which is specific to particular types of export ventures. In

such cases, collecting information on the entire export function of the firm may not be

appropriate. Indeed collecting information at the overall export function level and

inferring the results to the cohort of ventures within the firm’s export function

corresponds to what is called a downward cross level inference – an attempt to

make inferences at a lower level. Unless the relationships found at the export

function level hold at the cohort of ventures level, this type of cross level inference

will lead to biased recommendations for management and for theory development.

For example, the researcher may be interested in investigating the impact of export

market-oriented behavior on the performance of services ventures within firms’

export function portfolio. Cadogan et al. (2002) find that the relationship between

export market-oriented behavior and export performance is different for products and

for services exporters. Therefore, unless all the firm’s export ventures are service

ventures, collecting information at the export function level might mask the

relationships that exist for the service export venture cohort. Instead, the results

obtained at the export function level would reflect the effects of market orientation on

performance taking into account all the export ventures within the firm’s export

12
function (product, service and mixed ventures). In the case of this illustration,

collecting data at the overall export function level to test the effects of export market-

oriented behaviors on the export performance of firms’ service ventures would result

in biased findings.

In a similar vein, collecting performance data at the single export venture level to test

theory developed at the cohort of ventures level is not appropriate. Unless the single

venture is representative of the cohort of ventures as a whole, extrapolating

relationships found at the single venture level to the cohort of ventures will result in

upward cross level bias.

Collecting performance data from multiple export ventures within the firm’s export

function may also be unsuitable for the purposes of testing theory developed

regarding a cohort of ventures. Specifically, the sample of ventures would need to be

representative of the cohort, and unless this is the case, there is no guarantee that

the results obtained would be free from cross level bias.

The ideal type of performance data to collect to test theory developed for a cohort of

export ventures is data that is specific to the cohort itself. For economic

performance, for instance, this might include generating data on export sales across

the cohort of ventures (e.g., export sales generated across the entirety of the firm’

ventures of type X), or asking management to estimate the profits attributable to the

cohort (i.e., costs and revenues attributable across the entirety of the firm’ ventures

of type X).

13
Third type of research question: determinants of intra-firm variances in export

venture performance levels

As outlined earlier, the researcher may be concerned with studying how export

performance varies across multiple individual export ventures (i.e., the lower level

units) within the firm’s export function (i.e., the higher level unit) or, in other words,

the issue of intra-firm export venture performance variation.

Collecting performance data at the overall export function level is not suitable for

conducting this kind of study because it will not provide the researcher with within-

firm variance in export venture performance. It would also violate the principle that, in

order to obtain valid results, the level at which a theory is tested should match the

level at which that theory is developed. Collecting data at the cohort of ventures level

is also inappropriate here. Again, there is a cross level bias issue, and there is a

problem with the fact that the data will also fail to provide within-firm variance in

export venture performance. Obtaining data from a single export venture within the

firm’s export function is also unsuitable for the purposes of testing multiple venture

level theory. Again, the problem here is that performance data from only a single

export venture within firms will fail to provide data on intra-firm performance

variance.

In order to test theory about the determinants of intra-firm export venture

performance differences, one should collect data from a sample of multiple ventures

within the firm. Of course, one also must consider the generalizability of the sample

of ventures – if the sample is biased in important ways (e.g., only relatively

14
successful ventures are sampled), then the findings obtained my lack generalizability

to the entire export function level.

Fourth type of research question: determinants of the performance of a single export

venture within the firm’s export function across firms

It is possible that a researcher may be interested in testing theory which is

developed at the single export venture level. Specifically, in firms that are just

entering into export operations, and have only one export venture, researchers may

be interested in identifying the factors that eventually lead to venture success, and

greater internationalization (or venture failure, and retraction from export operations).

However, if the researcher is ultimately interested in explaining the overall

performance of the export function, data collection from only a single export venture

within the firm may not be appropriate, unless all the firm’s export ventures are

homogenous with regard to the impact of different antecedents on venture

performance.

