Economic Integration and Endogenous Trade Agreements: Analysing Global Trade Effects

Abstract

After the 1990s, global trade has seen a sudden proliferation of regional trade agreements (RTAs). Existing empirical literature argues about the differential impact of these trade agreements on high-income and lower-middle income countries’ trade. Further, economic and political arguments have developed to justify the proliferation of such trade negotiations. In this article, we have tried to build an argument taking economic and non-economic rationales of the formation of trade agreements as a complement to each other. Unlike the majority of existing literature in this field, our empirical analysis is an establishment of this debate which provides a platform to consider RTA formation to be an endogenous rather than exogeneous decision. Further, RTAs of varying nature are related differently to the unobservable which impede or facilitate trade. These unobservable also create endogeneity bias in the traditional estimation of RTAs’ impact on the trade flows of a country. In this study, we have attempted to investigate the endogeneity in the context of RTA engagements using the Heckman control function approach. This study ensures the endogeneity of RTAs with their positive and significant impact on trade flows. This results do not significantly differ from similar kinds of empirical studies available on international platforms.

JEL Classification: F1, F12, F14

Keywords

Regional trade agreements endogeneity gravity model Heckman control function treatment effects

Introduction

In a global arena, regional trade agreements (RTAs) have been formulated under the umbrella of GATT Article XXIV to facilitate multilateralism. From the perspective of both developing and developed nations, regional trade integration has become a profound phenomenon for economic development backed by the logic of the international political economy of GATT Article XXIV. The proliferation of regional and bilateral trade agreements is gaining rapid momentum by establishing various sets of rules for lower-middle income and high-income countries. Further, this regional integration affects both member and non-member countries differently and generally influences the nature and complexion of the relationship that exists between them.

A striking development is seen in the proliferation of trade agreements since the mid-1990s in the world’s trading system. Evidence reveals that lower-middle income countries are coming to the forefront to sign trade agreements among themselves, mostly dealing with commodity trade. Based on the classification of North-North, North-South and South-South (SS) varieties of existing RTAs, the World Trade Organization (WTO) RTA database reveals that almost 50 per cent of the RTAs that are in force are of SS variety, that is, signed between lower-middle income countries, with more recent ones covering WTO-Plus provisions. What is more interesting is that some lower-middle income countries are members of multiple trade agreements, where some agreements are signed only among lower-middle income countries, which is aptly termed a ‘spaghetti bowl’ of agreements. If we look into the new generation of RTAs, especially the ones involving high-income countries, they have become intercontinental and non-exclusive. These new generation RTAs are broader and deeper in terms of their regulatory measures’ harmonisation, easing inter- and intra-regional factor mobility, etc. Many agreements have limited the use of subsidies and quantitative restrictions along with the inclusion of regional investment rules, competition, non-tariff measures, labour and environment-related provisions (Nataraj, 2007). However, the existing literature on these trade agreements’ welfare gain or loss is providing mixed results. Particularly for lower-middle income countries, the impact of these RTAs is not so encouraging as trade diversion effects are dominating over trade creations. For high-income countries, this impact is ambiguous in most cases. Hence, it is justifiable to ask why the countries contract RTAs in this context. The arguments can be developed from economic and non-economic standpoints, which we have tried to build in the present article. The significant economic argument is whether or not RTAs stimulate trade between their members. The empirical investigation of this argument requires sophisticated economic and econometric modelling. Two broad, complementary avenues are available to measure the economic benefits of RTAs, namely. ex-post econometric analyses using standard gravity model and ex-ante general equilibrium simulation approaches.

Given this background, in this study, we have tried to explain the reasons behind this RTA proliferation from economic and non-economic standpoints and have supplemented the arguments with an econometric analysis using the gravity model. The non-economic argument of RTA formation has raised a debate between economists and political scientists, both theoretically and empirically, who have tried to explain RTA formation as an endogenous rather than exogenous decision. In this supplementary econometric analysis, we have tried to address this endogeneity bias using the Heckman control function technique, where our empirical investigation ensures the endogeneity of RTAs with its positive and significant impact on merchandise trade.

Further, these RTAs are differentiated rather than homogeneous, which we usually ignore during our empirical analysis. These unobservable heterogeneous effects of RTAs create the endogeneity bias, which, most of the time, is hard to explain empirically. The growing number of RTAs is caused by the new regionalism under which regional agreements often involve deep integration. The twenty-first century regional integration covers a broad spectrum beyond goods’ trade and tariff liberalisation—the WTO-plus provisions in RTAs include investment protection, competition policy, intellectual property rights, environmental protection, labour standards and TRIPS-plus obligations. These deeper RTAs extensively affect a country’s domestic policy with economic power asymmetries between negotiating nations. This leads to more complex trade associations and hence poses greater challenges to the member and also non-member countries, which, in a way, determines their positive and negative effect on trade.

Given this backdrop, the rest of the article is systemised as follows: In the next section, we have developed the economic justifications of RTAs, its theoretical mechanisms and empirics. In the following section, we have argued about the non-economic standpoints and the potential endogeneity of RTA formation. The subsequent section enumerates the model and the empirical estimation. The second last section deliberates the challenges of RTAs whereas the last section concludes the study.

Economic Justification of RTAs: Theoretical Mechanism and Empirics

The major economic justification is in the line of the expansion of intra-RTA trade flow irrespective of the choice of theoretical modelling that measures the economic gains of RTAs. Also, one can argue that tariff reduction and freer trade are primary incentives for signing RTAs between members. But, by 1991, due to GATT, most countries (barring South Asia) have already brought down their tariffs to fairly low levels so that the Vinerian benefits of RTAs to member countries (trade creating or diverting effects) become limited.

This can be explained further by arguing about the distributional effects of RTAs which occur at the cost of third countries or parties. If prices remain unchanged with tariff elimination among member countries as a result of trade agreement, each country substitutes away from procurement of her goods and buys more from one another. Further, all countries under a common RTA drive away from procurement of goods bought from non-member countries. This creates a positive term of trade effect for the member countries. Thus, RTA can become a more attractive proposition due to this potential ‘beggar thy neighbour’ effect, although it generates some adverse trade effects (diversion) on member countries.

Following Viner (1950), many empirical studies have centred around static gains and losses of member countries where much of the literature on regionalism implicitly assumes that RTAs are similar (see, for example, Bagwell & Staiger, 2010; Coulibali, 2007; Kandogan, 2008; Shahid, 2011; Winters & Chang, 2000; etc.). The underlying trade model in the Vinerian framework is the traditional Heckscher–Ohlin–Samuelson (HOS) model of trade, which is largely based on the supposition of heterogeneous countries. Now, this model and assumption are largely valid for the north-south trade. However, Krugman (1979) argued that a major proportion of international trade is transacted among developed nations, for which he has provided the theoretical basis. In his model, he has shown the welfare gain from the intra-industry trade type between high-income countries, where welfare gains from trade are based on the consumption selections of the consumers of trading partners rather than goods’ price changes.

