Page 1 of 14

Archives of Business Research – Vol. 9, No. 4

Publication Date: April 25, 2021

DOI:10.14738/abr.94.9991.

Alsaif, T. (2021). A Path Analysis On The Strategic Determinants Of The Average Revenue Per User In The Saudi Telecom Sector.

Archives of Business Research, 9(4). 43-56.

Services for Science and Education – United Kingdom

A Path Analysis On The Strategic Determinants Of The Average

Revenue Per User In The Saudi Telecom Sector

Dr. Talal Alsaif

Assistant Professor, University of Ha’il, College of Business Administration

ABSTRACT

This study compared a restructured hierarchical regression using structural

equation modeling (SEM) with a path analysis SEM Regression using The Rules of

Casual Order. The dataset originated from Jones & Alshammari (2017) which studied

the Value-Added Intellectual Coefficient (VAIC) determinants and capital

expenditures (CAPEX) effects on the average revenue per user (ARPU). The

comparisons showed CLE and CEEcap explained 61% of ARPU. For every 1 unit of

change in CLE and CEEcap combined, produces 2 units of change in ARPU. The results

on HCEcap and SCEcap where inconsistent, regression weights were insignificant at the

p ≤ .001 level, and both determinants did not correlate with Revenue. This study

showed that causation can be established prior to any multivariate or SEM statistical

procedures. The rules of casual order are an effective way of designing a model based

on reality and shows the true effects among observed variables.

INTRODUCTION

The difference between multiple regression and path analysis

The strength of association establishes a dependent variable’s interpretation using two or more

independent variables. Path analysis is considered an extension of multiple regression which

explicitly formulate casual models (Bryman & Cramer, 2011). Although Bryman & Cramer

(2011) strongly emphasize the path analysis cannot establish causality, it can be applied using

the rules of causal order to establish cause and effect (p. 309). Davis (1985) examines The Rules

of Casual Order as a precursor to any statistical applications. He further emphasizes that The

Rules of Casual Order, “...have nothing to do with statistics.” and are premised on the concept of,

“after cannot cause before” (Davis, 1985, p.11). This study will apply The Rules of Casual Order

to establish causality of the ARPU based on organizational activities that represent strategic

determinants. Upon structuring a model based of The Rules of Casual Order, path analysis is used

to examine and compare the model to a hierarchal regression model.

The ARPU strategic determinants

Before the application of casual order, a thorough discussion on the determinants is needed. The

activities of the organization that captures strategic implementation are recorded in the general

ledger. Therefore, to develop a casual order that shows the determinants with causality on ARPU,

accounting items are utilized. Both Kaplan & Norton (2008) and Porter (1985) discussion this

aspect of strategic implementation indirectly or not actually mentioning the role of these

accounting items directly. Kaplan & Norton (2008) mentions capital budgeting and the role

OPEX, CAPEX and STRATEX play in strategy execution. However, the accounting items are a part

Page 2 of 14

44

Archives of Business Research (ABR) Vol. 9, Issue 4, April-2021

Services for Science and Education – United Kingdom

of the financial systems used to classify spending on strategy execution (Kaplan & Norton, 2008,

p. 115).

Porter (1985) also acknowledge the accounting items but under the activities that transpire

inside the value chain. He describes the value chain as a basic tool that systematically examines

all activities an organization performs (p.33). Both Porter (1985) and Kaplan & Norton (2008)

agree that organizational activities that reflect strategy execution should be separated from

normal activities. This was clearly stated by Porter (1985) as a necessary isolation and no

grouped together with the normal operational activities under the same accounting

classifications (p.39).

ARPU is a key performance indicator that is globally recognized in all telecommunication

companies worldwide. Jones & Alshammari (2017) expands on the ARPU as a global KPI used to

determine network expansion and mediate between network and service provisioning of

telecom technologies (p. 161). Based Kaplan & Norton (2008) view of productivity and efficiency

alignment between the corporate, business, and functional level strategies, ARPU facilitates this

alignment. In addition to ARPU as a global KPI, and a strategic organizational alignment

component, it also acts as a key forecast indicator for the adoption rate of advanced technologies

(Stordahl & Elnegaard, 2007). This study identifies the causality based on the ARPU determinants

as represented as accounting line-items. The formula for the KPI is:

The objectives of this study are:

1. Demonstrate with path analysis the difference between association and causation.

2. The direct and indirect effects of CAPEX Labor on the ARPU.

3. What the levers of strategic implementation and how do they affect ARPU.

REVIEW OF LITERATURE

What defines association?

