The Impact of Bilateral Investment Treaties on FDI Inflows Into India: Some Empirical Results

Introduction

International investment treaties or as they are broadly referred to as international investment agreements (IIAs) allow individual private foreign investors to move legally against host states when they feel that the activities of the host state have had an adverse impact on their investment. The basis for such action lies in the investment treaties signed between states—such treaties can be in the form of bilateral investment treaties (BITs) or sometimes as a chapter of a bilateral or multilateral trade agreements between countries. These treaties typically consist of three parts (Dolzer & Schreuer, 2012): with the first part defining investment for the purpose of the treaty, the second listing substantive standards against which violations by the state in question are measured and the third identifies the dispute resolution mechanisms (often referred to as investor-state dispute settlement or ISDS) identifying the nature and scope of adjudicating international tribunals. In many quarters it has come to be felt that international investment treaties have been constructed to privilege investor rights over state rights that enable the regulation of economic activity and there have been several worldwide moves to counter this (see, for instance, Dhar et al., 2012; more generally the work of legal scholars associated with the TWAIL ¹ movement such as Chimni, 2017;also see Lavopa et al., 2013).

Turning to the Indian involvement in this system, commencing in the 1990s, India signed several BITs and by 2016 it had one of the largest investment treaty arrangements in the world, having signed 83 BITs and 13 other IIAs. India probably signed these treaties to signal that it was a reform-minded country that was seeking the inflow of foreign capital to develop its economy, without paying too much attention to the terms imposed by the treaties. About two decades after India signed its first BIT, much to the consternation of the Indian Government, starting around 2012 around a dozen or so cases came to be filed by foreign investors against the Indian state. Since the arbitral awards have awarded large damages against the Indian state, India has come to change its position with respect to international investment treaties—in 2017 it denounced the bulk of the treaties it had signed ² and furthermore insisted that all BITs must be renegotiated using a template provided by the new Model Treaty. The new Indian Model Treaty, which was finalised in December 2015, substantially privileges state rights over investor rights, reversing the orientation of the older treaties. This change in emphasis has meant that only a handful of new treaties have been signed by India. These include two BITs signed with Taiwan and Belarus in 2018, one treaty with the Kyrgyz Republic in 2019 and in 2020 India signed an Investment and Facilitation Treaty with Brazil. ³ India has also signed a BIT with Cambodia which is not in operation yet. There is little chance that these treaties will influence an inflow of foreign direct investment (FDI)—none of these countries are the usual major exporters of capital.

It is hard to predict the future impact of this new set of IIAs on FDI in India because currently we do not have sufficient points of data to discern any trends. However, we can get a sense of the impact of BITs by empirically studying their effect on foreign investment inflows into India over the period they were in force—that is, before they were denounced. It is important to fathom this effect because it helps us comprehend the role of institutional structures in governing individual investor risk in cross border investment. The advent of the new regime has perhaps subtracted this element of support for foreign investors, and the primary purpose of this study is to detect the magnitude and nature of the effect of IIA protection on the inflow of FDI into India. The analysis of the impact of IIAs has typically been performed using large cross-country data sets—our work here is distinct in that we try to capture the effects of IIAs on FDI inflows specifically into India.

We begin in Section 2 by providing a brief background and explaining the context of our study and then proceed to describe the model we use to pursue our empirical investigation in Section 3. Section 4 presents the results and discusses the implications of our findings. Section 5 concludes the paper with some comments that look out to the future in light of our results.

