Impact of trade liberalization on household welfare: an analysis using household exposure to trade indices

Impact of Trade Liberalization on Household Welfare An Analysis Using Household Exposure to Trade Indices Vol (0123456789) Social Indicators Research (2021) 153 503–531 https doi org10 1007s11205 020 02499 1 1 3 O R I G I N A L R E S E A R C H Impact of Trade Liberalization on Household Welfare An Analysis Using Household Exposure‑to‑Trade Indices Thang T Vo1 Dinh X Nguyen1 Accepted 17 September 2020 Published online 26 September 2020 © Springer Nature B V 2020 Abstract This paper quantifi.

ORIGINAL RESEARCH

Impact of Trade Liberalization on Household Welfare:

An Analysis Using Household Exposure‑to‑Trade Indices

or the expansion of labor demand Tariff reduction for exported goods is less favorable

to household welfare The impact of trade liberalization became smaller and less cant after the global downturn in 2008 Rural households suffered more vulnerability from trade, and the poor gained fewer benefits after the financial crisis in 2008

signifi-Keywords Trade liberalization · Trade exposure · Household · Vulnerability to poverty · Welfare · Vietnam

1 Introduction

International trade has long been accepted as a critical driver of economic development, especially over the long run (Onafowora and Owoye 1998; WTO 2001; Winters et al 2004; Wacziarg and Welch 2007; George 2010; Naito 2017; Fukuda 2018), even in times of cri-sis (Falvey et al 2012) Trade provides developing countries opportunities to accelerate economic growth, improve welfare, and reduce poverty, especially when major exports are labor-intensive products such as agricultural and basic manufactured goods With fewer trade barriers, the domestic market is enlarged with various products and better choices in

price and quality Exporting firms are more likely to employ additional low-skilled ers, which is expected to considerably impact poverty reduction.

work-However, the empirical impact of trade liberalization on poverty, welfare, and ity is far from consistent First, while the impact of trade liberalization on income is often positive at the national level, it may vary between regions within a country (Law 2019; Popli 2010) Regions with a higher proportion of sectors exposed to trade openness tend to experience a greater decline in poverty (Kis-Katos and Sparrow 2015; Popli 2010; McCaig

inequal-2011) However, urban areas with a greater share of workers in high tariff industries might have a smaller reduction in poverty and inequality once tariffs are reduced (Castilho et al

2012) Second, the skill and education level of workers also influences the actual impact

of trade openness on poverty and social welfare For instance, unskilled workers, and tors that mostly employ unskilled labor, can suffer during the opening process (Kis-Katos and Sparrow 2015), and trade liberalization has been shown to widen the skilled-unskilled wage gap in Latin American countries (Chaudhuri et al 2002; Hanson and Harrison 1999; Beyer et al 1999) to the detriment of poorer/less-skilled households (Naude and Coetzee

sec-2004)

It is recognized that the influence of trade on households varies in correspondence with two factors: (i) the impact of trade openness on domestic markets where the household is located, and (ii) the exposure of the household to the goods and factors that are involved in the opening process (Nicita 2009) Some empirical studies indicate that not every house-hold will benefit from trade liberalization (Nicita 2009; Marchand 2012), as trade open-ness can increase inequality while simultaneously reducing poverty overall (Winters et al

2004) Trade may even decrease welfare in the absence of insurance (Newbery and Stiglitz

1984) Therefore, a case study from an emerging country is required to enlarge the trade liberalization literature, wherein conclusive evidence at the household level is rare It is also important to empirically examine the channels through which trade openness contrib-utes to household welfare (Castilho et al 2012; Marchand 2012)

