1. INTRODUCTION
About 59% of Indonesia’s economic growth is contributed by SME (Marta, 2016). The productivity of Indonesians’ SME is far lower than the big enterprise. The implementation of best SCM is needed to enhance enterprises’ productivity (Daughtery et al., 2005). Supply chain management (SCM) can improve the efficiency and productivity, and also improve the competitive advantage and customer satisfaction (Njoku and Kalu, 2015). Customer Requirements and Design Requirements to implement SCM for SME should be established to help SME increase their competitiveness.
In 2016, non–oil and gas processing had the greatest contribution to Indonesia GNP, about 20.51%. Compared to the previous years there’s a significant diminishing development in the industry. The national statistical bureau stated that wood industry was one of the main cause of this unprecedented fall. In this paper, we tried to help wood industry SMEs by implementing SCM concepts to improve SMEs’ productivity and enhance their contribution to national GDP.
2. REVIEW OF LITERATURE
2.1 Productivity in Suply Chain Management
Based on literature study, productivity is one of the main objective when a company/organization implement supply chain management (Daughtery et al., 2005). Supply Chain Management ensured efficiency and productivity improvement, competitive advantage, dan customer satisfaction (Njoku and Kalu, 2015).
Logistic function and Supply Chain Management can potentially help the company/organization to achieve cost and value advantage.
2.2 Supply Chain Management Adoption in SME
Apart from the benefit of Supply Chain Management, consideration of Supply Chain Management implementation most of the time are based on the investment cost. There’re also indication of huge obstacles in implementing Supply Chain Management (Arend and Wisner, 2005).
Based on empirical study of Supply Chain Planning in manufacturing SMEs’ in South East Asia, it is concluded that: Organizational hierarchy couldn’t facilitate Supply Chain Management, relation of SMEs’ Supply Chain Management and their supliers highly impact the acuracy and realibility of supply chain, and poorly identified tactical operation might negatively impact SCM (Huin et al., 2002).
Strategic fit evaluation based on empirical study conducted to 200 SMEs in United States, Mexico and Europe, it was found that: Good performing SMEs might implement SCM; but SCM wasn’t the strategic fit in some performance measurements (Arend and Wisner, 2005).
2.3 Quality Function Deployment Method
Quality Function Deployment (QFD) method is used to acquire Customer Requirements (CR). QFD helps to identify and prioritize comparative advantage for every customer needs (Wang et al., 2018). Apart from considering customer needs, QFD can also associate those needs with the Design Requirement (DR). QFD approach is also recognized as a better multiattributes decisionmaking method because its capability to accommodate group decision making (Chakraborty and Prasad, 2016). House of Quality (HOQ) is the primary planning tool in Quality Function Deployment (QFD). House of Quality (HOQ) maps things that need to be done in the planning process. The factors that need to be done are called Customer Requirements and how we can achieve CR are called Design Requirements. In HOQ, there’s a part called body; this section connects the left side of the HOQ which are the Customer Requirements and the top part of the HOQ which are Design Requirements. The Roof of HOQ shows the internal interdependence relation of DRs while the left roof shows the internal interdependence relation of CRs. The CRs and DRs in this paper are factors that support the implementation of SCM in wood industry SME.
2.4 Interpretive Structural Modeling Method (ISM)
ISM is a process to effortlessly constructs knowledge structure and models the relation of interdependence between every element to enhance our understanding of a specific compound problem. In other words, ISM helps to identify the construction of a system and its components to analyze the problem from every aspects and point of view (Alawamleh and Popplewell, 2011). ISM helps to identify relations between variables to determine the complex relationships within a system and analyze the effect of each variable when compared to others (Dachyar et al., 2014;Gavareshk et al., 2017). ISM is used to identify the internal interdependence between CRs and DRs that support the implementation of SCM in wood industry SME.
2.5 Analytic Network Process Method (ANP)
ANP method is a framework used to solve decisionmaking problems without assuming the interdependence and interrelation of elements within a different level or the same level of hierarchy (Saaty, 2004). ANP is used because many decisionmaking problems cannot be constructed in hierarchymodel, ANP is the generalization of AHP that consider the interdependence of elements in the hierarchy (Saaty, 2006). To acquire the priority of each alternative in the decisionmaking model, we use pairwise comparison. Pairwise comparison matrices are constructed by comparing a pair of elements in regards to a specific component. ANP is used to generate the relative priority weight of DRs that support the implementation of SCM in wood industry SME based on the internal interdependence between CRs and DRs obtained from ISM.
2.6 ZeroOne Goal Programming Method (ZOGP)
Goal Programming model does not optimize the objective directly. Instead, it tries to minimize the deviation of the desired goals and the real achievement. The goals should be prioritized in the hierarchical priority. Deviation variables can be either positive or negative (Kim and Emery, 2000). ZOGP can be used in a model that has multiple objectives when we want to minimize the deviation of goals achieved. The results of ZOGP will shows which alternatives that we should take and which we should not (Wei and Chang, 2008). ZOGP generates the decisions of which a DR should be implemented or not, based on the relative priority weight of DRs obtained from ANP and other constraints concerned.
2.7 Research Gap
There are many analytical techniques ccombined with QFD to calculate the relative weight of Design Requirements, such as AHP, fuzzy set theory, linear programming, ANP in previous study. This research utilize ANP to consider the interrelation of Customer Requirements and Design Requirements in supply chain management model of SMEs’ wood industry. Effective design needs to consider the corporate goals, as such ZOGP are included in this reseach. ZOGP considers SMEs’ constraints in budgets, human resources, and implementation time.
3. METHODOLOGY
Data Collection started by calculating the significance of CRs and DRs for the implementation of SCM in wood industry SME by experts. The CRs and DRs that were previously acquired by literature review that has significance above a certain threshold will be used in this research. ISM was used to identify and illustrate the relation between CRs and DRs. The result of ISM was the reachability matrix, which recognizes whether a specific CR affects the emergence of other CR and a specific DR affect the emergence of other DRs, that will be used in ANP. ANP used the CRs and DRs relation matrices (reachability matrix) to create the ANP relationmodel. The ANP relationmodel will be used to do pairwisecomparisons between CRs and DRs. The results of these pairwisecomparisons were priority vectors that represent each CRs and DRs that will be one of the corporate goals in ZOGP model. Figure 1 shows the methodology used in this research.
3.1 Selection of CRs and DRs
Literature reviews obtained CRs and DRs that can help to implement the SCM for SMEs. Based on literature review we acquired 9 CRs and 54 DRs. The selection of CRs and DRs that will be used further in this research would be made by nine wood industry SME’ experts. The experts selected the CRs and DRs by filling in the questionnaire. The questionnaire contained Likert 5scale measurement, where the value of 1 indicates “Very unimportant,” and the value of 5 indicates “Very important.” The questionnaire was created to find the importance of CRs and DRs that affect the implementation of SCM in wood industry SME. The questionnaire was filled in by all nine experts.
The result of the questionnaire was calculated to acquire geomean values for each CRs and DRs. Using a threshold of 3.5, every CRs and DRs that has geomean values less than 3.5 will be considered unimportant and insignificance. The final results obtained 9 CRs and 27 DRs that have geomean values higher than 3.5 which thus will be further processed in this research. The CRs and DRs are the shown in Table 1.
3.2 InterDependence of CRs and DRs
This step is a part of ISM method. Experts identified the relation of 9 CRs and 27 DRs by constructing VAXO matrix. There are four symbols used in this matrix; the symbols used are the representation of the relation between CRs and DRs

