# Foreign project financing in segmented capital markets: equity versus debt.

It is well known in finance that capital market segmentation may
result in a breakdown of capital structure irrelevance propositions. It
should, therefore, be generally agreed upon that, under international
segmentation of capital markets, an optimal capital structure is
possible. Many papers that have dealt with this issue so far (e.g., see
Rubinstein [16], Adler and Dumas [1], [2], [3] and [4], Stapleton and
Subrahmanyam [18], Senbet [17], and Hodder and Senbet [12]) have
admitted certain limitations, such as nonstochastic exchange rates,
stochastic independence of exchange rates and cash flows (e.g., Adler
and Dumas [3]), or riskless debt (e.g., Senbet [17]). They have not
directly modeled internationally segmented capital markets in a way that
can calrify conditions for optimal international capital structure
choices.

In this paper, we propose to show sufficient conditions for the existence of an international capital structure when the capital markets of two countries are not perfectly integrated, and with the simultaneous presence of both exchange and default risks on debt. By an "international capital structure," we mean an optimal combination of equity issued in one market and debt is issued in another. The appearance of international capital structures in this sense is quite frequent in actual corporate decisions. Our sufficient conditions illuminate decision rules for such outcomes in a context of internationally segmented capital markets.

In order to keep the analysis simple and to present only the relevant points, we employ a "state preference" framework for two reasons. First, the model is simple and the conditions for the emergence of an international capital structure are straightforward and relatively intuitive. Second, the state preference result could also be generalized to more specific models (such as the CAPM or representations of contingent claims in the form of options). It also facilitates an easier understanding of the descriptions of segmentation and its consequences. (1)

In spite of its simplicity, our model yields several general economic intuitions. We assume, more specifically, that the two national capital markets are perfect and complete in the sense that pure contingent claims are priced in each one. This assumption is made in order to eliminate all possible causes for relevance of capital structure within each capital market, and to bring into sharp focus the effect of international segmentation. We also assume that the list of contingencies is the same in the two economies, but they are priced differently by each country's investors. Technically, this is feasible in the state preference framework. (2) Imagine two economies that enjoy a high degree of integration through the free movement of goods and services. This would mean that there is substantial commonality of contingencies in the two economies. Yet, if their capital markets are less integrated than their real sectors, the same contingencies will be priced differently in the respective capital narkets. It is easy to think of examples in the world economy or even within the European Economic Community that come close to fitting this general picture. Many national economies are currently more open to the trading of goods and services than to capital movements. They also highly sensitive to contingencies in the larger economies of the world. Yet, several nations retain trading restrictions on foreign financial assets and large groups of investors exhibit clear national preferences in portfolio selection, thus fostering the phenomenon of capital market segmentation.

What emerges from our analysis can be summarized as follows. First, starting with the premise of capital structure irrelevance in each market, we arrive at an optimal international capital structure comprised of equity in one country and debt in another. Let us be more explicit about the essence of the result: segmentation does not always yield an international capital structure. The sufficient condition is the simultaneous occurence of positive value differential for debt and negative value differential for equity issued in two different markets. Sufficient conditions for the emergence of such an international financing mix are derived as functions of international interest parity relations and covariances of project's cash flows with state contingent misalignments of pure security prices and exchange rates.

Second, given our segmented market structure, we show that foreign risky debt can increase shareholder wealth and , thus, should be issued even though the firm is indifferent between domestic and foreign default-free debt. The role played by default is essential in partioning the states such that differential valuations occur favoring equityholders in the country of ownership while debt is issued in the host country. (3) It allows for a sharing of exchange risk between stockholders in one country and bondholders in another, so that the total value created exceeds the project's value in either market. Finally, our results also include empirical predictions about the type of firms that might/might not issue foreign debt, or foreign equity.

I. Definitions and Perfect Integration

We assume two countries A and B with internal complete markets and (nominal) pure security prices [phi][sub.A](s), [phi][sub.B](s) denominated in their respective currencies. Pure security prices are defined as:

[Mathematical Expression Omitted]

where [pi](s) are (homogeneously believed) state probabilities, and [alpha][sub.A](s), [alpha][sub.B](s) are risk adjusting factors for each state s. By construction, we have

[Mathematical Expression Omitted]

where r[sub.A], r[sub.B] are riskless nominal rates of return in each country, and E represents expectations. In the absence of taxes and other internal imperfections, the MM propositions hold within each economy.

The present exchange rate is e[sub.0] (units of currency A/units of currency B) and the period [tau] exchange rate in state s is e[sub.tau](s) (all prices, interest rates and exchange rates are nominal).

Perfect capital market integration requires that arbitrage profits are zero for trading of any country B security and its perfect substitute in country A; therefore

[Mathematical Expression Omitted]

or equivalently,

[Mathematical Expression Omitted]

Essentially, conditions (3) or (3a) describe security price parity (SPP). "Security price parity" simply means that any security's price in one capital market equals its price in the other capital market translated by the prevailing exchange rates. This is equivalent to the definition of "purchasing power parity" (PPP) for commodities. Two immediate implications of SPP are noted. First, from condition (3) we obtain

[Mathematical Expression Omitted]

In other words, the ratio of pure security prices are predictors of the rate of change in exchange rates in each state. Second, adding condition (3) over all states we obtain

[Mathematical Expression Omitted]

and using definitions (1) and (2) yields

[Mathematical Expression Omitted]

The expression in (4) is the no-arbitrage condition for a riskless nominal asset in country B and its perfect substitute in country A, a risky asset. We refer to this relationship as "uncovered interest rate parity" (UIRP) in the present model. (4) Note that the price ratio of two nominal riskless assets (below) is not a predictor of the expected rate of change in exchange rates, unless exchange rates and domestic risk adjustment factors are independent or investors are risk-neutral. Thus, expression (4) implies (5)

[Mathematical Expression Omitted]

A graphical illustration of UIRP is presented in Exhibit 1. Note that the interest parity relationship in (4a) will deviate around the line defined by [r.sub.A]-[r.sub.B]=(E([e.sub.[tau]]) - [e.sub.0])/[e.sub.0] (forward discount/premium) due to the cov([e.sub.[tau]], [alpha][sub.A]) being positive or negative. In Exhibit 1, UIRP* illustrates a case where cov([e.sub.tau], [alpha.sub.A]>0, whereas UIRP** yields whenever cov(e.sub.[tau], [alpha][sub.A])<0.

II. International Segmentation

Given the previous description of SPP, an operational definition of segmentation can be obtained from possible violations of (3) in some states, i.e., arbitrage profits remain available in thosek states due perhaps to international investment restrictions, or national investor preferences. (6) Defining state-dependent arbitrage profits by h(s) where,

h(s)=[e.sub.0][alpha.sub.B](s)-[e.sub.tau](s)[alpha.sub.A](s),

segmentation implies that h(s)[is not equal to]0 for some states. In particular, h(s)>0 signifies the existence of arbitrage potential in sale (short) of a pure claim denominated in currency B and simultaneous purchase of its perfect substitute in country A, and, h(s)<0 signifies the existence of arbitrage potential in purchase of a pure claim in currency B and sale of its perfect substitute in country A.

Thus, the vector [h(s):s [epsilon] S] is a composite index of segmentation. Also note that, given our definition of state-dependent arbitrage profits,(h(s)), UIRP (4) can be written as

E(h(s))=0.

A special case of partial integration occurs when arbitrage can take place for riskless nominal assets of country B, like bank deposits or government bonds. If this is admitted, UIRP will hold and (4b) will be valid. However, (4b) is not a sufficient condition for (3) to hold in every state. Thus, whereas expected arbitrage profits across states are zero, they may still deviate from zero in single contingencies. (7)

Any composite claim in country B, such as a "project" offering cash flows denominated in currency B, [X(s)], will present arbitrage potential if E(X(s)h(s))[is not equal to]0. This quantity represents the difference in project value in the two markets and can be written as

[delta][V.sub.0]=[e.sub.0][V.sup.A.sub.0]-[V.sup.A.sub.0]=E(X)E(h)+cov(X,h),

where [V.sup.A.sub.0] and [V.sup.B.sub.0] are the project's values in country A and B, respectively. Clearly, if partial integration is valid in the sense that (4b) holds (i.e.,UIRP is valid), then (6) specializes to

[delta][V.sub.0]=cov(X,h).

Two observatons are immediately apparent. Firstly, from (6), the size and sign of [delta]V[sub.0] is not uniquely determined by a possible UIRP violation, unless cash flows and SPP violations are independent, since it also depends on the covariance term (the second term). Thus, although riskless country B claims are more valuable in country A, for example (E(h)<0), a strong positive covariance of X(s) to h(s) may lead to [delta][V.sub.0] (i.e.,project more valuable in country B market), or vice versa. Secondly, if the UIRP holds and [delta][V.sub.0] is given by (6a), the sign of value difference will depend exclusively on the pattern of cash flows across states in which SPP is violated. Note that it is possible that (4b) holds, whereas some states have positive (h(s)>0) and other have negative (h(s)<0) SPP violations. A project with high (low) cash flows in states with positive (negative) SPP violations will exhibit positive covariance, and vice versa.

If we were to presume that the nationality of stockholders were entirely determined by financial reasons and the project is to be all-equity financed, the implications of (6) and (6a) would be straightforward. The positive project value differential, ([delta][V.sub.0]), predicts that the project ends up in the hands of country B investors who could be the highest bidders. On the contrary, [delta][V.sub.0]<0 predicts that the project ends up in the hands of country A investors who would be the highest bidders, in this case. Thus, "home country" ownership of a "host country" project would require either a strong negative violation of UIRP (E(h)<0) and/or negative covariances of cash flows with SPP violations, the latter being the exclusive condition if the UIRP held strictly (E(h)=0).

III. Conditions for the Issuance

of Foreign Debt

Writers on international corporate finance have long recognized that international segmentation may give rise to profit opportunities by purchase or issuance of claims in different markets. Any security can be profitably issued in a foreign market if [delta]V[sub.0]>0 for the particular cash flows offered by the security. Specifically, the necessary condition for issuance of debt denominated in the same currency as the cash flows of the project (currency B) and sold to investors in country B (foreign debt) by firms in country A is

[delta][V.sub.D]=[e.sub.0][V.sup.B.sub.D]-[V.sup.A.sub.D]>0,

where [V.sup.B.sub.D] and [V.sup.A.sub.D] are the values of the same debt in country B and country A, respectively. However, this is not the sole condition for issuance of foreign debt, since it is not possible that the debt of the firm is not supported by some equity. In fact, we will show that segmentation could, in principle, imply an optimal capital structure, whereby a project will be financed by a mixture of securtities issued in two separate markets. To establish the conditions for optimal capital structure, we consider alternative financing possibilities for the project which promises cash flows [X(s)] denominated in currency B (foreign project).

