Drivers of and barriers against market orientation: a study of Turkish container ports

Abstract

Privatization transforms the operations-focused and stakeholder-independent approach of container port management into a competition and market-oriented perspective. This integrates a marketing philosophy with the management of the overall port organization. In this way, ports become more responsive to market demand and create profitable and sustainable customer value, so as to achieve competitive advantage. This change is caused both by the evolving port governance systems, as well as by changing market demand, requiring ports to respond to the needs of the supply chains they serve. An increasingly prevalent logistics perspective in maritime transport is accelerating pressure on ports—and other members of service supply chains—encouraging them to adopt market-oriented strategies. Accordingly, the purpose of this study is to discover both the drivers and the barriers that container ports face while adopting market-oriented strategies. Taking Turkish container ports as the main unit of analysis, this article combines the Delphi method with fuzzy analytic hierarchy processing to identify the prioritized opinions of industry stakeholders. The results indicate that customer relations are perceived to be very important for container ports, in terms of communicating value-added services, and even more important than economic concerns or possessing market knowledge. Often, however, ports define their customers within a limited scope, and fail to adopt a supply chain perspective. Therefore, new strategies, such as market resegmentation including support service providers, or evaluating end-customer demands by analyzing supply chains that include port services, should be developed to eliminate the prioritized barriers. Drivers, such as collecting market information, or building market-oriented teams, can be used more effectively to make market orientation a new competition tool for all ports. Finally, our results provide new variables to scholars studying port marketing and put forward recommendations for testing the relationships between these variables.

Appendices

Appendix I: Delphi study open-ended questions

• 1. How do port operators classify their clients? Who are the major client groups?

• 2. What are the market orientation activities of port operators (classifying clients into markets, researching customer expectations, measuring customer satisfaction, collecting market information, monitoring market share, tailoring special services for different customer categories, etc.)?

• 3. What do you think drives private port operators in Turkey to engage in these activities?

• 4. What do you think prevents private port operators in Turkey from engaging in market orientation activities?

Appendix II

The scale for AHP and fuzzy AHP is shown in Table 5:

Table 5 AHP and fuzzy AHP scales.

Appendix III

Fuzzy set is a class of objects containing a continuum of grades of membership, characterized by a membership function. A fuzzy number is a special fuzzy set which can be defined as $F = \left\{ {x,\mu_{F} \left( x \right),x \in R} \right\}$, where $x$ value ranges on real line $R_{1} : - \, \infty < x < + \infty$ and $\mu_{F} \left( x \right)$ is a continuous mapping from R₁ to the close interval $\left[ {0,1} \right]$. A triangular fuzzy number can be denoted by $\tilde{M}$, where the symbol “$\sim$” represents a fuzzy set for a triangular fuzzy number. Its membership function, $\mu \left( {x |\tilde{M}} \right)$ , assigns a grade of membership to each grade ranging between zero to one. It can be denoted as (l/m, m/u) or (l,m,u), where l represents the smallest value, m represents the most promising value and u represents the largest possible value in a fuzzy event (Mahmoodzadeh et al. 2007) (Fig. 1). It is important to note that the cases, where $l = m = u$, do not represent any fuzzy numbers (Chang 1996).

Fig. 1

Source Kahraman et al. (2003)

Triangular Fuzzy Numbers.

The membership function of $\tilde{M}$ in Fig. 1$\mu \left( {x |\tilde{M}} \right)$ can be defined as (Saaty 1980; Kahraman et al. 2003)

$$\mu \left( {x |\tilde{M}} \right) = \left\{ {\begin{array}{*{20}l} {0 } \hfill & { x < l\,or\,x > u,} \hfill \\ {\left( {x - l} \right)/\left( {m - l} \right), } \hfill & {l \le x \le m,} \hfill \\ {\left( {u - x} \right)/\left( {u - m} \right), } \hfill & {m \le x \le u.} \hfill \\ \end{array} } \right.$$

A fuzzy number has its left $\left( {l\left( y \right)} \right)$ and right $\left( {r\left( y \right)} \right)$ hand side representation of each degree of membership, and always given accordingly as

$$\tilde{M} = M^{l\left( y \right)} , M^{r\left( y \right)} = \left( {l + \left( {m - l} \right)y, u + \left( {m - u} \right)y} \right), y \in \left[ {0,1} \right]$$

The fundamental laws are followed for the operations between the two fuzzy numbers (Kaufmann and Gupta 1991).

