This chapter examines the differences between spot and forward currency interventions by way of a model that incorporates the motivations of different groups of currency market participants. The model concludes that, although spot and forward intervention have equivalent effects on the exchange rate, they have differing effects on forward-implied interest rate spreads and on capital flows. The model offers insights into the relative effectiveness of each type of intervention in response to different external shocks. Regressions that use available forward intervention data from Brazil, Mexico, and Peru support the model.
Introduction
Forward markets are tempting targets for central bank interventions. In contrast with spot interventions, forward-market interventions are self-sterilizing and do not require transfers of reserves. A further advantage is that forward interventions to sell foreign (and buy local) currency are not limited by the supply of foreign reserves. However, a central bank that proposes to use a forward intervention must consider whether that operation is likely to have the same effect on the exchange rate as the corresponding spot intervention. A related question is whether forward intervention may constitute an instrument that is distinct from spot intervention, with different effects on interest rates and capital flows.
Two recent developments highlight the importance of understanding forward-market intervention. First, a growing number of central banks have used forward intervention either in lieu of, or to supplement, spot market interventions. Among emerging markets, the Central Bank of Brazil is perhaps the most experienced in this regard, but other major central banks, including those of Mexico and Peru, have been increasingly active in forward markets. Second, substitutability between spot and forward markets, although still high, appears to have diminished, particularly since the global financial crisis. Consequently, the tight covered interest parity (CIP) relationship has weakened in many currency markets.
This study begins with brief accounts of forward-market intervention, before and after the financial crisis, followed by a review of the covered interest parity hypothesis. It offers possible reasons for the weakening of the CIP relationship, which—until recently—was thought to determine prices in currency forward markets. Subsequently, a model is presented that explicitly incorporates different sources of demand for currency forwards and explores its implications for spot and forward interventions. A brief empirical section tests some of the model’s implications against recent episodes of forward-market intervention in Brazil, Mexico, and Peru. The conclusion suggests a set of preliminary rules for monetary authorities to use when deciding between spot and forward-market intervention.
Forward Pricing and Covered Interest Parity
A currency forward (or, in an exchange with tradable contracts, a currency future) is a contract to exchange a given amount of one currency for another at a set price at a specific future date.1 The relationship between the spot exchange rate and the forward exchange rate is traditionally governed by CIP. In logarithmic form, this can be expressed as follows:
where i and i* are logs of the gross domestic and foreign interest rates, respectively, (approximately equivalent to the respective net interest rates), and e and f are logs of the spot and forward exchange rates (all of them for a common duration, such as one year), expressed in units of local currency per dollar. The relationship has been thought to hold a high degree of precision through an arbitrage argument. If f were to rise much above the value entailed by CIP (so that the local currency becomes weaker than implied by CIP), cross-border arbitrageurs would have an incentive to borrow domestically at i, trade for foreign currency (by convention, US dollars) at exchange rate e, invest abroad at interest rate i*, and purchase domestic currency back at the weaker rate f, which enables them to repay the domestic loan while realizing a riskless profit. In principle (and different from the case of uncovered interest parity), this constitutes a riskless arbitrage, given that the forward rate is already determined at the time of the trade. A parallel arbitrage argument would hold if f were to fall below the CIP-implied level.
Episodes of Forward Intervention
Some central banks have benefited from this relationship to conduct foreign exchange interventions in the forward rather than the spot market. If CIP is presumed to hold, a forward sale or the purchase of foreign currency contracts should have the same effect on the spot exchange rate as an equivalent amount of currency traded on the spot market. Some advanced economy central banks have followed this practice, including the Reserve Bank of Australia, which has sometimes used forward operations when intervening in currency markets (see Becker and Sinclair 2004).2 The advantage of this approach is that spot interventions need to be sterilized if the operation is not to affect domestic monetary conditions, whereas forward interventions have no effect on the money supply. A further possible advantage is that currency futures markets are, at times, more liquid than the corresponding spot markets.