Meade (1955) further extended the static theory of regional trade arrangements with feedback effects of international adjustments on member countries and focused on world economies’ economic welfare. Backed by these theories, much of the debate has centred on whether or not lower-middle income countries gain from RTAs. As expounded by Bhagwati and Panagariya (1996), the benefits are mainly focused on two characteristics that are common to lower-middle income countries: their relatively high tariff and trade dependency on one or more high-income country-partner. They have argued that small countries lose from an RTA due to their high dependency on trading partners and high tariffs. In the small country case, when the rest of the world (ROW) is a large supplier, prices in the small country do not change following tariff elimination under RTA. They argued that complementarities in trading nations, degree of initial protection, market size, domestic trade liberalisation measures and rules of origin (RoO) are additional factors on which the effectiveness of a trade agreement in creating additional trade depends.

Theoretical researchers have developed the conditions that illustrate the conditions under which the trade agreements induce welfare gains for participants. Wonnacott and Lutz (1989) and Krugman (1995) have advocated the ‘natural trading partner hypothesis’ as one of the most common theoretical arguments. This argument says that natural trading partner countries find trade agreements to be economically advantageous which are preferential in nature. The hypothesis contemplates that regional trade packs with nearby countries generate more trade-creating effects than trade-diverting effects. The logic behind this hypothesis is when a country imports a product from outside of its preferential trading area, the scope of trade diversion becomes limited pertaining to that product. Magee (2004) has argued about the optimistic assessment of natural trading partner hypothesis where he views that the welfare-augmenting or diminishing effect of regionalism critically depends on the trading partners engaged in preferential deals. Krishna (1998) has further described the political economy approach to regionalism, where he argues that trade diversion is more politically acceptable as it does not harm the domestic industries, whereas trade creation may displace them, which can generate political opposition towards the formation of an RTA.

Some studies discussed that gains from an agreement (trade creation and diversion) are not certain and depend upon the individual characteristics of the agreement (Bagwell & Staiger, 2010). Given that agreements differ vividly in scope and objectives and are extremely heterogeneous in nature, RTAs may enhance intra-regional trade at the stake of trade with third countries (diversion effect), which reduces the economic welfare of the trading partners. A study by Adams et al. (2003) revealed that non-member countries generate higher trade diversion effects compared to trade creation effects generated by members. They studied 18 RTAs altogether, among which major trade blocs like EU, NAFTA and MERCOSUR have not created any substantial trade within the region and hence have deteriorated the welfare effect. Krugman and Obstfeld (2003) illustrated the negative effects of MERCOSUR. They found that, following the agreement, the volume of trade among these nations has tripled in four years.

Vicard (2009) classified the form or depth of RTAs into four main tiers: preferential agreements (PTAs), free trade agreements (FTAs), customs unions (CUs) and common markets (CMs). With the help of gravity model, using panel data for 1960–2000, he investigated to what extent the depth or form of RTAs affects trade after controlling for agreements’ self-selection. His results reveal that the average treatment effect (ATE) of RTAs on bilateral trade does not significantly vary with respect to the form of agreements. The study of Shahid (2011) indicated that the existence of WTO-plus provisions in a given RTA is favourable for trade, whereas the presence of WTO-extra provisions impedes trade. MacPhee and Sattayanuwat (2014) examined the effect of 12 broad trade agreements on extra- and intra-regional trade flows of member and non-member lower-middle income countries’ trade from 1981 to 2008. Their result did not indicate any encouraging evidence of intra-bloc trade creation. In a recent study, Pant and Paul (2018) argued that the endogenous RTA has not proven to be a trade-promoting device for India.

Further, the computational general equilibrium (CGE) model is used to study the dynamic impact of RTAs. Studies by Baldwin and Venables (1995), Burfisher et al. (2002) and Robinson and Thierfelder (2002) are particularly noteworthy in this context. Although these models demonstrated greater gains from RTAs compared to static models, they still found that the benefits from some of the largest RTAs have a limited impact on their members. They performed an ex-ante impact assessment analysis of NAFTA in Mexico by using a multi-country CGE model. They disagreed with the view of Bhagwati and Panagariya and argued that domestic distortions of small lower-middle income countries lead to a welfare decline and prevent efficiency gains from an RTA. Domestic interest politics, undue adjustment and implementation expenses and inadequacy of negotiating and analytical capacities are potential costs faced by developing nations when they cut a deal for trade agreements with their developed trading counterparts (Thiratayakinant, 2010).

Theoretically, it has been further argued that a potential gainful effect from RTAs emerges from the market expansion which leads to higher productive efficiency with scale economy benefit for any industry. This factor may make trade agreements relatively lucrative for smaller countries as their firms had limited access to international market before the agreement. Hence, smaller countries would see the prosperity associated with enhanced competitiveness, which might provide them an incentive to join an RTA (see Berthelon, 2004, for example).

Non-economic Arguments and Endogeneity of RTAs

The non-economic arguments suggest that, as the WTO process appears to be unravelling, RTAs (commodity trade and WTO-plus agreements) offer new forms of plurilateralism. The decision to form RTAs is often political, aimed at being part of a political bloc for future multilateral negotiations. Countries may pursue RTAs for purely political reasons, making some RTAs strategic alliances that implicitly serve as part of security arrangements. In certain cases, RTAs are used to solidify domestic policy reforms. Additionally, some RTAs focus on interest group pressure, political leadership and strategic considerations in global politics and the world economy (see Baldwin, 2008; Bhagwati, 2008; Estevadeordal & Suominen, 2009; Mansfield, 1998; Schiff & Winters, 2003). As expounded earlier, political economy approach to regionalism advocates that trading partners may intend to sign trade-diverting PTAs as it does not harm domestic industry and hence is more acceptable politically (see, for example, Krishna, 1998). On the contrary, the trade creation effect of regional agreement demeans domestic industry, which generates considerable political opposition (Magee, 2004). This debate between economists and political scientists suggests political rationales for RTAs. Hence, whether these political reasons dominate over economic and geographical reasons is a matter in question as the economic justification of RTAs seems to become limited. Addressing this debate, theoretical and empirical studies have explained RTA formation as an endogenous rather than an exogenous decision (see, for example, the studies of Baier & Bergstrand, 2004; Baier et al., 2007; Egger et al., 2008; Magee, 2003). The argument is that the formation of RTAs is a consequence of increased trade among countries that choose to expand normal trade relations at a higher level, including WTO-plus provisions. This self-selection makes RTAs endogenous to the trading process, where RTAs are viewed as an outcome of trade itself.