Background and Context

As noted earlier, starting around 1991 India started signing a series of BITs and continued to do so over the next two decades. For several years, the treaties were not invoked by foreign investors except in a set of Enron related cases around 2003 where matters were settled for an undisclosed amount by 2004 (Kundra, 2008). In 2010 an Australian Coal mining firm White Industries, using the most-favoured-nation (MFN) clause in the Australia-India BIT invoked a clause in the India-Kuwait BIT to sue India and ended up receiving AUS$11 million in 2011. ⁴ After this a slew of cases invoking BITs came up against India, many of which are still pending final awards. ⁵ To illustratively point to one recently decided high-profile case involving action against India by Deutsch Telekom, an arbitration tribunal awarded $1 billion damages in favour of the company. ⁶ India has challenged the award. This and the many other similar cases against the Indian state have been perceived as not only imposing financial costs but also regulatory costs—inhibiting public policy by attacking the Indian judiciary, taxation policy and anti-corruption measures. The foremost reaction to this spate of cases has been to redraft the Model Treaty in 2015. The new Model Treaty works by giving the sovereign state much stronger rights than was the case earlier—it is structured to grant the government immunity against any liability for decisions it may take regarding taxation, government procurements, compulsory licenses, subsidies and national security, as well as from decisions taken by the judiciary. Furthermore, in 2016 India took the position that all existing BITs would be denounced (terminated) in 2017 and it has since proceeded to carry out this plan. In addition to denunciation, many treaties have also been allowed to lapse, with India taking the stance that any future treaties have to follow the new Model Treaty. As the work by legal scholar Prabash Ranjan (2019) argues, the excessive tilt towards state rights in the new Model Treaty means that very few countries seeking to protect its investors are willing to agree to the stringent terms required by the Model Treaty. One can only speculate about the empirical effects of denunciation and/or renegotiation of international investment treaties at this stage because empirical evidence on the effects will be visible only with the passage of time. To understand the role of international investment treaties on foreign investment that is empirically defensible, one is perforce obliged to study their impact over a period when these treaties have explicitly governed investment flows.

There is by now a good deal of research that has tried to quantify the impact of IIAs on FDI flows. The early literature—for instance Hallward-Driemieier (2003) and United Nations Conference on Trade and Development (1998)—did not find much of an empirical link between international investment treaties and FDI. Most probably, these inconclusive findings were on account of the fact that these empirical investigations had not allowed a sufficient passage of time to lapse from the time the treaties had been signed so that discernible empirical effects were not evident. However, subsequently a series of more recent studies show that the signing and ratification of international investment treaties does have an impact on FDI flows (see, for instance, Armstrong & Nottage, 2016; Berger et al., 2013; Busse et al., 2010; Egger & Merlo, 2007, 2012; Neumayer & Spess, 2005). It is standard practice in this literature to work with large cross-country data sets but few studies look at the impact of IIAs on particular countries—a broadly comparable study to ours looking at the impact of BITs on inflows/outflows of FDI in China can be found in Chen et al. (2015). Our attempt to study the impact of international investment treaties on foreign investment seeks to understand whether these treaties have had an impact on foreign investment flows specifically in the Indian case or not. Of course, this task throws up challenges as to how to precisely model and estimate relationships to get answers to our queries—it is much easier to work with cross country data and conceptualise empirical models in such settings.

Empirical Model

Generally speaking, the literature on the links between international investment treaties and inflows of FDI use the tenants of the Gravity Model to model relationships. To briefly recapitulate the Gravity Model, the model is formulated on the premise that the volume of trade between two countries is proportional to the product of their mass (measured as gross domestic product) and is inversely proportional to the distance between them. This simple formulation is surprisingly empirically robust and over the years it has been found that a variety of theoretical foundations are compatible with the standard empirical specification of the model. The Gravity Model was brought to the fore by Tinbergen (1962) and it is by now extensively used in empirical work pertaining to trade (for an overview, see Feenstra, 2002, and also see Bergstrand, 1985). While initially conceived as a model to study trade of goods, the model has been successfully extended to study trade in services—among many works that look at the model in terms of services see Kimura and Lee (2006) and Lennon et al. (2009).

The Gravity Model has also been used to study the flows of FDI. There is a common feature that runs across this literature on FDI flows, whether specifically addressing the role of BITs or not that, which is important to note before attempting to establish the empirical significance of BITs for FDI inflows into India. These studies are oriented towards looking at FDI as an outflow, typically conceived of as a decision made by multinational enterprises of home countries looking to invest in host countries. With this basic frame in place, this genre of work typically asks whether the FDI is aimed at horizontally duplicating production activities in host countries or seeks to establish separate vertical activities downstream in host countries (for instance, see Braconier et al., 2005; Markusen & Maskus, 2002). Other questions usually investigated include whether the outflow of trade and FDI should be viewed as substitutes or as complements (Martínez et al., 2012). In addition to this, of course, some studies take on institutional questions such as the role of BITs and Free Trade Agreements (FTAs) in influencing FDI (in this context apart from references cited earlier see particularly Brenton et al., 1999).

