![Sample our Global Development journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5929/TOC-Subject-Sample-2021-Global-Development.jpg"})
Abstract
In this paper, I make use of Vietnam’s enterprise surveys to estimate markups and product quality and investigate the effects of tariff preferences under the ASEAN and ASEAN+1 free trade areas (FTAs) on quality upgrading. The empirical results consistently show that Vietnam’s manufacturing firms have moved toward high-quality products and speeded up the rate of quality upgrading in response to the increasing competition from regional imports. The quality responses by domestic firms are much stronger in the competitive industries with a low degree of concentration, but are insignificant in highly concentrated industries. This indicates the importance of the domestic competition condition in transmitting the effects of tariff reductions. The empirical results and findings remain robust when the margins of tariff preferences or imports are used to proxy for import competition.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 The period between 2011 and 2015 is selected for two reasons. Firstly, all the ASEAN+1 FTAs took effect during this period with substantial tariff reductions taking place in Vietnam. Secondly, the enterprise surveys in the previous period used a 7-digit product classification, which cannot be merged with the 8-digit product classification employed in the period between 2011 and 2015. In recent years, Vietnam expanded its FTA network and signed free trade agreements with partners outside East Asia, including the Comprehensive and Progressive Agreement for Trans-pacific Partnership (CPTPP) in 2018 and EU-Vietnam Free Trade Area (EVFTA) in 2019. Because these FTAs only took effects recently, they are not taken into analysis.
2 Except for the censuses, which are conducted for every five years, only a proportion of small private firms are directly surveyed. These firms are selected within 4-digit industries according to different employment criterion, which vary by provinces and years. The threshold of 100 workers, which is used in this analysis, is the highest threshold used by the GSO during the period between 2011 and 2015.
3 Vietnam enterprise surveys have been used in several empirical studies on productivity and the effect of trade on productivity and employment, such as Ha and Kiyota (Citation2014), and Doan T.T. et al. (Citation2015).
4 Specifically I remove the observations with the deviation from the mean product prices below the 1st percentile or above 99th percentile. The observations with too small or too large physical capital and wage expenditures, and the observations with excessive growth in the physical capital are trimmed at the bottom and upper 1-percentile of the distribution. I also remove firm-year pairs, in which total product-level sales exceeds the revenues reported at the firm level.
5 Three industries related to printing, printing services, and reproduction of recorded media (i.e. codes 2221, 2222, and 2230 according to the ISIC Rev.3 classification) are also excluded. These industries have few observations and no records of imports. Other industries are not present because there are no observations.
6 The depreciation rate of 0.10 is close to the average depreciation rate calculated using the enterprise surveys. The real interest rate is estimated at the around 0.04 in the period on of study. I have tried different values of real interest rates and depreciation rates, and the empirical results remain robust to different imputed capital rents.
7 As in Loecker, et al. (Citation2016), I use the translog function of the following form:
8 I do not employ input tariffs in the production function estimation and the regression analysis. This is because the available Input-Output tables are quite aggregated and make it impossible to calculate input tariffs at the 4-digit industrial level. Furthermore, in Vietnam, tariffs and non-tariff barriers are largely employed to protect downstream industries. Capital goods and production materials are usually subject to zero or low tariffs and few non-tariff barriers.
9 Because the information on exports is not available at the product level, the export dummy is defined at the firm-level. The export dummy takes the value of unity for the firms that report positive export revenues and zero otherwise.
10 I use STATA software to estimate the production function. My STATA program adopts several features from the program provided in the online appendix in Loecker, et al. (Citation2016).
11 I use the fourth-order polynomial expansion in production inputs and third-order expansion in input price control functions using the following formula: ; where
is the order of the polynomial. Adding more terms and higher orders does not improve the R-squared.
12 I use CONOPT algorithm to solve the system of equations. More details about GAMS software can be found in the website: http://www.gams.com.
13 The Herfindahl index is the sum of squares of firms’ market shares. In our analysis, market shares are defined in decimal terms, so the Herfindahl index runs from 0 to 1. The concentration ratio is the sum of the market shares of four largest firms in each industry.
14 Amiti, and Khandelwal (Citation2013) show that the effects of tariffs on product quality are much lower in magnitude and significance in the countries with weak business environment than the countries with strong business environment.
15 I also conduct firm-level regressions, which relate tariff variables to firms’ performance indicators. Because the results of firm-level regressions are qualitatively similar to those obtained from industry-level regressions, they are not reported to save space.
16 Because the number of observations is relatively low in some industries in the sub-sample of single-product firms, I aggregate certain 2-digit industries into single ones. Specifically, processed foods (Vietnam’s industry code 10), beverages (11) and tobacco (12) are aggregated into processed foods, beverages and tobacco; Oil refinery (19) and chemicals and chemical products (20) are aggregated into Chemicals; Metals (24) and metal products (25) are aggregated into Metals; Electronics (26), electrical equipments (27), and other equipments and machinery (28) are aggregated into Machinery and Equipments; Automobile (29) and other transportation means (30) are aggregated into Transportation Means; Other Manufactures consist of pharmaceuticals (21), furniture (31), and other manufacturing activities (32).
17 Mayer,, Melitz, and Ottaviano (Citation2014) show that the number of products is a weakly increasing function of firms’ productivity.
18 I use Rauch, (Citation1999)’s liberal classification to identify differentiated imports and aggregate imports at the industrial level. The industries featuring differentiated imports are those with the record of differentiated imports.
19 Since the mid of 2000s, state-owned conglomerates have been established in key industries in Vietnam. These state conglomerates comprise of state general corporations and large state enterprises, and have nearly monopoly control in certain industries. According to Perkins and Tu Anh (Citation2009), the state conglomerates have been formed partly to provide support to their member enterprises to deal with foreign competition when tariffs and non-tariff barriers have been lowered or removed under international agreements.
20 In these regressions, Vietnam’s trading partners are grouped into seven aggregated partners, i.e. ASEAN countries, Japan, China, Korea, India, Australia and New Zealand, and the rest of the World. Imports from each of these partners are subject to the same schedule of tariff reductions. The zero margins of preferences are applied to the rest of the world.
21 I also use the elasticity of substitution estimated by Broda, and Weinstein (Citation2006), but the results are qualitatively and quantitatively similar and not reported here.