Further, to hinder or facilitate trade, heterogeneous RTAs are related divergently to the unobservable. The intensification of trade negotiations involves various aspects, such as the expansion of domestic markets that provide benefits from economies of scale, geographical factors stemming from the negotiating nations’ capabilities, and the portrayal of these negotiations as political signals to domestic economic agents. Additionally, it includes the coordination of macroeconomic policies with trade facilitation agendas. These unobservable factors can introduce endogeneity bias in the traditional estimation of RTAs’ impact on a country’s trade flows. However, such studies are limited as the conventional assessments of the effect of RTAs on bilateral trade flows have been impelled to be biased due to this endogeneity (Baier et al., 2007).

Model Specification and the Empirical Estimation

For the last 40 decades or so, international trade economists have been quantifying whether the trade agreements are having trade-creation or diversion effects using the gravity model, which has been the workhorse for estimating such effects for cross-country empirical analysis. Tinbergen (1962) analysed this model first to estimate bilateral trade flows within the EU countries. Tinbergen’s study evaluated the effect of FTAs on trade where he found insignificant effects of ATEs of FTAs on trade flows. However, Aitken’s (1973) estimate revealed PTAs’ impact on bilateral trade between member economies is positive and significant.

During the post-1975 period, developments of new trade theories have generated further room for benefits and losses of trade considering differentiated goods in a monopolistically competitive framework. The ramification of these specifications had been articulated in the seminal works of Krugman (1979, 1980), Helpman and Krugman (1995), and others. Based on these theoretical foundations, further improvements in the models were done by Anderson (1979), Bergstrand (1985), Sanso et al. (1993), Matyas (1997, 1998) and Anderson and van Wincoop (2003). These models were later used in various empirical works including Sharma and Chua (2000), Lee and Park (2005) and Pusterla (2007) in the Asian context also.

Econometric specifications were built upon the basis of abovementioned theoretical foundations to estimate gravity equations. For example, studies by Matyas (1997, 1998) and Egger (2000) argue that a panel data approach captures the overall business cycle aspect encountered by the trading economies more efficiently over cross-section approach. Further, Martinez-Zarsoso and Nowak-Lehman (2003), Egger and Pfaffermayr (2003) and Cheng and Wall (2005) have also advocated panel estimation over cross-section, which allows country-pair heterogeneity more effectively.

Carrere (2006) used Hausman–Taylor specification, which included three RTA dummies covering countries’ exports, imports and intra-regional trade flows in a panel data framework. His study suggests a more reliable impression of trade effects of RTAs. Covering seven RTAs, his study shows that those RTAs have enhanced the intra-regional trade beyond the levels conceived by the standard gravity model. At times, this is coupled with a reduction in exports and imports from the ROW. This evidences the trade diversion effect. Coulibaly (2009) combined gravity regressions with non-parametric estimation techniques in a two-step estimation procedure. They imposed minimal structure on the model by capturing the non-monotonic trade effects.

In all these studies mentioned above, RTA dummies were taken as typically exogenous variables into the system. However, our economic and non-economic arguments mostly provide the theoretical rationale for RTAs being endogenous. Existing econometric works advocate that political and economic foundations persuade preferential trade liberalisation through the membership of trade agreements. This is along the lines attributed to economic theories as enumerated by Baier and Bergstrand (2004, 2007), Magee (2003), Egger et al. (2008), etc. Baier and Bergstrand (2004) introduced why a typical FTA dummy variable is potentially ‘endogenous’. To address this potential endogeneity, they formulated a general equilibrium framework with ‘pure economic’ determinants or factors of FTAs. The model was persistent with Heckscher–Ohlin and new trade (intra-industry) theories. Their study paved a benchmark for future political economy empirical models of trade agreements in a general equilibrium framework. In later attempts, Baier et al. (2007) and Magee (2003) used Heckman’s control function approach in an across-section framework to address the endogeneity of trade agreements. In his paper, Magee (2003) used instrumental variables to adjust for the endogeneity of FTAs in a simultaneous-equations system. He too has found large positive and negative effects of FTAs on trade flows. Baier et al. (2007) applied treatment effects approach to estimate the effects of FTAs on bilateral trade flows in a panel data setup to address the potential endogeneity bias in estimation. Before them, Trefler (1993) had used an instrumental variable approach to account endogeneity of trade policies. Egger et al. (2008) estimated the impact of endogenous new RTA membership on trade by formulating an empirical model that reveals the likelihood of new RTA membership controlled by country size, factor endowments and trade and investment costs.

Liu (2008) has empirically tested the political economy motivations for trade agreement formation. He used the median voter model to review the political economy factors for FTA formations. In a panel setup, he used the duration analysis to derive policy implications in the context of dynamic time path cases of trade agreements and static welfare analysis.

All these studies have provided the theoretical and empirical background to review the impact of endogenous RTA on trade flows. Based on the major argument of endogeneity of these studies, here we have followed the standard gravity model to explain the relationship between RTAs with trade. Following Anderson and van Wincoop (2003), under the new trade theory, the operational structural model of bilateral aggregate nominal trade flows between the trading partners can be written as:

X_{i j} = \frac{Y_{i} . Y_{j}}{Y_{w}} {(\frac{τ_{i j}}{P_{i} . P_{j}})}^{1 - σ} with,

(1)

{(P_{i})}^{1 - σ} = \sum_{j} {(\frac{τ_{i j}}{P_{j}})}^{1 - σ} \frac{Y_{j}}{Y_{w}} and {(P_{j})}^{1 - σ} = \sum_{i} {(\frac{τ_{i j}}{P_{i}})}^{1 - σ} \frac{Y_{i}}{Y_{w}}

(2)

where, $X_{i j}$ is nominal merchandise trade flow from home (j) to its partner country i at period t. $τ_{i j}$ is the trade costs associated with trade flow from j to its partner country i at period t. $Y_{j}, Y_{i}$ and $Y_{w}$ are nominal gross domestic products of country j, country i and world, respectively. $σ$ is elasticity of substitution of all product varieties. $P_{i}$ and $P_{j}$ are inward and outward multilateral resistance (MR) as they depend upon $τ_{i j}^{1 - σ}$ , $Y_{j}$ and $Y_{i}$ combined with a normalisation. Intuitively, the outward and inward MR of the partner countries can be measured by the incidence of trade costs of average sellers and buyers (Anderson & Yoto, 2010).