Association rules are defined as correlations between attributes (Hair, Joseph F., Anderson,

Rolph E., Tatham, Ronald L., Black, 1998). Correlations such as the Pearson Correlation estimates

the linear association between attributes or variables (Kline, 2016). Given that correlations play

and important role in SEM and establishes the degree of association between variables, there are

factors that affect correlations (Schumacker, Randall E; Lomax, 2004). Three factors that affect

correlations, but have no effect on causation are level measurement, nonlinearity, and missing

values. There are more factors that affect correlations, but the point of mentioning the factors is

to highlight that associations measured by correlation can be spurious (Bryman & Cramer, 2011,

p.9). These factors to do not remove the possibilities of a spurious relationship or causality

between two or more variables. This is the major difference between associations and causation.

Page 3 of 14

45

Alsaif, T. (2021). A Path Analysis On The Strategic Determinants Of The Average Revenue Per User In The Saudi Telecom Sector. Archives of

Business Research, 9(4). 43-56.

URL: http://dx.doi.org/10.14738/abr.94.9991

What defines causation?

Causation must be established prior to the statistical measure of association such as correlation

or regression. Both Bryman & Cramer (2011) and Davis (1985) acknowledge this point. There

is a key point of establishing causality, mentioned by Bryman & Cramer (2011) which is, “...cause

precedes effect” (p.10). Davis (1985) mentions this same point in the form of rules necessary to

establish the cause and effect between two variables (p.8). These ten rules precede statistics and

are based on the third principle of casual order - “after cannot cause before” (Davis, 1985, p.11).

Davis (1985) three principles of casual order are:

1. Causal analysis in social research depends on assumptions about causal direction.

2. Assumptions depend on empirical beliefs about how the world works.

3. Assumptions are not arbitrary or whimsical. They are based on ten rules. Davis (1985)

ten rules of causal order are presented in Table 1.

This study will design a causal design of a composite ARPU of four leading telecom companies in

the Kingdom of Saudi Arabia. A path analysis on this design is compared to hierarchal regression

model from Jones & Alshammari (2017), re-created using SEM. The purpose of this procedure is

to test Davis (1985) rules of causal order. Since both models are created in SEM, the three criteria

for a statistical significant comparison between the two models is based on Schumacker & Lomax

(2004) which states:

1. The model must have a non-statistically significant chi-square.

2. A root-mean-square error (RMSEA) value of .05 or less.

3. A statistical significance of individual parameter estimates for the path models.

4. The magnitude and direction of parameter estimates must make sense.

SEM is used in this comparison between models, but there are necessary differences in the

determinants used in the path analysis model. The path analysis model will use subscribers and

Revenue determinants that comply with the rules of casual order as opposed to statistically

significant Pearson Correlation results. This means that the nature of accounting line-items, any

calculations between them, and value as represented by special accounting treatments will

dictate determinant inclusion.

VAIC, Capex labor, and accounting line-items strategic determinants

At this point value-add from an accounting and intangible perspective must be discussed because

the variables in both models warrants justification. Pulic (2004)leverages accounting line-items

to calculate the VAIC to reflect intangible value of an organization. There has been criticism

regarding Dr. Pulic’s VAIC indicator and inconsistencies when correlated with traditional

financial measures (Andriessen, 2004; Iazzolino & Laise, 2013; Stahle et al., 2011). However,

Jones & Alshammari (2017) address this criticism upon observing the behavior of VAIC versus

homogenous effects and establish as posit which states:

When VAIC measured in terms of productivity and equity ratios such as ROA and ROE,

given all organizations in a market index or exchange as a sample size, a moderating

effect measuring customers’ responses to the wide range of products and services in

terms of Revenue is needed (p. 167).