V: Represents that elements of E_{i} affect the emergence of elements of E_{j}, doesn’t work both ways

A: Represents that elements of E_{j} affect the emergence of elements of E_{i}, doesn’t work both ways

X: Represents that elements E_{i} and E_{j} affect each other

O: Represents that E_{i} dan E_{j} does not affect each other
Nine wood industry SME experts were asked to fill the VAXO matrices and the matrices obtained are shown in Table 2 and Table 3.
The second step is to construct Reachability Matrix (RM) table. RM table was constructed by translating the SSIM symbols to the binary matrix (0, 1), the rules are as follow:

If the relation of E_{i} to E_{j} = V in VAXO, then element E_{ij} = 1 and E_{j}i = 0 in RM

If the relation of E_{i} to E_{j} = A in VAXO, then element E_{ij} = 0 and E_{j}i = 1 in RM

If the relation of E_{i} to E_{j} = X in VAXO, then element E_{ij} = 1 and E_{j}i = 1 in RM

If the relation of E_{i} to E_{j} = O in VAXO, then element E_{ij} = 0 and E_{j}i = 0 in RM
The result of the transformation from VAXO Table to RM Table for CRs and DRs are shown in Table 4 and Table 5.
3.3 Relative Priority of DR
The ANP model is created based on the RM table obtained. The model consists of CRs and DRs, the arrows that link one CR other CRS and one DR to other DRs represent the interrelation between those CRs obtained in Table 4 and DRs obtained in Table 5. The ANP model is shown in Figure 2.
Nine wood industry SME experts were asked to fill in the questionnaire to do a pairwisecomparison. The questionnaire contained Likert 9scale measurement. The scale represents the weight of importance of one CR to other CRs. For example, we have two elements, a and b. In the questionnaire we filled in 2, this means that element a is two times more important than element b. If the questionnaire was filled in with the value of ½, then element b is two times more important than element a. All inconsistency ratio of each CR and DR are less than 0.1, meaning all experts inputs are validated and acceptable as shown in Table 6. Table 7 shows the unweighted supermatrix, it was obtained by combaining all decision priority matrix of every CRs and DRs to one single Supermatrix. The weighted supermatrix in Table 8 is obtained by normalizing the Unweighted Supermatrix. The final result of the ANP is the normal values of each DRs. The higher their value the higher their priority to be implemented in SCM for wood industry SME. The final weight of each DR in Table 9 is obtained by normalizing the Weighted Supermatrix.
3.4 Construction of ZOGP Model
We use four constraints, which are the relative prioritie of ANP that have inconsistency ratio less than 0.1, the budget, the number of employees, and the implementation time needed to implement every DR
Where w_{anp} is the priority weight of ANP in the objective function, w_{B} is the priority weight of budget in the objective function, w_{E} is the priority weight of the number of employees in the objective function, w_{T} is the priority weight of the number of employees in the objective function, ${w}_{{j}^{ANP}}$ is the ANP decision priority of DRj, x_{j} is j^{th} DR, cj is the cost spend to implement DRj, B is the budget limitation, ej is the number of employees needed to implement DRj, E is the number of employees limitation. tj is the time spent to implement DRj, and T is the implementation time limitation.
To determine the resource needed to implement every DR, we created a questionnaire. This questionnaire contained how many or how much resources would be needed to implement every DR. The result of the questionnaire is shown in Table 10.
We created eight simulation scenarios to be analyzed; each simulation has different constraints or limitations, the simulation represent the budget, the number of employees and the implementation time that SME may have as their constraints. The constraints are depend on SME scenarios are as follow :