The log-linear form of the gravity equation (1) gives:

ln X_{i j} = - \ln Y_{w} + ln Y_{i} + ln Y_{j} + ln {(τ_{i j})}^{1 - σ} - ln {(P_{i})}^{1 - σ} - ln {(P_{j})}^{1 - σ}

(3)

As the direct determinants of trade frictions $τ_{i j}$ are not available, following existing literature, we have considered the available proxy variables which can account for the trade costs between the countries. We assume that $τ_{i j}$ is a function of time-invariant factors like bilateral log-distance between two partner countries ( $D_{i j}$ ), common colony ( $c o m_c o l o n i z e r_{i j})$ , common official language ( $c o m_l a n g_{i j})$ , common boarder ( $c o n t i_{i j})$ , common language that 9 per cent of each country speaks ( $c o m_l a n g_e t h n o_{i j}),$ colony after year-1945 ( $c o l_a f t_1945_{i j})$ and two countries are/were the same country ( $s a m e_c o u n t r y_{i j}) .$ In most of the empirical literature, trade policy variable has been accounted as an element of trade costs. Following those conventional empirical models, RTAs between two trading partners $(R T A_{i j})$ can be considered as an important indicator variable or element of $τ_{i j}$ . Hence, the commonly adopted specification or functional form of $τ_{i j}$ is:

τ_{i j}^{1 - σ} = \exp (β_{3} . ln D_{i j} + β_{4} . c o m_c o l o n i z e r_{i j} + β_{5} . c o m_l a n g_{i j} + β_{6} . c o n t i_{i j} + \dots + δ . R T A_{i j})

(4)

Substituting (4) into (3) we get,

ln (\frac{X_{i j}}{Y_{i} . Y_{j}}) = ln E_{i j} = β_{0} + β_{3} ln D_{i j} + β_{4} . c o m_c o l o n i z e r_{i j} + β_{5} . c o m_l a n g_{i j} + β_{6} . c o n t i_{i j} + \dots + δ . R T A_{i j}

(5)

where, $β_{0} = - ln Y_{w}$ , $α_{s i} = - ln {(P_{i})}^{1 - σ}$ and $γ_{j} = - ln {(P_{j})}^{1 - σ}$ .

Now, if we relax the restrictions of unitary income elasticities, then the model becomes (in its stochastic form):

ln X_{i j} = β_{0} + β_{1} ln Y_{i} + β_{2} ln Y_{j} + β_{3} ln D_{i j} + β_{4} . c o m_c o l o n i z e r_{i j} + β_{5} . c o m_l a n g_{i j} + β_{6} . c o n t i_{i j} + \dots + δ . R T A_{i j} + α_{i} + γ_{j} + u_{i j}

(6)

where $α_{i} and γ_{j}$ are fixed effects of country i and j.

As explained earlier, existing literature has raised serious concern about the exogeneity of the trade agreements and thus the consistency and efficiency of the estimated parameters of the standard gravity model. Potential endogeneity of $R T A_{i j}$ implies its correlation with the white noise error term $u_{i j}$ . The three roots of endogeneity generally recognised: omitted variables, simultaneity and errors of measurement.¹ Omitted variables appear when we miss out on some important regressors due to data unavailability in a regression model. To address the endogeneity bias, instrumental variable and/or Heckman control function techniques are the possible methods based on joint distributions of $u_{i j}$ and $R T A_{i j}$ (Baier et al., 2007; Egger et al., 2011).

To address this econometrically, following Maddala (1993) and Wooldridge (2002), simultaneous equations can be written as (using conventional matrix notation):

x_{i j} = Γ_{i j}^{'} . β + δ R T A_{i j} + α_{i} + γ_{j} + u_{i j}

(7)

R T A_{i j}^{*} = x'_{i j} . μ + Γ_{i j}^{'} . a + z'_{i j} . b + ϵ_{i j}

(8)

Since we cannot observe $R T A_{i j}^{*}$ , we assume the indicator variable $R T A_{i j}$ is a function of $R T A_{i j}^{*}$ , that is,

\begin{matrix} R T A_{i j} & = 1 if R T A_{i j}^{*} > 0 \\ = 0 otherwise \end{matrix}

Here, $R T A_{i j}^{*} R T A_{i j}^{*}$ is the latent variable that represents utility gain/loss for the representative country of an RTA. $x_{i j}$ is $ln X_{i j} .$ $Γ_{i j}^{'} = (1, . ., ln D_{i j} \dots)'$ is the vector of all explanatory variables including trade cost factors except $R T A_{i j}$ . $β$ and $a$ are vectors of coefficients corresponding to $Γ_{i j}^{'}$ . $z_{i j}$ is the set of instruments (not in $Γ$ ) and $b$ is the vector of associated parameters. We have explained the potential elements of $z_{i j}$ subsequently.

Further, the structural equations need some restriction on coefficients to be logically consistent,² that is, either $δ or μ$ must be equal to zero. $δ = 0$ makes the study of RTA’s effectiveness on trade flow irrelevant. Setting $μ = 0,$ the reduced form for $R T A_{i j}^{*}$ is:

R T A_{i j}^{*} = Γ_{i j}^{'} . a + z'_{i j} . b + ϵ_{i j}

(9)

Now, the conditional expectation function of model (7) is:

E (x_{i j} | Γ_{i j}, R T A_{i j}, α_{i}, γ_{j}) = Γ_{i j}^{'} . β + δ R T A_{i j} + α_{i} + γ_{j} + E (u_{i j} | Γ_{i j}, R T A_{i j})

(10)

\begin{matrix} With R T A_{i n d, j} = 1 {if Γ}_{i j}^{'} \cdot a + z_{i j}^{'} \cdot b \geq ϵ_{i j} \\ = 0 otherwise \end{matrix}

[from (9)]

Under the assumption of bivariate normality of $ϵ_{i j}$ and $u_{i j}$ , we know:

E (u_{i j} | ϵ_{i j}) = η . E (ϵ_{i j} | Γ_{i j}^{'}, z'_{i j}, R T A_{i j})

Then the above conditional mean function can be written as:

\begin{matrix} E (x_{i j} | Γ_{i j}, R T A_{i j}, α_{i}, γ_{j}) & = Γ_{i j}^{'} . β + δ R T A_{i j} + α_{i} + γ_{j} + η . E (ϵ_{i j} | Γ_{i j}^{'}, z'_{i j}, R T A_{i j}) \\ = Γ_{i j}^{'} . β + δ R T A_{i j} + α_{i} + γ_{j} + η . φ_{i j} \end{matrix}

(11)

where, $φ_{i j}$ = $E (ϵ_{i j} | Γ_{i j}^{'}, z'_{i j}, R T A_{i j}) = R T A_{i j} . [\frac{ϕ (Γ_{i j}^{'} . a + {z^{'}}_{i j} . b)}{Φ (Γ_{i j}^{'} . a + {z^{'}}_{i j} . b)}] - (1 - R T A_{i j}) [\frac{- ϕ (Γ_{i j}^{'} . a + {z^{'}}_{i j} . b)}{1 - Φ (Γ_{i j}^{'} . a + {z^{'}}_{i j} . b)}]$ is the inverse Mills ratio which serves as a ‘control’ function for the unobserved heterogeneity and potential selection bias.

