Like most classic relationships in economics, the IRP
relationship has its origins in the writings of the early political
economists of the nineteenth century. It was John
Maynard Keynes, however, who popularized the expression.
In its most general form, IRP considers the
returns available to investors (or, equivalently, the funding
costs available to borrowers) from the following
three related alternatives:
Alternative 1: Borrowing or lending in domestic
currency at rate: (1þRi,t).
Alternative 2: Borrowing or lending in foreign
currency and hedging the proceeds with foreign
exchange forwards at rate: (1þR* )( i,t Fi,t/St).
Alternative 3: Borrowing or lending in foreign
currency without hedging the proceeds at expected
rate: E(St þ i/St)(1þR* ). i,t
In the expressions above, R* is the foreign currency i,t
interest rate for borrowing or lending for i periods at
time, t; St is the exchange rate at time t, expressed in units
of domestic currency per foreign currency (i.e., as a direct
quote) and Fi,t is the forward rate at time t for purchasing
foreign currency i periods in the future.
As Richard Levich discusses elsewhere in this Handbook,
IRP in its canonical form is equivalent to the proposition
that alternatives 1 and 2 should be equal.
Equation (15.1) summarizes this equality:
ð1 þ Ri;tÞ¼ð1 þ R
i;tÞ
Fi;t
St
ð15:1Þ
In other words, investors should earn precisely the
same returns – and borrowers should incur precisely
the same costs – whether they invest (or borrow) at the
domestic interest rate or at the foreign interest rate
whenever they use forwards to hedge (or ‘cover’) their
foreign exchange rate risk. For this reason, this type of
IRP is often referred to as covered interest parity (CIP).
The logic behind the CIP is simple. Assume, for example,
that investors can earn higher interest rates in the domestic
currency, so that R* is less than i, t Ri,t. If the foreign
currency did not trade at a premium in the
forward market, so that Fi,t was higher than St, arbitrageurs
could make a riskless profit by borrowing money
in the foreign currency, exchanging it in the spot market,
investing the proceeds in domestic currency, and
ultimately paying off their loan at maturity with the foreign
currency they purchase in the forward market at
the undervalued rate.
While this fairly straightforward arbitrage logic underpins
CIP, the IRP argument is often taken one step
further by asserting that alternatives 1 and 3 should be
equal as well. In other words, investors (or borrowers)
should earn the same expected returns (or incur the same
expected costs) in foreign and domestic currency even if
they do not purchase forward contracts and instead simply
sell their foreign currency principal at maturity at the
prevailing spot exchange rate. Equation (15.2) summarizes
this equality:
ð1 þ Ri;tÞ¼ð1 þ R
i;tÞE Stþi
St
ð15:2Þ
Since investors (or borrowers) do not purchase forward
contracts, they inevitably leave their foreign currency
exposure unhedged (or ‘uncovered’). For this
reason, this type of IRP is often referred to as uncovered
228 15. OPPORTUNISTIC FOREIGN CURRENCY DEBT ISSUANCE
II. FORCES BEHIND GLOBALIZATION
interest parity (UIP). While the CIP rests on fairly clear
no-arbitrage logic, the UIP relies on much less firm theoretical
footing. Leaving foreign currency unhedged entails
considerable risks given the inherent uncertainty of
future exchange rates. If borrowers and investors are risk
neutral, then the UIP follows from the same logic outlined
for the CIP above. If not, however, Eq. (15.2) need
not hold in equilibrium.
The implications of IRP for opportunistic FC
debt issuance are simple. Obviously, if IRP held, there
would be no cross-currency borrowing cost differences
for potentially opportunistic bond issuers to exploit.
Assume, for example, that foreign currency interest
rates, Ri
* , are lower than domestic currency interest ,t
rates, Ri,t. If the UIP held, Eq. (15.2) would imply that
any borrowers who sought to take advantage of relatively
low foreign currency interest rates by issuing
bonds abroad would ultimately be disappointed. By
the time their bonds get matured, they would find themselves
having to repay their debt in a currency that had
appreciated by exactly enough to offset the lower interest
payments they had previously enjoyed. This relationship
is readily apparent by dividing both sides of
Eq. (15.2) by (1þRi
* ), subtracting 1, and rearranging ,t
terms to get
Ri;t R
i;t
1 þ R
i;t
¼ E Stþi St
St
ð15:3Þ
The left-hand side of Eq. (15.3) is the percentage interest
cost savings that borrowers would enjoy by borrowing
in foreign currency. The right-hand side, on the other
hand, is the percentage loss they would take in the currency
markets by having to repay their debt with more
expensive foreign currency.