Furthermore, export function level theory tested on data collected from only a single

venture also may suffer from selection bias problems. For example, selection bias

might occur if, on choosing a single venture to provide data on, respondents are

drawn to those ventures that are more successful, say (either for social desirability

reasons, or because such ventures are top of mind). In the latter cases, the ventures

may be more successful because strategies and plans (such as adaptation of the

marketing mix) have paid off. Conversely, in those ventures where performance is

poorer, poor performance might be attributable to the idiosyncratic venture-specific

15
factors that are beyond management control, such as exchange rate fluctuations,

regulatory upheavals, or changes in competitive profiles within specific markets: the

strategies and plans the firms implement (such as an adapted marketing mix) might

have little effect on success. If respondents select data from ventures that are

responsive to the firms’ strategies, the researcher may overestimate the impact of

the antecedent variables in the theoretical model. In this latter situation, the

researcher might incorrectly conclude that a strategy (e.g., adaptation of the

marketing mix) has a strong performance impact across all export ventures, when

the real relationship between the strategy and performance is weaker than predicted.

Discussion

As our examples above demonstrate, export performance theory can be undertaken

at various different implied levels:

1. Export function-wide theory: the export performance of the firm is viewed in its

totality, and no distinction is made between the performance levels achieved by

different export ventures.

2. Export cohort-wide theory: the cohort is believed to behave differently from other

types of ventures, and so performance is assessed across the cohort, but no

distinction is made between the performance levels achieved within the cohort of

export ventures.

3. Intra-firm export venture level theory: intra-firm variances in export venture

16
performance levels are the subject of interest.

4. Single export venture level theory: a single venture within the firm is the focus of

study. Notionally, the study is not interested in the overall performance of the export

function or of a cohort of ventures, and is also not interested in explaining intra-firm

export venture performance variation.

The data that are required to test theories developed at these different levels must

be appropriate to the level of the theory. And of course, research studies may be

interested in developing and testing multi level theories of export success, (e.g., a

study may contain theory at both the export function-wide level and intra-firm export

venture level), and so would require data from multiple levels for theory testing

purposes.

We also argue that attention is best focused on theories that have potential

usefulness for managerial practice, and that can assist managers in the task of

boosting the overall export performance of companies. Export function-wide theory

has strong potential to aid in this respect: factors that are shown to determine the

overall performance of the export function can be flagged by researchers as potential

factors to add to managers’ dashboard of malleable controls that enhance export

outcomes. Similarly, export cohort-wide theory can be used to develop more fine-

grained models of the determinants of export performance, by taking into account

the idiosyncrasies of the different cohorts of ventures which constitute firms’ export

venture portfolios. The results can be used to provide export practitioners with

recommendations about how they can improve the performance of strategically

17
important venture cohorts (e.g., service export ventures; product export ventures;

high tech export ventures; commodity export ventures; new product export ventures),

and this may ultimately lead to increases in the overall level of performance achieved

at the export function level.

Theories which are developed at the intra-firm export venture level analyze why,

within the same firm, different export ventures exhibit different levels of performance.

Accordingly, multiple venture level theories of export performance offer managers

insights into the specific “levers” that influence the successes of specific ventures.

Accordingly, we argue that research at the intra-firm export venture level is a critical

piece of the jigsaw in terms of understanding the drivers of the overall, export

function-wide, success of exporting companies.

However, theories constructed at the single venture level have less obvious practical

relevance, since they ignore the fact that the export venture is nested within a

company, and instead treat the export venture as though its performance is of

ultimate interest. These theories exclude from consideration the overall export

success of the company, or of cohorts of export ventures, or of the relative success

of other export ventures within the firm. Yet, our earlier discussions highlight the fact

that enhanced venture performance (e.g., market share objectives, or the

achievement of venture-specific sales goals) will come at a cost. Since these costs

will often be partly or wholly incurred outside of the venture (e.g., by production, or

operations, or marketing, or at the broader export function level), success at the

individual export venture level may come at a price at the functional level that

outweighs the benefits attained at just one single venture. However, since only a

18
single venture is the focus of study, no information is provided about its broader

context. Ultimately, extrapolating from findings uncovered at a single export venture

level to the export function as a whole, or to the cohort-based or intra-firm export

venture level, can produce biased findings and potentially lead to the presentation of

perverse and harmful recommendations for managerial practice. As a result, we

strongly recommend that results obtained in studies conducted at the single export

venture level be validated at higher levels (i.e., at the export function, cohort and

intra-firm levels) before recommendations are made to managers.