$Φ (.)$ denotes the standard normal cumulative distribution function (cdf) and $ϕ (.)$ is the probability density function (pdf).

A two-stage method of moment estimation is used to get consistent and efficient estimators of this model (see Heckman, 1997; Vella & Verbeek, 1999 and Wooldridge, 2002, Ch. 18). In the first step, we estimate $\hat{b}$ and $\hat{a}$ from a probit regression and compute $φ_{i j}$ . Then in the second step, using ordinary least squares (OLS) we estimate the following to estimate the remaining parameters $β, δ and η$ :

x_{i j} = Γ_{i j}^{'} . β + δ R T A_{i j} + α_{i} + γ_{j} + η . φ_{i j} + u_{i j}

(12)

If $η = 0$ , then the errors are independent, and $φ_{i j}$ = 1 implies $E (ϵ_{i j} | Γ_{i j}^{'}, z'_{i j}, R T A_{i j}) = 0$ which signifies $R T A_{i j}$ is exogeneous. However, if $η \neq 0$ , then we will get inconsistent $β and δ$ if we neglect $φ_{i j}$ .

The existing theoretical and empirical literature advocates the endogenous and interdependent formation of RTAs. The proposed hypotheses of Baier and Bergstrand (2004) and others have justified the standard determinants like size of the economies, trade cost measures, bilateral distance, border sharing, remoteness, sharing colonial links, etc. to determine the likelihood of country-pairs forming RTAs. $Γ_{i j}^{'}$ captures these variables.

Further, we have also tried to capture some omitted determinants, which play a crucial role in determining the likelihood of RTA formation. These omitted determinants mainly capture the political arguments that we have tried to bring earlier in the non-economic argument. The demand and supply side logic of the political economy of trade agreements has emphasised the role played by the societal interest groups and the effects of state institutions on trade agreements and/or trade policies. Political economists began focusing on more specific domestic political institutions while analysing trade policy and/or trade agreements. Good quality of governance removes ambiguity about contract enforcement reducing the cost of transaction, increasing the safety of property, and thus indirectly increasing the degree of trust between two countries in case of bilateral transactions. Homogenous governance may lead to like standards in conventions and business practices along with familiarity with formal procedures in a country pair (Beugelsdijk & Van Schaik, 2001). What policymakers perceive plays an important role in trade negotiations. This brings down the cost of adjustment that they would have to incur to familiarise themselves with each other’s institutional environment. Homogeneity of institutional structure may enhance bilateral trust and hence bilateral trade. The world governance indicator, government effectiveness,³ can be proxied, which reflects the capacity of a government to effectively formulate and implement sound policies that are conducive to any trade negotiations. The absolute difference of the scores of the government effectiveness of the partner countries has been considered here as an indicator variable.

Studies of Baldwin (1989), Hillman and Riley (1989), Rodrik (1994), Haggard and Kaufman (1995), etc. advocated the importance of electoral component, that is, the median voters’ preferences in determining trade agreements. The world governance indicator political stability⁴ is a proxy variable that captures the techniques through which governments are selected or elected, monitored and replaced. In our current analysis, we have considered these two variables as potential instruments of trade policy. We have taken the absolute difference of the scores as an indicator variable here.

Data

The trade data and the gravity model variables are collected from the World Integrated Trade Solution (WITS) database and ARTNeT Secretariat for the years 2013–2019 for 67 countries (unbalanced panel). The data of RTA and two governance indicator variables have been supplemented from Design of Trade Agreements (DESTA) and World Governance Indicator (WGI) database. The descriptions and the descriptive statistics of the data used in the model are given in Appendix 1.

Empirical Estimation

Table 1 describes the estimation results of our model mentioned by Equations 10 and 11. Column 1 represents the results of the gravity model (Equation 7) where RTA variable is considered as exogeneous explanatory variable in the system. Columns 2 and 3 are the results of the two-step estimation process where we have addressed the endogeneity related to RTA variable in the system. Column 2 is the 1st step estimation of probit regression of the control equation and column 2 shows the results of the 2nd step regression result of the outcome equation.

Table 1.

Empirical Estimation.

Variables	(1)	(2)	(3)
	FE	1st Step Probit	2nd Step FEIV
	Exo-RTA	Endogenous-RTA
Dep. Variables	$ln X_{i j}$ +1	$R T A_{i j}$	$ln X_{i j}$ +1
$R T A_{i j}$	0.0452		0.0593*
	(0.0559)		(0.7564)
$φ_{i j}$			–3.334**
			(1.9670)
$ln (Y_{i})$	0.4700***	0.0288**	0.5640***
	(0.0858)	(0.0144)	(0.0998)
$ln (Y_{j})$	0.8540***	0.1520***	0.9380***
	(0.0655)	(0.0127)	(0.0860)
$ln D_{i j}$		–0.8020***
		(0.0394)
$c o n t i_{i j}$		–0.1890
		(0.1440)
$c o m_l a n g_{i j}$		–0.0452
		(0.1850)
$c o m_l a n g_e t h n o_{i j}$		0.0677
		(0.1830)
$c o m_c o l o n i s e r_{i j}$		0.3000**
		(0.1360)
$c o l_a f t_1945_{i j}$		0.3580
		(0.2350)
$s a m e_c o u n t r y_{i j}$		–0.2870
		(0.2880)
$g e_{i j}$	–	–0.0678**	–
		(0.0287)
$p s_{i j}$	–	0.0881**	–
		(0.0443)
Constant	–	1.1390*	–
		(0.5990)
Adj. R²	0.956		0.957
Pseudo R²		0.277
Country-FE	Yes	No	Yes
Time-FE	Yes	Yes	Yes
Wald chi²		622.20
Prob > chi²		0.00
F-stat	77.02		58.06
Prob > F	0.00		0.00
Sargan–Hansen Stat of OIR			2.528
p Value of OIR			.468
Observations	19,705	19,904	19,705

Source: Author’s calculation by using data from ARTNeT, DESTA and WGI databases.

Notes: ***, ** and * denote the level of significance at 1%, 5% and 10%, respectively. Cluster-robust standard errors have been calculated at country level and are given in parentheses. Time invariant variables are omitted in fixed effect estimation results.

From an econometric point of view, the parameters in the model (12), namely. $β$ , $δ$ , $α_{i}$ and $γ_{j}$ , are the nuisance parameters. Thus, the model is a two-way country-specific effects model where $α_{i}$ and $γ_{j}$ subsume the effects of other country-specific MR terms. The relationship between $(α_{i}$ , $γ_{j})$ and the regressors $R T A_{i j}$ and $Γ_{i j}$ determine the econometric technique. Random effect estimation produces consistent and efficient estimators if $(α_{i}$ , $γ_{j})$ are independent of $R T A_{i j}$ and $Γ_{i j}$ . But in this study, the underlying economic model suggests that $(α_{i}$ , $γ_{j})$ are not independent of $R T A_{i j}$ and $Γ_{i j}$ . Hence, we are treating the model as a two-way fixed effect model. Further, we have run the Hausman test which again revalidates the appropriateness of the fixed effect model over random effect in our case.