As noted before, when national markets are internally complete and perfect, the MM propositions hold and capital structure is irrelevant. Consequently, if the project is financed entirely in either of the two markets, variations in capital structure will not matter. Thus, if the project is owned by either country A or country B investors, its owners do not gain by issuing debt in their respective domestic markets. (8)

If the project is financed by a mix of debt and equity in the same economy, values in the two national markets will be [V.sup.B.sub.0]=[V.sup.B.sub.E]+[V.sup.B.sub.D], and [V.sup.A.sub.0]=[V.sup.A.sub.E]+[V.sup.A.sub.Dd]. It follows that

[Mathematical Expression Omitted]

An important implication of capital structure irrelevance within each national economy can be stated: the difference in the total value of the project in the two countries will equal the sum of differences in the values of debt and equity securities issued in the two markets.

Next, we derive the conditions under which segmented markets can lead to an "international" capital structure, i.e., debt and equity securities are held respectively by investors in different countries. If total cash flows from the project are apportioned between the two classes of securityholders such that the difference in values, [delta]V[sub.E] and [delta][V.sub.D], have opposite signs, then an international capital structure is optimum. More specifically, the condition under which country A investors would retain equity ownership but issue debt to investors in country B is

[delta][V.sub.E]<0 and [delta][V.sub.D]>0.

Combining (9) and (8) imply that

[delta]V[sub.D]>[delta]V[sub.0].

Thus, under the conditions above, country A investors who own the project initially, could gain only [delta][V.sub.0] by selling it off to country B investors. However, if they retain an equity claim to the project, they could gain more by selling debt securities to country B investors. This condition, along with (7), now becomes sufficient for the issuance of foreign debt by shareholders domiciled in country A, against a project located in country B. (9)

Our main conclusion here is that segmentation does not automatically imply the optimality of foreign debt. Foreign debt will only appear if the advantages it bestows on country A shareholders are superior to the alternative of selling off the entire project's equity to investors in country B.

IV. Analysis of Sufficient Conditions

for Foreign Debt

In this section, we link teh sufficient conditions (7) and (10) (or, equivalent, conditions in (9) to the project and market parameters using differential valuation functions similar to (6).

Consider an issue of foreign debt with promised payments D denominated in the currency of the cash flows of the project (currency B). Since D is promised against project cash flows X(s), the states can be categorized into two subsets [and] and [and]', where

X(s) [is greater than or equal to] D for states in [and], and X(s) <D for states in [and]'.

In the subject of states ([and]), the debt obligation is fulfilled and shareholders receive residual income from the project. In the subset of states ([and]'), the debt obligation cannot be met and debt is therefore in default. In the latter case, shareholders receive nothing and bondholders simply obtain total project cash flows. When the possibility of [Lambda]' exists (i.e., when [Lambda]' is a nonempty set), debt is ex-ante risky, otherwise debt is ex-ante default-free.

Let us define [X.sub.D(s)] and [X.sub.E(s)] as cash flows to be received by debt and equity holders respectively, in state s. It follows that by definition

[X.sub.D(s)] = D for states in [Lambda], and [X.sub.D(s)] for states in [Lambda]', and

[X.sub.E(s)] = X(s) - D for states in [Lambda], and [X.sub.E(s)] = 0 for states in [Lambda]'.

It also follows that, in every state

[X.sub.D(s) + [X.sub.E(s) = X(s)

Using previous definitions, we can now determine the differential valuation function for debt issued to country B investors: (10)

[Mathematical Expression Omitted]

Correspondingly, the differential function for equity is

[Mathematical Expression Omitted]

Conditions (9) for the emergence of an international capital structure now become:

[Mathematical Expression Omitted]

From the definition of expected values, these can be equivalently stated as

[Mathematical Equation Omitted]

Conditions (14a) and (14b) imply that an international capital structure will emerge if the covariances of debt (equity) cash flows with SPP violations obey certain respective lower (upper) bounds. We examine two special cases below.

A. The Case of Riskless Foreign Debt

The inclusion of only riskless foreign debt in capital structure is described by a promised payment D to country B investors, such thast it can be fully honored by project cash flows in every state. This case is interesting because it corresponds to the inclusion of "high quality" foreign debt in capital structure. Since the promised payment is always honored, we have:

[X.sub.(s)] = D for every state s, and cov([X.sub.D, h]) = cov(D, h) = 0.

In this case, (14a) and (14b) speialize to

[Mathematical Expression Omitted]

Condition (15a) is easy to interpret: Riskless foreign debt can be a part of capital structure only if there is a violation of uncovered interest parity (UIRP). Thus, the condition of e(h) > 0 is necessery, since it implies that riskless debt issued in country B is always more valuable than its perfect substitute in country A. However, it should be noted that since (15a) is not a sufficient condition, the existence of a violation of UIRP does not by itself imply an international capital structure. Fulfillment of (15b) is also required, and this is only possible if project cash flow with SPP violations is negative and large.

The point in this case is that the strength of the UIRP violation may undermien sufficient conditions for an international capital structure. In fact, the more positive E(h) is, the more stringent condition (15b) becomes, and the more negative cov(X, h) must be for both (15a) and (15b) to be satisfied simultaneously. Thus, we note that very large violations of UIRP would warrant a transfer of whole project ownership from country A to country B investors, and would restrict the number of projects that could support an international capital structure. (11) On the other hand, relatively mild UIRP violations (i.e., small positive E(h)) would admit a larger number of projects as candidates for supporting an international capital structure, where foreign riskless debt is included.

B. The Case of Partial Integration

As noted previously, partial integration is taken to mean that UIRP holds (E(h) = 0), but SPP is violated (h(s) [is not equal to] 0) in particular states. It is obvious that no riskless debt could be part of an international capital structure, since condition (15a) in the previous section would not hold. Risky debt is admissible. Conditions (14a) and (14b) specialize to (12)

[Mathematical Expression Omitted]

For condition (16a) to hold, [X.sub.D] must be variable across states, (i.e., debt must be risky). Since cash flows to debt and equity ([X.sub.D], [X.sub.E], respectively) add up to project cash flow in every state, the emergence of an international capital structure requires a decomposition of project cash flows into two parts with opposite covariation to SPP violations. Thus, cash flows accruing to debt (residual) claimants ought to covary positively (negatively) with SPP violations. Whether these conditions could be satisfied would depend on the firm's financial policy and the project cash flow characteristics. Financial policy determines the separation of states in two subsets ("default" and "nondefault" states). It is clear from the definitions of [X.sub.E] and [X.sub.D] that each is invariant over a different subset of states. Thus, [X.sub.E] does not vary across "default" states, being always zero in those states. [X.sub.D] does not vary across "nondefault" states, being always D in those states.

The covariances in (16a) and (16b) depend both on the variation between subsets and on the variation within subsets, for [X.sub.E] and [X.sub.D]. The variation within subsets, for both [X.sub.E] and [X.sub.D], closely tracks project cash flow variation, since in the "default" states [X.sub.D] equals project cash flows, and in the "nondefault" states [X.sub.E] equals project cash flow less a constant, D. It is, therefore, easy to see that the likelihood of both conditions (16a) and (16b) being satisfied depends to a significant extent on the behavior of project cash flows within each subset of states. Thus, for example, if project cash flows exhibit positive covariation to SPP violations in the "default" subset, as well as in the "nondefault" subset of states, this will make (16a) easy to satisfy but (16b) less likely to be satisfied. Or again, if project cash flows covary negatively with SPP violations within both subsets of states, this will facilitate fulfillment of (16b) but make harder the simultaneous fulfillment of (16a). A desirable project type which is most likely (although not exclusively) to support an international capital structure is one whose cash flows covary positively to SPP violations within the "default" set but negatively within the "nondefault" set of states. (13)

V. The Simple Case of Exchange Rate

Misalignments

SPP violations, h(s), is a complex measure depending on exchange rates and risk adjustment factors, where h(s) = [e.sub.0[alpha].sub.B(s) - e[tau](s)[alpha].sub.A(s)] < 0. A partial analysis that isolates only the effect of exchange rates is possible if we assume risk neutrality. In this case, we obtain, [14]

[Mathematical Expression Omitted]

Note that under risk neutality, h(s) will be zero in any state only if the future exchange rate is perfectly predicted by the differential in interest rates between countries A and B, [e.sub.[tau].(s)] = [e.sub.0([r.sub.A/r.sub.B).]

In contrast, the condition for partial integration, which is analyzed below, E(h) = 0, now implies that the interest rate differential predicts the future exchange rate on average, but that deviations can occur in specific states.

[Mathematical Expression Omitted]

Assuming E(h) = 0, we can restate the conditions for the issuance of foreign debt (16a), (16b)) under risk neutrality as

[Mathematical Expression Omitted]

Recall that (19a) refers to the residual claims of country A stockholders and (19b) refers to the default claims of country B bondholders, and by construction, [Mathematical Expression Omitted]

It is worth considering the type of projects which could satisfy conditions (19a) and (19b). Firstly, we observe that if project cash flows vary statistically with exchange rates in a monotonic fashion, it will not be possible to satisfy (19a) and (19b) simultaneously, for any partition of states. Thus, for example, a strongly export-oriented (import-competing) project would tend to do well in states with undervaluation and poorly in states with overvaluation of the foreign currency, so that

[Mathematical Expression Omitted]

While an import-dependent project would exhibit the opposite:

[Mathematical Expression Omitted]

A strongly export-oriented project may be a foreign project owned by a multinational that produces goods to be exported to the home country (country A). The goods are invoiced in currency B and have a perfectly elastic demand in country A. For example, a U.S. multinational has a subsidiary in Mexico that produces shoes from materials and labor produced in Mexico to be exclusively sold in the U.S. market. This is a strongly export-oriented foreign project and our analysis predicts that, under segmentation, it should be financed by equity issues in Mexico.

An example of a strongly import-based project would be a subsidiary of IBM in Hungary that sells U.S. manufactured computers. In general, this will be a foreign project that imports goods from country A to be sold in country B market where the goods are invoiced in A's currency and demand in country B is perfectly elastic. Our prediction is that this type of project is optimally financed by country A equity. In short, the export-oriented project would be optimally financed by country B equity and the import-based project would be optimally financed by country A equity. This is an interesting result, which suggests that cross-country ownership will tend to appear in projects which are heavily import-based rather than export-oriented.