Conclusions

Export performance research has been topical for decades, and during this time, a

plethora of alternative ways of measuring this construct has proliferated. Historically,

much attention was paid to the content of export performance measurement,

including debates about economic versus non-economic measures, or objective

versus subjective measures. More recently, however, researchers now recognize

that export performance is not a single construct, but is multi-facetted, and that the

determinants of different kinds of export performance may differ. In the same vein,

we argue that researchers need to explicitly recognize that export performance can

occur at different levels in the firm, and that each level will require its own theory.

Furthermore, theory developed at one level may not hold for another level, and

export performance measured at one level may not be appropriate for testing models

of determinants of export success developed at a different level.

19
Reflecting on the state of the art in terms of the export performance literature, we

notice some interesting patterns, and several important areas where research is

needed. First, while many studies examine export performance using either a

venture level approach (e.g., Murray et al., 2007) or a function level approach (e.g.,

Calantone et al., 2006), most seek inter-firm comparisons, in order to aid managers

in their bids to improve the performance of their export operations.

Second, however, when researchers develop and test theory that is about a single

export venture performance, they rarely recognize explicitly that the research cannot

automatically be generalized to a higher level. As a result, the findings often are

interpreted inappropriately (by the researchers and potentially by readers) as being

valid beyond the venture-specific level to the level of the entire export function.

Third, a problem that export performance researchers may encounter when

submitting papers to scholarly journals is that their work can be rejected by reviewers

who insist on performance assessment at an incorrect level, and who criticize

performance assessment that is undertaken at the correct level. For example, in a

recent review from a leading journal, one of the authors received the following

comment from a reviewer who stated that the work suffered from problems with the

measurement of export performance because: “it is indeed problematic in not asking

the respondents to focus on one product-market in responding to the survey”. The

reviewer’s problem is that, in their opinion, all export performance research should

use the export venture level, regardless of the fact that the theory being tested is

developed at the export function level. Thus, the reviewer is trying to impose the idea

that a theory along the lines of “something that firms’ export functions do to varying

20
levels causes the export functions to have different levels of export success” should

not be tested by collecting data about the success of export functions. The illogicality

of the suggestion is obvious.

More subtle, but equally worrying, a second reviewer (of the same paper) had a

slightly different take on the performance measurement issue. This time, the

reviewer commented that the problem with the model was that it was developed at

the level of the export function. “[T]he study is undertaken at the overall [export

function] level. I am concerned about this[. E]xport performance …[is] likely to differ

from one export market to another… the use of the export venture as the unit of

analysis would have taken account of such differences across export markets that is

not possible in the firm level investigation employed in this study.” The logic of the

reviewer’s argument implies that, since ventures can vary in performance levels, only

intra-firm venture performance information can be used to study export success.

Indeed, by implication, the reviewer is stating that theories about the determinants of

export function performance should not be developed and tested (because ventures

within an export function may have different performance levels) – only intra-firm

export venture performance theories can be studied. We disagree wholeheartedly

with this stance. We believe that it is perfectly valid to try to identify broad factors that

might make one firm better at exporting than another.

Unfortunately, invalid criticisms of the kinds illustrated here are not rare incidents, but

happen regularly in the review process.

21
To put matters into context, within the broader marketing literature, business success

is often an important dependent variable. In these studies, performance metrics are

typically assessed at the level of the entire business (e.g., customer satisfaction,

ROI, Tobin’s q, etc). It would be unusual to see researchers measuring performance

at a specific sub level, such as a single product-market level, and using that

information as though it were a valid measure of the success of the whole of the

business. It would likewise be worrying if research that investigates the determinants

of business success, and that measures performance at the firm level (e.g., using a

measure such as Tobin’s q), were rejected because of reviewers’ personal

preferences for single product-market performance data!

