Importing Institutions: The Political Economy of BITs and ...ipec.gspia.pitt.edu/Portals/7/Papers/Importing Institutions The... · Prepared for the Conference on Globalization and

Importing Institutions: The Political Economy of BITs and

FDI Flows∗

B. Peter Rosendorff

New York University †

December 21, 2010

INTRODUCTION

Institutions matter for development. The institutions that protect private property from

expropriation, or excessive taxation of effort, income or profit are fundamental to economic

growth and prosperity (Acemoglu, Johnson and Robinson, 2001). Yet many states find it

difficult, if not impossible, to build, maintain and entrench these crucial institutions in the

fabric of the polity and the economy.

Barriers to the formation and entrenchment include may be political or historical in origin.

Historical barriers to institutional development are likely when norms of property rights

protection have failed to take hold, or the state has developed and emerged without sufficient

capacity to protect property rights. Colonial history, legal systems (and the structure of the

judiciary), ethnic conflict have all been identified with weak states and weak institutional

development, failing to adequately reign in the extractive power of the state. Institutional∗Prepared for the Conference on Globalization and the Politics of Poverty and Inequality at the Indian

Institute of Management Bangalore, January 5-6, 2011.†Wilf Department of Politics, 19 West 4th Street, New York, NY 10012. Thanks to James Hollyer and

Kong Joo Shin for their help and advice and to Tim Buthe and Jennifer Tobin for sharing their data. Thisis a very preliminary draft - all comments are welcome.

1

structures that have emerged over long periods have a tendency to continue to survive,

making institutional change difficult or unlikely.

Political barriers to reform may be distributional - new rules require cooperation of en-

trenched interests to be implemented. Reforms may be blocked or delayed, because groups

with influence can’t agree on how the costs associated with the new institutional structure

should be allocated. (Alesina and Drazen, 1991), or they don’t know ex ante if they will

bear the costs of institutional shift (Fernandez and Rodrik, 1991). Alternatively the politi-

cal barriers have to do with the structure of the polity. Leaders constrained, say by multiple

veto players, or accountable to a large electorate, will find that changing rules, or building

institutions, or changing norms very difficult to achieve, requiring the consent of various

political interests (Tsebelis, 2002). Alternatively, leaders with no veto players, and maximal

discretion could attempt to build situations that foster investment and development - but

such a leader suffers from a credibility problem. Just as easily as the rules were written,

the judges appointed, the tax rates set, these could be overturned and reversed. Knowing

this, potential investors may shy away from investing in what is perceived to be a highly

unstable environment (Drazen and Masson, 1994).

Internal institution building is likely to be blocked, compromised, and prevented by all

these internal (and external) constraints. Leaders however, often have more autonomy

or discretion when it comes to international agreements. International agreements often

incorporate, build or establish a set of rules, norms and behaviors that are considered ac-

ceptable, and by implication specifies those behaviors to be deemed non-compliant. These

agreements regulate international interactions - they generate focal points, coordinate ex-

pectations, eliminate or reduce incomplete information, offer commitment devices etc. By

signing such an international agreement, the leader effectively imports the institutions that

can enhance development; these institutions are adopted, having the jumped the internal

barriers that prevented their emergence domestically.

Bilateral investment treaties (BITs) are devices for institution-poor countries to import

2

institutions that can credibly commit the host governments to minimizing arbitrary and

capricious policy shifts, punitive tax rates, and outright expropriation. Not only do rules

governing property rights and tax treatment get adopted, but violations of those rules have

consequences at the international level - enhancing the credibility of the government to

limit discretionary and arbitrary changes in policy, and thereby encouraging higher levels

of investment.

BITs are legal instruments signed between states that take on the force of international

law, and govern the rights and and obligations of states that host foreign capital within their

jurisdictions. In ratifying a BIT, a state incorporates the terms of the treaty as part of its

legal system (Salacuse and Sullivan, 2005). Most significantly however, and in contrast to

international trade agreements in particular, recent BITs provide for private standing - they

allow individual firms to initiate binding arbitration proceedings at an international tribunal

without the consent of either state’s government, making enforcement the responsibility of

the parties most affected by violations of the agreement.

To varying degrees, BITs provide a compelling mechanism to credibly import a set of

institutions that commit a state not to expropriate, over-regulate, over-tax, or otherwise

excessively interfere in the market, and endangers the signatories with “swift, substantial

compensation” in the instance of violation.

This paper explores two questions simultaneously. We explore the determinants of a

state, desirous of foreign capital, to decide to accede to a BIT. We develop (and test) a

theory that states with weak domestic institutions are indeed the states most likely to sign

a BIT - exactly because they seek to import stronger property rights protections than they

are able to adopt at home. Secondly, once we have a theory of BIT accession, we are in a

position to address the effect of BITs on FDI flows, taking account of the fact that states

have self-selected into these treaties in the first place. We find that when controlling for this

endogenous selection into the set of countries that have signed, the effect of BIT accession

is indeed to increase the inflows of FDI for those countries most in need of an improved

3

institutional environment.

LITERATURE

Institutions matter for economic performance. So have political scientists and economists

argued since North (1989, 1990), Olsen (1993), Coase (1960) , Williamson (1985) and others.

The state has a crucial role in enabling and facilitating the private contracts that enable

private exchange to occur. But simultaneously, the power to enforce contracts comes with

the risk that state can use that power to transfer resources from one group to another.

A successful state needs institutions to permit trade, while at the same time must have

adequate checks and balances to against expropriation by the government.

Consider a policymaker in an institution-poor country facing a potential foreign investor.