The simultaneous fulfillment of conditions (19a) and (19b) requires a special project which in some states (in [Lambda]') behaves as an export-oriented (or import-competing) one, while in other states (in [Lambda]) behaves as an import-based one. In other words, the states of export-oriented behavior must, on the whole, be states of inferior project performance, whereas the states of import-based type behavior must, on the whole, be states of superior project performance. Schematically, the behavior of cash flows for such a special type of project can be represented as in Exhibit 2.

The type of specialized project, as depicted by cash flow patterns in Exhibit 2, is not an unlikely occurrence. It is an import-based project whose advantages decline rapidly in states when the currency of country B is undervalued. As undervaluation becomes deeper (to the left of point B in the diagram) the project's advantage switches toward a moderate export orientation (import-competing). Such a project could be one which is planned to sell its output in country B but uses primarily inputs imported to that country. As country B currency undervalues, however, some switching to domestic inputs takes place, and this makes possible the avoidance of cash flow decline, and even a reversal into a cash flow increase as currency B undervalues further and further.

Other, more complicated examples of X, e[tau] variations could also yield an international mix of financing. Quite possibly, this type of specialized project could be more common since cash flows rarely depend exclusively on exchange rates. Since there are many factors that can affect cash flows across states, e.g., economic conditions within the particular national economy, technological features, or even national policy decisions, it would not be practical to come up with a complete listing of projects that can support an international capital structure. Specific projects, however, could be analyzed to see whether they can support such a capital structure.

VI. An Illustrative Example

To demonstrate the results found by our analysis, a numerical example will be presented. The hypothetical data relating to market conditions and state contingent cash flows of three foreign projects [X.sub.P], [X.sub.M], and [X.sub.V] are shown in Exhibit 3. Note that Project P is strongly export-oriented (i.e., cov(X.sub.P], [e.sub.[tau]]) is strongly negative), Project M is heavily import-based (i.e., cov(X.sub.M], [e.sub.[tau]]) is strongly positive), and Project V is a special or mixed project (i.e., cov(X.sub.V], [e.sub.[tau]]) does not have any strong tendency). The market data also reflect a case where UIRP holds (i.e., E(h) = 0).

For each project, we compute the differential values of the project ([[delta]V.sub.0]), equity ([delta]V.sub.E]), and debt ([delta.V.sub.D), from [[delta]V.sub.0] = cov(X,h), [[delta].V.sub.E] = cov([X.sub.E,h]) and [[delta]V.sub.0] = cov(X, h), [[delta]V.sub.E] = cov([X.sub.E], h) and [[delta.V.sub.D] = cov([X.sub.D, h), at various levels of foreign debt issued (D).

Exhibit 4 presents the values of [[delta]V.sub.E], [[delta]V.sub.D] and [[delta].V.sub.0] for all three projects. In the all-equity cases (D = 0), [[delta]V.sub.E] = [[delta]V.sub.0]. (Since E(h) = 0, default-free foreign debt would not change the results for any of the cases, and thus is ignored.) Looking at Column 3 for Project P, we see [[delta]V.sub.0] = +2.978. (At any debt level, it is not possible for [[delta]V.sub.E] and [[delta]V.sub.D] to have opposite signs.) This project (strongly export-oriented) is best financed by country B (host) equity, since country B stockholders will bid a value 2.978 more than country A investors would. (15) Project M, a heavily import-based project, reports [[delta]V.sub.E] = -3.002, which signifies that country A investors value the project higher by 3.002, a result predicted by our analysis, in which this type of a project should be optimally financed by country A (home) equity. At all levels of debt issued, the [[delta]V.sub.D] values are negative, favoring debt from country A.

However, Project V is a different case. If it were all-equity financed, [[delta]V.sub.E] = [[delta]V.sub.0] = -0.418. This means stockholders in country A would bid 0.418 higher than those in

[TABULAR DATA OMITTED]

country B. However, if foreign risky debt is issued instead, our sufficient conditions (16a) and (16b) for an international capital structure are satisfied. At all debt levels, we observe [[delta][V.sub.D]] > O and [[delta][V.sub.D]] > [[delta][V.sub.O]]. The optimal debt level is when [[delta][V.sub.D]] is at its maximum [[delta][V.sub.D]]=0.540) which occurs at a debt level of 100. As predicted, this project should be financed by country B risky debt (foreign debt) and country A (domestic) equity and the optimal debt level is around 100.

VII. Conclusions, Predictions and

Some Financial Management

Implications

Our study offers new insights into international financial decisions, when capital markets are intentionally segmented. Since several empirical studies find that segmentation is present in world financial markets, the analysis in our study is quite relevant. We define segmentation broadly as the condition in which identical securities are not indentically priced in two markets. We also categorize segmentation into "severe" or "mild" types, depending on whether uncovered interest rate parity (UIRP) is violated or not. We note that even when UIRP holds, "mild" segmentation is possible, in the sense that securities are identically priced on average but not in reference to specific contingencies.

Our first fundamental proposition is that an international capital structure is not automatically implied as the optimal outcome in segmented markets. Thus, it is not sufficient for debt issues to be "cheaper" in country B than in country A in order to have an international capital structure. It must be simultaneously true that equity issues are "cheaper" in country A than in B. Otherwise, the optimal outcome would be either an all-equity financed firm, or a firm with debt and equity both issued in one national market or the other.

The empirical implication of the proposition is, if interest parity is violated, then international capital structure will more likely be observed in situations of small rather than large violations. (16) Large deviations from interest parity will influence all securities' value differentials in the same direction, and thus equity will also tend to be cheaper where cheaper debt is also found. Small deviations, on the other hand, will allow greater differentiation in capital sourcing via issuance of various securities. Thus, the incidence of international capital structure is greater under mild market segmentation.

Our second fundamental proposition is that an international capital structure does not have to be limited to inclusion of riskless foreign debt. Specifically, even when interest parity holds and riskless debt enjoys no differential price advantage in one market relative to the other, risky debt may still enjoy such a differential advantage, depending on the pattern of cash flows and its relation to the size of price misalignments in particular contingencies. This will occur if default contingencies (inherent in risky debt) overlap significantly with those market contingencies which are more highly priced in the foreign market.

This second proposition predicts what type of project will be empirically observed to have an international capital structure. Since riskless debt is exclusively associated with an UIRP violation, international capital structures which are limited to riskless debt will more likely surface in firms or projects with cash flows negatively dependent on price misalignments, or, in our more specific approximation, positively related to exchange rate variation. On the other hand, capital structures which include foreign risky debt, will be more likely supported by projects whose cash flows are weakly related or even unrelated to exchange rates. Thus, a strong exchange rate influence on cash flows may create incentives for riskless debt, if interest rate parity does not hold, but will create disincentives for risky debt, if interest parity does hold. Thus, the results of our analysis shed light on the cross-sectional variations in international debt/equity ratios.

Finally, here are our practical suggestions for the international financial managers. First, when interest rates are out of line and debt appears cheaper in the host country, it is not sufficient to settle for an issue of high quality foreign debt. They should examine whether selling the project is a more profitable alternative for home country shareholders, even before financing policy is decided for a foreign project or subsidiary.

Second, if interest rates appear to be aligned by the IRP, it is not an obvious conclusion that a debt issue in the host country is superfluous. It is worth examining whether issues of lower quality debt (risky) in the host country could be a more desirable alternative for projects whose payoffs are weakly related to exchange rates, or more generally, projects which are not primarily driven by foreign exchange fluctuations.

Differential valuations for a project or a subsidiary suggesting that maximum benefits are best attained by the sale of the project, may be in conflict with the firm's desire, out of nonfinancial reasons, for maintaining ownership. Such reasons could derive from proprietary technology embodied in the project, or from strategic considerations and linkages between the project and the owner-firm. Our third suggestion is that financial managers should consider the issue of securities in the host country as a "second best" solution to resolve this conflict. There are at least three possibilities: these are selling a minority equity stake in the host country, selling debt, or a mixture of the two. The second best solution would depend on specific problem parameters, such as project features and price misalignments. We surmise that this type of problem is quite frequent, where financial policies are subject to an "ownership constraint."

(1) As will be seen in our analysis that follows, the definition of segmentation is evident if one is dealing with pure state-contingent claims.

(2) Uncertainty in the (two-country) world economy is modeled by uncertain states of nature to be realized at time [tau]. We denote the collection of all possible states by S and an element of S as s. Suppose there is a single state variable in each country denoted by [q.sub.A] and [q.sub.B] (such as GNP in country A and country B) and there are five states of names [s.sub.1], [s.sub.2], [s.sub.3], [s.sub.4], [s.sub.5]. The values of [q.sub.A] and [q.sub.B] are described by

[TABULAR DATA OMITTED]

We see that there are five distinct states in the world economy commonly viewed by the individuals of the two open economies. Note that if the two economies were closed, the number of distinct states would be four in economy A and three in economy B. Since our paper assumes open economies, the individuals see the five-state common space. We further assume that the number of linearly independent securities in each economy is equal to the number of common contingencies observed. Thus, both markets are complete and contingent claims can be uniquely priced. However, due to differences in individuals and economies and market imperfections (such as restrictive trading, limitations on short selling) the contingent claims in the two markets are not perfect substitutes and their prices are different.

(3) It is well known that risky debt can create value by completing the market. The role default plays in our paper parallels that notion.

(4) UIRP is usually stated as [r.sub.A]/[r.sub.B] = E([e.sub.[tau])/[e.sub.0], and is derived from purchasing power parity (PPP) and the Fisher Effect. It is sometimes denoted [i.sub.A]/[i.sub.B] = E([e.sub.[tau])/[e.sub.0], where [i.sub.A] and [i.sub.B] where [i.sub.A] and [i.sub.B] are nominal rates of any two assets that are similar in all aspects (including default risk) except currency of denomination. In our study, we distinguish between this general form of UIRP and the relationship that concerns only default-free (riskless) nominal assets. In both cases, risk aversion parameters are heeded, whereas they are nonexistent in the common usage of these conditions. Thus our, SPP conditions, (3), is a more accurate statement of this general UIRP.