Vietnam provides an interesting case for trade literature as it has experienced successful economic development, major expansion of trade, and a history of trade policy reform and institutional change Since the start of Doi Moi, which began replacing the former ineffi-cient centralized economic system with a market-oriented economy in 1986, Vietnam has achieved immense success in promoting trade and economic growth The country’s GDP rose from only 61 billion USD to 175 billion in 2017 at an average rate of around 6.1% Vietnam joined ASEAN in 1995; acceded to the WTO in 2007; and has signed trade agree-ments with a number of countries, including China, India, Japan, Korea, and Chile, which have greatly reduced the tariff barriers to Vietnamese goods entering these markets As a result, exports as a share of GDP increased from 6.6% in 1986 to 50% in 2000 before sur-passing 100% in 2017 (World Bank 2019)

However, only around 7% percent of Vietnamese firms are involved in exports to eign markets; these exporters employ more than 90% of the workforce (SME 2015) Fur-thermore, most of Vietnam’s export industries are dominated by large firms Consequently, performance related to the expansion of employment is less satisfactory (Abbott et  al

for-2009), unemployment remains at around 6.9% in urban areas, and underemployment sists in rural areas Therefore, whether wage workers have gained from the trade boom remains unclear

per-To address this issue, our paper combines the household data extracted from the nam Household Living Standard Survey (VHLSS) and trade data from United Nations International Trade Statistics Database (UNCOM-TRADE) and Trade Analysis Infor-

Viet-welfare in Vietnam We construct two indices to represent household exposure to trade

as two proxies for trade liberalization, while welfare is considered through the scope of household vulnerability, income, and expenditure The study makes a unique contribution

to literature in four ways First, the households’ exposure to trade indices are unique and provide an incisive tool for empirical analysis The advantage of these indices is that they captures the influence of trade liberalization at the national level and the response of house-holds Second, the study provides rare evidence regarding the impact of trade liberalization

at the household level Third, using a wide range of data from 2002 to 2016, which mostly captures the period of trade policy reforms in Vietnam, we compare the impact of trade opening at different points in time Fourth, this study is the first attempt to compare the impact of trade liberalization on household vulnerability before and after the global finan-cial crisis in 2008

The rest of this paper is organized as follows Section 2 reviews the analytical work for trade liberalization and household welfare, including the research gaps Sections 3and 4 discuss empirical strategies and data description, respectively, while Sect. 5 pre-sents the results and findings The final section offers policy implications and concluding remarks

frame-2 Analytical Framework for Trade Liberalization and Household

2.1.1 The Price Channel

In the simplest form, the household welfare effect ( ΔWh ) is expressed as a function of the change in the price of a single good, i, used by the household h:

where qi represents production, ci represents consumption, and pi represents the price of good i.

In pure accounting terms, the retail domestic price of good i is the product of a string of

transmission from its world price, tariffs, exchange rates, etc., down to domestic taxes and distribution costs If the product’s final price is lower than its previous price, households

that are net consumers of good i will gain, but households that are net producers of good

(1)

ΔW h= (q i−c i) ⋅ Δpi

i will suffer loss If imported products compete with local products and result in lower

prices, the welfare effects will be similar Previous research investigating the effect of trade

on welfare via this channel includes Porto (2006), Nicita (2009), and Marchand (2012)

2.1.2 The Factor Market Channel

Trade can affect household welfare via the factor market, as illustrated in the left-hand side

of Fig. 1 The effect depends on the elasticity of factor supplies Under particular ditions, if the factor supplies are fixed, the real wage will increase if the production of goods increases By contrast, when factor supplies are infinitely elastic, the wage will be fixed and an adjustment will take place in terms of employment, and workers will switch between sectors in pursuit of higher wages In the real world, both these ends–wholly fixed

con-or wholly elastic labcon-or supplies–rarely exist Therefcon-ore, determining the elasticity of the labor supply is an important precondition of investigating the effects of trade on poverty Still, in both cases, if the conditions are met, workers always gain from trade liberalization Some studies that have investigated labor market mechanisms include Seshan (2013) and Kis-Katos and Sparrow (2015)

