Importing Inequality: Trade Liberalisation, Technology and Women’s Employment

Abstract

We investigate the impact of trade liberalisation on female workers in manufacturing using arguably exogenous variation in Indonesia’s tariff reductions in the 1990s and 2000s. This study utilises output and input tariff changes to examine two different channels through which trade liberalisation affects women’s employment: import competition and imported goods. Our results confirm the findings of previous literature that showed that increased competition driven by a reduction in output tariffs encourages women’s employment; however, we also provide new evidence that shows that a reduction in input tariffs may hurt women’s employment. We surmise that the latter is driven by importing firms’ shifting demands for skilled workers.

Keywords:

• trade liberalisation
• gender inequality
• import competition
• skill-biased technological change

JEL classifications:

• F14
• F66
• J16
• J24

FUNDING AND ACKNOWLEDGEMENTS

We thank the Japan Society for Promotion of Science for financial support under the Young Scientist (B) Grant (#16K17113). We thank Jong-Wha Lee, Stephan Litschig and two anonymous referees for their valuable comments. We also thank participants of the RIEM-IDDPRP Workshop of Southwestern University of Finance and Economics, the Osaka Workshop on Economics of Institutions and Organization at the University of Tokyo and the Singapore Economic Review Conference.

DISCLOSURE STATEMENT

The authors have no competing interests to declare.

Notes

1 For fast-track products with tariffs higher than 20%, the tariff on the commodities would be immediately reduced to 20%. They promised that the upper bar would be further decreased to below 5% within ten years. For commodities with tariff rates of 20% or less, tariffs were mandated to be reduced to 0–5% within seven years.

2 Indonesia’s Annual Manufacturing Survey employs industry codes based on ISIC rev2 and rev3. However, these industry codes cannot be perfectly matched with ISIC codes. Therefore, approximately 23% of firms were dropped from our sample.

3 Moreover, we could not obtain the unpublished concordance product code and industry code employed by Amiti and Konings (2007).

4 Tariff measures are different in the base units calculated: product code in the case of Amiti and Konings (2007) and industry code in ours. The statistics presented in table 1 are aggregated at ISIC3 in our case and ISIC2 in Amiti and Konings (2007).

5 Two reasons can potentially explain the sharp decrease in the number of employees in the manufacturing sector. The first is that 1997 was a year of economic recession. The second is the sharp increase in the real value of the minimum wage (Rama 2001; Comola and Mello 2011).

6 We define exporters as firms that have any positive amount of exportation each year. The data provide two pieces of information: the percentage of outputs exported and the total number of exports. As the total number of exports has several missing values, we use the percentage of outputs exported to define firms as exporters.

7 The data provide information related to imported material. Therefore, we define importers as firms whose share of imported material is greater than 10% of the total materials reported. From the data, it is impossible to distinguish if firms are indirect importers (forwarders) or not.

8 Indirect import occurs when a company purchases products from foreign countries through a local intermediary. In this case, firms may report that they are not importers, although they utilise imported inputs for their production.

9 As the Annual Manufacturing Survey only provides each firm’s unique identification number in the data before 2000, we can only use their tariff measure to analyse the data from 1993 to 1999. For more analyses using Amiti and Konings’s (2007) tariff measure in 1993, 1995, 1997 and 1999, please refer to Oishi (2017).

10 In table 6, we do not control for log relative wages; nevertheless, the results are robust and similar.

11 It would be ideal to use the Herfindahl index calculated using earlier rounds as we utilise the data since 1993. We have the data for 1990; however, we have a relatively small number of observations, making it difficult to calculate the Herfindahl index within all four-digit industries.

12 The Annual Manufacturing Survey in 1995, 1996, 1997 and 2006 provides detailed employment information disaggregated by skill and gender. However, the information lacks accuracy compared to the aggregate formal employment figures we used in this study.

13 To construct the provincial-level minimum wage, we acquired pre-1997 data from the appendix of Rama (2001) and post-1997 data from the Ministry of Manpower in Indonesia.

14 NTBs include regulations such as import quota; requirements related to rules of origin; and various regulations on licensure, packaging and the way products are manufactured. We constructed an indicator if there were any NTBs within four-digit industries that the firm belonged to.

15 We downloaded NTB data reported by Indonesia in 2017. No NTB was reported until 2002. NTBs are applied depending on the country or origin. In terms of the HS6 product code, 4, 11, 13 and 7 products were reported in 2002, 2003, 2004 and 2005, respectively. From 2006 to 2009, 15, 13, 18 and 18 products, respectively, were reported to have NTBs.