Brazil has frequently used forward intervention, but has not always treated it as a close substitute for spot intervention. During two periods of strong capital inflows, in 2006–08 and 2009–11, the Central Bank of Brazil accompanied spot market US dollar purchases with a buildup of its long US dollar forward position, which increased its net forward US dollar position from –$12 billion to +$11 billion between January 2009 and April 2011. These interventions were intended not only to keep the spot exchange rate from strengthening too much but also to respond to a rising demand for the Brazilian real in the forward market, in some cases from “carry trade” investors. In conducting its forward operations, the Central Bank of Brazil was also mindful of the effect on the cupom cambial, the implied US dollar interest rate available to domestic investors that borrow in local currency to obtain dollars and repurchase the domestic currency in the forward market to close out the loan (for a detailed empirical and analytic account, see Garcia and Volpon 2014).Under CIP this rate should equal the US dollar interest rate at the appropriate term, although in practice, this has never been the case; the implied US dollar interest rate in Brazil has always been higher than the onshore US dollar interest rate, which means that Brazil’s currency has been stronger in the forward market than would be implied by CIP. By purchasing US dollars forward during periods of capital inflows, the Central Bank of Brazil could reduce the differential between onshore and offshore US dollar interest rates.
Korea’s central bank has also intervened in currency forward markets in response to imbalances (Baba and Shin 2011). As in Brazil, the domestic currency in Korea, the won, has generally been stronger in the forward market than would be implied by CIP. However, in Korea, the demand to sell US dollars and to buy local currency forwards arises largely from exporters (shipbuilders, in particular) who wish to match anticipated US dollar revenues from sales abroad to their local-currency production costs. The forward contracts they purchase are sold by domestic banks, who match the resulting long forward US dollar positions by borrowing US dollars in the spot market from foreign banks. During the 2007–09 global financial crisis, when foreign banks were temporarily unwilling to lend US dollars to Korean banks, the deviation from CIP intensified, which left the won much stronger in the forward market than would be implied by CIP. The Bank of Korea responded with a combination of policy measures, including the sale of US dollar swaps, which had the effect of filling spot market demand for US dollars while reducing the deviation from CIP.3
In recent years, intervention by emerging market central banks in forward markets has often been in response to negative external shocks. This was the case for Brazil in 2013–14—during a period of economic slowdown and some capital outflows—when the Central Bank of Brazil undertook net forward US dollar sales of almost $90 billion, equivalent to one-fourth of its reserve stock. Faced with negative external shocks, Mexico’s central bank has, since the end of 2016, engaged in some forward peso purchases. Peru’s central bank has managed a forward position for several years. The section on empirical results examines the forward operations of these three central banks in detail.
Cost of Arbitrage and the Weakening of Covered Interest Parity
The tight adherence to CIP that formerly characterized major currency markets has markedly weakened since the global financial crisis (Figure 5.1). At the same time, divergences from CIP,4 prevalent in emerging markets since well before the crisis, have not diminished, and in some cases have intensified. One widely cited explanation for the change in market behavior is that the cost of arbitraging CIP divergences has increased with the postcrisis increase in the cost of bank capital caused by tighter regulations. Tighter controls on cross-border banking exposures in emerging markets may also work to limit arbitrage. Concurrently, demand for currency hedges by nonfinancial corporations has risen, which increases the amount of arbitrage activity needed to restore CIP. These influences are reflected in the model of spot and forward interventions subsequently discussed.
Presentation of Model
The model here recalls the approach of Ghosh, Ostry, and Chamon (2016), who make the case for treating spot intervention as an instrument distinct from the policy interest rate. In lieu of the standard uncovered interest parity assumption, the authors adopt a modified condition that relates capital flows to departures from uncovered interest parity, but it does not assume that uncovered interest parity holds instantaneously and at all times. Their approach is extended here with a similar flow condition in the forward market. An important consequence of these joint assumptions is that neither uncovered interest parity nor CIP is constrained to hold at all times. The possibility of significant departures from CIP reflects conditions that have been prevalent in financial markets since the global financial crisis.
The model is presented in static form, but it presumes a long-term equilibrium, including a long-term equilibrium log real exchange rate ē, assumed to be zero.5 CIP is not presumed to hold in the long-term equilibrium, although divergences from CIP drive arbitrage flows. Both spot and forward interventions are presumed to be zero in the long-term equilibrium. Scalar values that are not presumed to equal zero in the long-term equilibrium are designated as δ or α. Values corresponding to temporary external shocks are designated by ε and are set to zero in the long-term equilibrium.
The following eight equations provide the basic model structure. Exchange rates are shown in units of local currency per US dollar, so that higher values of e and f correspond to local currency depreciation in spot and forward markets, respectively. Spot and forward exchange rates e and f, and domestic and foreign interest rates i and i* are given in logs (i and i* are logs of the gross domestic and foreign real interest rates). The real exchange rate e may differ from the long-run real exchange rate ē.
Current Account (CA)
A depreciation (e up) improves the CA. The eCS term reflects the effect of commodity price shocks on the CA balance.
