• 沒有找到結果。

2.1 Logistics on Global Trade

The definition of Logistics is the activity that manages the flows of goods, cash, information between the point of supply and the point of demand. It includes activities like transportation, warehousing, packaging, and material handling, etc. (Gunner, et al. 2012). In this global world, logistics is becoming more important aspect for companies as well as countries. As Burmaoglu and Sesen (2011) suggested that firm level logistics activities has been affected by national and global environment.

Arvis, et al. (2012) proposed that the most important elements of national competitiveness is the national logistics. The quality of logistics services and infrastructure definitely has impact on the transportation of goods among countries. The definition of efficient delivery in term of logistics services is the ability to move goods immediately, reliably and at low cost mentioned by Hollweg and Wong (2009),while Korinek and Sourdin (2011) shows that inefficient logistics structure will cause additional costs for a company in terms of time and money. This kind of condition will affect the country and company competitiveness in a negative way.

Some empirical studies have revealed the positive impact of logistics performance on international trade flows. Nordas and Piermartini (2004) found that quality of infrastructure has a significant relationship with trade flows and among of all infrastructure indicators shows that port efficiency has the largest relationship with trade flow. Hausman et al. (2005) have demonstrated that the significant relationship between transportation costs and international trade flow indicates that weak logistics performance causes a decrease in trade volumes. Also Limao and Venables (2001) proved the significant relationship between transportation costs, the quality of transportation infrastructure and countries trade volumes.

2.2 Logistics Performance Index (LPI)

Logistics Performance Index (LPI) is an index developed by The World Bank. It is a comprehensive index which is created to help countries identify the challenges and opportunities they face in trade logistics performance. Arvis, et al. (2012), this is a survey based index that is assessed by the logistics expert and conducted every two years. The first LPI survey was conducted in 2007, updated in 2010, 2012 and the latest edition is in 2014.

The LPI survey consists of two major parts offering two different prospective. First is the international LPI that provides qualitative evaluations of a country by its trading partners-logistics professionals working outside of the country and then the domestic LPI which

provides both qualitative and quantitative assessments in the country by logistics professionals works inside the country including detailed information on the logistics environment, core logistics processes, institutions, performance time and cost data.

The LPI can also show countries logistics bottleneck based on its component and the country can identify where their weak aspect from all of the LPI area. Below is the six areas of LPI assessments of a country’s logistics performance:

• Customs: Efficiency of the customs clearance process.

• Infrastructure: Quality of trade and transport-related infrastructure.

• International Shipments: Ease of arranging competitively priced shipments.

• Logistics Competence: Competence and quality of logistics services.

• Tracking & Tracing: Ability to track and trace consignments.

• Timeliness: Frequency with which shipments reach consignee within the scheduled or expected time.

The LPI is constructed from these six indicators using principal component analysis (PCA), a standard statistical technique used to reduce the dimensionality of a dataset. The result is a single indicator the LPI that is a weighted average of those scores. The weights are chosen to maximize the percentage of variation in the LPI’s original six indicators.

According to Arvis, et al. (2012) on survey respondent data collection as described in the LPI report on how to measure the international LPI. They described in the LPI report that to measure the international LPI, each survey respondent rates eight oversea markets on six core components of logistics performance. The eight countries are chosen based on the most important export, import markets of the country where the respondent is located, on random selection, and for landlocked countries on neighboring countries that form part of the land bridge connecting them with international markets. The method used to select the group of countries rated by each respondent varies by the characteristics of the country where the respondent is located.

The 2009 respondent demographics of the LPI survey constitutes of one thousand professionals from various international companies in 130 countries who participated in the survey. 69 out of the 130 countries participated in the 2009 LPI survey in Canada.

2.3 Seemingly Unrelated Regression

Regression analysis is a statistical analysis that plays a major role in helping to model many economic phenomenon in the form of mathematical equations: y = Xβ + ε. This model is a linear form. When the variables in X values is determined earlier (pre-determined) as

variables that can be measured or assessed. β is an unknown parameter and its value will predicted, and ε which represents the error.

Based on these assumptions, the β parameter estimation methods are often used is the least squares method. This method can only be used to estimate the model parameters in a single regression equation. However, in certain cases there is a model that consists of multiple regression equations or systems that are simultaneously, thus forming a system of equations.

In this model, the equations were related to each other where there is a relationship between the dependent variable, resulting in a correlation between the errors between equations. The system of the equation is called the Seemingly Unrelated regressions (SUR).

Even though based on the literature review there has been no research that apply SUR in logistic performance measurement area. This approach is already used in some research to analyze multivariate case. Budiwinarto (2013) applies SUR to linear demand system model to analyze the food demand model in Indonesia where the case is the food demand of a household related to others. Again Wilde, et al. (1999) apply SUR to identify the effect of income and food programs on dietary quality. Both cases are multivariate problems where the model is consist of more than one equation.

2.3.1 SUR Basic Equation

The SUR procedure was originally developed by Zellner (1962). SUR equation system contains a set of interrelated equations. The relationship between equations can be seen from the error correlations between equations (contemporaneous correlation). Basic SUR model is defined as:

Since it is a multivariate approach then the equation (1) could be expand into a matrix as equation (2). Where Y1,Y2,…,Yj is represent the dependent variables where in this study is the LPI input indicators, and X1, X2, …, Xm is the set of predictor for each equation where in this study is the social, economic, and infrastructure factors, while β1, β2,…, βm are the coefficient for each predictors and ε1, ε2,…, εm is the error for each equations.

For SUR model, the assumptions are the errors are uncorrelated and that the error for any individual model have constant variance but that the errors in different models are

correlated. Where it can be describe mathematically by these equation below:

Where I is the identity matrix m x m. Value above shows the covariance of two equations in the system with the m equation. In general:σ11

Ω = E(εi ε’j ) = �

All information about the error covariance is in Ω matrix. The most efficient estimator from Equation (2) is Generalized Least-Square estimation, (GLS) Pindyck and Rubinfeld (1991), Greene (1991):

β= (XΩ−1 X)−1 (XΩ−1 Y) (5) Because Ω elements is unknown, then the element must be alleged. This estimation conducted by using any residual from each equations obtained and applying the least squares method,

σ′ii = sii = ei .ei

𝑁𝑁−𝐾𝐾𝐾𝐾

σ′ij = sij = ei .ei

�(𝑁𝑁−𝐾𝐾𝐾𝐾)(𝑁𝑁−𝐾𝐾𝐾𝐾)

ei = Yi − Xiβi

In the SUR equation, it can be seen that between one equation and other are interrelated. This is indicated by the correlation between the errors of each equations.

2.3.2 The advantage of SUR

As already proposed in previous chapter that in this research the measurement of countries logistics performance is using six main criteria from LPI component. Thus the prediction model is a multivariate model with six dependent variable which related one and another. According to this specification the SUR model is the most suitable approach to model this logistics performance measurement. Compared to other approaches like multiple regression or multivariate regression, SUR approach is more advance because it has different sets of predictors that can be used as independent variables for every dependent variables.

Unlike the multivariate regression where the model use same set of predictors to explain dependent variables. This characteristics makes the approach suitable to be applied in the study, since the framework shows that the problem is a multivariate where the model is

consist of more than one dependent variables that related to one and another. Moreover each dependent variables are unique so it is necessary to use different sets of predictors to model this logistics performance measurement problem.

相關文件