2.1.3 The Fiscal Channel

Despite the welfare gained by poor households, trade liberalization definitely reduces ernment revenue by decreasing tariff, import regulations, imposing losses on state monop-oly firms, and so on As government revenue is reduced, spending on social support may also decrease and, therefore, harm the poor However, for Winters (2002), falling revenue does not inevitably lead to reduced poverty reduction performance, as expected revenue loss is usually overstated by governments Even if revenue is lost, the government is pri-marily affected and not the poor The government’s ability to distribute revenue efficiently for anti-poverty campaigns before trade liberalization is often lacking In case of Vietnam, during the 1990s, aid for economic growth and direct aid for the poor was imbalanced, biased toward urban areas, and not conductive to poverty alleviation (Huong and Winters

gov-Quantities Exchange rate Tariffs, Quotas Tariff Revenue

Pass-through, competition

Border price

Taxes, regulation, distributors, procurement

Distribution, taxes, regulation, cop-ops

Retail price Factor markets

Welfare

Fig 1 Winters’ framework for the effect of trade liberalization on household welfare

2.2 Measurements, Level of Analysis, and Research Gap

2.2.1 Measuring Trade Openness

Empirical studies employ different definitions and measurements of trade openness due to the distinct context examined by each study However, according to McCulloch

et al (2002), the concept of openness can be divided into two main facets: openness in practice and openness in policy The former focuses on the influence of trade on eco-nomic activities and actual price distortions, which may not be controlled Conversely, the latter focuses on policies designed to curb trade and is comprehensively control-lable by the government

The concepts of openness and trade openness have been the subject of heated debate

in the last decade; hence, the appropriate way to measure trade openness is not yet well-established The confusion and ambiguity surrounding trade openness can be seen

in the numerous definitions used in various papers Some dominant measures include the share of trade (exports, imports, or both) in gross domestic product (Avelino et al

2005; Bussmann 2009; Aizenman and Noy 2009) and the adjusted trade share (Li et al

2004; Alcala and Ciccone 2004) Much effort has been made to improve these urements throughout the years For example, some studies proposed using five trade openness criteria dummy variables to represent trade openness (Cigno et  al 2002; Neumayer and de Soysa 2005) Other studies developed a Trade Restrictive Index, which measures the degree of trade protection for each country’s economy with respect

meas-to income, expenditure conditions, and balance of trade responses (Anderson and Peter Neary 1996; Bach and Martin 2001; Manole and Spatareanu 2010) One advanced measure of trade openness, called the Composite Trade Share, attempted to capture two dimensions of trade simultaneously, income from trade and interaction in trade activi-ties (Squalli and Wilson 2011) Another study relied on the correlation of a region’s export with its value- added production and exports to calculate a Regional Openness Index (Marjit et al 2007) Finally, one study utilized various regional indices such as infrastructure, institutions, market efficiency, industrial labor share, retail sales, and a Provincial Competitive Index to represent regional trade openness (Le 2014)

2.2.2 Data and Level of Analysis

Many types of data sources at different levels have been used to conduct empirical research on the links between trade liberalization and household welfare Several stud-ies focus on the global level (Baye 2017; Falvey et al 2012; Onafowora and Owoye

1998; Wacziarg and Welch 2007; Tanaka and Hosoe 2011), while many others focus

on the country level wherein each observation represents a geographic unit within the country such as a region, province, or district (Castilho et  al 2012; Kis-Katos and Sparrow 2015; Law 2019; Marchand 2012; McCaig 2011; Nicita 2009; Topalova

2007) Far fewer analyses address the household level, and these tend to study trends in trade liberalization by identifying dates before and after the removal of trade barriers, evaluating the impact on household welfare rather than using actual trade data (Gold-berg and Pavcnik 2007; Coello et al 2010; Popli 2010) This approach limits insights

on how trade liberalization affects households through tariff reduction

in recent years, none of the new measurements were applicable at the household level, thus omitting substantial household variations in response to new opportunities offered by trade reforms In the following section, we propose two indices to quantify household exposure-to-trade openness through their labor supply in response to increased labor demand and tariff reduction Our study provides rare evidence regarding the impact of trade liberalization on vulnerability at the household level.