Scenario 1: budget 75 million rupiahs, 35 employees, dan 317 implementation time

Scenario 2: budget constraint three times from the scenario 1

Scenario 3: number of employees constraint three times from the scenario 1

Scenario 4: implementation time three times from the scenario 1

Scenario 5: combination of scenario 1 and 2

Scenario 6: combination of scenario 2 and 4

Scenario 7: combination of scenario 3 and 4

Scenario 8: all constraints three times from the scenario 1
Nine Wood SMEs experts decided that all deviations in the objective function are wiegthed equally thus w_{anp} = w_{B} = w_{E} = w_{T} =1
Based on the data obtained, we showed mathematical model of scenario 1, as follows:
Priority 1. Vector priority of ANP, with the sum of 1.
0,039X_{1.1} + 0,036_{1.2}+0,022X_{2.1}+0,029X_{2.2}+ 0,039X_{2.3} + 0,027X_{2.4} + 0,024X_{2.5} + 0,081X_{3.1} + 0,064X_{4.1} + 0,069X_{5.1} + 0,035X_{6.1} + 0,037X_{6.2}+ 0,035X_{6.3} + 0,023X_{7.1}+0,049X_{7.2} + 0,037X_{7.3}+ 0,048X_{7.4}+ 0,035X_{7.5}+ 0,034X_{7.6}+ 0,036X_{7.7} + 0,020X_{8.1}+ 0,025X_{8.2}+ 0,015X_{8.3}+ 0,019X_{8.4}+ 0,027X_{8.5} + 0,052X_{9.1} +0,041X_{9.2} + ${\text{d}}_{{\text{1}}^{\text{}}}\hspace{0.33em}\text{}\hspace{0.33em}{\text{d}}_{{\text{1}}^{\text{+}}}$ = 1
Priority 2. Minimize the cost to 75 (in million rupiah)
100X_{1.1} +130X_{1.2} +25X_{2.1} +25X_{2.2} +25X_{2.3} +25X_{2.4} +0,3X_{2.5}+1,5X_{3.1}+20X_{4.1}+0,455X_{5.1}+130X_{6.2}+35X_{6.3} +60X_{7.1}+4,3X_{7.2}+140X_{7.3}+130X_{7.4} +10X_{7.6}+130X_{7.7} +25X_{8.1} +5X_{8.4} +X_{8.5} +1,68X_{9.1} + ${\text{d}}_{{\text{2}}^{\text{}}}\hspace{0.33em}\text{}\hspace{0.33em}{\text{d}}_{{\text{2}}^{\text{+}}}$ = 75
Priority 3. Minimize the number of employees to 35
15X_{1.1}+15X_{1.2}+8X_{2.1}+8X_{2.2}+8X_{2.3}+8X_{2.4}+8X_{2.5}+2X_{3.1} +8X_{4.1} +X_{5.1} +2X_{6.2} +15X_{6.3} +3X_{7.1} +5X_{7.2} +15X_{7.3} +15X_{7.4} +15X_{7.7} +15X_{8.4} +X_{8.5} +2X_{9.1} + ${\text{d}}_{{\text{3}}^{\text{}}}\hspace{0.33em}\text{}\hspace{0.33em}{\text{d}}_{{\text{3}}^{\text{+}}}$ = 35
Priority 4. Minimize the implementation time to 317
24X_{1.1}+24X_{1.2}+14X_{2.1}+14X_{2.2}+14X_{2.3}+14X_{2.4} +14X_{2.5}+7X_{3.1}+24X_{4.1}+7X_{5.1}+7X_{6.2}+24X_{6.3}+24X_{7.1} +14X_{7.2} +24X_{7.3} +24X_{7.4} +24X_{7.7} +2X_{8.4} +0,037X_{8.5} + 24X_{9.1} + ${\text{d}}_{{\text{4}}^{\text{}}}\hspace{0.33em}\text{}\hspace{0.33em}{\text{d}}_{{\text{4}}^{\text{+}}}$ = 317
Binary model
X_{1.1}, X_{1.2}, X_{2.1}, X_{2.2}, X_{2.3}, X_{2.4}, X_{2.5}, X_{3.1}, X_{4.1}, X_{5.1}, X_{6.1}, X_{6.2}, X_{6.3}, X_{7.1}, X_{7.2}, X_{7.3}, X_{7.4}, X_{7.5}, X_{7.6}, X_{7.7}, X_{8.1}, X_{8.2}, X_{8.2}, X_{8.3},X_{8.4}, X_{8.5}, X_{9.1}, X_{9.2} = 0, 1
Nonnegativity