Capital Flow Condition (Modified Uncovered Interest Parity)
A portion of capital inflows is determined by the modified interest parity condition. A higher real domestic interest rate I, relative to the real foreign rate i*, attracts inflows from nonhedged yield-seeking investors. These investors adjust their expected returns by the expected real appreciation of the exchange rate e, equal to e – ē (or simply e with ē set to zero). Additional capital flows not directly linked to interest rates—such as foreign direct investment—are represented by δOFF whereas ϵFS is a financial shock, assumed to be zero in long-term equilibrium and negative in the event of an externally based financial crisis. A positive figure for g corresponds to an inflow on the capital/financial account.
Synthetic Dollar Borrowing Condition
Market participants that need US dollar funding but lack access to offshore capital markets can borrow US dollars synthetically by taking out a loan in local currency, exchanging the proceeds for foreign currency, and repurchasing the local currency necessary to repay the loan in the forward market. This operation generates an outflow on the capital/financial account, and an equivalent inflow in the forward market.
Arbitrage Trade
These flows arbitrage divergences from CIP. If CIP holds, then the quantity in the parentheses is zero. To the extent that the interest rate available to a hedged foreign investor purchasing assets in the domestic market (i + e —f) exceeds the rate at which that investor borrows in the foreign market (i*), these flows will be larger. A positive value of hAT corresponds to an inflow on the capital account and an outflow in the forward market.
Hedging Demand
Demand for forward hedges (such as from exporters or importers) is taken to be independent of exchange rate expectations or interest rates. A positive value for hHD in the model is assumed to entail hedging demand to buy dollars and to sell local currency. This would normally be the case for an importer who seeks to ensure the US dollar value of expected future receipts. Accordingly, a positive value of hHD corresponds to an outflow in the forward market. It is also possible, however, that exporters could dominate the hedging market (see the earlier Korean example), in which case, δHD would take a negative value, as would hHD in long-term equilibrium or in the absence of a shock to hedging demand.
Forward-Market Demand
This combines the three sources of demand for currency forwards, equations (5.3) through (5.5).
Balance of Payments Condition
This is the standard balance of payments identity adapted to the model.
Forward-Market Equilibrium Condition
where W represents the central bank’s net forward position.
Algebraic manipulation of equations (5.1) through (5.8) allows derivation of separate closed-form expressions for e and f, in terms of exogenous variables (see annex 5.1). These are summarized in the following (Jacobian) matrix of first derivatives (5.13):
Note that the term
Results from the Model
Spot and Forward Interventions Have the Same Effect on the Exchange Rate
From equation (5.9), it follows that the effect of a unit change in reserves (that is, in a spot intervention) on the exchange rate is the same as that of a unit change in the central bank’s forward position. That is, the model implies that forward intervention is as effective as spot intervention in shifting the spot exchange rate.6 It also follows from (5.13) that spot intervention has the same effect on the forward rate as it has on the spot rate. An immediate consequence is that spot intervention should have no effect on the divergence from CIP (i* – i – e + f), or “basis spread.”
Spot and Forward Interventions Have Different Effects on the Forward Rate and Therefore on the Divergence from CIP
By contrast, a forward intervention does not have the same effect on spot and forward rates. A change in the central bank’s forward position has a greater effect on the forward rate than on the spot rate. This implies that a central bank purchase of US dollar forwards will increase the basis spread (i* – i – e + f). In contrast, a central bank sale of US dollar forwards reduces the basis spread, and if the basis spread is negative (as is often the case), it increases the divergence from CIP. This acts to discourage synthetic US dollar borrowing (an outflow) and encourage arbitrage trading. Forward intervention, although it is as effective as spot intervention in shifting the exchange rate, also affects synthetic US dollar borrowing costs (essentially the onshore US dollar interest rate) and capital flows.
