INDUSTRIAL ESTATE, FIRMS’ PRODUCTIVITY, AND INTERNATIONAL TRADE RELATIONSHIP: THE CASE OF INDONESIAN MANUFACTURING FIRMS Hubungan Kawasan Industri, Produktivitas Perusahaan, dan Perdagangan Internasional: Studi Kasus Perusahaan Manufaktur Indonesia

Many believe that the industrial estate could encourage the industrial sector in developing countries due to its better infrastructure, access to supporting industries, and the market as well as technology and information spillover. These factors could lead to a higher productivity level and export activities of manufacturing firms inside the industrial estate. Some previous studies still provide a mixed result regarding the relationship between these three variables. Thus, this paper contributes to the related study by examining the relationship between an industrial estate and both productivity level and export activity in the case of Indonesian Manufacturing Firms. The paper introduces the practice of entropy balancing, one of matching methods along with firm-level data as a unit of analysis. A significant difference in the number of observations between firms inside and outside the industrial estate motivates the usage of matching methods technique, so the data become balanced. The treatment is when the firms being in the industrial estate. There are two outcomes variables, which are productivity level and export activity. The result found that being industrial estate improves firms’ productivity, yet it fails to promote export activity.

A successful industrial estate initially should be a tool for promoting regional industrial development (Lee et al., 2017). Because of that, it should answer the regional business problems.

Outcome Variable
The concept of productivity is related to firms' utilization of resources.
Generally, productivity is defined as a ratio output to input. Hence, productivity The main problem of estimating production function is an unobservable factor that leads to productivity shock. Levinsohn & Petrin (2003)  Futhermore, Petrin et al. (2004) added that the production function covers the relationship between productivity shock and its input. If there is a positive productivity shock, firms will expand the output, and requiring more input. On the other hand, firms will decrease employing the input once there is a negative shock. Using input as a proxy will overcome the truncating of firms' zero investment.
Following Levinsohn & Petrin (2003) and Beveren (2012), now the raw material variable is not exogeneous, but as a function of capital and productivity. The function is changed into: Using, Levinsohn & Petrin (2003) approach, the as the logarithm of a  (2003) approach, the paper is also using physical productivity calculation. The physical productivity ( ) is a productivity measurement in terms of real output.
The calculation follows Ryzhenkov (2016), which stated that physical productivity as a ratio of actual production to its input usage.
The paper also intends to capture the relationship between an industrial estate and firms' export activity. The proportion of export to total output indicates the export activity of the firms.
All the outcome variables are in the real value, which is already weighted by using the wholesale price index for each sector and consumer price index.

Many variables influence firms'
decision to join the industrial estate as well as the outcome variables: productivity and export activity. The paper applies tariff, firms' characteristics, and location as the covariates that might correlate both on treatment and outcome variables.
Many studies that examine the relationship between tariffs and productivity as well as export. Amiti & Konings (2007) are the first to use both input and output tariffs to firms' productivity levels. Input tariffs are tariffs that are enacted to imported raw material, while output tariff is an import tariff on final goods. Reducing input tariffs will lower input purchasing from abroad, so the cost of production will be lower. The output tariff's impact on the firm is related to the competition with import of the final goods in the market.
Reducing the output tariff could lead a stricter competition with final import goods so that it could drive higher productivity (Amiti & Konings, 2007).
Tariff reduction also happens to encourage export. Fan (2015)  The measurement of input and output tariff follows Amiti & Konings (2007). The output tariff will be calculated based on its HS 6-digit code from the concordance list of the Indonesian Industrial Code 2009. As a result, the tariff will only differ across the industry (k). Output tariff will be weighted average goods that are being produced by firm i. Since the code of specific goods data is challenging to find, so it is assumed that produced goods are the same as industry k.
Hence industry k can be seen as good that the firm produces.
Nevertheless, because of nonexistence firms' input level data, the paper utilizes a weighted average tariff instead that is varied across the industry (k) for using the input industry Jakarta is a location indicator. It is a binary variable, one if a firm is in Jakarta city and 0 otherwise.

Entropy Balancing
The condition of the treatment and control group sometimes not equal because it is not randomly assigned.
Thus, looking up the causal relationship between the treatment and the outcome might be difficult. The outcome might correlate with some covariates that relate to the outcome variables.
The paper utilizes one of the matching methods approach, entropy balancing, to overcome the covariate imbalanced.
Abadie & Cattaneo (2018)  The weight's usage is to make the covariates between the treatment and control variable as close as possible.
Thus, the target is to minimize the entropy distance among the groups.
With the constrains: The means a set of R

Descriptive Analysis
The difference between the treatment group and the control group is quite noticeable based on their average value of outcome variables ( Table 2) Usually, firms inside the zone have a high dependency on importing and exporting goods. Therefore, an authority sometimes gives a free import tariff to support their business activity (Moberg, 2018). For the firms' characteristics, such as ratio capital to labor, the number of workers, and the average wage, the values of the treatment group are also bigger than those of the control group. Due to incoming investment, the ratio capital to labor is higher in the zone through technology spillover (Lu et al., 2017).
The firms inside the industrial zones also tend to pay higher wages since they applied a good regulation of employment, including working hours, safety and health, and overtime fee (Cirera & Lakshman, 2017;Thanh et al, 2018). Foreign firms are also more likely to choose industrial estate as a production base. One of the reasons is to link between global connectedness and local innovation performance (Turkina & Assche, 2018). While Puig et al. (2019) assumed that the MNEs' foreign investment usually pick a location with strategic characteristics.     Indonesia is a big size country with a huge population. Undoubtedly, the consumption portion was high, even reach around 55% of total GDP in 2018.
Furthermore, due to its growing demand, the domestic market is still potential. Nazarczuk & Uminski (2018a) stated that the role of an industrial estate on international trade seems higher on import activity than export.
The higher number of foreign companies is usually linked to the global value chain, which allows them to acquire raw materials from abroad.