Fourth, as the above discussions show, relying on data from a single export venture

to shed light on functional export performance is dangerous, and should not be

encouraged. Rather, we recommend that researchers explicitly address the level of

analysis issue when developing their theory, and explain in their methodological

section how their data match the theory developed. For instance, if the theory is

implicitly or explicitly a theory of the determinants of export function performance,

then the appropriate level of measurement is at the functional level, and the

researcher should explain this to the reader.

A number of areas for future research are also apparent from our previous

discussions.

First, much research has been undertaken that uses single ventures as the basis for

making recommendations at the export function level. Without validation studies, we

22
remain unsure as to the appropriateness of this practice. Future researchers need to

focus on testing the assumptions the research community has built up based on the

findings of single venture level research studies.

Second, there is an opportunity to expand the boundaries of export performance

research. Specifically, relatively few research studies assess export performance at

the cohort level or at the intra-firm level. Cohort research involves studying

meaningful groups of ventures (e.g., services exports) to identify determinants of

inter-firm export success at the cohort level, while intra-firm performance research

involves studying the determinants of performance variance across ventures within

the firm. Undertaking research of this kind, potentially, may contribute substantially to

theory and practice. On the one hand, it would open the door to the possibility of

identifying new antecedent variables, specific to the level of analysis chosen. On the

other hand, insights from research undertaken at the cohort and intra-firm levels may

result in adjustments to the nomological network explaining how the dimensions of

export success and their determinants are related. For instance, in the context of an

intra-firm model of export venture performance, venture level moderators (moderator

variables that vary across ventures in the firm) of the relationships between venture

performance determinants and venture performance dimensions may be uncovered.

Finally, extending the reasoning above, it becomes clear that there is a need for

researchers to integrate higher and lower level models of export performance into

multi level models of export success. For instance, a multi level model of export

performance might simultaneously model factors that explain intra-firm export

venture performance variance and factors that explain inter-firm variances in export

23
function performance. A model of this kind is likely to have stronger explanatory

power, and be able to provide more valid and precise information that can be used

by scholars to develop recommendations for practicing export managers.

24
Endnote

1. Our list of potential theories of interests and associated levels of analysis

presented in the rows in Figure 1 is not exhaustive. For instance, an additional row

could be added for researchers interested in identifying causes of performance

variation between cohorts of ventures within the same firm. However, the list covers

the main research questions we believe researchers are interested in answering.

25
References

Aulakh, P.S., Kotabe, M. and Teegen, H. (2000), “Export strategies and performance

of firms from emerging economies: evidence from Brazil, Chile, and Mexico”,

Academy of Management Journal, Vol. 43 No. 3, pp. 342-361.

Brouthers, L.E. and Xu, K. (2002), “Product stereotypes, strategy and performance

satisfaction: the case of Chinese exporters”, Journal of International Business

Studies, Vol. 33 No. 4, pp. 657-677.

Cadogan, J.W., Kuivalainen, O. and Sundqvist, S. (2009), “Export market-oriented

behavior and export performance: quadratic and moderating effects under

differing degrees of market dynamism and internationalization”, Journal of

International Marketing, Vol. 17 No. 4, pp. 71-89.

Cadogan, J.W., Sundqvist, S., Salminen, R.T. and Puumalainen, K. (2002), “Market-

oriented behavior: comparing service with product exporters”, European Journal

of Marketing, Vol. 36 No. 9/10, pp. 1076-1102.

Calantone, R.J., Kim, D., Schmidt, J.B. and Cavusgil, S.T. (2006), “The influence of

internal and external firm factors on international product adaptation strategy

and export performance: A three-country comparison”, Journal of Business

Research, Vol. 59, pp. 176-185.

Cavusgil, S.T. and Zou, S. (1994), “Marketing strategy-performance relationship: an

investigation of the empirical link in export market ventures”, Journal of

Marketing, Vol. 58 No. 1, pp. 1-21.

Douglas, S.P. and Wind, Y. (1987), “The myth of globalization”, Columbia Journal of

World Business, Vol. 22 No. 4, pp. 19-29.