Why should this asymmetry between spot and forward interventions exist? Consider the equations for the balance of payments equilibrium (5.9) and the forward-market equilibrium (5.10), depicted in Figure 5.2. If the central bank sells reserves and takes no other action, as panel 1 in Figure 5.2 shows, then the balance of payments equilibrium line shifts to the left. However, the forward-market equilibrium line is unchanged—forward equilibrium will continue to hold only if any change in e is matched by a change in
Forward Dollar Sales Induce Equivalent Capital Inflows
Forward US dollar sales induce equivalent capital inflows, whereas spot US dollar sales do not have this effect. Both operations cause the spot exchange rate to strengthen, which reduces both the CA surplus and the modified uncovered interest parity inflows. In the case of spot US dollar sales, the loss in inflows is equivalent to the reserves supplied to the market. Selling US dollars forward, on the other hand, has the same effect on CA and modified uncovered interest parity inflows, but it brings arbitrage flows onshore and does not cost reserves. Equation (5.8) shows a decline of $100 million in the central bank’s forward position (ΔW = –100), which entails a decline in (hSDB – hAT) of $100 million, relative to the counterfactual in which there is no forward intervention. Equation 5.7 shows that (CA + g) must fall by $100 million, since ΔR = 0, and (hAT – hSDB) increases by $100 million. By contrast, a reserves sale of $100 million has no effect on (hAT – hSDB), since the basis spread does not change, so it must induce the same decline in (CA + g) that is prompted by a forward sale.8
The Determination of Whether to Intervene in the Spot or Forward Market Depends on the Shock
Comparison of equations (5.11) and (5.12) in the annex shows that both commodity (εCS) and financial (εFS) shocks have the same effect on spot and forward rates in the model. Neither of them, therefore, affects the basis spread. However, a shock to hedging demand (εHS) has a greater effect on the forward rate than on the spot rate.
Economic considerations would, in general, preclude intervention to support the domestic currency in response to an adverse commodity shock, unless there are financial stability reasons to mitigate the shock. Otherwise, the model points to an intervention strategy of using either spot or forward intervention in response to a financial shock, and of favoring forward intervention in response to a hedging shock. A fourth potential shock (not represented in the model by an ε variable) would correspond to an increase or reduction in the marginal propensity to undertake synthetic dollar borrowing, γSDB, perhaps because of a structural change in onshore funding conditions. Here the model leads to a clear recommendation, as this term does not appear in equation (5.11)(which expresses the exogenous determinants of the exchange rate), but does appear in equation (5.12) for the forward intervention. The correct response to a change in this propensity would be forward rather than spot intervention.
Empirical Results
To gauge the empirical applicability of the model, tests were conducted using recent daily spot and forward intervention data from Brazil (Figure 5.3), Mexico (Figure 5.4), and Peru. Several factors tend to limit the empirical investigations of the effects of forward intervention. Until recently, few central banks have published data on forward interventions, particularly at a daily frequency. In addition, the exact form of forward operation varies considerably by country, as do the instruments used (such as nondeliverable forwards, options, swaps, futures, and so on). Tests of intervention in either the spot or forward market tend to be subject to high endogeneity—it is often difficult to separate the effects of the circumstances that prompted the intervention (such as capital flight) from the effects of the intervention itself.
Brazil: Exchange Rate, Spot Dollar Sales, Swap Sales, and Forward Dollar Sales, by Quarter, 2012–17
Source: Central Bank of Brazil.Brazil: Exchange Rate, Spot Dollar Sales, Swap Sales, and Forward Dollar Sales, by Quarter, 2012–17
Source: Central Bank of Brazil.Brazil: Exchange Rate, Spot Dollar Sales, Swap Sales, and Forward Dollar Sales, by Quarter, 2012–17
Source: Central Bank of Brazil.
Mexico: Exchange Rate, Basis Spread, Spot Dollar Sales, and Forward Dollar Sales, by Quarter, 2012–17
Source: Bank of Mexico.Mexico: Exchange Rate, Basis Spread, Spot Dollar Sales, and Forward Dollar Sales, by Quarter, 2012–17
Source: Bank of Mexico.Mexico: Exchange Rate, Basis Spread, Spot Dollar Sales, and Forward Dollar Sales, by Quarter, 2012–17
Source: Bank of Mexico.One set of tests focuses on the model’s conclusion that spot and forward operations have similar effects on the exchange rate. A second set examines whether there is an identifiable effect of forward intervention on basis spreads, as the model predicts. The sample period of 2013–17 for all three countries corresponds to a phase of capital outflows, particularly for Brazil, and for Mexico during the latter part of the period. Accordingly, intervention was generally aimed at supporting the domestic currency, although in Brazil it was also focused, at times, on affecting the implied onshore US dollar interest rate.
The tests offer stronger confirmation of the basis spread hypothesis than they do of the equivalence of spot and forward intervention. For Brazil and Mexico, forward intervention variables are correctly signed (Table 5.2). For Brazil, the sale of currency swaps is shown to affect the basis spread, within a 99 percent confidence interval, with the expected sign.9 For Mexico, forward US dollar sales have a significant effect, within a 95 percent confidence interval and with the expected sign, on the basis spread. As forward interventions in Mexico did not begin until late 2016, this result reflects the experience of 2016–17. Tests for Peru do not show a significant effect of spot or forward sales on the basis spread.