(5) It is also implied in (4) that f = [e.sub.0][r.sub.A]/[r.sub.B] = E([E.sub.[tau]) + [r.sub.A] cov([e.sub.[tau]], [[alpha].sub.A]), where f is the forward rate implied by interest rate parity. Thus, our arguments and results are consistent with the theoretical and empirical evidence for the existence of a risk premium in the presence of risk aversion (see, for example, Cumby and Obstfld [7] and Hodrick and Srivastava [13]). The risk premium is a function of the covariance of exchange rates and country A risk adjustment factors in the above relationship.

(6) Empirical evidence of market segmentation based on these factors is by now well documented. See, for example, Jorion and Schwartz [14], Errunza and Losq [8], Alexander, Eun and Janakiramanan [5], Gultekin, Gultekin and Penati [10], Hietala [11], and Bonser-Neal, Brauer, Neal and Wheatley [6].

(7) Note that this occurrence resembles arbitrage pricing theory in the sense that APT prices securities correctly, on the average, but pricing errors can exist for individual securities. We thank Lemma Senbet for pointing this out of us.

(8) Note that segmentation and exchange and default risks play no role in this case and we end up with the MM irrelevance in the international context. It does not matter where the project is located and in which market it is financed. Financial structure is irrelevant for a foreign project as well as a domestic one. See Thomadakis and Usmen [19] for detailed derivations.

(9) There are two special cases for these sufficient conditions. One occurs when [[delta][V.sup.E]] = 0, as might be expected in dual listings of the company's stock. [delta]pV.sub.D]] > 0, then, becomes the sole condition in issuance of foreign debt. Another case is [[delta][V.sub.D]] = 0 which is relevant when riskless debt is issued and E(h) = 0. Here, [[delta]V.sub.E] [unkeyable] 0, alone determines the country of equity issues.

(10) We can reexpress (12a) as follows,

[Mathematical Expression Omitted]

The first term on the righ hand side of (12c) represents the differential value of a default-free claim in country P promising D. The second term is the differential value of the default option held by owners of the residual claim in the project, recalling that A' is the set of states where D > X(s), the states in which default occurs. Also note that [D - X(s)] is the payoff at maturity to 2 put option written on the project with a strike price equal to the promised debt payment. Hence, the first term in brackets is the value of a put option due to default in the foreign market converted into domestic currency, [e.sub.0] [P.sub.B]. The second term, on the other hand, is the value of the same put option in the domestic market, [P.sub.A]. Note that in the valuation of [P.sub.A], the underlying asset is the perfect substitute of the foreign project in market A and the strike price is no longer a constant. Remembering that risky debt involves a default option held by shareholders who can sell the project for the amount of the promised payment, it bocomes clear that the term in brackets is the differential value of the put option held by shareholders. Given segmentation, [P.sub.B] [is not equal to] [P.sub.A] (in a CAPM framework this would imply that different default premiums are assessed in the two markets for the same default risk).

(11) IF (15b) is not satisfied, country A investors are better off selling the whole project to country B investors.

(12) We should note that the sufficient conditions [[delta][V.sub.D]] > 0 and [[delta][V.sub.D] > 0 and their analysis that follows are closely tied to the location of the project, the location of the firm (its equityholders), and the location of the debt issues. The present paper analyzes the case of international ownership where equityholders are in country A, the project is in country B (thus, a foreign project), and debt holders are in country B in whose currency the debt is denominated (thus, foreign debt). (The same set of sufficient conditions (but with opposite signs) would have obtained had the project been located in country A but equityholders were in country B and had issued foreign debt in country A.)

Assuming the project is located in country B, observing equity issues in country B and debt issues in country A would require a different set of sufficient conditions, specifically [[delta][V.sub.E]] > [[delta][V.sub.D]] < 0 and [[delta][V.sub.D]] > [[delta][V.sub.O]]. In this case, the project can be considered domestic (located in the country of the equityholders), and debt can be termed "Eurobond," since payments are promised against cash flows in a currency other than the home currency of the bondholders. It should be clear that, in this case, a different partition of states into default and nondefault sets [psi] and [psi]' respectively) will be the result. Although not symmetric, the conditions for the emergence of "Eurobond" debt are analogous to those derived for foreign debt. For extensive analysis of "Eurobond" debt, see Thomadakis and Usmen [19].

(13) The preceding discussion is derived from simple algebraic arguments. It is easily shown that,

[Mathematical Expression Omitted]

where [Mathematical Expression Omitted]

By definition,

E[Lambda]'(h) + E[Lambda](h) = 0 and cov[Lambda]'(X, h) + cov[Lambda](X, h) = cov(X, h).

Thus, the first terms on the right hand side of (16c) and (16d) must necessarily be of opposite signs and can be regulated by setting D [unkeyable] E(X). Although, the second terms on the right hand side of (16c) and (16d) do not have to be of opposite sign, it is move likely that the project will support an international capital structure, if they are, being respectively positive and negative (cov[Lambda]'(X, h) > 0 and Cov[Lambda](X, h) < 0).

(14) Under risk neutrality, we have

[Mathematical Expression Omitted]

since risk adjustment factors no longer serve as differential value-weights for different states.

(15) Even if foreign equity is a more valuable alternative to foreign debt, it may not exclude firms to maintain ownership in the foreign project due to reasons such as proprietary technology or control of ownership. These motivations may lead to a venture instead.

(16) Empirical evidence of differential valuation of debt issues is obtained by studies which have documented positive stock price reaction (in the U.S.) to international bond issues of U.S. firms abroad. For example, Kim and Stulz [15] study default-free Eurobond issues and find average positive stock market reactions to the announcement of those issues. It is notable that in their results, although positive on average, there is substantial presence of cases with negative reactions, despite the fact that in most cases there was a favorable yield differential for Eurobond issues.

References

[1] M. Adler, "The Cost of Capital and Valuation of a Two-Country Firm," Journal of Finance (March 1974), pp. 119-132.

[2] M. Adler and B. Dumas, "Optimal International Acquisitions," Journal of Finance (March 1975), pp. 1-19.

[3] M. Adler and B. Dumas, "The Long Term Financial Decisions of the Multinational Corporation," in International Capital Markets, E.J. Elton and M.J. Gruber (eds.), Amsterdam, North Holland, 1975, pp. 360-387.

[4] M. Adler and B. Dumas, "International Portfolio Choice and Corporate Finance: A Synthesis," Journal of Finance (June 1983), pp. 925-976.

[5] G.J. Alexander, C.S. Eun, and S. Janakiramanan, "International Listings and Stock Returns: Some Empirical Evidence," Journal of Financial and Quantitative Analysis (June 1988), pp. 135-151.

[6] C. Bonser-Neal, G. Brauer, R. Neal, and S. Wheatley, "International Investment Restrictions and Closed-End Country Fund Prices," Journal of Finance (June 1990), pp. 523-547.

[7] R.E. Cumby and M. Obstfeld, "A Note on Exchange Rate Expectations and Norminal Interest Differentials: A Test of the Fisher Hypothesis," Journal of Finance (June 1981), pp. 697-704.

[8] V. Errunza and E. Losq, "International Asset Pricing Under Mild Segmentation: Theory and Test," Journal of Finance (March 1985), pp. 105-124.

[9] C.S. Eun and S. Janakiramanan, "A Model of International Asset Pricing With a Constraint on the Foreign Equity Ownership," Journal of Finance (September 1986), pp. 897-914.

[10] M.N. Gultekin, N.B. Gultekin, and A. Penati, "Capital Controls and International Capital Market Segmentation: The Evidence From the Japanese and American Stock Markets," Journal of Finance (September 1989), pp. 849-869.

[11] P.K. Hietala, "Asset Pricing in Partially Segmented Markets: Evidence from the Finnish Market," Journal of Finance (July 1989), pp. 697-718.

[12] J.E. Hodder and L.W. Se bet, "International Capital Structure Equilibrium," Journal of Finance (Vol. 45, 1990), pp. 1495-1516.

[13] R.J. Hodrick and S. Srivastava, "An Investigation of Risk and Return in Forward Foreign Exchange," Journal of International Money and Finance (Vol. 3:1, 1984), pp. 5-29.

[14] P. Jorion and E. Schwartz, "Integration vs. Segmentation in the Canadian Stock Market," Journal of Finance (July 1986), pp. 603-616.

[15] Y.C. Kim and R.M. Stulz, "The Eurobond Market and Corporate Financial Policy: A Test of the Clientele Hypothesis," Journal of Financial Economics (December 1988), pp. 189-205.

[16] M.E. Rubinstein, "Corporate Financial Policy in Segemented Securities Markets," Journal of Financial and Quantitative Analysis (December 1973), pp. 749-761.

[17] L.W. Senbet, "International Capital Market Equilibrium and the Multinational Firm Financing and Investment Policies," Journal of Financial and Quantitative Analysis (September 1979), pp. 455-480.

[18] R.C. Stapleton and M.G. Subrahmanyam, "Market Imperfections, Capital Market Equilibrium and Corporation Finance," Journal of Finance (May 1977), pp. 307-319.

[19] S. Thomadakis and N. Usmen, "International Market Segmentation and the Corporate Borrowing Decision," Unpublished manuscript, Baruch College, CUNY and Rutgers University, 1990.

[20] N. Usmen, "Essays on Segmentation on International Currency and Asset Markets: Implications for Corporate Finance," Unpublished Ph.D dissertation, Baruch College, CUNY, 1988.

Appendix

In this appendix, we prove that if (16a) and (16b) are valid, they imply there exists necessarily an optimal level of foreign risky debt specified by a promised payment [unkeyable] which lies between the minimum and maximum promised payment levels allowed by the project. The minimum promised payment, [D.sub.min], is the minimum state contingent cash flow: [D.min] = min{X(s)}. This minimum level defines a default-free claim and since UIRP holds, [[delta][V.sub.D]] (D = [D.sub.min]) = 0. The maximum possible promised payment, [D.sub.max], is the maximum state contingent cash flow: [D.sub.max] = max{X(s)}. In effect, this represents an issue of unlevered equity in country B. Therefore, [[delta][V.sub.D]] (D = [D.sub.max]) = [[delta][V.sub.O]]. Provided that (16a) and (16b) hold, there exists at least one level of promised payment [unkeyable], such that [D.sub.min] < [unkeyable] < [D.sub.max], and

[Mathematical Expression Omitted]

Hence, there exists an optimal level of foreign risky debt.

Stavros Thomadakis is a Professor of Finance at Baruch College, CUNY, New York. Nilufer Usmen is an Assistant Professor of Finance at Rutgers University, Newark, New Jersey.