3 Empirical Method

3.1 Measuring Vulnerability as Expected Poverty

Vulnerability as expected poverty (VEP) is a measure of vulnerability first proposed by Chaudhuri (2003) and applied to Vietnam by Vo (2018) and Vo and Van (2019) Household vulnerability is defined as the likelihood that a household will fall into poverty in the next period VEP can be estimated using the following procedures, beginning with computing the consumption function:

where ci is per capita consumption expenditure for household i; Xi represents a vector of

observable household characteristics (age, gender, ethnic, and marital status of the hold head, the household’s female ratio and dependent ratio, the highest educational quali-fication of family members, land ownership, and electricity access) and communal char-acteristics (household location, provincial poverty rate, interaction between provincial

house-poverty rate and household location, and administrative region); 𝛽 is a vector of parameters

to be estimated; and ei is a mean-zero disturbance term that captures idiosyncratic shocks

that lead to different levels of per capita consumption

With cross-sectional data, Chaudhuri (2003) suggests using a three-step Feasible

General-ized Least Squares technique to produce the expected log consumption, ̂E[ln ci|Xi] , and the

variance of log consumption, ̂ Var[ln c i|Xi] Assuming that ln ci is normally distributed, the

estimated probability that a household will be in poverty in the future (for example, at time

t + 1 ) is given by:

(2)

ln c i=𝛼 + 𝛽 ⋅ X i+e i

where Φ(.) is the cumulative function of the standard normal and z is the actual poverty

line The poverty lines used in this study are based on expenditure and use the definitions from the General Statistical Office—World Bank (World Bank 2018) Details of the pov-erty line are provided in Table 8

3.2 Estimation Strategy

In this paper, we apply the framework described in Winters (2002) to investigate the link between trade exposure and household welfare through the factor market channel The base-line model is as follows:

where yht are indicators of welfare outcomes for household h in year t, 𝛼h are the household fixed effects, 𝛼t are the year fixed effects, Zht is a vector of household controls (including

age, gender, marital status, and years of education for the household head, dependent ratio,

and location of household), and 𝜖ht is an error term assumed to be independently

distrib-uted normal

The variable of interest Lht represents trade liberalization at the household level From Winters’ framework, Lht could be the household’s participation in international trade, which

is likely to be correlated with welfare related outcomes captured in the dependent variable

y ht We, therefore, construct a proxy variable measuring changing household trade

participa-tion resulting from factors exogenous to the household We first construct the index of labor

demand due to exports for industry i in year t:

All households are exposed to the same labor demand due to exports in industry i

How-ever, the household’s decision to join the labor market varies in accordance with their bility to respond to the opportunities created by trade liberalization Additionally, the dis-tance from their home to urban areas where most trade-related activities occur may affect their decision to join the labor force Therefore, in this study, we weight the labor demand

capa-index by the percentage of the household’s members working in industry i and the

house-hold’s distance from urban areas Only individuals 15 years or older and who have income from their first main job are considered National trade shocks are transformed into house-hold trade shocks, and we call this index the household exposure to exports:

Another indicator that we use as a proxy for household exposure to trade is the tariff index

In this case, national labor demand is modified by the average value of the Most Favored Nation tariff rate applied to Vietnamese goods exported to China, the U.S., Japan, and Korea, the four main destinations for Vietnamese exports over the last 10 years, which is represented by TariffMFN

With these two exposure-to-trade indices in the place of Lht , we expect to capture the

effects of trade liberalization through labor demand at the industry level and labor supply

at the household level The higher the household exposure index, the higher is the exposure

of that household to trade Following the theory of international trade, we expect that, as exposure to trade increases, the household gains more benefits from trade, including higher income, higher expenditure, and lower vulnerability to poverty