The policy-maker understands that in order to make more efficient use of domestic resource

endowments, foreign capital may be necessary. Such a policymaker faces a time consistency

problem however, for after the FDI has arrived, there are ex post incentives to expropriate

the investment and redistribute it domestically for the purposes of political survival. This

“hold-up” results in less FDI actually occurring, for the investor is reluctant to put capital

at risk in such an environment (Drazen, 2000).

Domestic institutions that enforce property rights are characteristic of richer nations.

Such rights not only allow the investor to reap the rewards from that investment, but help

create a stable business environment in which more domestic and international investment

can occur. Reductions in the risk associated with arbitrary expropriation reduces the costs

of investment and makes more economic activity feasible. Simply putting in place the

domestic legal systems and institutional structures that treat all investment equally is a long

and costly process that may not always be in the political interests of leaders accountable

to special interests.

Leaders have considerably freer reign when it comes to acceding to an international treaty.

4

These international international institutions provide opportunities to more credibly enforce

property rights protections and to reduce the transaction risk associated with FDI. Bilateral

investment treaties provide one such avenue, and guarantee a high standard of treatment,

legal protection under international law, and access to international dispute resolution.

BITs offer precision of obligations along a variety of dimensions crucial to lowering the

transactions costs of foreign investment: they require a well-defined standard of treatment,

the free transfer of funds and repatriation of capital and profits, transparency of national

laws, equal treatment across investors, compensation for war and other civil disturbances.

Most significantly, they offer dispute-settlement provisions that permit both investor and

state standing 1

The innovation that has given the BITs their bite is that both investor-state and state-

state disputes can be brought before an international tribunal for adjudication. Such bodies

include the World Bank Group’s International Center for the Settlement of International

Disputes (ICSID), or the International Chamber of Commerce (ICC). The United Nations

Commission on International Trade Law (UNCITRAL) has a framework document that

can govern arbitrations but does not operate an arbitration institution. The basis in in-

ternational law for the enforcement of arbitral decisions is provided by the 1958 New York

Convention on the Recognition and Enforcement of Foreign Arbitral Awards.

There is considerable debate in the literature as to whether FDI is enhanced by the pres-

ence of BITs. Early studies established a “conventional wisdom” in which BITs appeared to

have little effect on FDI (Vandevelde, Aranda and Zimmy, 1998). Tobin and Rose-Ackerman

(2005) followed the early studies, and confirmed the overall negative finding, studying US

FDI flows to developing countries. They found little effect of BITs on FDI. Salacuse and

Sullivan (2005) find that US BITs do increase FDI inflows (subsequently modified by Haftel1Aside from BITs, there are other instruments of international law that have some of these investment-

protecting features, such as Trade and Investment Framework Agreements, Investment Guarantee Agree-ments, protections embedded in Preferential Trade Agreements, Friendship, Navigation and Commercetreaties and others (UNCTAD, 2000).

5

(2008)); while Gallgher and Birch (2006) find the opposite result. These studies were all

dyadic in design - studying the FDI flows between signatories of BITs; monadic studies -

looking at total FDI flows into signatories with and without BITs, generated an equally

confused picture. Tobin and Rose-Ackerman, using 5 year averages of FDI flows could find

no significant effects of BITs; Neumayer and Spess (2005) on the other hand looking at large

sample of 120 developing countries between 1970 and 200 found a positive and significant

effect for the number of BITs on FDI flows. Yackee (2007) expands the Neumeyer and

Spess dataset, recodes BITs by weighting them by size of the capita-exporting countries,

and counts broader treaties as BITs if they include an investment related chapter. Yackee

find no relationship between BITs and FDI. Finally, Buthe and Milner (2009) look at non-

OECD countries’ inflows of FDI as a percentage of GDP, and find a positive correlation

with the number of BITs that are signed. Recently, Jandyhala, Henisz and Mansfield (2007)

have a comprehensive study in which they find that BITS do affect FDI when the BIT is

between a lesser developed and more developed pair, rather than between two poor or two

rich countries.

All these empirical approaches however suffer from the same problem of inference - the

treatment of all countries in the sample as having the same propensity to sign BITs in the

first place (or that propensity varies in a random fashion). In fact, some countries are more

likely to sign these treaties, and the propensity to sign these treaties is correlated with

the dependent variable (FDI flows), and hence the error term in these regressions. That

is because states are self-selecting into the treatment, we have a problem of selection bias

that has not been adequately addressed2

In this paper we study the decision by a capital-importing country to sign a BIT. They

will do so when they desire to import “better” institutions to protect from excessive state2Most of these authors do study only developing countries, avoiding the obvious pitfalls of mixing the

behaviors of capital-importing with capital-exporting countries. Nevertheless, the motives of developingcountries to sign BITs will vary to the degree which they have incentives to import the institutions in thefirst place.

6

taking. Which countries will therefore sign BITs? Those with weak domestic institutions,

ex ante. Hence we have a theory of selection into treatment. This of course has important

consequences for the effects of BITS on FDI flows. Consider a state with very strong

domestic institutions: such state has little to gain from a BIT, and signing it will have little

effect on FDI flows; on the other hand, a state with weak institutions will sign in order to

import credibility - such a state will see a dramatic improvement in FDI inflows.

Empirically, one possibility would be to interact a quality measure of domestic institutions

with the presence or the absence of a BIT; this however still treats BITs signing as having

a constant and independent effect across countries. Instead in what follows, we build a

theoretical model of the decision to accede to a BIT, and explore the effect on the BIT on

FDI flows in equilibrium. Using the theoretical foundations that establish the determinants

of the BIT accession decision, we now have a theoretically sound basis for instrumenting the

number of BITS in the FDI equation. We start by building a theoretical model of selection

into BITs and FDI. We develop a number of theoretical predictions which we then bring to

the data.