In this paper, we propose to show sufficient conditions for the existence of an international capital structure when the capital markets of two countries are not perfectly integrated, and with the simultaneous presence of both exchange and default risks on debt. By an "international capital structure," we mean an optimal combination of equity issued in one market and debt is issued in another. The appearance of international capital structures in this sense is quite frequent in actual corporate decisions. Our sufficient conditions illuminate decision rules for such outcomes in a context of internationally segmented capital markets.

In order to keep the analysis simple and to present only the relevant points, we employ a "state preference" framework for two reasons. First, the model is simple and the conditions for the emergence of an international capital structure are straightforward and relatively intuitive. Second, the state preference result could also be generalized to more specific models (such as the CAPM or representations of contingent claims in the form of options). It also facilitates an easier understanding of the descriptions of segmentation and its consequences. (1)

In spite of its simplicity, our model yields several general economic intuitions. We assume, more specifically, that the two national capital markets are perfect and complete in the sense that pure contingent claims are priced in each one. This assumption is made in order to eliminate all possible causes for relevance of capital structure within each capital market, and to bring into sharp focus the effect of international segmentation. We also assume that the list of contingencies is the same in the two economies, but they are priced differently by each country's investors. Technically, this is feasible in the state preference framework. (2) Imagine two economies that enjoy a high degree of integration through the free movement of goods and services. This would mean that there is substantial commonality of contingencies in the two economies. Yet, if their capital markets are less integrated than their real sectors, the same contingencies will be priced differently in the respective capital narkets. It is easy to think of examples in the world economy or even within the European Economic Community that come close to fitting this general picture. Many national economies are currently more open to the trading of goods and services than to capital movements. They also highly sensitive to contingencies in the larger economies of the world. Yet, several nations retain trading restrictions on foreign financial assets and large groups of investors exhibit clear national preferences in portfolio selection, thus fostering the phenomenon of capital market segmentation.

What emerges from our analysis can be summarized as follows. First, starting with the premise of capital structure irrelevance in each market, we arrive at an optimal international capital structure comprised of equity in one country and debt in another. Let us be more explicit about the essence of the result: segmentation does not always yield an international capital structure. The sufficient condition is the simultaneous occurence of positive value differential for debt and negative value differential for equity issued in two different markets. Sufficient conditions for the emergence of such an international financing mix are derived as functions of international interest parity relations and covariances of project's cash flows with state contingent misalignments of pure security prices and exchange rates.

Second, given our segmented market structure, we show that foreign risky debt can increase shareholder wealth and , thus, should be issued even though the firm is indifferent between domestic and foreign default-free debt. The role played by default is essential in partioning the states such that differential valuations occur favoring equityholders in the country of ownership while debt is issued in the host country. (3) It allows for a sharing of exchange risk between stockholders in one country and bondholders in another, so that the total value created exceeds the project's value in either market. Finally, our results also include empirical predictions about the type of firms that might/might not issue foreign debt, or foreign equity.

I. Definitions and Perfect Integration

We assume two countries A and B with internal complete markets and (nominal) pure security prices [phi][sub.A](s), [phi][sub.B](s) denominated in their respective currencies. Pure security prices are defined as:

[Mathematical Expression Omitted]

where [pi](s) are (homogeneously believed) state probabilities, and [alpha][sub.A](s), [alpha][sub.B](s) are risk adjusting factors for each state s. By construction, we have

[Mathematical Expression Omitted]

where r[sub.A], r[sub.B] are riskless nominal rates of return in each country, and E represents expectations. In the absence of taxes and other internal imperfections, the MM propositions hold within each economy.

The present exchange rate is e[sub.0] (units of currency A/units of currency B) and the period [tau] exchange rate in state s is e[sub.tau](s) (all prices, interest rates and exchange rates are nominal).

Perfect capital market integration requires that arbitrage profits are zero for trading of any country B security and its perfect substitute in country A; therefore

[Mathematical Expression Omitted]

or equivalently,

[Mathematical Expression Omitted]

Essentially, conditions (3) or (3a) describe security price parity (SPP). "Security price parity" simply means that any security's price in one capital market equals its price in the other capital market translated by the prevailing exchange rates. This is equivalent to the definition of "purchasing power parity" (PPP) for commodities. Two immediate implications of SPP are noted. First, from condition (3) we obtain

[Mathematical Expression Omitted]

In other words, the ratio of pure security prices are predictors of the rate of change in exchange rates in each state. Second, adding condition (3) over all states we obtain

[Mathematical Expression Omitted]

and using definitions (1) and (2) yields

[Mathematical Expression Omitted]

The expression in (4) is the no-arbitrage condition for a riskless nominal asset in country B and its perfect substitute in country A, a risky asset. We refer to this relationship as "uncovered interest rate parity" (UIRP) in the present model. (4) Note that the price ratio of two nominal riskless assets (below) is not a predictor of the expected rate of change in exchange rates, unless exchange rates and domestic risk adjustment factors are independent or investors are risk-neutral. Thus, expression (4) implies (5)

[Mathematical Expression Omitted]

A graphical illustration of UIRP is presented in Exhibit 1. Note that the interest parity relationship in (4a) will deviate around the line defined by [r.sub.A]-[r.sub.B]=(E([e.sub.[tau]]) - [e.sub.0])/[e.sub.0] (forward discount/premium) due to the cov([e.sub.[tau]], [alpha][sub.A]) being positive or negative. In Exhibit 1, UIRP* illustrates a case where cov([e.sub.tau], [alpha.sub.A]>0, whereas UIRP** yields whenever cov(e.sub.[tau], [alpha][sub.A])<0.

II. International Segmentation

Given the previous description of SPP, an operational definition of segmentation can be obtained from possible violations of (3) in some states, i.e., arbitrage profits remain available in thosek states due perhaps to international investment restrictions, or national investor preferences. (6) Defining state-dependent arbitrage profits by h(s) where,

h(s)=[e.sub.0][alpha.sub.B](s)-[e.sub.tau](s)[alpha.sub.A](s),

segmentation implies that h(s)[is not equal to]0 for some states. In particular, h(s)>0 signifies the existence of arbitrage potential in sale (short) of a pure claim denominated in currency B and simultaneous purchase of its perfect substitute in country A, and, h(s)<0 signifies the existence of arbitrage potential in purchase of a pure claim in currency B and sale of its perfect substitute in country A.

Thus, the vector [h(s):s [epsilon] S] is a composite index of segmentation. Also note that, given our definition of state-dependent arbitrage profits,(h(s)), UIRP (4) can be written as

E(h(s))=0.

A special case of partial integration occurs when arbitrage can take place for riskless nominal assets of country B, like bank deposits or government bonds. If this is admitted, UIRP will hold and (4b) will be valid. However, (4b) is not a sufficient condition for (3) to hold in every state. Thus, whereas expected arbitrage profits across states are zero, they may still deviate from zero in single contingencies. (7)

Any composite claim in country B, such as a "project" offering cash flows denominated in currency B, [X(s)], will present arbitrage potential if E(X(s)h(s))[is not equal to]0. This quantity represents the difference in project value in the two markets and can be written as

[delta][V.sub.0]=[e.sub.0][V.sup.A.sub.0]-[V.sup.A.sub.0]=E(X)E(h)+cov(X,h),

where [V.sup.A.sub.0] and [V.sup.B.sub.0] are the project's values in country A and B, respectively. Clearly, if partial integration is valid in the sense that (4b) holds (i.e.,UIRP is valid), then (6) specializes to

[delta][V.sub.0]=cov(X,h).

Two observatons are immediately apparent. Firstly, from (6), the size and sign of [delta]V[sub.0] is not uniquely determined by a possible UIRP violation, unless cash flows and SPP violations are independent, since it also depends on the covariance term (the second term). Thus, although riskless country B claims are more valuable in country A, for example (E(h)<0), a strong positive covariance of X(s) to h(s) may lead to [delta][V.sub.0] (i.e.,project more valuable in country B market), or vice versa. Secondly, if the UIRP holds and [delta][V.sub.0] is given by (6a), the sign of value difference will depend exclusively on the pattern of cash flows across states in which SPP is violated. Note that it is possible that (4b) holds, whereas some states have positive (h(s)>0) and other have negative (h(s)<0) SPP violations. A project with high (low) cash flows in states with positive (negative) SPP violations will exhibit positive covariance, and vice versa.

If we were to presume that the nationality of stockholders were entirely determined by financial reasons and the project is to be all-equity financed, the implications of (6) and (6a) would be straightforward. The positive project value differential, ([delta][V.sub.0]), predicts that the project ends up in the hands of country B investors who could be the highest bidders. On the contrary, [delta][V.sub.0]<0 predicts that the project ends up in the hands of country A investors who would be the highest bidders, in this case. Thus, "home country" ownership of a "host country" project would require either a strong negative violation of UIRP (E(h)<0) and/or negative covariances of cash flows with SPP violations, the latter being the exclusive condition if the UIRP held strictly (E(h)=0).

III. Conditions for the Issuance

of Foreign Debt

Writers on international corporate finance have long recognized that international segmentation may give rise to profit opportunities by purchase or issuance of claims in different markets. Any security can be profitably issued in a foreign market if [delta]V[sub.0]>0 for the particular cash flows offered by the security. Specifically, the necessary condition for issuance of debt denominated in the same currency as the cash flows of the project (currency B) and sold to investors in country B (foreign debt) by firms in country A is

[delta][V.sub.D]=[e.sub.0][V.sup.B.sub.D]-[V.sup.A.sub.D]>0,

where [V.sup.B.sub.D] and [V.sup.A.sub.D] are the values of the same debt in country B and country A, respectively. However, this is not the sole condition for issuance of foreign debt, since it is not possible that the debt of the firm is not supported by some equity. In fact, we will show that segmentation could, in principle, imply an optimal capital structure, whereby a project will be financed by a mixture of securtities issued in two separate markets. To establish the conditions for optimal capital structure, we consider alternative financing possibilities for the project which promises cash flows [X(s)] denominated in currency B (foreign project).

As noted before, when national markets are internally complete and perfect, the MM propositions hold and capital structure is irrelevant. Consequently, if the project is financed entirely in either of the two markets, variations in capital structure will not matter. Thus, if the project is owned by either country A or country B investors, its owners do not gain by issuing debt in their respective domestic markets. (8)

If the project is financed by a mix of debt and equity in the same economy, values in the two national markets will be [V.sup.B.sub.0]=[V.sup.B.sub.E]+[V.sup.B.sub.D], and [V.sup.A.sub.0]=[V.sup.A.sub.E]+[V.sup.A.sub.Dd]. It follows that

[Mathematical Expression Omitted]

An important implication of capital structure irrelevance within each national economy can be stated: the difference in the total value of the project in the two countries will equal the sum of differences in the values of debt and equity securities issued in the two markets.