Spot Exchange Rate Regression
Spot Exchange Rate Regression
| Brazil | Mexico | Peru | |
|---|---|---|---|
| Exchange rate, lagged | -0.07*** (0.02) |
-0.02 (0.02) |
0.06** (0.02) |
| Dollar sale | -0.000001 (1.80) |
-0.000016 (0.000069) |
0.000006 (0.000004) |
| Dollar sale, lagged | -0.0000004 (0.000002) |
-0.002*** (0.000070) |
-0.0000034 (0.000004) |
| Forward dollar sale | 0.000002 (0.000001) |
-0.000005 (0.000044) |
0.0000024**” (0.000001) |
| Forward dollar sale, lagged | 0.0000002 (0.000001) |
0.000012 (0.000043) |
0.0000009 (0.000001) |
| Local currency swaps | 0.0000007 (0.000002) |
||
| Local currency swaps, lagged | -0.00000291* (0.000002) |
||
| Australian dollar spot exchange rate | -0.83*** (0.18) |
-5.20*** (0.66) |
-0.12** (0.05) |
| Colombian peso exchange rate | 0.0002*** (0.00004) |
0.001*** (0.0001) |
0.00012*** (0.00001) |
| Mexican peso exchange rate | 0.05*** (0.01) |
0.007** (0.002) |
|
| Peruvian sol exchange rate | 0.28*** (0.09) |
0.98*** (0.33) |
|
| Turkish lira exchange rate | 0.17*** (0.04) |
1.51*** (0.15) |
0.023 (0.01) |
| Brazilian real exchange rate | 0.89*** (0.10) |
0.02*** (0.01) |
|
| Euro exchange rate | -0.41 (0.52) |
-0.0663 (0.0430) |
|
| Japanese yen exchange rate | -0.01*** (0.004) |
||
| Constant | 0.000387 (0.001) |
-0.00018 (0.003) |
0.00013 (0.0002) |
| R2 | 0.31 | 0.45 | 0.24 |
| Adjusted R2 | 0.30 | 0.44 | 0.24 |
| N | 1163 | 1203 | 1193 |
| F-statistic | 44.0 (0.00) | 81.27 (0.00) | 35.52 (0.00) |
Spot Exchange Rate Regression
| Brazil | Mexico | Peru | |
|---|---|---|---|
| Exchange rate, lagged | -0.07*** (0.02) |
-0.02 (0.02) |
0.06** (0.02) |
| Dollar sale | -0.000001 (1.80) |
-0.000016 (0.000069) |
0.000006 (0.000004) |
| Dollar sale, lagged | -0.0000004 (0.000002) |
-0.002*** (0.000070) |
-0.0000034 (0.000004) |
| Forward dollar sale | 0.000002 (0.000001) |
-0.000005 (0.000044) |
0.0000024**” (0.000001) |
| Forward dollar sale, lagged | 0.0000002 (0.000001) |
0.000012 (0.000043) |
0.0000009 (0.000001) |
| Local currency swaps | 0.0000007 (0.000002) |
||
| Local currency swaps, lagged | -0.00000291* (0.000002) |
||
| Australian dollar spot exchange rate | -0.83*** (0.18) |
-5.20*** (0.66) |
-0.12** (0.05) |
| Colombian peso exchange rate | 0.0002*** (0.00004) |
0.001*** (0.0001) |
0.00012*** (0.00001) |
| Mexican peso exchange rate | 0.05*** (0.01) |
0.007** (0.002) |
|
| Peruvian sol exchange rate | 0.28*** (0.09) |
0.98*** (0.33) |
|
| Turkish lira exchange rate | 0.17*** (0.04) |
1.51*** (0.15) |
0.023 (0.01) |
| Brazilian real exchange rate | 0.89*** (0.10) |
0.02*** (0.01) |
|
| Euro exchange rate | -0.41 (0.52) |
-0.0663 (0.0430) |
|
| Japanese yen exchange rate | -0.01*** (0.004) |
||
| Constant | 0.000387 (0.001) |
-0.00018 (0.003) |
0.00013 (0.0002) |
| R2 | 0.31 | 0.45 | 0.24 |
| Adjusted R2 | 0.30 | 0.44 | 0.24 |
| N | 1163 | 1203 | 1193 |
| F-statistic | 44.0 (0.00) | 81.27 (0.00) | 35.52 (0.00) |
Basis Spread Regression
Basis Spread Regression
| Brazil | Mexico | Peru | |
|---|---|---|---|
| Basis spread, lagged | -0.39*** (0.02) |
-0.13*** (0.03) |
-0.35*** (0.03) |
| Spot exchange rate | -0.7 (0.72) |
0.040** (0.02) |
0.57 (2.70) |
| Spot exchange rate, lagged | -0.98 (0.74) |
0.14*** (0.02) |
9 2*** (2.70) |