By precisely the same logic, if the CIP held, Eq. (15.1)
would imply that borrowers who sought to take advantage
of relatively low foreign currency interest rates by
borrowing abroad and using forward contracts to hedge
their currency risk would be disappointed as well. In this
case, they would find that the forward premium that
they had to pay to purchase foreign currency in the forward
market would exactly offset the interest savings
they enjoyed. As above, this relationship is readily apparent
by dividing both sides of Eq. (15.1) by (1þRi
* ), ,t
subtracting 1, and rearranging terms to get
Ri;t R
i;t
1 þ R
i;t
¼ Fi;t St
St
ð15:4Þ
As Eq. (15.4) makes clear, the CIP relationship requires
that the forward rate is set so that any apparent
interest cost savings are immediately offset by the premium
borrowers must pay in the forward market to purchase
the foreign currency that they need to repay their
foreign currency loans at maturity.
Evidence on the UIP
There are ample reasons to suspect, however, that IRP
does not hold over the time periods relevant to bond issuers’
borrowing decisions. Even at short horizons, tests
of the UIP routinely fail. In fact, the persistent failure of
the UIP is one of the most enduring mysteries in international
finance. There is an extensive literature dating
back to the early studies, by Bilson (1981), Longworth
(1981), and Meese and Rogoff (1983), that demonstrate
that as a predictor of future exchange rate movements,
the interest rate premium is at best useless and at worst
perverse. In a survey of 75 published estimates, in fact,
Froot and Thaler (1990) find that the vast majority of estimates
suggest that relatively low-interest-rate currencies
actually depreciate over time. In other words,
opportunistic borrowers in these currencies would benefit
not only from lower interest costs during the life of
their loans but also by being able to repay these loans
with relatively less expensive foreign currency in the
future. (For a discussion of the various theoretical
explanations for this finding, see elsewhere in this
Handbook.)
While the failure of the UIP is almost universally accepted
among economists, as Chinn and Meredith
(2004) point out, virtually all empirical analyses consider
relatively short-term interest rates and corresponding
currency movements. In the case of the early research
into the UIP completed in the 1980s and 1990s, this focus
was inevitable. Exchange rates for major currencies only
began floating after the collapse of the Bretton Woods exchange
rate regime in the 1970s, so researchers lacked
sufficiently historical data with which to examine potential
longer horizon trends. With the benefit of an additional
two decades’ worth of data points, however,
Chinn and Meredith investigate whether or not the
data are more consistent with the UIP at much longer
horizons. For most major currencies, they run simple
regressions that compare the ultimate appreciation or
depreciation rate observed over the subsequent 5- and
10-year periods to the difference between local government
benchmark bond yields and the corresponding
US rates. Unlike virtually all tests on short-term data,
they are unable to reject the hypothesis that the coefficient
on the interest differential is 1 (the standard test
for the UIP). Instead, they find coefficients that are positive
and often statistically indistinguishable (though
smaller in magnitude) than 1. While these results are
more consistent with the UIP, they are perhaps best
interpreted as failing to provide additional evidence
against the UIP rather than providing evidence in favor
of it. In any case, given the overwhelming body of evidence
against the UIP that has been built up over decades
of research, even well-informed borrowers
would seem to have ample reason to doubt that their
LONG-TERM IRP 229
II. FORCES BEHIND GLOBALIZATION
potential interest costs savings from opportunistic uncovered
FC debt issuance would be offset by foreign exchange
losses when they ultimately went to repay their
foreign currency principal at maturity.
Evidence on the CIP
With regard to the CIP, the evidence is more complex
and potentially more interesting. Unlike the case of the
UIP, the overwhelming body of evidence on the CIP over
the past 50 years has been supportive – so much so, in
fact, that the CIP has come to be taken as a benchmark
for evaluating the mobility of capital between markets.
However, even in situations in which currencies are
freely convertible and the flow of capital is unrestricted
between borders, the CIP holds only as an approximation.
As Frenkel and Levich (1975, 1977) note, the CIP
is enforced by arbitrage transactions, and real-life arbitrageurs
inevitably face a variety of costs in executing
the strategies they employ. In the presence of these transaction
costs, the CIP holds only within the range of noarbitrage
‘neutral bands,’ whose width is determined
by the transaction costs that covered interest arbitrageurs
face.
To illustrate the point, consider the simple example
outlined above in which the foreign currency interest
rate appears too low, given domestic currency interest
rates and the pri