A MODEL OF SELECTION IN BITS AND FDI LEVELS

We consider a game between a home firm and the host government, where the firm exports

capital k from home to host. It employs labor l in host and produces x in host according

to

x = x(k, l) = min{k, l}

It then brings the x it produced back home, and uses x as an input into the production

of y. It produces y at home according to

y =12

(ln s+ lnx)

where s might be thought of as skilled labor or human capital only available at home,

7

specific to the firm and in inelastic supply at price σ. The firm is a price taker. We assume

the world price of k is given and fixed, κ. That is capital that is exported to the host is

employed on world markets at world price. Labor in the host country is paid w. Let the

world price of of y be 1.

The host government applies a tax t on every unit of capital that is employed at home.

We model this in the “iceberg” form. That is for every k units of capital that is shipped

to the host country, only (1− t)k are available for production. The firm must still pay the

cost κ for each unit of capital it ships.

Profit for the firm: Π = 12s + 1

2 ln(min{(1 − t)k, l}) − κk − wl − σs. If we let K denote

“after tax” capital, we can write this as : Π = 12s+ 1

2 ln(min{(K, l})−κ K1−t −wl−σs. Note

that the firm must still pay the rental rate on the capital that is taxed away and no longer

available for production.

This form has the feature, that the higher is the tax on capital, the less capital is available

for production, and therefore the less labor the firm will employ.

Host government utility is G = tk + awl. That is the host government taxes the foreign

capital imported by the home firms produced. It taxes the firm at rate t. The host gov-

ernment also values the employment generated by the foreign investment. If we assume, to

keep things simple, that the host country has no other production, wl is national income,

and hence the second term is the value to the host government of national income. It

weights these by a. We assume that a is private information - the host government knows

a, but home firm does not. We denote the home firm’s prior cdf over a as α(·). As α varies,

the political influence exerted by the social welfare concerns varies relative to tax revenues.

The greater is this variation, the more will tax rates jump around, as government tries to

balance these two interests. Hence the domestic political shock, α is our proxy for “political

risk”.

8

DOMESTIC INVESTOR PROTECTION

The general form of investor protection is a “promise” by the host government not to tax

the home firm at any rate higher than p. This “promise” can be codified in the tax laws,

it can be the rate at which domestic firms are taxed, or some other mechanism by which

the foreign firms have coordinated their expectations about the reasonable tax rate likely

to apply to their activities.

However the host state can renege on this “promise” and attempt to tax the foreign firm

at a rate that is higher than p, something it might do if its interests in social welfare are

sufficiently low (α is small)

The strength of the credibility of the promise is designated π. That is if the host gov-

ernment taxes the home firm at a rate t that is higher than p, with probability π, the host

government will be required to, and will, reimburse the home firm for the excess takings.

This probability, or strength of the credibility of the promise, is assumed to be monotonic

in the strength of the institutions that protect investment. These institutions include the

host country’s domestic legal system, the strength of its courts and judiciary in reigning in

excessive behavior by the executive. The set set of domestic legal institutions that protect

property rights as summarized by the statistic π, with more protections characterized by

larger values of π.

The Game with Protection

If the host country breaks its promise and applies a tax rate larger than p, the home firm

appeals to the relevant domestic institutions for arbitration. Let Pr(win|violation ) = π,

which is exogenous. If the plaintiff wins, the plaintiff pays the contracted tax rate under the

agreement, p and not the violation tax rate t. That is we assume no problem of enforcement

of an institutional finding. The sequence of moves:

1. nature reveals the value of a to the host government. This is private to the host.

9

2. The home firm chooses k.

3. The host government chooses t.

4. if t ≤ p, home firm employs local labor, production occurs and game ends.

5. If t > p, then foreign, host government is taken to court. Nature determines the

outcome of the case with Pr(win|violation) = π. If plaintiff wins, the tax rate reverts

to p; if not, the tax rate that is applied is t. The home firm then employs local labor

l, production occurs and game ends.

Notice that the firm makes its investment decision before it knows what the government

is going to do, with respect to the tax rate. Firm chooses how much capital and labor to

employ in the host country. Assume for the moment that the firm knows what the tax rat

t will be, and takes that as given. From standard microeconomic principles, the firm’s cost

function is

C(w, κ, x) = x

(κ

1− t+ w

)(1)

Then the contingent demand for capital and labor in this fixed proportions production

function is, by Shepard’s lemma:

k =x

1− tl = x

where x is any given level of output. We also know that l = k(1 − t). These demand

functions are contingent - this is the amount of capital and labor need to produce a given

output x. If the tax rate goes up, the firm must ship more capital since taxes are reducing

the amount of capital that can be put into production.

We begin by characterizing the equilibrium to the game with domestic property rights

protection. The proof is in the appendix.

Proposition 1. The equilibrium to the domestic protection game is t =

1 if a < 1w

0 if a > 1w

and x = 12

1−α+πα(κ+w(1−α+πα)) .

10

The government observes its private shock a. If a is low enough, the government has little

weight on the worker’s welfare in its payoff function, and instead chooses to expropriate for

private gain. It sets the tax rate at 1, above its promise of p. If instead, the political benefits

of public welfare are large enough (a large), the government forgoes the private benefits of

expropriation for the public benefit of social welfare.

The firm, on the other hand, must make its investment decision before the private in-

formation is revealed to the government, and before the government sets its tax rate. It

chooses a level of investment taking into account both the strength of the domestic institu-

tions π and the the likelihood of expropriation, captured by the ex ante probability that a

to low, α. It chooses the level of investment that maximizes the expected profit given these

two sources of uncertainty.