The estimation results of column 1 reveal that the treatment effect of RTA on world trade is 0.045 which is insignificant. This means RTAs had no impact on world trade during 2013–2019 when it was treated as an exogeneous explanatory variable. The result reveals that RTAs are not generating any welfare gain as trade creation and diversion effects are netting out each other. This result is quite similar to the other existing empirical studies that we have mentioned in the previous section.

Now, considering the endogeneity of RTAs, we have used the instrumental variable approach which is shown in columns 2 and 3. Here, we have found a different result. Considering the strong exogeneity of the instruments,⁵ we have found that both instruments are statistically significant at the chosen level of significance. Further, Sargan–Hansen test statistic for over-identified restriction reveals the relevance of chosen instruments. Hence, coming to the interpretation of the coefficients, we can infer that the negative coefficient of government effectiveness indicator reveals, the higher the difference between the partner countries in terms of the effectiveness of governments, the lower the probability of signing an RTA between them. Again, the closer the countries in terms of their political stability, the higher the probability of signing an RTA between them.

The other economic factors, like GDPs of the trading partners and other trade cost factors like distance and common colony, are significant factors in the formation of trade agreements between two countries. Rest of the factors like common language, same country, post-1945 colonial relation, etc. are not important factors in RTA formation.

We have computed the inverse Mills ratio ( $φ_{i j}$ ) from the probit regression to run the 2nd stage regression as explained before. Column 3 represents the results where RTA variable is instrumented in the outcome equation. As explained, the coefficient of $φ_{i j}$ , that is, $\hat{η}$ , is a measure for potential endogeneity of RTA. The estimated result of $η$ signifies the evidence for selection into RTAs on unobservables. We have found negative and significant $\hat{η}$ , which rejects the exogeneity of $R T A_{i j}$ . Negative values signify that the country-level heterogeneity (i.e., bilateral economic and political factors which we have not included in our model) complementing the formation of RTA get on with unobservables which impact bilateral trade negatively. This negative self-selection, depending on the unobservables, creates downward bias in the estimated RTA parameter as we abandon the exogeneity assumption. Remaining indicator variables are fairly stable in column 3. The ATE of RTA is found to be 0.06, significant at 10 per cent level of significance. This indicates a 6.18 per cent significant increase in world trade due to RTA during the period under consideration. These findings are in line with international studies done by Baier and Bergstand (2004), Egger et al. (2008, 2011), etc.

Even though the trade agreements have significant trade-enhancing effects on the partner countries, the negotiations face some challenges. This we have delineated in our next section.

Challenges of RTAs

The negotiators of regional trading arrangements face several challenges due to their second-best nature as their balance of costs and benefits is not a given. This can be dealt with in economic analysis through the measurement of creation and diversion of trade. The negative effect on world trade efficiency is coined as ‘trade diverting’ for a partner country, where controversies lie in the potential welfare-decreasing effect by switching from a relatively low-cost producer to high-cost product. Nonetheless, Krugman has argued about low welfare loss from trade diverting effects as geographically close countries that are already engaged in a sizeable amount of trade are more intending to go for trade negotiations.

Many multilateral discussions are taking place on the market access rules of the non-member countries of any RTAs which generates potential discrimination over the member countries. Further discriminatory effects are arising from it in the areas of regulatory harmonisation among RTA member countries complicating the accretion to further RTAs with divergent rules (World Trade Organization, 2011).

Further, the growing number of RTAs is caused by the new regionalism under which regional agreements often involve deep integration. The problem of measurement becomes further complicated because increasingly deeply integrated RTAs have now included services and other elements. But the analysis of effectiveness and implementary effects of these deeper RTAs are complex as data are poorer along with less developed analytical tools.

Another challenge of RTAs is related to the RoO. RoOs have grown manifold along with the proliferation of trade agreements. Many economists have advocated them as distortionary due to their product-by-product criteria and for overlapping presence in the same member, country is corresponding to different RTAs. In this context, Estevadeordal and Suominen (2005) and Krishna and Krueger (1995) argued that the changing process of production to meet RoO requirements majorly distorts patterns of trade and investment flows of member countries. The size and complexity of RoO across markets are largely addressed in existing empirical works which also provide their ad valorem tariff equivalent estimate. Particularly in the case of North-South agreements, it reveals that they are quite extortionate. According to Cadot et al. (2005), RoOs have reduced the gains to Mexico from NAFTA almost by 50 per cent. Cadot and de Melo (2008) indicated that compliance costs related to the fulfilment of RoO requirements range from 3 to 5 per cent of final goods prices, which primarily offset the small preferences earmarked to high-income countries.

Further, the spaghetti-bowl nature of overlapping RTAs generates internal friction. This creates transaction and administrative costs for firms that operate across various RTAs. These costs could be higher or lower than their normal expected level if operating under a uniform set of trade rules. In this context, RoO protocols are a case in point. Estevadeordal et al. (2009) mentioned a study in their report. The studies of the Asian Development Bank (ADB) and Inter-American Development Bank (IDB) have reported that in countries like Peru, Singapore, Mexico and Thailand, approximately 60–80 per cent of their large firms prefer a single agreement pattern or mega-regional agreements with a common set of RoO over the current type of multiple and overlapping trade agreement patterns. These existing trading patterns also generate complex and troublesome custom procedures to authenticate the RoO of countries.

Concluding Remarks

RTAs have proliferated during the last two decades or so. The existing empirical literature has shown little economic benefits of those RTAs irrespective of the economic model choice. It can be argued that very often, RTA is a consequence of the growing trade between the partner countries, so it exemplifies the de facto justification of trade relations. This self-selection of RTAs is more a political decision rather than an economic decision. Thus, endogenous selection into RTA in a political economy framework can revise the estimates of its economic benefits. In this analysis, we have found that RTAs have a significant and positive impact on trade flows after considering their endogeneity.

However, these estimates are rather simplistic and may suffer from an upward bias as they do not consider the country-level heterogeneity while implementing the agreements. Practically, RTAs face some challenges during their implementation, which are more prominent in the case of SS cooperation as there are too many overlapping RTAs operating simultaneously. Further, over time, RTAs have moved from a shallow integration to a deeper integration which has to make the agreement more complex. In any econometric analysis, it is very difficult to capture these challenges separately due to the non-availability of appropriate proxies. Nevertheless, the fixed effect estimation process can absorb these unobserved heterogeneities collectively and give us an unbiased estimate of RTAs’ impact on trade flows. A future direction of work can be suggested by incorporating country-level heterogeneity of RTAs to get further revised estimates of its impact on trade flows.