$$M_{1} + M_{2} = \left( {l_{1} + l_{2} , m_{1} + m_{2} , u_{1} + u_{2} } \right),$$

$$M_{1} \oplus M_{2} = \left( {l_{1} l_{2} , m_{1} m_{2} , u_{1} u_{2} } \right),$$

$$\lambda \oplus M_{1} = \left( {\lambda l_{1} , \lambda m_{1} , \lambda u_{1} } \right), \lambda > 0, \lambda \in R,$$

$$M_{1}^{ - 1} = \left( {1/l_{1} , 1/m_{1} , 1/u_{1} } \right).$$

In Fuzzy AHP calculations, first the fuzzy synthetic value is calculated by (Mahmoodzadeh et al. 2007, p. 138)

the object set $X = \left\{ {x_{1} ,x_{2} , x_{3} , \ldots ,x_{n} } \right\},$

the goal set $B = \left\{ {b_{1} ,b_{2} , b_{3} , \ldots ,b_{t} } \right\}$ and

the TFNs for each goal $g_{i}$ is represented by $M_{{g_{i} }}^{j} \left( {j = 1,2, \ldots ,t} \right)$ such that

$$M_{{g_{i} }}^{1} ,M_{{g_{i} }}^{2} , \ldots ,M_{{g_{i} }}^{t} \;i = 1,2, \ldots , n.$$

This can therefore be represented as

$$S_{i} = \mathop \sum \limits_{j = 1}^{t} M_{{g_{i} }}^{j} \otimes \left[ {\mathop \sum \limits_{i = 1}^{n} \mathop \sum \limits_{j = 1}^{t} M_{{g_{i} }}^{j} } \right]^{ - 1}.$$

(1)

$\mathop \sum \limits_{j = 1}^{t} M_{{g_{i} }}^{j}$ is obtained by performing the fuzzy addition operation of $t$ values for a particular matrix such that

$$\mathop \sum \limits_{j = 1}^{t} M_{{g_{i} }}^{j} = \left( {\mathop \sum \limits_{j = 1}^{t} l_{j} , \mathop \sum \limits_{j = 1}^{t} m_{j} \mathop \sum \limits_{j = 1}^{t} u_{j} } \right)$$

(2)

while $\left[ {\mathop \sum \limits_{i = 1}^{n} \mathop \sum \limits_{j = 1}^{t} M_{{g_{i} }}^{j} } \right]^{ - 1}$ is obtained by performing the fuzzy addition operation of $M_{{g_{i} }}^{j} \left( {j = 1,2, \ldots ,t} \right)$ values such that

$$\mathop \sum \limits_{i = 1}^{n} \mathop \sum \limits_{j = 1}^{t} M_{{g_{i} }}^{j} = \left( {\mathop \sum \limits_{i = 1}^{n} l_{i} , \mathop \sum \limits_{i = 1}^{n} m_{i} \mathop \sum \limits_{i = 1}^{n} u_{i} } \right).$$

(3)

The inverse vector $\left[ {\mathop \sum \limits_{i = 1}^{n} \mathop \sum \limits_{j = 1}^{t} M_{{g_{i} }}^{j} } \right]^{ - 1}$ is then computed by

$$\frac{1}{{\mathop \sum \nolimits_{i = 1}^{n} u_{i} }}, \frac{1}{{\mathop \sum \nolimits_{i = 1}^{n} m_{i} }}, \frac{1}{{\mathop \sum \nolimits_{i = 1}^{n} l_{i} }}$$

(4)

(Saaty 1980; Chang 1996; Kahraman et al. 2003).

In defuzzification, a procedure similar to the approach of determining the left and right score by fuzzy min and fuzzy max is followed to determine the corresponding fuzzy scores as crisp numbers. Then, the total score is determined as a weighted average with respect to the membership functions.

In Chen and Hwang (1992)’s method, for every i-th criterion, alternatives are evaluated with fuzzy numbers $\tilde{f}_{ij}$ for $j = 1,2, \ldots , J$, and $J$ is the number of alternatives. The TFNs for this set are $\tilde{f}_{ij} = \left( {l_{ij} , m_{ij} , u_{ij} } \right)$, $j = 1,2, \ldots , J$ and i-th (one) criterion is included in the defuzzification computation. The algorithm to determine the crisp values for the i-th criterion is employed in four steps.