X_{1.1}, X_{1.2}, X_{2.1}, X_{2.2}, X_{2.3}, X_{2.4}, X_{2.5}, X_{3.1}, X_{4.1}, X_{5.1}, X_{6.1}, X_{6.2}, X_{6.3}, X_{7.1}, X_{7.2}, X_{7.3}, X_{7.4}, X_{7.5}, X_{7.6}, X_{7.7}, X_{8.1}, X_{8.2}, X_{8.2}, X_{8.3}, X_{8.4}, X_{8.5}, X_{9.1}, X_{9.2}, ${\text{d}}_{{\text{1}}^{\text{}}},\hspace{0.33em}{\text{d}}_{{\text{1}}^{\text{+}}},{\text{d}}_{{2}^{}},{\text{d}}_{{2}^{\text{+}}},{\text{d}}_{{3}^{}}{\text{d}}_{{3}^{\text{+}}},{\text{d}}_{{4}^{}},{\text{d}}_{{4}^{\text{+}}}>0$
Objective function
The results of scenario 1 ZOGP is shown in Table 11
Value of x_{j} = 1 indicates that certain DR is implemented and value of x_{j} = 0 indicates that certain DR is not implemented in that particular scenario.
4. RESULTS
4.1 The results of DR selection based on ZOGP Model
The results of ZOGP model from all scenarios are presented in Table 12.
It can be seen that different scenario resulted in the different decision, but 13 DRs can be implemented, in any scenario. Those DRs are DR2.3, DR3.1, DR4.1, DR5.1, DR6.1, DR7.2, DR7.5, DR7.6, DR8.2, DR8.3, DR8.5, DR9.1, dan DR9.2.
4.2 Analysis of ZOGP Scenarios
The result objective function from each scenario can be seen at Table 13.
Based on the result obtained from scenario 1 through 8, it can be seen that implementation time is not necessary to be considered when implementing the SCM in SME because there’s no difference in the objective function. When comparing scenario 2 through 4, where only a specific constraint is subjected to change, it can bee is seen that budget gives the best objective function. When comparing scenario 5 through 7, where two constraints are subjected to change at the same time, it can be seen that the combination of budget and number of employees gives the best objective function. There’s no difference of objective function between scenario 5 and 8; this further proves that implementation time has doesn’t affect the selection of DR that need to be implemented in SCM.
5. CONCLUSION
This research found 9 CRs and 27 DRs, CR factors are Inventory Management; Suppliers Management; Demand Planning and Forecasting; Logistic and Distribution Planning; Product Flexibility; Design Quality; Capacity, Sales and Production Planning; Information Sharing; Human Resource Management
Having the capability to identify factors (promotion, discount, etc.) that affect demand and measure how each factor affects the future demand (DR3.1), being able to reduce the time needed for products to reach market (DR5.1) and having logistics capability to achieve competitive advantage in price by using efficient and effective logistics (DR4.1) are the three DRs with the highest decision priority
Thirteen DRs can be implemented in any scenario based on the results of ZOGP. Those DRs are DR2.3, DR3.1, DR4.1, DR5.1, DR6.1, DR7.2, DR7.5, DR7.6, DR8.2, DR8.3, DR8.5, DR9.1, dan DR9.2.
Budget and Number of employees constraints, need to be considered when selecting which DRs to implement because we can see from scenario 5, that these two constraints can result in the best objective function.