Appendix 1 Descriptive Statistics

Variables	Description	Unit	Obs	Mean	Std. Dev.	Min	Max
$ln X_{i j}$	Log of exports, current $	$ million	22,684	17.872	4.052	0.000	26.588
$R T A_{i j}$	RTA enforced between partners	Dummy	22,684	0.188	0.391	0.000	1.000
$ln (Y_{i})$	Log of GDP of reporter (million), current $	$ million	20,005	26.217	2.121	18.605	30.338
$ln (Y_{j})$	Log GDP of partner (million), current $	$ million	20,017	25.610	2.470	18.462	30.338
$ln D_{i j}$	Log of bilateral distance	km	22,684	8.559	0.920	4.088	9.883
$c o n t i_{i j}$	Two countries are contiguous	Dummy	22,684	0.042	0.202	0.000	1.000
$c o m_l a n g_{i j}$	Two countries share official language	Dummy	22,684	0.099	0.299	0.000	1.000
$c o m_l a n g_e t h n o_{i j}$	Two countries share common language based on the fact that at least 9% of each of the country speaks	Dummy	22,684	0.093	0.291	0.000	1.000
$c o m_c o l o n i s e r_{i j}$	Two countries have had a common coloniser after 1945	Dummy	22,684	0.042	0.200	0.000	1.000
$c o l_a f t_1945_{i j}$	Two countries have had a colonial relationship after 1945	Dummy	22,684	0.016	0.127	0.000	1.000
$s a m e_c o u n t r y_{i j}$	Two countries were/are the same country	Dummy	22,684	0.014	0.119	0.000	1.000
$g e_{i j}$	Absolute difference in government effectiveness score of partner countries	Index	22,684	1.196	0.825	0.000	3.875
$p s_{i j}$	Absolute difference in political stability score of partner countries	Index	22,684	1.081	0.831	0.000	4.176

Footnotes

Acknowledgements

The author would like to thank one referee of the journal for useful comments. Usual disclaimers apply.

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The author received no financial support for the research, authorship and/or publication of this article.

Notes

References

Adams

, Dee

, Gali

, & McGuire

(2003, May). The trade and investment effects of preferential trading arrangements—Old and new evidence (Productivity Commission Staff Working Paper). Australian Government, Productivity Commission.

Aitken

N. D.

(1973). The effects of the EEC and EFTA on European trade: A temporal cross-section analysis. The American Economic Review , 63(5), 881–892.

Anderson

J. E.

(1979). A theoretical foundation for the gravity equation. American Economic Review , 69(1), 106–116.

Anderson

, & van Wincoop

(2003). Gravity with gravitas: A solution to the border puzzle. American Economic Review , 93(1), 170–192.

Anderson

J. E.

, & Yoto

V. Y.

(2010). The changing incidence of geography. American Economic Review , 100(5), 2157–2186.

Bagwell

, & Staiger

(2010). The World Trade Organization: Theory and practice. Annual Review of Economics , 2 (1), 223–256.

Baier

S. L.

, & Bergstrand

J. H.

(2004). Economic determinants of free trade agreements. Journal of International Economics , 64(1), 29–63.

Baier

S. L.

, Bergstrand

J. H.

, & Vidal

(2007). Free trade agreements in the Americas: Are the trade effects larger than anticipated? World Economy , 30(9), 1347–1377.

Baldwin

R. E.

(1989). The political economy of trade theory. Journal of Economic Perspective , 3(4), 119–135.

10.

Baldwin

(2008). Managing the noodle bowl: The fragility of East Asian regionalism. The Singapore Economic Review , 53 (3), 449–478.

11.

Baldwin

R. E.

, & Venables

A. J.

(1995). Regional economic integration. In

Grossman

G. M.

, & Rogoff

(Ed.), Handbook of international economics (Vol. 3, pp. 1597–1644). Elsevier.

12.

Bergstrand

(1985). The gravity equation in international trade: Some macroeconomic foundations and empirical evidence. The Review of Economics and Statistics , 67(3), 474–481.

13.

Berthelon

(2004). Growth effects of regional integration agreements (Central Bank of Chile Working Papers No. 278).

14.

Beugelsdijk

, & van Schaik

A. B. T. M.

(2001). Social capital and regional economic growth (CentER Discussion Paper 2001-102). Tilburg University. https://research.tilburguniversity.edu/en/publications/social-capital-and-regional-economic-growth

15.

Bhagwati

(2008). Termites in the trading system: How preferential agreements undermine free trades . OUP.

16.

Bhagwati

, & Panagariya

(1996). Preferential trading areas and multilateralism strangers, friends, or foes?

Bhagwati

, & Panagariya

(Eds.), The economics of preferential trade agreements . The AEI Press.

17.

Burfisher

M. E.

, Robinson

, & Thierfelder

(2002). Developing countries and the gain from regionalism. American Journal of Agricultural Economics , 84(3),736–748.

18.

Cadot

, & de Melo

(2008). Why OECD countries should reform rules of origin. World Bank. https://openknowledge.worldbank.org/handle/10986/4416

19.

Cadot

, Carrère

, Melo (De)

, & Portugal-Pérez

(2005). Market access and welfare under free trade agreements: Textiles under NAFTA. World Bank Economic Review , 19(3), 379–405.

20.

Carrere

(2006). Revisiting the effects of regional trade agreements on trade flows with proper specification of the gravity model. European Economic Review , 50(2), 223–247.

21.

Cheng

I.-H.

, & Wall

H. J.

(2005). Controlling for heterogeneity in gravity models in trade and integration. Federal Reserve Bank of St. Louis Review , 87(1), 49–63.

22.

Coulibali

(2007). Evaluating the trade effect of developing regional trade agreements: A semi-parametric approach (World Bank Policy Research Working Paper 4200). World Bank.

23.

Coulibaly

(2009). On the complementarity of regional and global trade. World Bank. https://openknowledge.worldbank.org/handle/10986/9181

24.

Egger

(2000). A note on the proper econometric specification of the gravity equation. Economics Letters , 66(1), 25–31.

25.

Egger

, & Pfaffermayr

(2003). The proper panel econometric specification of the gravity equation: A three-way model with bilateral interaction effects. Empirical Economics , 28(3), 571–580.

26.

Egger

, Egger

, & Greenaway

(2008). The trade structure effects of endogenous regional trade agreements. Journal of International Economics , 74(2), 278–298.

27.

Egger

, Larch

, Staub

K. E.

, & Winkelmann

(2011). The trade effects of endogenous preferential trade agreements. American Economic Journal: Economic Policy , 3(3), 113–143.

28.

