Observation of Japan's FTA Utilization Rate

Purpose of the Analysis

In an international trade, firms may benefit through utilizing Free Trade Agreement (FTA) by reducing their tariff burdens. However, there are also costs associated such as learning cost of rules of origin and certification procedures for each FTA, Certification costs to customs etc.. Therefore, if costs exceed benefits, there is a high possibility that FTA will not be utilized. The purpose of the project is to analyze Japan's FTA utilization rate on imports and explore its determinants through regression analysis. For the determinants, the tariff margin and the monthly imports are used.

Datasets

Japan's total import data

• Source: Japan Custom
• CIF import value
• national tariff line level (HS (Harmonized System) nine-digit level)
• annual data of 2012 to 2021
• HS2017 version

Japan's FTA utilized import data

• Source: Japan Custom
• CIF import value
• national tariff line level (HS nine-digit level)
• annual data of 2012 to 2021
• data for 2012 to 2016: HS2012 version, data for 2016 to 2021: HS2017 version

Japan's MFN tariff rate and FTA tariff rate

• Source: World Trade Organization
• national tariff line level (HS nine-digit level)
• HS2017 version

Note: As of May 2023, Japan has 20 FTAs with 50 countries in force. However, the latest FTA, RCEP (Regional Comprehensive EPA), which was concluded in January 2022 will not be included in the analysis since the annual import data is only available until 2021.

Regression Model

The specific regression model used in this project is as follows. FTA Utilization Rate is the dependant variable and Tariff Margin and Monthly Imports are the independant variables.

FTA Utilization Rate = αTariff Margin + βlnMonthly Imports

Definition of the variables:

• FTA Utilization Rate: the FTA utilization rates in imports from country i for product p in year t. Product is defined at national tariff line which is the most detailed data we can get from trade data. As explained before, FTA utilization rate is defined as the share of imports of products that are eligible under to receive such preferential tariff rates. 0% indicates FTA is not being utilized and 100% indicates FTA is at perfect utilization.

• Tariff Margin: the absolute difference between FTA tariff rate and MFN tariff rate on product p from country i in year t. As explained above, products with zero tariff margin, which are ineligible to receive preferential tariff rate, is not included. It shall be noted that simplification is made regarding to the calculation of tariff margin. Some countries like Australia, Brunei, Chile, Indonesia, Malaysia, Mexico, Peru, Philippines, Singapore, Thailand and Vietnam have sevral FTAs concluded with Japan. Since tariff margin may differ for each FTA, I defined the tariff margin as the average tariff margin of all FTAs. Although firms may intuitively chose to use the FTA with the largest tariff margin, there may be other factors that effect their decision. To this end, I decided to use the average instead of using the maximum tariff margin of all FTAs option.

• Monthly Imports: the average of monthly imports of products p from country i in year t. This variable is suppose to define as a firm-level transaction sizes. Due to data availability, this was the nearest estimation that can be made for firm-level transaction size.

Explanation of the Scripts

Running order of the script shall be:

1. '01_tariffmargin.py'
2. '02_epa_imports.py' or '03_total_imports.py'
3. '04_epa_utilization_rate.py' or '05_regression_analysis.py'

1. '01_tariffmargin.py'

This script aims to create a dataframe 'master_jp_tariff': dataframe of tariff margin for each tariff line.

The first section reads an excel file of Japan's MFN and FTA tariff rate and clean the data to only the necessary datas for calculating the tariff margin. Then the next section calculates the tariff margin for eachtariff line.

2. '02_epa_imports.py'

This script aims to create a dataframe 'master_jp_epa': dataframe of FTA utilized annual import data for each tariff line for 2012 through 2021.

The first section reads excel files of Japan's FTA import datas for 2012 through 2021. Since a column name and column values for files before 2020 and after 2021 is different, the reading process is done seperately. In addition, it cleans the data by (a) translate the country name from Japanese to English, (b) exclude GSP or LDC utlized import datas, and (c) exclude import datas where tariff margin is zero.

2. '03_total_imports.py'

This script aims to create a dataframe 'master_jp_im': dataframe of total annual import data for each tariff line for 2012 through 2021.

The first section read an excel files of Japan's total import datas for 2012 through 2021. In addition, it cleans the data by (a) convert the import value into thousands yen, (b) translate the country name from Japanese to English, (c) exclude non-FTA partner countries' import datas, and (d) exclude import datas where tariff margin is zero etc..

3. '04_epa_utilization_rate.py'

This script aims to calculate the FTA utilization rate from different perspectives by using dataframes 'master_jp_epa' and 'master_jp_im'. To note, this script is independent from the purpose of this project analysis. Rather, it is to observe the overview of the FTA utilization datas and get insights.

It calculates the FTA utilization rate by 3 different perspectives: (1) by country level and product at section level (* section level is defined by the WCO), (2) by country level, and (3) by product at section level. It also creates a pivot table by product section and country with the FTA utilization rate of year 2021.

3. '05_regression_analysis.py'

This script aims to conduct regression analysis by using 'master_jp_epa' and 'master_jp_im'.

First, it prepares the data for each variable in the regression model: FTA utilization rate, monthly import, and tariff margin. Then, conduct regression analysis by simple ordinary least squares (OLS) method using the variable datas.

Regression Results

Below is the result of the regression model estimated by the OLS method. The regression model had a good overall fit as indicated by the high R-squared value of 0.692.

Coefficient for both Tariff Margin and Monthly Imports shows positive relation with the FTA Utilization Rate. This result is reasonable by a common sense that as the tariff margin become larger, it implies the gain from applying FTA tariff rate will increase, which leads to greater incentive to use FTA for firms. Also, similarly, as the transaction size is bigger for the firm, incentive to use FTA will be greater.

OLS Regression Results for Japan's FTA Utilizarion Rate

Variable	Coefficient
Const	0.0260
ln_Monthly Import	0.0910
Tariff Margin	0.0048

Conclusion

The regression analysis suggests two policy options to improve FTA utilization rate: to increase tariff margin or to increase the size of the import. However, since import size depends on the decision of the importer (the firms), it is not realistic for the government to intervene to change firm's decision making. Likewise, tariffs are needed to protect the domestic industry, therfore it is not ideal to lower the FTA tariff rate to increase the tariff margin.

In hence, policies to promoting positive factors (tariff margin and import size) are limited. It is assumed that the negative factors are the main obstacle for utilizing FTA as seen from the past research papers.

Therefore, for the future research, negative factors such as restrictiveness of rules of origin and other various costs shall be included as independent variables in the regression model and explore policies to improve the FTA utilization rate.