Next, we derive the conditions under which segmented markets can lead to an "international" capital structure, i.e., debt and equity securities are held respectively by investors in different countries. If total cash flows from the project are apportioned between the two classes of securityholders such that the difference in values, [delta]V[sub.E] and [delta][V.sub.D], have opposite signs, then an international capital structure is optimum. More specifically, the condition under which country A investors would retain equity ownership but issue debt to investors in country B is

[delta][V.sub.E]<0 and [delta][V.sub.D]>0.

Combining (9) and (8) imply that

[delta]V[sub.D]>[delta]V[sub.0].

Thus, under the conditions above, country A investors who own the project initially, could gain only [delta][V.sub.0] by selling it off to country B investors. However, if they retain an equity claim to the project, they could gain more by selling debt securities to country B investors. This condition, along with (7), now becomes sufficient for the issuance of foreign debt by shareholders domiciled in country A, against a project located in country B. (9)

Our main conclusion here is that segmentation does not automatically imply the optimality of foreign debt. Foreign debt will only appear if the advantages it bestows on country A shareholders are superior to the alternative of selling off the entire project's equity to investors in country B.

IV. Analysis of Sufficient Conditions

for Foreign Debt

In this section, we link teh sufficient conditions (7) and (10) (or, equivalent, conditions in (9) to the project and market parameters using differential valuation functions similar to (6).

Consider an issue of foreign debt with promised payments D denominated in the currency of the cash flows of the project (currency B). Since D is promised against project cash flows X(s), the states can be categorized into two subsets [and] and [and]', where

X(s) [is greater than or equal to] D for states in [and], and X(s) <D for states in [and]'.

In the subject of states ([and]), the debt obligation is fulfilled and shareholders receive residual income from the project. In the subset of states ([and]'), the debt obligation cannot be met and debt is therefore in default. In the latter case, shareholders receive nothing and bondholders simply obtain total project cash flows. When the possibility of [Lambda]' exists (i.e., when [Lambda]' is a nonempty set), debt is ex-ante risky, otherwise debt is ex-ante default-free.

Let us define [X.sub.D(s)] and [X.sub.E(s)] as cash flows to be received by debt and equity holders respectively, in state s. It follows that by definition

[X.sub.D(s)] = D for states in [Lambda], and [X.sub.D(s)] for states in [Lambda]', and

[X.sub.E(s)] = X(s) - D for states in [Lambda], and [X.sub.E(s)] = 0 for states in [Lambda]'.

It also follows that, in every state

[X.sub.D(s) + [X.sub.E(s) = X(s)

Using previous definitions, we can now determine the differential valuation function for debt issued to country B investors: (10)

[Mathematical Expression Omitted]

Correspondingly, the differential function for equity is

[Mathematical Expression Omitted]

Conditions (9) for the emergence of an international capital structure now become:

[Mathematical Expression Omitted]

From the definition of expected values, these can be equivalently stated as

[Mathematical Equation Omitted]

Conditions (14a) and (14b) imply that an international capital structure will emerge if the covariances of debt (equity) cash flows with SPP violations obey certain respective lower (upper) bounds. We examine two special cases below.

A. The Case of Riskless Foreign Debt

The inclusion of only riskless foreign debt in capital structure is described by a promised payment D to country B investors, such thast it can be fully honored by project cash flows in every state. This case is interesting because it corresponds to the inclusion of "high quality" foreign debt in capital structure. Since the promised payment is always honored, we have:

[X.sub.(s)] = D for every state s, and cov([X.sub.D, h]) = cov(D, h) = 0.

In this case, (14a) and (14b) speialize to

[Mathematical Expression Omitted]

Condition (15a) is easy to interpret: Riskless foreign debt can be a part of capital structure only if there is a violation of uncovered interest parity (UIRP). Thus, the condition of e(h) > 0 is necessery, since it implies that riskless debt issued in country B is always more valuable than its perfect substitute in country A. However, it should be noted that since (15a) is not a sufficient condition, the existence of a violation of UIRP does not by itself imply an international capital structure. Fulfillment of (15b) is also required, and this is only possible if project cash flow with SPP violations is negative and large.

The point in this case is that the strength of the UIRP violation may undermien sufficient conditions for an international capital structure. In fact, the more positive E(h) is, the more stringent condition (15b) becomes, and the more negative cov(X, h) must be for both (15a) and (15b) to be satisfied simultaneously. Thus, we note that very large violations of UIRP would warrant a transfer of whole project ownership from country A to country B investors, and would restrict the number of projects that could support an international capital structure. (11) On the other hand, relatively mild UIRP violations (i.e., small positive E(h)) would admit a larger number of projects as candidates for supporting an international capital structure, where foreign riskless debt is included.

B. The Case of Partial Integration

As noted previously, partial integration is taken to mean that UIRP holds (E(h) = 0), but SPP is violated (h(s) [is not equal to] 0) in particular states. It is obvious that no riskless debt could be part of an international capital structure, since condition (15a) in the previous section would not hold. Risky debt is admissible. Conditions (14a) and (14b) specialize to (12)

[Mathematical Expression Omitted]

For condition (16a) to hold, [X.sub.D] must be variable across states, (i.e., debt must be risky). Since cash flows to debt and equity ([X.sub.D], [X.sub.E], respectively) add up to project cash flow in every state, the emergence of an international capital structure requires a decomposition of project cash flows into two parts with opposite covariation to SPP violations. Thus, cash flows accruing to debt (residual) claimants ought to covary positively (negatively) with SPP violations. Whether these conditions could be satisfied would depend on the firm's financial policy and the project cash flow characteristics. Financial policy determines the separation of states in two subsets ("default" and "nondefault" states). It is clear from the definitions of [X.sub.E] and [X.sub.D] that each is invariant over a different subset of states. Thus, [X.sub.E] does not vary across "default" states, being always zero in those states. [X.sub.D] does not vary across "nondefault" states, being always D in those states.

The covariances in (16a) and (16b) depend both on the variation between subsets and on the variation within subsets, for [X.sub.E] and [X.sub.D]. The variation within subsets, for both [X.sub.E] and [X.sub.D], closely tracks project cash flow variation, since in the "default" states [X.sub.D] equals project cash flows, and in the "nondefault" states [X.sub.E] equals project cash flow less a constant, D. It is, therefore, easy to see that the likelihood of both conditions (16a) and (16b) being satisfied depends to a significant extent on the behavior of project cash flows within each subset of states. Thus, for example, if project cash flows exhibit positive covariation to SPP violations in the "default" subset, as well as in the "nondefault" subset of states, this will make (16a) easy to satisfy but (16b) less likely to be satisfied. Or again, if project cash flows covary negatively with SPP violations within both subsets of states, this will facilitate fulfillment of (16b) but make harder the simultaneous fulfillment of (16a). A desirable project type which is most likely (although not exclusively) to support an international capital structure is one whose cash flows covary positively to SPP violations within the "default" set but negatively within the "nondefault" set of states. (13)

V. The Simple Case of Exchange Rate

Misalignments

SPP violations, h(s), is a complex measure depending on exchange rates and risk adjustment factors, where h(s) = [e.sub.0[alpha].sub.B(s) - e[tau](s)[alpha].sub.A(s)] < 0. A partial analysis that isolates only the effect of exchange rates is possible if we assume risk neutrality. In this case, we obtain, [14]

[Mathematical Expression Omitted]

Note that under risk neutality, h(s) will be zero in any state only if the future exchange rate is perfectly predicted by the differential in interest rates between countries A and B, [e.sub.[tau].(s)] = [e.sub.0([r.sub.A/r.sub.B).]

In contrast, the condition for partial integration, which is analyzed below, E(h) = 0, now implies that the interest rate differential predicts the future exchange rate on average, but that deviations can occur in specific states.

[Mathematical Expression Omitted]

Assuming E(h) = 0, we can restate the conditions for the issuance of foreign debt (16a), (16b)) under risk neutrality as

[Mathematical Expression Omitted]

Recall that (19a) refers to the residual claims of country A stockholders and (19b) refers to the default claims of country B bondholders, and by construction, [Mathematical Expression Omitted]

It is worth considering the type of projects which could satisfy conditions (19a) and (19b). Firstly, we observe that if project cash flows vary statistically with exchange rates in a monotonic fashion, it will not be possible to satisfy (19a) and (19b) simultaneously, for any partition of states. Thus, for example, a strongly export-oriented (import-competing) project would tend to do well in states with undervaluation and poorly in states with overvaluation of the foreign currency, so that

[Mathematical Expression Omitted]

While an import-dependent project would exhibit the opposite:

[Mathematical Expression Omitted]

A strongly export-oriented project may be a foreign project owned by a multinational that produces goods to be exported to the home country (country A). The goods are invoiced in currency B and have a perfectly elastic demand in country A. For example, a U.S. multinational has a subsidiary in Mexico that produces shoes from materials and labor produced in Mexico to be exclusively sold in the U.S. market. This is a strongly export-oriented foreign project and our analysis predicts that, under segmentation, it should be financed by equity issues in Mexico.

An example of a strongly import-based project would be a subsidiary of IBM in Hungary that sells U.S. manufactured computers. In general, this will be a foreign project that imports goods from country A to be sold in country B market where the goods are invoiced in A's currency and demand in country B is perfectly elastic. Our prediction is that this type of project is optimally financed by country A equity. In short, the export-oriented project would be optimally financed by country B equity and the import-based project would be optimally financed by country A equity. This is an interesting result, which suggests that cross-country ownership will tend to appear in projects which are heavily import-based rather than export-oriented.

The simultaneous fulfillment of conditions (19a) and (19b) requires a special project which in some states (in [Lambda]') behaves as an export-oriented (or import-competing) one, while in other states (in [Lambda]) behaves as an import-based one. In other words, the states of export-oriented behavior must, on the whole, be states of inferior project performance, whereas the states of import-based type behavior must, on the whole, be states of superior project performance. Schematically, the behavior of cash flows for such a special type of project can be represented as in Exhibit 2.