Estevadeordal

, & Suominen

(2005). Rules of origin in preferential trading arrangements: Is all well with the spaghetti bowl in the Americas? Economíca , 5(2), 63–103.

29.

Estevadeordal

, & Suominen

(2009). Bridging regional trade agreements in the Americas . Inter-American Development Bank.

30.

Haggard

, & Kaufman

R. R.

(1995). The political economy of democratic transitions . Princeton University Press.

31.

Heckman

(1976). Simultaneous equation models with continuous and discrete endogeneous variables and structural shifts. In

Goldfeld

& Quandt

(Eds.), Studies in nonlinear estimation (pp. 235–272). Ballinger.

32.

Heckman

(1978). Dummy endogenous variables in a simultaneous equation system. Econometrica , 46(4), 931–961.

33.

Heckman

J. J.

(1997). Instrumental variables: A study of implicit behavioral assumptions used in making program evaluations. Journal of Human Resources , 32(3), 441–462.

34.

Helpman

, & Krugman

(1995). The politics of free trade agreements. American Economic Review , 85 (4), 667–690.

35.

Hillman

A. L.

, & Riley

J. G.

(1989). Politically contestable rents and transfers. Economics and Politics , 1(1), 17–39.

36.

Kandogan

(2008). Consistent estimates of regional blocs trade effects. Review of International Economics , 16(2), 301–314.

37.

Kaufmann

, Kraay

, & Mastruzzi

(2010). The worldwide governance indicators: Methodology and analytical issues. World Bank policy research working paper , (5430).

38.

Krishna

(1998). Regionalism and multilateralism: A political economy approach. Quarterly Journal of Economics , 113(1), 227–251.

39.

Krishna

, & Krueger

(1995). Implementing free trade areas: Rules of origin and hidden protection (NBER Working Paper No. 4983).

40.

Krugman

(1979). Increasing returns, monopolistic competition, and international trade. Journal of International Economics , 9(4), 469–479.

41.

Krugman

(1980). Scale economies, product differentiation, and the pattern of trade. The American Economic Review , 70(5), 950–959.

42.

Krugman

P. R.

(1995). The move toward free trade zones. In

King

(Ed.), International economics and international economic policy : A reader (pp. 163–182). McGraw-Hill.

43.

Krugman

P. R.

, & Obstfeld

(2003). International economics: Theory and policy . World Student Series.

44.

Lee

J.-W.

, & Park

(2005). Free trade areas in East Asia: Discriminatory or non-discriminatory. The World Economy , 28(1), 21–48.

45.

Liu

(2008). The political economy of free trade agreements: An empirical investigation. Journal of Economic Integration , 23(2), 237–271.

46.

MacPhee

C. R.

, & Sattayanuwat

(2014). Consequence of regional trade agreements to developing countries. Journal of Economic Integration , 29(1), 64–94.

47.

Maddala

G. S.

(1993). Limited-dependent and qualitative variables in economics . Cambridge University Press.

48.

Maddala

G. S.

, & Lee

L.-F.

(1976). Recursive models with qualitative endogenous variables. Annals of Economic and Social Measurement , 5(4), 168–188.

49.

Magee

(2003). Endogeneous preferential trade agreements: An empirical analysis. Contributions to Economic Analysis & Policy , 2(1), article 15.

50.

Magee

(2004). Trade creation, trade diversion, and endogenous regionalism . Econometric Society 2004 North American Summer Meetings, Series Number 289.

51.

Mansfield

(1998). The proliferation of preferential trading agreements. Journal of Conflict Resolution , 42(5), 523–543.

52.

Martinez-Zarsoso

, & Nowak-Lehman

(2003). Augmented gravity model: An empirical application to MERCOSUR-EU trade flows. Journal of Applied Economics , 6(2), 291–316.

53.

Matyas

(1997). Proper econometric specification of the gravity model. The World Economy , 20(3), 363–368.

54.

Matyas

(1998). The gravity model: Some econometric considerations. The World Economy , 21(3), 397–401.

55.

Meade

J. E.

(1955) Trade and welfare . Oxford University Press.

56.

Nataraj

(2007). Regional trade agreements in the Doha round: Good for India? (ADB Institute Discussion Paper No. 67).

57.

Pant

, & Paul

(2018). Do regional trade agreements increase trade of India?—An econometric approach. Journal of Economic Integration , 33(3).

58.

Pusterla

(2007, January). Regional integration agreements: Impact, geography and efficiency (IDB-SOE Working Paper No. 1).

59.

Robinson

, & Thierfelder

(2002). Trade liberalization and regional integration: The search for large numbers. Australian Journal of Agricultural and Resource Economics , 46(4), 585–604.

60.

Rodrik

(1994). What does the political economy literature on trade policy (not) tell us that we ought to know ? (NBER Working Paper No. 4870).

61.

Sanso

, Cuairan

, & Sanz

(1993). Bilateral trade flows, the gravity equation, and functional form. The Review of Economics and Statistics , 75(2), 266–275.

62.

Schiff

, & Winters

L. A.

(2003). Regional integration and development . A co-publication of the World Bank and Oxford University Press.

63.

Shahid

(2011). India-Korea CEPA: An assessment. Korea and the World Economy , 12 (1), 45–98.

64.

Sharma

S. C.

, & Chua

S. Y.

(2000). ASEAN: Economic integration and intra-regional trade. Applied Economics Letters , 7(3), 165–169.

65.

Thiratayakinant

K. L.

(2010). Multilateral supervision of regional trade agreements: Developing countries’ perspectives (Doctoral Dissertation). London School of Economics, ProQuest LLC 2014.

66.

Tinbergen

(1962). Shaping the world economy: Suggestions for an international economic policy . Twentieth Century Fund.

67.

Trefler

(1993). Trade liberalization and the theory of endogenous protection: An econometric study of U.S. import policy. Journal of Political Economy , 101(1), 138–160.

68.

Vella

, & Verbeek

(1999). Two-step estimation of panel data models with censored endogenous variables and selection bias. Journal of Econometrics , 90 (2), 239–263.

69.

Vicard

(2009). On trade creation and regional trade agreements: Does depth matter? Review of World Economy , 145, 167–187.

70.

Viner

(1950). The customs union issue . Carnegie Endowment for International Peace.

71.

Winters

L. A.

, & Chang

(2000). Regional integration and import prices: An empirical investigation. Journal of International Economics , 51(2), 363–377.

72.

Wonnacott

, & Lutz

(1989). Is there a case for free trade areas? In

Schott

J. J.

(Ed.), Free trade areas and US trade policy . Institute for International Economics.

73.

Wooldridge

J. M.

(2002). Econometric analysis of cross section and panel data . The MIT Press.

74.

World Trade Organization. (2011). The WTO and preferential trade agreements: From co-existence to coherence . World Trade Report.