4 Data

Household data for this study are drawn from the Vietnam Household Living Standard veys (VHLSS) These nationally representative surveys have been conducted by the Gen-eral Statistics Office of Vietnam every two years since 2000 with technical assistance from the World Bank with both household and commune survey The household survey collects detailed information about households, such as demographic information, education, employ-ment and labor force participation, income, expenditure, health, housing, durable goods, fixed assets, and participation in poverty programs The commune survey collects basic information regarding demography, socioeconomic characteristics, and the infrastructure of communes The surveys cover approximately 9300 households biennially 2002 onward The sample sizes and descriptive statistics of the basic variables in this study are provided in Table 9 and Table 10

Sur-Panel data for the VHLSS is only viable when the investigation spans over two years, as the VHLSS re-surveys only 50% of its observations every two years For panel data extended for more than two periods, the number of valid panel observations is reduced massively The inconsistency of matching between VHLSS waves prior to 2008 were addressed using McCaig (2009) revised version of the identification codes The number of observations for the panels used in this study are provided in Table 11

The trade value in goods for all the trade partners of Vietnam and tariff data, hereafter referred to as Vietnam Trade Data, were taken from UNCOMTRADE (United Nation 2019) and TRAINS (UNCTAD and WB 2019), respectively Both data are coded based on the Har-monized Commodity Description and Coding System (HS code) by degree of aggregation, starting from sections (one digit) to Chapters (two digits), Headings (four digits), and Sub-headings (six digits) However, the levels beyond six digits (eight and ten digits) are not har-monized between countries In this study, we use the data aggregated at the two-digit level.The tariff data used in this study is the mean rate of tariff Vietnam’s main trade part-ners–China, Japan, South Korea, and the U.S.–to imported Vietnamese goods This collective and comprehensive measure of import tariff rates for more than 9828 tariff lines is expected to represent a less-biased measure of openness than using Vietnam’s import tariff rate for foreign goods

(7)HHExposureTRht= 1

Distanceht ⋅

∑

i

1TariffMFN

i

⋅WorkerihtTotal Workerht

5 Results

5.1 Descriptive Statistics

Figure 2a illustrates the evolution of Vietnam’s exports to its top five partners from the early 2000s After the Vietnam-U.S BTA was signed in 2001, Vietnam’s exports increased dramatically until the global financial crisis in 2008 In 2008–2009, exports to major part-ners, except China, significantly declined Exporting turnover recovered and expanded quickly after 2010, especially to the U.S., China, Japan, and Korea

However, data from the period show that the average Most Favored Nation tariff rates of the U.S and Japan varied only slightly Korea’s MFN tariff fluctuated, while that of China fell sharply between 2000 and 2006 and has remained around 10% since then (Fig. 2b).Figure 3 depicts the evolution of Vietnamese household welfare from 2002 to 2016 Income and expenditure increased in this period, but there was a disparity between house-holds in urban and rural areas Growth rates for income and expenditure in urban areas were faster The gap between income and expenditure was also wider for urban households, representing a higher level of saving These differences partly explain urban households’ slightly lower levels of income inequality, as measured by the GINI index and much lower vulnerability to poverty (Ward 2016) The figure also indicates that the economic cri-sis affected income inequality and vulnerability in urban and rural areas However, rural households tended to suffer greater losses from the economic downturn than their peers in urban areas, as they had less effective means to cope with shocks Panel estimation from VHLSS confirms that household vulnerability is inversely associated with income and expenditure (Table 12)

Table 1 describes the mean of the two exposure indices across surveys by region, erty status, and industry sector In general, tariff reduction and exposure to exports cre-ated more opportunities for households over time, but the former shrunk and the latter sub-sequently fluctuated following the 2008 crisis Both indices indicate that those above the poverty line were more exposed to international trade than the poor Households in urban areas were more likely to be exposed to international trade opportunities created by tariff reduction than their peers in rural areas They, therefore, had more chances to consume less-expensive imported products However, urban households might likewise have been subject to greater competitive pressure from imported goods Similarly, urban households responded to opportunities in the labor force created by export industries In term of work-force sectors, those who worked in mining and quarries responded better to tariff reduc-tion while households having jobs in agriculture, forestry, aquaculture, and manufacturing industries were more exposed to exports Our decomposition of the two indices in Table 2shows that the shares of labor supply in the export index are higher than those figures in the tariff index, implying that households are more responsive to export increases than to tariff reductions Notably, exports generated more opportunities for the poor as the export increase in Vietnam is mostly related to agriculture and related services (Table 13) and the poor’s income mostly depend on this dominant sector (Table 14)

pov-5.2 Impacts of Trade on Vulnerability, Income and Expenses

Table 3a reports the impact of trade liberalization on vulnerability, income, and expenses

at the household level In general, the average effects of trade on Vietnamese households