Notice that while π is the exogenous probability of property rights enforcement after

expropriation, ψ = 1− a+πa is the effective, equilibrium ex ante probability that property

rights will be enforced. That is after taking account of government’s incentive to expropriate.

For the purposes of the home firm, this is is the statistic that matters most: it tells them

the likelihood that, in equilibrium, their investment will be protected.

Also, some investment always occurs when there is some domestic protection; absent

protection, there are circumstances ( a < 1w ) where FDI does not occur.

STRENGTHENING PROPERTY RIGHTS PROTECTION

Recall that π is our measure of the strength of the domestic institutional property rights

protection. In this simple form of the game, the government’s decision to expropriate is a

function only of the domestic political pressure a, and not a function of strength of domestic

institutions, the government’s behavior does not change.

Lemma 1. FDI rises with improvements in property rights protection. That is dkdπ > 0.

11

Proof. Recall from Proposition 1 that x = 12

1−α+πα(κ+w(1−α+πα)) . Then

dx

dπ=

ακ

4 ((κ+ w) (1− α+ πα))2> 0

Now a rise in x requires a rise in k. Hence dkdπ > 0

A rise in the probability that any excessive takings will be reimbursed to the firm makes

the firm more willing, ex ante, to commit capital for investment. Clearly the firm prefers

a more certain or less risky investment climate. But what is in the best interests of a

politically astute policy-maker?

Does Government always want stronger property rights?

The host government’s payoffs in equilibrium:

G =

[tk + awk(1− t)] if t ≤ p

(1− π)[tk + awk(1− t)

]+ π[pk + awk(1− p)] if t > p

The government behavior in equilibrium is to set the tax rate to 1 if a < 1w , and zero

otherwise. Together these imply

G = k[aw(1− pα(w−1)) + α(w−1) (1− π + πp)

]Consider the government’s expected utility before the shock is revealed:

EG = k[aw(1− pα(w−1)) + α(w−1) (1− π + πp)

]dEG

dπ= kπ

[aw(1− pα(w−1)) + α(w−1) (1− π + πp)

]− kα(1− p)

d2EG

dπ2= kππ

[aw(1− pα(w−1)) + α(w−1) (1− π + πp)

]− kπkα(1− p)− kπα(1− p) < 0

Now kπ > 0, kππ < 0. But p < 1; hence there exists some value of π that is optimal for

the home government. Denote this optimal π as Π. Then

kπ[aw(1− pα(w−1)) + α(w−1) (1−Π + Πp)

]− kα(1− p) = 0

12

Note that Π < 1 for k large enough and/or kπ small enough. Otherwise Π = 1. In

what follows we assume that the politically optimal institutional quality is interior, i.e.

Π < 1.

We assume in what follows that Π is NOT a choice variable for the government in the sense

that is is structurally, or politically unable to unilaterally build the appropriate institutions

to generate the optimal degree of property rights enforcement.

INTERNATIONAL PROPERTY RIGHTS ENFORCEMENT

Bilateral investment treaties signed between the host and home governments have some

or all of the following characteristics.

1. Fair and equitable treatment, most favored nation treatment and national treatment:

A commitment not to tax foreign investment at rates larger than applied to domestic

firms, or firms from any other country. That is, national treatment and most favored

nation treatment.

2. Investment Dispute Settlement: Resort to international arbitration in the instance of

any dispute. We assume that international arbitration is binding in that a finding by

the arbitrator of a violation results in compliance by the host government.

We assume that as before, the maximal tax rate (treated as given by national treatment

or most favored nation treatment) is p. We assume that the probability of a finding by

the arbitrator in the instance of a violation is φ. That is Pr(win|violation) = φ, which is

exogenous. If the plaintiff wins, the plaintiff pays the national treatment tax rate, p and

not the violation tax rate t. That is as before we assume no problem of enforcement of an

institutional finding. The sequence of moves:

1. Host government chooses whether or not to sign a BIT, with institutional strength φ.

2. Nature reveals the value of a to the host government. This is private to the host.

3. The home firm chooses k.

13

4. The host government chooses t.

5. if t ≤ p, home firm employs local labor, production occurs and game ends.

6. If t > p, then foreign, host government is taken to court. If they have signed a BIT,

the court is the international arbitrator with strength φ; if no BIT is in effect, the

domestic courts are appealed to with strength π.

7. Nature determines the outcome of the case. If plaintiff wins, the tax rate reverts to

p; if not, the tax rate that is applied is t. The home firm then employs local labor l,

production occurs and game ends.

The structure of the game is identical with that of the previous section, with the addition

of a first move by the host government where it decided whether to sign a BIT. From the

analysis in the previous section the host will sign the BIT if the strength of the international

arbitrator φ lies closer to its ideal institutional strength Π, than do domestic courts.

Proposition 2. A host country signs a BIT if and only if |Π− π| ≥ |Π− φ|.

WHO SIGNS?

In what follows, let us make the reasonable assumption that international arbitration is

the strongest institution, i.e. Π < φ < 1. Now consider the set of potential host governments

indexed by the strength of their institutions, π ∈ (0, 1). Then there are four groups of host

countries: In the states with the lowest capacities to protect property rights, these leaders

have much to gain by importing the institutional quality of the international arbitrator, and

are most likely to sign the BIT. In the next group, the domestic property rights regime is

inadequate even in the eyes of the leader, but if the leader signs the BIT, the international

institutional strength may be “too” strong in that it drives the strength of the property

rights enforcement regime to a level that is worse for the leader than the imperfect domestic

regime. Hence the BIT is not signed, even though the leader would prefer a stronger property

rights regime. In the next group, the domestic regime is stronger than that desired by the

14

leader; hence there is no incentive to import an even stronger regime, so no BITs are signed.