The type of specialized project, as depicted by cash flow patterns in Exhibit 2, is not an unlikely occurrence. It is an import-based project whose advantages decline rapidly in states when the currency of country B is undervalued. As undervaluation becomes deeper (to the left of point B in the diagram) the project's advantage switches toward a moderate export orientation (import-competing). Such a project could be one which is planned to sell its output in country B but uses primarily inputs imported to that country. As country B currency undervalues, however, some switching to domestic inputs takes place, and this makes possible the avoidance of cash flow decline, and even a reversal into a cash flow increase as currency B undervalues further and further.

Other, more complicated examples of X, e[tau] variations could also yield an international mix of financing. Quite possibly, this type of specialized project could be more common since cash flows rarely depend exclusively on exchange rates. Since there are many factors that can affect cash flows across states, e.g., economic conditions within the particular national economy, technological features, or even national policy decisions, it would not be practical to come up with a complete listing of projects that can support an international capital structure. Specific projects, however, could be analyzed to see whether they can support such a capital structure.

VI. An Illustrative Example

To demonstrate the results found by our analysis, a numerical example will be presented. The hypothetical data relating to market conditions and state contingent cash flows of three foreign projects [X.sub.P], [X.sub.M], and [X.sub.V] are shown in Exhibit 3. Note that Project P is strongly export-oriented (i.e., cov(X.sub.P], [e.sub.[tau]]) is strongly negative), Project M is heavily import-based (i.e., cov(X.sub.M], [e.sub.[tau]]) is strongly positive), and Project V is a special or mixed project (i.e., cov(X.sub.V], [e.sub.[tau]]) does not have any strong tendency). The market data also reflect a case where UIRP holds (i.e., E(h) = 0).

For each project, we compute the differential values of the project ([[delta]V.sub.0]), equity ([delta]V.sub.E]), and debt ([delta.V.sub.D), from [[delta]V.sub.0] = cov(X,h), [[delta].V.sub.E] = cov([X.sub.E,h]) and [[delta]V.sub.0] = cov(X, h), [[delta]V.sub.E] = cov([X.sub.E], h) and [[delta.V.sub.D] = cov([X.sub.D, h), at various levels of foreign debt issued (D).

Exhibit 4 presents the values of [[delta]V.sub.E], [[delta]V.sub.D] and [[delta].V.sub.0] for all three projects. In the all-equity cases (D = 0), [[delta]V.sub.E] = [[delta]V.sub.0]. (Since E(h) = 0, default-free foreign debt would not change the results for any of the cases, and thus is ignored.) Looking at Column 3 for Project P, we see [[delta]V.sub.0] = +2.978. (At any debt level, it is not possible for [[delta]V.sub.E] and [[delta]V.sub.D] to have opposite signs.) This project (strongly export-oriented) is best financed by country B (host) equity, since country B stockholders will bid a value 2.978 more than country A investors would. (15) Project M, a heavily import-based project, reports [[delta]V.sub.E] = -3.002, which signifies that country A investors value the project higher by 3.002, a result predicted by our analysis, in which this type of a project should be optimally financed by country A (home) equity. At all levels of debt issued, the [[delta]V.sub.D] values are negative, favoring debt from country A.

However, Project V is a different case. If it were all-equity financed, [[delta]V.sub.E] = [[delta]V.sub.0] = -0.418. This means stockholders in country A would bid 0.418 higher than those in

Exhibit 3. Parameter Values of Market Conditions and Project Cash Flows Used Throughout the Illustrative Example s [pi](s) e(s) h(s) Xp XM XV 1 0.20 0.40 0.114 150 50 100 2 0.20 0.47 0.052 130 80 130 3 0.20 0.50 0.025 100 100 80 4 0.20 0.60 -0.064 80 130 50 5 0.20 0.67 -0.127 50 150 150 Note: rA = 1.12, rB = 1.06, eO = 0.50, E(e[tau]) = 0.528, and E(h) = O.

[TABULAR DATA OMITTED]

country B. However, if foreign risky debt is issued instead, our sufficient conditions (16a) and (16b) for an international capital structure are satisfied. At all debt levels, we observe [[delta][V.sub.D]] > O and [[delta][V.sub.D]] > [[delta][V.sub.O]]. The optimal debt level is when [[delta][V.sub.D]] is at its maximum [[delta][V.sub.D]]=0.540) which occurs at a debt level of 100. As predicted, this project should be financed by country B risky debt (foreign debt) and country A (domestic) equity and the optimal debt level is around 100.

VII. Conclusions, Predictions and

Some Financial Management

Implications

Our study offers new insights into international financial decisions, when capital markets are intentionally segmented. Since several empirical studies find that segmentation is present in world financial markets, the analysis in our study is quite relevant. We define segmentation broadly as the condition in which identical securities are not indentically priced in two markets. We also categorize segmentation into "severe" or "mild" types, depending on whether uncovered interest rate parity (UIRP) is violated or not. We note that even when UIRP holds, "mild" segmentation is possible, in the sense that securities are identically priced on average but not in reference to specific contingencies.

Our first fundamental proposition is that an international capital structure is not automatically implied as the optimal outcome in segmented markets. Thus, it is not sufficient for debt issues to be "cheaper" in country B than in country A in order to have an international capital structure. It must be simultaneously true that equity issues are "cheaper" in country A than in B. Otherwise, the optimal outcome would be either an all-equity financed firm, or a firm with debt and equity both issued in one national market or the other.

The empirical implication of the proposition is, if interest parity is violated, then international capital structure will more likely be observed in situations of small rather than large violations. (16) Large deviations from interest parity will influence all securities' value differentials in the same direction, and thus equity will also tend to be cheaper where cheaper debt is also found. Small deviations, on the other hand, will allow greater differentiation in capital sourcing via issuance of various securities. Thus, the incidence of international capital structure is greater under mild market segmentation.

Our second fundamental proposition is that an international capital structure does not have to be limited to inclusion of riskless foreign debt. Specifically, even when interest parity holds and riskless debt enjoys no differential price advantage in one market relative to the other, risky debt may still enjoy such a differential advantage, depending on the pattern of cash flows and its relation to the size of price misalignments in particular contingencies. This will occur if default contingencies (inherent in risky debt) overlap significantly with those market contingencies which are more highly priced in the foreign market.

This second proposition predicts what type of project will be empirically observed to have an international capital structure. Since riskless debt is exclusively associated with an UIRP violation, international capital structures which are limited to riskless debt will more likely surface in firms or projects with cash flows negatively dependent on price misalignments, or, in our more specific approximation, positively related to exchange rate variation. On the other hand, capital structures which include foreign risky debt, will be more likely supported by projects whose cash flows are weakly related or even unrelated to exchange rates. Thus, a strong exchange rate influence on cash flows may create incentives for riskless debt, if interest rate parity does not hold, but will create disincentives for risky debt, if interest parity does hold. Thus, the results of our analysis shed light on the cross-sectional variations in international debt/equity ratios.

Finally, here are our practical suggestions for the international financial managers. First, when interest rates are out of line and debt appears cheaper in the host country, it is not sufficient to settle for an issue of high quality foreign debt. They should examine whether selling the project is a more profitable alternative for home country shareholders, even before financing policy is decided for a foreign project or subsidiary.

Second, if interest rates appear to be aligned by the IRP, it is not an obvious conclusion that a debt issue in the host country is superfluous. It is worth examining whether issues of lower quality debt (risky) in the host country could be a more desirable alternative for projects whose payoffs are weakly related to exchange rates, or more generally, projects which are not primarily driven by foreign exchange fluctuations.

Differential valuations for a project or a subsidiary suggesting that maximum benefits are best attained by the sale of the project, may be in conflict with the firm's desire, out of nonfinancial reasons, for maintaining ownership. Such reasons could derive from proprietary technology embodied in the project, or from strategic considerations and linkages between the project and the owner-firm. Our third suggestion is that financial managers should consider the issue of securities in the host country as a "second best" solution to resolve this conflict. There are at least three possibilities: these are selling a minority equity stake in the host country, selling debt, or a mixture of the two. The second best solution would depend on specific problem parameters, such as project features and price misalignments. We surmise that this type of problem is quite frequent, where financial policies are subject to an "ownership constraint."

(1) As will be seen in our analysis that follows, the definition of segmentation is evident if one is dealing with pure state-contingent claims.

(2) Uncertainty in the (two-country) world economy is modeled by uncertain states of nature to be realized at time [tau]. We denote the collection of all possible states by S and an element of S as s. Suppose there is a single state variable in each country denoted by [q.sub.A] and [q.sub.B] (such as GNP in country A and country B) and there are five states of names [s.sub.1], [s.sub.2], [s.sub.3], [s.sub.4], [s.sub.5]. The values of [q.sub.A] and [q.sub.B] are described by

[TABULAR DATA OMITTED]

We see that there are five distinct states in the world economy commonly viewed by the individuals of the two open economies. Note that if the two economies were closed, the number of distinct states would be four in economy A and three in economy B. Since our paper assumes open economies, the individuals see the five-state common space. We further assume that the number of linearly independent securities in each economy is equal to the number of common contingencies observed. Thus, both markets are complete and contingent claims can be uniquely priced. However, due to differences in individuals and economies and market imperfections (such as restrictive trading, limitations on short selling) the contingent claims in the two markets are not perfect substitutes and their prices are different.

(3) It is well known that risky debt can create value by completing the market. The role default plays in our paper parallels that notion.

(4) UIRP is usually stated as [r.sub.A]/[r.sub.B] = E([e.sub.[tau])/[e.sub.0], and is derived from purchasing power parity (PPP) and the Fisher Effect. It is sometimes denoted [i.sub.A]/[i.sub.B] = E([e.sub.[tau])/[e.sub.0], where [i.sub.A] and [i.sub.B] where [i.sub.A] and [i.sub.B] are nominal rates of any two assets that are similar in all aspects (including default risk) except currency of denomination. In our study, we distinguish between this general form of UIRP and the relationship that concerns only default-free (riskless) nominal assets. In both cases, risk aversion parameters are heeded, whereas they are nonexistent in the common usage of these conditions. Thus our, SPP conditions, (3), is a more accurate statement of this general UIRP.

(5) It is also implied in (4) that f = [e.sub.0][r.sub.A]/[r.sub.B] = E([E.sub.[tau]) + [r.sub.A] cov([e.sub.[tau]], [[alpha].sub.A]), where f is the forward rate implied by interest rate parity. Thus, our arguments and results are consistent with the theoretical and empirical evidence for the existence of a risk premium in the presence of risk aversion (see, for example, Cumby and Obstfld [7] and Hodrick and Srivastava [13]). The risk premium is a function of the covariance of exchange rates and country A risk adjustment factors in the above relationship.