As we seek to look at the role of BITs and FTAs on inducing FDI inflows to India, it is impossible to directly mimic these studies because they are oriented to studying outflows, trying to capture the forces that influence the decisions of multinational enterprises in the home country. One is tempted to apply the Gravity Model directly to the specific Indian case by trying to see the relationship between the inflows of FDI and the usual correlates suggested by the Gravity Model, such as gross domestic product and the distance between countries, and then going on to append institutional features. This is perhaps not the correct way to approach the issue because only a fraction of FDI outflow of home countries shows up as Indian FDI inflows and while this fraction can be thought of in terms of feeding a general Gravity Model, the causality fuelling the model will be picked up only very weakly if we were to force a Gravity equation directly on the FDI inflows into India.

To overcome these problems, we exploit one of the key findings suggested in the FDI-Gravity Model literature, namely that the empirical evidence suggests that FDI and trade tend to be complementary rather than substitutes. Thus, in the model we suggest that FDI inflows follow trade, and therefore we place a trade variable on the right-hand side of the equation. The volume of trade itself is presumably a product of the forces that make up the Gravity Model but additionally indicates the tenor of the unilateral policy regime, marking the degree of openness, exchange rate policy and capital control regimes. If the trade variable captures dynamic interactions between India and other nations alongside signalling aspects of the unilateral policy regime, a bilateral policy variable in the form of BITs and FTAs can be thought of as signifying the certainty with which investors can expect the protection of their rights.

Using this broad sense, we specify three empirical models to test the impact of BITs on FDI, as well as more broadly the impact of a system governed by international investment treaties on FDI.

The model specifications are:

Model 1

Y i t = α + β 1 + Y i t − 1 + β 2 X i t + β 3 B i t + β 4 C C i t + ε i t

Model 2

Y i t = α + β 1 + Y i t − 1 + β 2 X i t + β 3 B i t + β 4 C B I i t + β 5 C C i t + ε i t

Model 3

Y i t = α + β 1 + Y i t − 1 + β 2 X i t + β 3 B i t + β 4 C B P i t + β 5 C C i t + ε i t

where

Y_it is the inflow of FDI into India from country i in year t.

The three model specifications differ from each other, in that while Model 1 incorporates the presence of a BIT as a binary variable, Model 2 includes the total number of BITs signed by India in year t as an additional cumulative variable and Model 3 contains the total number of BITs signed by partner country in year t as the additional cumulative variable. Thus, each Model incorporates the BIT variable differently. The other variables reflect relationships that follow from our general discussion above. Following the observation that FDI follows trade in a complementary manner, we include the volume of bilateral trade as an explanatory variable. Capturing an institutional dimension of this relationship, the models also include a binary variable as to whether a trade agreement in the form of a CECA or CEPA has been signed with a trading partner or not. We have only included trade agreements of the CECA/CEPA variety because they have a chapter dedicated to investment that is more or less along the lines of a BIT. (It may be noted that standard trade agreements were not included for technical reasons discussed below.) Finally, all models incorporate the lagged inflow of FDI, which is commonly thought to influence current levels of similar investment.

Turning to the data sets used in our estimation, ⁷ to capture the inflow of FDI into India from country i in year t, we used FDI data published by the Department of Industrial Policy & Promotion, Government of India. Our bilateral trade data (total exports and imports from country i in year t) were obtained from the World Integrated Trade Solutions, World Bank database ⁷ . The information on the various international investment treaties signed by India was gathered from the website of the Ministry of Finance, Government of India.

Since we have a dynamic panel data set and are studying FDI inflows into India from its partner countries over time, it is likely that the model would have been subject country specific fixed effects, which might have led to endogeneity in the regression model. Additionally, since we expect the FDI inflow from country i in year t to be determined in part by the FDI inflows from country i in year t-1, in effect, it is likely that the error terms will be serially correlated. To correct for serial correlation and country-specific fixed effects, we use the generalised methods of moments (GMM) to estimate this model. We use the Arellano-Bond GMM estimator, which is commonly used for dynamic panel data. Furthermore, to correct for any heteroscedasticity, we report robust standard errors.