In the rare and unlikely case which we shall ignore as unlikely, when the leader prefers a

weak regime, but the domestic regime is even stronger than the international courts, then

BIT is signed in an attempt to weaken the authority of the domestic courts. However, the

home, capital exporting country is unlikely to offer a BIT in this instance. So this case is

moot. Therefore we have

If π < 2Π− φ then BIT

If π ∈ [2Π− φ,Π] then no BIT

If π > Π then no BIT

Proposition 3. BIT signing is monotonic in domestic institutional quality. Only host

states with poor domestic property rights enforcement regimes (will be offered and) sign

BITS.

This can be seen in Figure 1 below. Low property rights enforcement regimes sign the

BIT; the higher types don’t.

BITS AND FDI FLOWS

We can now consider the effect of BITs on FDI flows. We can write xB = 12

1−α+(Bφ+(1−B)π)iα(κ+w(1−α+(Bφ+(1−B)π)α))

and B is an indicator function as to whether a BIT is signed (B = 1) or not (B = 0). That

is x1 is the firm’s output if a BIT is signed; x0 if not. Notice that x1 is not a function of

π, but that x0 rises with π. We will write x0 = x0(π). FDI flows are characterized by the

following proposition:

Proposition 4.

If π ≤ 2Π− φ then x = x1

If π > 2Π− φ then x = x0(π)

The diagram below, Figure 1, we plot the effects of BITS on FDI flows.

15

Fig. 1. FDI and BITs by the Strength of Domestic Property Rights Institutions

EMPIRICAL CONSEQUENCES

One observation from Figure 1 is immediate: the presence of a BIT does not necessarily

imply higher FDI flows. Consider two countries, one with a low π and one with a high

value of π. FDI flows could be very similar. Hence any empirical strategy that compares

countries that have signed BITs with ones that have not will be unlikely to find any effect

of BITs on FDI flows. And if they do find an effect of BITs on FDI flows, all they may

really be capturing is the effect of institutions on FDI, and not the effect of BITs.

This is true even if the empirical strategy attempts to control for the strength of domestic

16

institutions. The reason of course is that such an empirical approach fails to take account

of the fact that states are self-selecting into these treaties - selection is endogenous.

Our key theoretical results are that

1. BITs are signed by countries whose domestic institutions are weak.

2. FDI rises with BITs when domestic institutions are weak

3. FDI is unaffected by BITs when domestic institutions are strong

So institutions affect FDI levels directly and indirectly - institutional quality influences

the decision to sign a BIT; and conditional on NOT signing, institutions affect FDI directly.

If a state signs a BIT, domestic institutions matter less. BIT signing is therefore endogenous

to at least one of the independent variables - institutional quality.

In what follows we put these predictions to an empirical test, trying to take account of

the theoretically motivated problem of endogeneity.

EMPIRICAL INVESTIGATION

We consider 112 developing countries between the years of 1970 and 2004. The level of

analysis is the country-year, and we measure FDI by the inward flow in US dollars. All

dollar variables are at constant 2005 dollar values. The count of BITs by country and year

is from UNCTAD and a BIT is coded as present if it has entered into force by the year in

question. GDP and GDP per capita are also from the UNCTAD Investment Instruments

Online Database.

Our measure of institutions will be at first the standard polity score (from POLITY IV)

which captures the constraints on the political establishment. We will also use Polity’s

measure of executive constraints, and the law and order measure from ICRG Risk Ser-

vices.

17

Table 1. Descriptive Statistics

Variable N Mean Std. Dev. Min Max UnitsFDI 3155 796.29 3707.69 -4887.92 60630 2005 Constant (mill) $GDP per capita 2450 4054.35 4227.04 150.807 41000.23 2005 Constant $GDP 2472 107 314 0.163 4820 2005 Constant $ (bill)Polity 3256 -1.18 6.79 -10 10 -10: Autoc +10:DemocExecutive Constraints 3214 3.48 2.2 1 7 1: none 7: constrainedLaw and Order 1623 3.03 1.29 0 6 0:poor 6:goodBITs 3919 2.05 3.46 0 19 Count

Sources: BITs, FDI, GDP per capita and GDP are from UNCTAD; Polity and Executive Constraints arefrom POLITY IV; Law and Order is from ICRG.

BITs are signed by countries with weak institutions

Consider the first the effect of institutions on BIT signings. We have the summed count

of the number of BITs signed by each country in each year. If BITs are designed to replace

domestic institutions with stronger international institutions, then we should see, consistent

with Proposition 3 above, that those states with weak institutions are more likely to sign

more BITs. We run a Poisson regression of BITs on Polity and other covariates, with fixed

effects. The results are in Table 2.

Table 2 identifies that while not strongly robust, the preliminary empirical findings are

consistent with the model. That is, states with poorer institutions, proxied here by Polity,

are signing BITS less frequently. Note too that poorer countries are more likely to sign

more BITs than are richer countries. It seems on the face of it, that more FDI leads to

more BITS - this could of course be either that states sign more BITs when there is already

more FDI rather than less; alternatively, the causality is running the opposite way - the

BITs are leading to more FDI. And of course, in this regression FDI and BITs are both

endogenous to the institutional parameters.