(6) Empirical evidence of market segmentation based on these factors is by now well documented. See, for example, Jorion and Schwartz [14], Errunza and Losq [8], Alexander, Eun and Janakiramanan [5], Gultekin, Gultekin and Penati [10], Hietala [11], and Bonser-Neal, Brauer, Neal and Wheatley [6].

(7) Note that this occurrence resembles arbitrage pricing theory in the sense that APT prices securities correctly, on the average, but pricing errors can exist for individual securities. We thank Lemma Senbet for pointing this out of us.

(8) Note that segmentation and exchange and default risks play no role in this case and we end up with the MM irrelevance in the international context. It does not matter where the project is located and in which market it is financed. Financial structure is irrelevant for a foreign project as well as a domestic one. See Thomadakis and Usmen [19] for detailed derivations.

(9) There are two special cases for these sufficient conditions. One occurs when [[delta][V.sup.E]] = 0, as might be expected in dual listings of the company's stock. [delta]pV.sub.D]] > 0, then, becomes the sole condition in issuance of foreign debt. Another case is [[delta][V.sub.D]] = 0 which is relevant when riskless debt is issued and E(h) = 0. Here, [[delta]V.sub.E] [unkeyable] 0, alone determines the country of equity issues.

(10) We can reexpress (12a) as follows,

[Mathematical Expression Omitted]

The first term on the righ hand side of (12c) represents the differential value of a default-free claim in country P promising D. The second term is the differential value of the default option held by owners of the residual claim in the project, recalling that A' is the set of states where D > X(s), the states in which default occurs. Also note that [D - X(s)] is the payoff at maturity to 2 put option written on the project with a strike price equal to the promised debt payment. Hence, the first term in brackets is the value of a put option due to default in the foreign market converted into domestic currency, [e.sub.0] [P.sub.B]. The second term, on the other hand, is the value of the same put option in the domestic market, [P.sub.A]. Note that in the valuation of [P.sub.A], the underlying asset is the perfect substitute of the foreign project in market A and the strike price is no longer a constant. Remembering that risky debt involves a default option held by shareholders who can sell the project for the amount of the promised payment, it bocomes clear that the term in brackets is the differential value of the put option held by shareholders. Given segmentation, [P.sub.B] [is not equal to] [P.sub.A] (in a CAPM framework this would imply that different default premiums are assessed in the two markets for the same default risk).

(11) IF (15b) is not satisfied, country A investors are better off selling the whole project to country B investors.

(12) We should note that the sufficient conditions [[delta][V.sub.D]] > 0 and [[delta][V.sub.D] > 0 and their analysis that follows are closely tied to the location of the project, the location of the firm (its equityholders), and the location of the debt issues. The present paper analyzes the case of international ownership where equityholders are in country A, the project is in country B (thus, a foreign project), and debt holders are in country B in whose currency the debt is denominated (thus, foreign debt). (The same set of sufficient conditions (but with opposite signs) would have obtained had the project been located in country A but equityholders were in country B and had issued foreign debt in country A.)

Assuming the project is located in country B, observing equity issues in country B and debt issues in country A would require a different set of sufficient conditions, specifically [[delta][V.sub.E]] > [[delta][V.sub.D]] < 0 and [[delta][V.sub.D]] > [[delta][V.sub.O]]. In this case, the project can be considered domestic (located in the country of the equityholders), and debt can be termed "Eurobond," since payments are promised against cash flows in a currency other than the home currency of the bondholders. It should be clear that, in this case, a different partition of states into default and nondefault sets [psi] and [psi]' respectively) will be the result. Although not symmetric, the conditions for the emergence of "Eurobond" debt are analogous to those derived for foreign debt. For extensive analysis of "Eurobond" debt, see Thomadakis and Usmen [19].

(13) The preceding discussion is derived from simple algebraic arguments. It is easily shown that,

[Mathematical Expression Omitted]

where [Mathematical Expression Omitted]

By definition,

E[Lambda]'(h) + E[Lambda](h) = 0 and cov[Lambda]'(X, h) + cov[Lambda](X, h) = cov(X, h).

Thus, the first terms on the right hand side of (16c) and (16d) must necessarily be of opposite signs and can be regulated by setting D [unkeyable] E(X). Although, the second terms on the right hand side of (16c) and (16d) do not have to be of opposite sign, it is move likely that the project will support an international capital structure, if they are, being respectively positive and negative (cov[Lambda]'(X, h) > 0 and Cov[Lambda](X, h) < 0).

(14) Under risk neutrality, we have

[Mathematical Expression Omitted]

since risk adjustment factors no longer serve as differential value-weights for different states.

(15) Even if foreign equity is a more valuable alternative to foreign debt, it may not exclude firms to maintain ownership in the foreign project due to reasons such as proprietary technology or control of ownership. These motivations may lead to a venture instead.

(16) Empirical evidence of differential valuation of debt issues is obtained by studies which have documented positive stock price reaction (in the U.S.) to international bond issues of U.S. firms abroad. For example, Kim and Stulz [15] study default-free Eurobond issues and find average positive stock market reactions to the announcement of those issues. It is notable that in their results, although positive on average, there is substantial presence of cases with negative reactions, despite the fact that in most cases there was a favorable yield differential for Eurobond issues.

References

[1] M. Adler, "The Cost of Capital and Valuation of a Two-Country Firm," Journal of Finance (March 1974), pp. 119-132.

[2] M. Adler and B. Dumas, "Optimal International Acquisitions," Journal of Finance (March 1975), pp. 1-19.

[3] M. Adler and B. Dumas, "The Long Term Financial Decisions of the Multinational Corporation," in International Capital Markets, E.J. Elton and M.J. Gruber (eds.), Amsterdam, North Holland, 1975, pp. 360-387.

[4] M. Adler and B. Dumas, "International Portfolio Choice and Corporate Finance: A Synthesis," Journal of Finance (June 1983), pp. 925-976.

[5] G.J. Alexander, C.S. Eun, and S. Janakiramanan, "International Listings and Stock Returns: Some Empirical Evidence," Journal of Financial and Quantitative Analysis (June 1988), pp. 135-151.

[6] C. Bonser-Neal, G. Brauer, R. Neal, and S. Wheatley, "International Investment Restrictions and Closed-End Country Fund Prices," Journal of Finance (June 1990), pp. 523-547.

[7] R.E. Cumby and M. Obstfeld, "A Note on Exchange Rate Expectations and Norminal Interest Differentials: A Test of the Fisher Hypothesis," Journal of Finance (June 1981), pp. 697-704.

[8] V. Errunza and E. Losq, "International Asset Pricing Under Mild Segmentation: Theory and Test," Journal of Finance (March 1985), pp. 105-124.

[9] C.S. Eun and S. Janakiramanan, "A Model of International Asset Pricing With a Constraint on the Foreign Equity Ownership," Journal of Finance (September 1986), pp. 897-914.

[10] M.N. Gultekin, N.B. Gultekin, and A. Penati, "Capital Controls and International Capital Market Segmentation: The Evidence From the Japanese and American Stock Markets," Journal of Finance (September 1989), pp. 849-869.

[11] P.K. Hietala, "Asset Pricing in Partially Segmented Markets: Evidence from the Finnish Market," Journal of Finance (July 1989), pp. 697-718.

[12] J.E. Hodder and L.W. Se bet, "International Capital Structure Equilibrium," Journal of Finance (Vol. 45, 1990), pp. 1495-1516.

[13] R.J. Hodrick and S. Srivastava, "An Investigation of Risk and Return in Forward Foreign Exchange," Journal of International Money and Finance (Vol. 3:1, 1984), pp. 5-29.

[14] P. Jorion and E. Schwartz, "Integration vs. Segmentation in the Canadian Stock Market," Journal of Finance (July 1986), pp. 603-616.

[15] Y.C. Kim and R.M. Stulz, "The Eurobond Market and Corporate Financial Policy: A Test of the Clientele Hypothesis," Journal of Financial Economics (December 1988), pp. 189-205.

[16] M.E. Rubinstein, "Corporate Financial Policy in Segemented Securities Markets," Journal of Financial and Quantitative Analysis (December 1973), pp. 749-761.

[17] L.W. Senbet, "International Capital Market Equilibrium and the Multinational Firm Financing and Investment Policies," Journal of Financial and Quantitative Analysis (September 1979), pp. 455-480.

[18] R.C. Stapleton and M.G. Subrahmanyam, "Market Imperfections, Capital Market Equilibrium and Corporation Finance," Journal of Finance (May 1977), pp. 307-319.

[19] S. Thomadakis and N. Usmen, "International Market Segmentation and the Corporate Borrowing Decision," Unpublished manuscript, Baruch College, CUNY and Rutgers University, 1990.

[20] N. Usmen, "Essays on Segmentation on International Currency and Asset Markets: Implications for Corporate Finance," Unpublished Ph.D dissertation, Baruch College, CUNY, 1988.

Appendix

In this appendix, we prove that if (16a) and (16b) are valid, they imply there exists necessarily an optimal level of foreign risky debt specified by a promised payment [unkeyable] which lies between the minimum and maximum promised payment levels allowed by the project. The minimum promised payment, [D.sub.min], is the minimum state contingent cash flow: [D.min] = min{X(s)}. This minimum level defines a default-free claim and since UIRP holds, [[delta][V.sub.D]] (D = [D.sub.min]) = 0. The maximum possible promised payment, [D.sub.max], is the maximum state contingent cash flow: [D.sub.max] = max{X(s)}. In effect, this represents an issue of unlevered equity in country B. Therefore, [[delta][V.sub.D]] (D = [D.sub.max]) = [[delta][V.sub.O]]. Provided that (16a) and (16b) hold, there exists at least one level of promised payment [unkeyable], such that [D.sub.min] < [unkeyable] < [D.sub.max], and

[Mathematical Expression Omitted]

Hence, there exists an optimal level of foreign risky debt.

Stavros Thomadakis is a Professor of Finance at Baruch College, CUNY, New York. Nilufer Usmen is an Assistant Professor of Finance at Rutgers University, Newark, New Jersey.

Printer friendly Cite/link Email Feedback | |

Title Annotation: | International Finance Special Issue; includes appendix |
---|---|

Author: | Thomadakis, Stavros; Usmen, Nilufer |

Publication: | Financial Management |

Date: | Dec 22, 1991 |

Words: | 8290 |

Previous Article: | The wealth effect of international joint ventures: the case of U.S. investment in China. |

Next Article: | Cross-border liability, border taxes, and capital structure. |

Topics: |