• 1. Normalization:

$$u_{i}^{ \hbox{max} } = \mathop {\hbox{max} }\limits_{j} u_{ij} , l_{i}^{ \hbox{min} } = \mathop {\hbox{min} }\limits_{j} l_{ij}$$

(5)

$$n_{ \hbox{min} }^{ \hbox{max} } = u_{i}^{ \hbox{max} } - l_{i}^{ \hbox{min} }$$

(6)

compute for all criteria $n_{j} , j = 1, 2, \ldots , J$.

$$x_{lj} = \left( {l_{ij} - l_{i}^{ \hbox{min} } } \right)/\Delta_{ \hbox{min} }^{ \hbox{max} }$$

(7)

$$x_{mj} = \left( {m_{ij} - l_{i}^{ \hbox{min} } } \right)/\Delta_{ \hbox{min} }^{ \hbox{max} }$$

(8)

$$x_{uj} = \left( {u_{ij} - l_{i}^{ \hbox{min} } } \right)/\Delta_{ \hbox{min} }^{ \hbox{max} }$$

(9)

• 2. Compute left (ls) and right (us) normalized values, for $j = 1, 2, \ldots , J$.

$$x_{j}^{ls} = x_{mj} /\left( {1 + x_{mj} - x_{lj} } \right)$$

(10)

$$x_{j}^{us} = x_{uj} /\left( {1 + x_{uj} - x_{mj} } \right)$$

(11)

• 3. Compute total normalized crisp value, for $j = 1, \,2\,, \ldots ,\, J$.

$$x_{j}^{crisp} = \left[ {x_{j}^{ls} \left( {1 - x_{j}^{ls} } \right) + x_{j}^{us} x_{j}^{us} } \right]/1 - x_{j}^{us} + x_{j}^{us}$$

(12)

• 4. Compute crisp values for $j = 1, \,2\,, \ldots , \,J$.

$$f_{ij} = l_{i}^{ \hbox{min} } + x_{j}^{\text{crisp}} \Delta_{ \hbox{min} }^{ \hbox{max} } .$$

(13)

This computation is done for all criteria, $\tilde{f}_{ij} , i \in \tilde{n}$. Here, $\tilde{n}$ denotes the set of criteria evaluated with fuzzy numbers (Opricovic and Tzeng 2003).

The fuzzy evaluation matrix for respondent 1 is shown in Tables 6 and 7. The comparisons were then transferred into a comparison matrix through the preference scale. In the calculations, in Tables 6 and 7, Eqs. (1–4), were followed. All matrices of the remaining 19 respondents were made in a similar way.

Table 6 Drivers—fuzzy evaluation matrix for respondent 1

Table 7 Barriers—fuzzy evaluation matrix for respondent 1

The fuzzy synthetic values for Respondent 1 are calculated and the results are shown in Table 8.

Table 8 Fuzzy synthetic values $\left( {S_{ij} } \right)$ for respondent 1—drivers and barriers

$$\overline{{\tilde{W}_{i} }} = \left( {\mathop \prod \limits_{d = 1}^{D} \tilde{W}_{i}^{d} } \right)^{{\frac{1}{D}}} , \forall d = 1, 2, \ldots , D$$

(14)

Here, $\overline{{\tilde{W}_{i} }}$ represents the combined fuzzy weight of decision element $i$ of $D$ respondents, whereas $W_{i}^{k}$ is the fuzzy weight of decision element $i$ of respondent $d$ and $D$ illustrates the number of decision makers.

Table 9 showing the calculations are as follows:

Table 9 CFCS defuzzification method_Δ_(max–min) calculation

Normalized crisp values showing the final results are shown in Table 10 for drivers and Table 11 for barriers.

Table 10 CFCS defuzzification_ crisp values/normalized crisp values calculation_drivers

Table 11 CFCS defuzzification_ crisp values/normalized crisp values calculation_barriers

Appendix IV

The consistency of data is measured and reported to ensure that the degree of consistency is satisfactory to confirm that the AHP results are meaningful. Perfect consistency rarely occurs; however, the pairwise comparisons in a judgment matrix can be considered as consistent if the corresponding CR is less than 10% (Saaty 1980).

$$\delta_{ \hbox{max} } = \frac{1}{n}\mathop \sum \limits_{i = 1}^{n} \frac{{i{\text{th entry in}} Aw}}{{i{\text{th entry in}} w}}.$$

(15)

Then, CI is calculated for each matrix by the following formulae:

$${\text{CI}} = \frac{{\delta_{ \hbox{max} } - n}}{n - 1}.$$

(16)

Later, CI is divided by random index (RI) value to calculate the CR.

$${\text{CR}} = \frac{\text{CI}}{\text{RI}}.$$

(17)

RI is the average value of CI for randomly chosen entries in $A$, which is obtained from a large number of simulation runs; thus it depends on the number of criterion. The value of RI for matrices with 1–10 criterions is shown in Table 12:

Table 12 Average random index (RI) based on matrix size.