The results of our estimation across all three specifications can be seen in Table 1 and the figures for the post-estimation diagnostics can be seen in Table 2. Table 2 shows the results of the statistical test to check if the Arellano-Bond assumptions ⁸ were satisfied while estimating the models. Since the results of the test support the inference that errors are autocorrelated in order 1 but not in order 2, it can be concluded that using the Arellano-Bond GMM estimator was indeed appropriate for our estimation exercise. It is indeed the case that while the use of GMM to estimate parameters resolves issues of endogeneity; however, we further investigated the possibility of endogeneity by plotting the residuals against the independent variables and found no evidence of any correlation between them. Thus, we were able to rule out endogeneity, ensuring the robustness of our results. It also needs to be mentioned that while performing our estimation exercises (as stated above) we were forced to not include the variable as to whether India had signed a standard foreign trade agreement with the partner country because inclusion of this variable raised the problem of multicollinearity.

OPEN IN VIEWER

Table 1.

Parameters Estimated Using Generalised Method of Moments.

Serial No.	Variable	Model 1	Model 2	Model 3
1.	FDI Lagged (Yit-1)	0.122** (0.053)	0.078 (0.051)	0.096*** (0.051)
2.	Bilateral Trade (Xit)	0.508* (0.14)	0.206*** (0.122)	0.272** (0.11)
3.	BIT Signed (Bit)	0.271 (0.33)	–0.56 (0.432)	–0.589 (0.441)
4.	Cumulative BITs signed by India (CBIit)		0.034* (0.007)
5.	Cumulative BITs signed by partner country (CBPit)			0.068* (0.01)
6.	CECA/CEPA Signed (CCit)	1.88* (0.67)	1.16*** (0.616)	1.56** (2.40)
7.	Constant	–1.81*** (-0.98)	–0.96 (0.72)	–2.95* (0.95)

Source: The authors.

Notes: *Significant at 1% level of significance; **significant at 5% level of significance; ***significant at 10% level of significance. The values in brackets are the robust standard errors.

OPEN IN VIEWER

Table 2.

Result of Diagnostic Test: Arellano-Bond Test for Zero Autocorrelation in First-Differenced Errors.

Order	Model 1		Model 2		Model 3
	z	Prob > z	z	Prob > z	z	Prob > z
1	–4.815	0.0000	–4.6501	0.0000	–4.6219	0.0000
2	1.3439	0.1790	1.0736	0.2830	1.2329	0.2176

Source: The authors.

Results

Concluding Comments: Looking out Into the Future

As we look out to the future, the Indian Government has stated, at the beginning of 2020, that it is thinking of replacing the BITs oriented protection of investment with a domestic investment law that is aimed at protecting foreign investments (Shah & Ahmed, 2020). To the extent this is tantamount to India shying away from international law, this is an act that goes against improving the system of governance of investment flows globally (Ranjan, 2020). In fact, the current Indian government has, on many occasions, voiced complete antipathy to any multilateral governance of international investment. Instead, the faith seems to reside in soliciting foreign investment, supported by indices such as an escalation in the ease of business ranking. However, mere improvement in this kind of ranking may not be sufficient to encourage substantive FDI. One way of interpreting our results—that the collective presence of an overall investor protection is positively and significantly linked to foreign investment flows—is to invoke support for the view that the risk borne by individual foreign investors is quite importantly served by ISDS protection, rather than the many other values that go into the construction of economy wide indices. Our empirical study shows that over a period when the ISDS protection was in place, though India may have had to confront some adverse rulings against its regulatory actions, the overall participation in a system governed by IIAs did influence the inflow of FDI positively. This in itself tells us that in the future we should be supporting multilateral international institutions that benefit the global economy—institutions that can simultaneously take away the negative aspects of the current ISDS system that overly impose restrictions on the regulatory space of host states but retain overall international law protection to cross country investors. As we confront the current world-wide pandemic crisis, it may be useful to think of the world that will emerge after the crisis—if the world is to be a better place, then multilateral institutions that are able to govern across borders are going to be an essential ingredient of the new order—or alternately we will all have to work with a more insular economy.