26
Ito, K. and Pucik, V. (1993), “R&D spending, domestic competition, and export

performance of Japanese manufacturing firms”, Strategic Management Journal,

Vol. 14 No. 1, pp. 61-75.

Katsikeas, C.S., Leonidou, L.C. and Morgan, N.A. (2000), “Firm-level export

performance assessment: review, evaluation, and development”, Journal of the

Academy of Marketing Science, Vol. 28 No. 4, pp. 493-511.

Katsikeas, C.S., Samiee, S. and Theodosiou, M. (2006), “Strategy fit and

performance consequences of international marketing standardization”,

Strategic Management Journal, Vol. 27 No. 9, pp. 867-890.

Klein, K.J., Dansereau, F. and Hall, R.I. (1994), “Levels issues in theory

development, data collection, and analysis”, Academy of Management Review,

Vol. 19 No. 2, pp. 195-229.

Kohli, A.K. and Jaworski, B.J. (1990), “Market orientation: the construct, research

propositions, and managerial implications”, Journal of Marketing, Vol. 54 April,

pp. 1-18.

Lages, L.F., Jap, S.D. and Griffith, D.A. (2008), “The role of past performance in

export ventures: a short-term reactive approach”, Journal of International

Business Studies, Vol. 39 No. 2, pp. 304-325.

Madsen, T.K. (1998), “Executive insights: managerial judgment of export

performance”, Journal of International Marketing, Vol. 6 No. 3, pp. 82-93.

Morgan, N.A., Kaleka, A. and Katsikeas, C.S. (2004), “Antecedents of export venture

performance: a theoretical model and empirical assessment”, Journal of

Marketing, Vol. 68 No. 1, pp. 90-108.

27
Mossholder, K.W. & Bedeian, A.G. (1983), “Cross-level inference and organizational

research: perspectives on interpretation and application”, Academy of

Management Review, Vol. 8 No. 4, pp. 547-558.

Murray J.Y., Gao, G.Y., Kotabe, M., and Zhou, N. (2007), “Assessing measurement

invariance of export market orientation: A study of Chinese and non-Chinese

firms in China”, Journal of International Marketing, Vol. 15 No. 4, pp. 41-62.

Narver, J.C. and Slater, S.F. (1990), “The effect of market orientation on business

profitability”, Journal of Marketing, Vol. 54 October, pp. 20-35.

Rousseau, D.M. (1985), “Issues of level in organizational research: multi-level and

cross-level perspectives”, in Cummings L. L. and Staw, B.M (Eds.), Research in

Organizational Behavior, Greenwich, CT: JAI Press, Vol. 7, pp 1-37.

Sousa, C.M.P., Martínez-López, F.J.and Coelho, F. (2008), “The determinants of

export performance: a review of the research in the literature between 1998 and

2005”, International Journal of Management Reviews, Vol. 10 No. 4, pp. 343-

374.

28
View publication stats

Figure 1: Matching levels of theory and data

Level at which performance data is needed given a research
question with a specific level of theory
Performance data Performance data Performance data Performance data
collected at the level of collected at the level of collected at the export collected at the export
the entire export a cohort (i.e., a subset) venture level venture level
function of export ventures (from a single venture) (from multiple ventures)

Export function-wide theory Tentatively suitable Tentatively suitable

Provided the cohort of Provided the sample of
What variables determine how the Ideal
ventures generalizes to Not suitable ventures generalizes to
performance of firms’ entire export the population of the population of
functions varies across firms? ventures in the firm. ventures in the firm.

Tentatively suitable
Export cohort-wide theory
Provided the sample of
What variables determine how the
ventures generalizes to
performance of a cohort of firms’ Not suitable Ideal Not suitable
the cohort of ventures in
export ventures varies across
Level the firm.
firms?
of
Theory Intra-firm export venture level Ideal
theory Provided the sample of
What variables determine how the Not suitable Not suitable Not suitable ventures generalizes to
performance of individual export the population of
ventures varies within the firm? ventures in the firm.

Single export venture level theory

What variables determine how the Suitable
Not suitable Not suitable Ideal
performance of a single export but resource inefficient
venture varies across firms?

29