Hence this preliminary table offers both preliminary evidence in support of the hypothe-

ses, but suffers from the very endogeneity problems the model is designed to help us ad-

18

Table 2. Determinants of BITs

Model 1 Model 2 Model 3 Model 4GDP Per Capita -0.00006** -0.00006** -0.00006* -0.00006***

(0.0001) (0.00001) (0.00001) (0.00001)GDP −2.3× 10−13*** −2.4× 10−12*** −2.0× 10−12*** −2.0× 10−13

(4.8× 10−13) (4.9× 10−13) (5.3× 10−13) (8.8× 10−14)Year 0.08*** 0.08*** 0.08*** 0.09***

(0.003) (0.003) (0.003) (0.004)FDI 0.00002*** 0.00002*** 0.00001 7.43× 10−6

(0.00000) (0.00000) (0.00001) (5× 10−6)Polity -0.052 -0.04**

(0.042) (0.013)Executive Constraints -0.00015 0.1**

(0.012) (0.04)Law and Order 0.086*** 0.08***

(0.012) 0.018N 1922 2055 1486 1286

* p < 0.05, ∗ ∗ p < 0.01, ∗ ∗ ∗p < 0.001. Poisson regression on a panel of all years and countries with fixedeffects. Standard errors in parentheses

dress.

Effect of BITs on FDI, Taking Endogeneity (More) Seriously

The predictions of the model are summarized in Figure 1: that countries with poor in-

stitutions will sign BITs; countries with better institutions will not; and that since FDI

responds to both institutions and BITs, and that BITS themselves are determined by in-

stitutions, and we should not expect to find much effect of BITs alone on FDI.

To deal with the fact that BITs are endogenous, we run an instrumental variable regres-

sion on the entire panel. We let BITs be instrumented by the institutional variables (polity

and executive constraints and law and order); we then explore the effects of the instrument

for BITs (interacted with institutional-quality) on FDI flows.

The model predicts that states with poor institutions sign BITs in order to import better

institutions. States with better institutions don’t bother to sign, and have large FDI flows.

Those states with poor institutions that do sign see a jump in FDI; those states with poor

19

institutions that do not sign BITs have low FDI.

To test this prediction, we do two things. Firstly, we interact the number of BITs signed

with the institution variable - to establish that poor institutions and BITS together have

an impact on FDI independent of BITS and institutions alone; and secondly, since BITs

is endogenous, we instrument for this variable using the institutions variable (and other

identifying variables) and use the instrumented values for BITs in the FDI regression.

We make some other adjustments to improve the fit. The number of BITs has been rising

steadily over time - so we generate a lowess smoothed version of the BITs variable and

use that to take account of the time trend in the selection equation. Then in the outcome

equation, we make use of decade specific dummies to control for the effects of time on FDI

flows. Finally we split the sample, where we examine democracies (polity ≥ 6) vs. non-

democracies in the first case, and in the second, we compare autocracies (polity ≤ −6) vs

non-autocracies.

Overall, for country i in year t, we estimate the outcome equation, using fixed ef-

fects:

FDIit = b0 + b1regimeit + b2regime ∗BITsit + b3BITsit + b4GDPit

+ b5GDPpercapitait + b6ExecutiveConstraintsit +∑d

γddecadeid + ε1it + ν1i

where the endogenous variable BITs is instrumented by the predicted values of the regression

(the selection equation):

BITsit = a0 + a1Polityit + a2GDPit + a3GDPpercapitait + a4ExecutiveConstraintsit

+ a5LawOrderit +BITLowessit + ε2it + ν2i

The central prediction of the model is captured in the coefficient on the interaction term,

b2. We expect to find that for societies with poor institutions - such as autocracies - when

they sign a BIT, there is a big jump in FDI. We expect therefore the coefficient on the

autocracy × BIT variable to to be positive and significant.

20

The model also predicts that for high quality institutional countries, the effect of a BIT

on FDI is zero, or insignificant. So we expect to find that the coefficient on democracy ×

BIT is insignificant. The results are in Table 3 below.

Table 3. Institutions and BITs Together Matter for FDI

FDI Model 1 Model 2 Model 3BITs -127.574 -278.934 -1089.918*

(272.46) (281.05) (759.40)GDP 0.000*** 0.000*** 0.000***

(0.00) (0.00) (0.00)GDP per capita 0.491*** 0.494*** 0.409***

(0.06) (0.06) (0.09)Executive Constraints -168.857 -198.906* -452.134**

(147.29) (138.05) (240.67)Polity 75.155 11.399 54.949

(52.88) (57.01) (56.90)Autocracy × BITs 288.958* 845.721**

(129.93) (435.17)Year 105.079 134.319* 266.299**

(98.46) (64.49) (142.08)Democracy × BITs 148.386 751.571

(195.99) (547.29)Constant -212473.390 -270294.868* -531035.942*

(195873.27) (127966.06) (281184.17)N 1407 1407 1407R2 0.67 0.64 0.68F 14.26*** 13.91*** 42.42***

* p < 0.05, ∗ ∗ p < 0.01, ∗ ∗ ∗p < 0.001. Instrumental variables regressionon a panel of all years and countries with fixed effects. Standard errors inparentheses. Decade dummies are suppressed.

The results bear out the core predictions of the model - that states with poor quality

institutions will import better ones by signing BITs with the express intent of improving

FDI. Countries with higher quality institutions have no need to import better institutions

- they don’t bother to sign, and the effect of any BITs that are signed by these countries is

insignificant.

21

The remaining variables have the expected signs. Poorer constraints on the executive

lead to less FDI; GDP and GDP per capita are positively related to FDI - richer countries

get more investment flows.

Table 3 presents the coefficients on the variables in the outcome equation - the determi-

nants of FDI flows. While not presented, we can make use of the coefficients in the selection

equation (the determinants of BITs) as a final check of the argument. In those regressions,

the polity variable is negative in all three cases and significant in two out of the three. That

is autocrats - societies with poor institutions protecting property rights - are more likely to

sign BITs - as compared with democracies.

