DETERMINANTS OF INDONESIA’S EXPORTS OF MANUFACTURED PRODUCTS: A PANEL DATA ANALYSIS Faktor Penentu Ekspor Produk Manufaktur Indonesia: Analisis Data Panel

Indonesia’s export has been decreasing since 2012. This problem has raised government’s attention to increase the export performance. One sector that can be improved is manufacturing. This study analyzes the determinants of Indonesia’s manufacturing export from 2005 to 2014. The major factors examined in this study include real exchange rate, foreign direct investment (FDI), gross domestic product (GDP) and trade policies. Those factors are examined by using panel data regression with a random effect model. The results revealed that relative change of exchange rate, real GDP, distance between two countries and average tariffs significantly affected the Indonesia’s manufacturing export. It is recommended that Indonesian government maintains the exports to countries which have high GDP, expand the export market, stabilize Rupiahs exchange rate, encourage local industries to use advanced technologies, and facilitate the simplification of import procedures.


INTRODUCTION
Indonesia's export value has been shrinking since 2012.This is shocking because it was experienced after Indonesia made history by doubling its export value over a period of five years in 2011.This fall was mainly due to the financial crisis in 2011 caused export demand from Indonesia's trading partner countries to decline; consequently Indonesia's export value fell to USD 190 billion (Sukarno, 2012).Another reason for the decline in export value was the drop in export mining commodities' prices, as shown by data of Statistics Indonesia (Syafputri, 2013).The export value, then, dropped consistently in the following years to approximately USD 176 billion in 2014 (Figure 1) (Ministry of Trade, 2015).To address the decline in export trade, Indonesia's government, through the Ministry of Trade, established a target which was to increase Indonesia's export value by three folds in a period of five years starting from 2014.However, the Indonesian government has also introduced a policy prohibiting export of raw materials in order to guarantee natural resource sustainability and develop local industries (Gunawan, 2014).Therefore, in order to meet the target, the government aims to focus on increasing export performance in the manufacturing sector rather than resource-based export.
In the case of manufacturing export performance in Indonesia, it gradually increased from 2004 to 2011 except for a small drop in 2009 (Figure 2).This In contrast, total export value suffered decline after the 2011 economic crisis, but the manufacturing sector seemed to be robust enough to stand up with this economic shock (Soderbom & Teal, 2003).
Moreover, from 2004 to 2014, the manufacturing sector contributed supply and demand sides to avoid bias which commonly occurs when estimating export performance of developing countries based on only one side and disregard another side (Riedel, 1998).Hence, manufacturing export determinants examined in this study come from supply and demand side.
Based on the availability of data, supply factors examined consist of foreign direct investment (FDI) and other factors between 35 and 50 percent of Indonesia's total export.The highest percentage was in 2004 which was just over 50%.
In 2014, the manufacturing sector contributed to the aggregate export by about 40%.However, the former Minister of Trade, Rachmat Gobel wanted to increase this contribution to 65% to fulfill the international demand of manufacturing products (Pusat Hubungan Masyarakat, 2015).
This study aims to find determinants of manufacturing export performance in Indonesia. Deliarrnov (1995) stated that countries do export if they have an excess of domestic supply of goods and services.
On the other hand, Goldstein & Khan (1985)

Method of Analysis
Manufacturing (1) 1 The equation excludes a factor of Indonesia's GDP, because this study only focuses on Indonesia.The use of time dummies takes up the effect of Indonesia's GDP changes.
Where is trade value from country j to k, and are nominal GDP of country j and k, is distance between country j and k, and represents other factors that may affect trade between country j and k.
In this study, specification by Sheldon, Mishra, & Thompson (2013) with is modified by disaggregating variable to some more variables which will be explained later.(3) test is required.One of test that can be considered is the Hausman's test (Gujarati, 2003).Hausman (1978), as cited in Baltagi (2008), stated that the null hypothesis .
Hausman argued that is consistent without considering whether is true or not.However, is only consistent and asymptotically efficient under .Gujarati (2003) concluded that if the null hypothesis is rejected, the REM approach is not appropriate; consequently, it is better to use the FEM approach.
This study utilises yearly panel data, dating from 2005 to 2014, which consists of 28 countries resulting in a total of 280 observations.Countries, chosen in this study, are the top 28 importing countries in 2014, which contribute approximately 90 percent of the total of Indonesia's manufacturing exports (Table 1). (4) FEM, , this model does not treat as fixed, but assumes that is a random variable having mean value of , hence the intercept value for each individual is: Where is a random error term with a zero mean value and variance of .
By combining equation ( 4) and ( 5), the equation becomes (Gujarati, 2003): The composite error term consists of error from the cross-section ( ) and the combination between time series and cross-section error component ( ).
To choose the more appropriate approach between FEM and REM, a

Finding the best approach
According to Gujarati (2003) Consequently, the more appropriate approach for this study is REM.Gross domestic product (GDP) is the total production and expenditure of goods and services in a country (Mankiw, 2010).According to

Figure 1 .
Figure 1.Indonesia's Total Export Value and its Change From 2004 to 2014 Source: Ministry of Trade (2015) trend was almost similar to trends in Indonesia's total export.In 2004, the manufacturing export value was about USD 36 billion, then it rose consistently to slightly above USD 50 billion in 2008.Although it dropped to about USD Non 2009, it recovered in the following year and remained stable at around USD 70 billion from 2011 to 2014.
However, this finding only covered some particulars industries which cannot explain Indonesia's manufactures as a whole.Therefore, it is needed to study further what factors significantly affect the whole Indonesia's manufacturing export performance, hence it can provide recommendations to improve Indonesia's export value.This paper consists of four sections commencing with brief background of this study.The second section briefly describes the methodology and data for this study, followed by a discussion of the results in section three.The final section draws conclusions and offer some recommendations for the future (for MoT leaders in making policies to increase Indonesian manufacturing export performance).
to capture particular effects in each year affecting Indonesia's manufacturing export value.This gravity model, then, is analyzed by panel data regression which has two approaches, Fixed Effect Model (FEM) and Random Effect Model (REM).The fixed effect model (FEM) considers the individuality of each crosssection unit and let the intercepts differ for each individual ( ).Yet, this model still assumes that the slope coefficients are constant across individuals.Gujarati (2003) formulated the model as: The intercept has subscript to show that the intercepts of individuals may be different because of particular characteristics of each individual.However, it has no subscript t to suggest that each individual is time invariant.Although the random effect model (REM) has the same basic model as (2) Data of nominal FDI inflows to Indonesia are acquired from UNCTAD.According to UNCTAD (2013), these data are on a net basis, therefore, their value each year might be positive representing investment or negative representing disinvestment.There is a problem when these data are transformed to logarithm value.Hence, to address this problem, these data are divided into two variables: investment, consisting of positive value of FDI inflows, and disinvestment, consisting of the absolute value of the negative value of FDI inflows.According to Cavallari & d'Addona (2013), nominal FDI data are scaled by the GDP deflator of each partner country from the World Bank's World Development Indicator (2015) to obtain real values.Data of distances between each country and Indonesia are from CEPII (2015) and the average tariff data from the World Bank's WITS (2015).

Manufacturing Export Value and Total Export Value in Indonesia from 2004 to 2014
Source: Ministry of Trade (2015) Figure 2.(FDI), GDP growth and exchange rate.

Table 3 . Impact of Each Determinant on the Indonesia's Manufacturing Export (Random Effects)
* and ** are statistically significant at level of five and one per cent, respectively.