(a) Vietnam’s exports to top five countries ($ million, 2010’s value)

(b)Most Favored Nation tariff rates of Vietnam’s major export partners

Fig 2 Export values and tariff rates

were positive from 2002 to 2012, but the impact became insignificant or even negative from 2012 to 2016 As the two trade exposure indices use two different units of measure-ment, we cannot compare the magnitude of the estimated coefficients of the two indices For income and expenditure, the positive signs of the coefficients indicate that the more the households were exposed to trade, the better was their welfare The results show that for most of the period of 2002–2012, trade liberalization was a net-positive for Vietnam-ese households Notably, exposure to exports positively contributed to households’ income and expenditure Similarly, the expansion of exports mitigated households’ vulnerability

to poverty (the probability of becoming poor) from 2002 to 2004 and 2006 to 2008, while tariff reduction augmented the vulnerability during 2002 and 2004 In contrast, interna-tional trade in Vietnam from 2012 to 2016 had limited influence on households’ welfare, as all coefficients are negative or insignificant It appears that trade negatively influenced the expenditure of those who were more involved

There is reason for the slight concern about the reverse causality between households’ welfare and the household labor supply, although this is just one of the three components

in the household exposure indices (Table 2) Therefore, we use communal labor supply to

replace the household labor supply The communal labor supply for industry i is calculated

by dividing the total labor in industry i in commune c by the total labor in the commune

The exposure indices thus capture labor demand at the country level, labor supply at the commune level, and the geographic disadvantage of households that are restrained from participating in international trade

In general, the estimated coefficients of household export exposure show similar effects However, in this case, tariff reduction appears to have had more negative impact on house-holds’ vulnerability and welfare The unwanted effects of tariff reduction can be explained

GINI VEP Per income Per expense

Fig 3 Households’ welfare during 2002–2016

in two ways First, lower tariff for Vietnamese products was offered in exchange for lower tariff on imported products Households would, therefore, spend less money on similar total consumption Second, if cheaper imported products pressured household production, household income would decrease, and in turn, household expenditure would shrink.

We check the robustness of our estimation using communal labor supply as an mental variable for household labor supply The first stage regressions of this instrumental approach are presented in Table 4a The estimated results in Table 4b are almost identical

instru-to those in Table 3b The labor market at the communal level shows smaller effects when the labor supply capacity of households is taken into account

Table 1 Statistics of household exposure indices at the household level Source: Author’s calculation from

We further examine the impact of trade liberalization on sub-groups of households Our estimated results divided according to household location are presented in Table 5, and they indicate that both rural and urban households obtained more benefits from increased exports Rural households derived larger share of income from exports than their peers in the urban areas, but they registered more vulnerability variation from trade The results filtered by poverty status in Table 6 show that both the poor and the non-poor gained from export liberalization Although the non-poor were negatively impacted by tariff cuts, they received more advantages on income and expenditure after the financial crisis in 2008.

We then estimate the impact of trade on households at the provincial level With this approach, we divide our data into the periods before and after the global financial crisis The results show that export exposure and tariff exposure improved household welfare in terms of higher income, higher expenditure, and less vulnerability After the global down-turn in 2008, the impact of international trade was smaller and less significant, regardless

of whether labor demand was estimated at the household or provincial level (Table 7)

Table 2 Decomposition of household exposure indices at the household level Source: Author’s calculation

Table 3 Fixed effects on vulnerability, income and expense