CONCLUSION

BITs are signed by countries that need them the most - countries with poor domestic

institutions, environments where property rights are weak, where risk of expropriation is

high and where the transactions costs of doing business are large. The BIT is a simple

device to replace a set of less efficient norms and institutions with a credible guarantee of a

reduction in risk. Hence for these countries, the signing of BIT has positive effects on FDI

flows.

Those countries that don’t need better institutions, or for whom the international prop-

erty rights enforcement regime is too politically costly, don’t bother to sign. But their

investment climate is ex ante better, and FDI flows are not noticeably affected by the ab-

sence of a BIT. Hence for these states, any BITS that are signed do not have substantial

effects on FDI flows.

Interestingly, it is the autocratic states that are more likely to sign BITs in the first

place. In the context of the literature on international trade agreements this finding is

surprising. Democracies have been shown to be more cooperative when it comes to signing

preferential trade agreements, and are more likely to have lower barriers to trade than are

22

autocracies (Mansfield, Milner and Rosendorff, 2000, 2002). Here the effect of regime type

is the opposite of what has been established in that literature.

APPENDIX

Proof of Proposition 1: The equilibrium to the domestic protection game is t = 1 if a < 1w

0 if a > 1w

and x = 12

1−α+πα(κ+w(1−α+πα)) .

Firm chooses how much capital and labor to employ in the host country. Assume for

the moment that the firm knows what the tax rat t will be, and takes that as given. From

standard microeconomic principles, the firm’s cost function is

C(w, κ, x) = x

(κ

1− t+ w

)(2)

Then the contingent demand for capital and labor in this fixed proportions production

function is, by Shepard’s lemma:

k =x

1− tl = x

where x is any given level of output. We also know that l = k(1 − t). These demand

functions are contingent - this is the amount of capital and labor need to produce a given

output x. If the tax rate goes up, the firm must ship more capital since taxes are reducing

the amount of capital that can be put into production. The government learns its value of

a and can choose to apply a tax rate t ≤ p and receive the payoff in accord with complying

with the initial commitment; or the government can hold-up the firm by applying a higher

tax, and dragged to court for arbitration. The firm wins restitution for the excessive takings

with probability π.

EG =

[tk + awl] if t ≤ p

(1− π)[tk + awl

]+ π[pk + awl] if t > p

23

Now solving for optimal t∗. If t ≤ p ∈ (0, 1):

EGt = k − awk

t = 0 if a > 1w

That is, for t = 0 < p it must be that a > 1w If t > p:

EG = (1− π)[tk + awl

]+ π[pk + awl]

= (1− π)[tk + awk(1− t)

]+ π[pk + awk(1− p)]

EGt = (1− π)[k − awk]

t = 1

which occurs when a < 1w . Consider the behavior of the firm. The expected profit of the

firm is:

EΠ = (1− α(w−1))[12

ln s+12

lnx− κx− wx− σs]

+α(w−1)[(1− π)[−κx] + π[12

ln s+12

lnx(1− p)− κx− wx− σs]]

∂Π∂s

= ψ12

1s− σ = 0

s =ψ

2σ∂Π∂x

= (α− 1)(κ+ w − 1

2x

)− π α

(κ+ w − 1

2x

)+ ακ (π − 1) = 0

x =12

1− α+ πα

(κ+ w(1− α+ πα))

24

APPENDIX 2: TOTAL NUMBER OF BILATERAL INVESTMENT

TREATIES CONCLUDED BY INDIA, 1 JUNE 2010

Partner Date of signature Date of Entry in ForceArgentina 20-Aug-99 12-Aug-02Armenia 23-May-03 30-May-06Australia 26-Feb-99 4-May-00Austria 8-Nov-99 1-Mar-01Bahrain 13-Jan-04 5-Dec-07Bangladesh 9-Feb-09 —Belarus 26-Nov-02 23-Nov-03Belgium and Luxembourg 31-Oct-97 8-Jan-01Bosnia and Herzegovina 12-Sep-06 13-Feb-08Brunei Darussalam 22-May-08 18-Jan-09Bulgaria 29-Oct-98 23-Sep-99China 21-Nov-06 1-Aug-07Colombia 10-Nov-09 —Croatia 4-May-01 19-Jan-02Cyprus 9-Apr-02 12-Jan-04Czech Republic 11-Oct-96 6-Feb-98Denmark 6-Sep-95 28-Aug-96Djibouti 19-May-03 —Egypt 9-Apr-97 22-Nov-00Ethiopia 5-Jul-07 —Finland 7-Nov-02 9-Apr-03France 2-Sep-97 17-May-00Germany 10-Jul-95 13-Jul-98Ghana 18-Aug-02 —Greece 26-Apr-07 10-Apr-08Hungary 3-Nov-03 2-Jan-06Iceland 29-Jun-07 16-Dec-09Indonesia 10-Feb-99 22-Jan-04Israel 29-Jan-96 18-Feb-97Italy 23-Nov-95 26-Mar-98Jordan 30-Nov-06 22-Jan-09Kazakhstan 9-Dec-96 26-Jul-01Korea, Republic of 26-Feb-96 7-May-96Kuwait 27-Nov-01 28-Jun-03Kyrgyzstan 16-May-97 10-Apr-98Lao, PDR 9-Nov-00 5-Jan-03Latvia 18-Feb-10 —Libyan Arab Jamahiriya 26-May-07 —

25

Partner Date of signature Date of Entry in ForceMacedonia, TFYR 17-Mar-08 17-Nov-08Malaysia 1-Aug-95 12-Apr-97Mauritius 4-Sep-98 20-Jun-00Mexico 21-May-07 23-Feb-08Mongolia 3-Jan-01 29-Apr-02Montenegro 31-Jan-03 —Morocco 13-Feb-99 22-Feb-01Mozambique 19-Feb-09 —Myanmar 24-Jun-08 8-Feb-09Netherlands 6-Nov-95 1-Dec-96Oman 2-Apr-97 13-Oct-00Philippines 28-Jan-00 29-Jan-01Poland 7-Oct-96 31-Dec-97Portugal 28-Jun-00 19-Jul-02Qatar 7-Apr-99 15-Dec-99Romania 16-Feb-09 —Russian Federation 23-Dec-94 5-Aug-96Saudi Arabia 25-Jan-06 20-May-08Senegal 3-Jul-08 —Serbia 31-Jan-03 24-Feb-09Slovakia 25-Sep-06 27-Sep-07Spain 30-Sep-97 15-Dec-98Sri Lanka 22-Jan-97 13-Feb-98Sudan 22-Oct-03 —Sweden 4-Jul-00 1-Apr-01Switzerland 4-Apr-97 16-Feb-00Syrian Arab Republic 18-Jun-08 22-Jan-09Taiwan Province of China 17-Oct-02 28-Nov-02Tajikistan 13-Dec-95 14-Nov-03Thailand 10-Jul-00 13-Jul-01Trinidad and Tobago 12-Mar-07 7-Oct-07Turkey 17-Sep-98 18-Oct-07Turkmenistan 20-Sep-95 27-Feb-06Ukraine 1-Dec-01 12-Aug-03United Kingdom 14-Mar-94 6-Jan-95Uruguay 11-Feb-08 —Uzbekistan 18-May-99 28-Jul-00VietNam 8-Mar-97 1-Dec-99Yemen 1-Oct-02 10-Feb-04Zimbabwe 10-Feb-99 —

26

References

Acemoglu, Daron, Simon Johnson and James A. Robinson. 2001. “The Colonial Originsof Comparative Development: An Empirical Investigation.” The American EconomicReview 91(5):1369–1401.

Alesina, Alberto and Allan Drazen. 1991. “Why Are Stabilizations Delayed?” AmericanEconomic Review 81(5):1170–1188.

Buthe, Tim and Helen Milner. 2009. Bilateral Investment Treaties and Foreign DirectInvestment: A Political Analysis. In The Effect of Treaties on Foreign Direct Invest-ment: Bilateral Investment Treaties, Double Taxation Treaties, and Investment Flows,ed. Karl P. Sauvant and Lisa E. Sachs. New York, NY: Oxford University Press.

Drazen, Allan. 2000. Political Economy in Macroeconomics. Princeton, NJ: PrincetonUniversity Press.

Drazen, Allan and Paul Masson. 1994. “Credibility of Policies Versus Credibility of Policy-makers.” Quarterly Journal of Economics 109(3):735–754.

Fernandez, Raquel and Dani Rodrik. 1991. “Resistance to Reform: Status Quo Bias in thePresence of Individual Specific Uncertainty.” American Economic Review 81(5):1146–1155.

Gallgher, Kevin P. and Melissa B. L. Birch. 2006. “Do Investment agreements attractinvestment? Evidence from LAtin America.” Journal of World Investment and Trade7(6):961–973.

Haftel, Yoram Z. 2008. “The effect of U.S. BITs on FDI inflows to developing countries:signaling or credible commitment?” Unpublished, U of Illinois .

Jandyhala, Srividya, Witold Henisz and Edward D. Mansfield. 2007. “Pooling Dyads is aBIT Inappropriate.” Unpublished, University of Pennsylvania .

Mansfield, Edward D., Helen V. Milner and B. Peter Rosendorff. 2000. “Free to Trade:Democracies, Autocracies and International Trade.” American Political Science Review94(2):305–322.

Mansfield, Edward D., Helen V. Milner and B. Peter Rosendorff. 2002. “Why DemocraciesCooperate More: Electoral Control and International Trade Agreements.” InternationalOrganization 56(3):477–514.

Neumayer, Eric and Laura Spess. 2005. “Do bilateral investment treaties increase foreigndirect investment to developing countries?” World Development 33(10):1567–1585.

North, Douglass C. 1990. Institutions, Institutional Change, and Economic Performance.New York: Cambridge University Press.

North, Douglass C. and Barry R. Weingast. 1989. “Constitutions and Commitment: TheEvolution of Institutions Governing Public Choice in Seventeenth-Century England.”Journal of Economic History 69:803–32.

Olsen, Mancur. 1993. “Dictatorship, Democracy and Development.” American PoliticalScience Review 87(3):567–76.

Salacuse, Jeswald W. and Nicholas P. Sullivan. 2005. “Do BITs really work? An evaluationof bilateral investment treaties and their grand bargain.” Harvard International LawJournal 46(1):67–130.

27

Tobin, Jennifer and Susan Rose-Ackerman. 2005. “Foreign direct investment and the busi-ness environment in developing countries: The impact of bilateral investment treaties.”Yale Law and Economics Research Paper Series (293).

Tsebelis, George. 2002. Veto Players: How Political Institutions Work. Princeton: Prince-ton University Press.

UNCTAD. 2000. Bilateral Investment Treaties 1959 -1999. United Nations.Vandevelde, Kenneth J., Victoria Aranda and Zbigniew Zimmy. 1998. Bilateral Investment

Treaties in the Mid-1990s. New York - Geneva: UNCTAD.Yackee, Jason Webb. 2007. “Do BITs really Work? Revisiting the Empirical Link between

Investment Treaties and Foreign Direct Investment.” Legal Studies Research Paper Series,University of Wisconsin (1054).

28