| Spot dollar sale | -0.0000011 (0.0001) |
0.00016*** (0.0001) |
0.0005 (0.0004) |
| Spot dollar sale, lagged | 0.000047 (0.0001) |
0.000082 (0.0001) |
0.0006 (0.0004) |
| Forward dollar sale | -0.000027 (0.00004) |
0.000037 (0.00004) |
0.00016* (0.0001) |
| Forward dollar sale, lagged | -0.000021 (0.00004) |
-0.000071** (0.00004) |
0.00021* (0.0001) |
| Local currency swaps | 0.00019*** (0.00005) |
||
| Australian dollar basis spread | 0.04 (0.12) |
0.01 (0.01) |
0.08 (0.10) |
| Australian dollar basis spread, lagged | 0.033 (0.12) |
||
| Colombian peso basis spread | 0.049* (0.03) |
||
| Colombian peso basis spread, lagged | 0.055** (0.03) |
||
| Peruvian sol basis spread | 0.0003 (0.00250) |
||
| Constant | 0.001 (0.02) |
-0.00028 (0.0024) |
-0.00017 (0.03) |
| R2 | 0.17 | 0.07 | 0.13 |
| Adjusted R2 | 0.16 | 0.06 | 0.12 |
| N | 1035 | 1178 | 1167 |
| F-statistic | 16.23 (0.00) | 10.17 (0.00) | 19.69 (0.00) |
Basis Spread Regression
| Brazil | Mexico | Peru | |
|---|---|---|---|
| Basis spread, lagged | -0.39*** (0.02) |
-0.13*** (0.03) |
-0.35*** (0.03) |
| Spot exchange rate | -0.7 (0.72) |
0.040** (0.02) |
0.57 (2.70) |
| Spot exchange rate, lagged | -0.98 (0.74) |
0.14*** (0.02) |
9 2*** (2.70) |
| Spot dollar sale | -0.0000011 (0.0001) |
0.00016*** (0.0001) |
0.0005 (0.0004) |
| Spot dollar sale, lagged | 0.000047 (0.0001) |
0.000082 (0.0001) |
0.0006 (0.0004) |
| Forward dollar sale | -0.000027 (0.00004) |
0.000037 (0.00004) |
0.00016* (0.0001) |
| Forward dollar sale, lagged | -0.000021 (0.00004) |
-0.000071** (0.00004) |
0.00021* (0.0001) |
| Local currency swaps | 0.00019*** (0.00005) |
||
| Australian dollar basis spread | 0.04 (0.12) |
0.01 (0.01) |
0.08 (0.10) |
| Australian dollar basis spread, lagged | 0.033 (0.12) |
||
| Colombian peso basis spread | 0.049* (0.03) |
||
| Colombian peso basis spread, lagged | 0.055** (0.03) |
||
| Peruvian sol basis spread | 0.0003 (0.00250) |
||
| Constant | 0.001 (0.02) |
-0.00028 (0.0024) |
-0.00017 (0.03) |
| R2 | 0.17 | 0.07 | 0.13 |
| Adjusted R2 | 0.16 | 0.06 | 0.12 |
| N | 1035 | 1178 | 1167 |
| F-statistic | 16.23 (0.00) | 10.17 (0.00) | 19.69 (0.00) |
The tests offer no more than very qualified support to the hypothesis that spot and forward interventions have equvalent effects on the exchange rate (Table 5.1). The case of Peru does support this hypothesis to some extent, in that the responses of the exchange rate to each type of intervention are similar. The regressions for Brazil and Mexico in this regard are inconclusive.
Conclusion
This chapter elucidates the features of forward intervention with a model that incorporates the motivations of different actors in forward markets. This model permits analysis of the potentially different effects of spot and forward intervention in currency markets. Key conclusions are that spot and forward intervention should have similar effects on the exchange rate, that forward interventions influence the basis spread, and that forward interventions are capable of inducing capital flows, whereas spot interventions do not have this effect. The chapter offers data on when either spot or forward intervention is most appropriate, an area that can be expanded with future research.
Annex 5.1. Algebraic Manipulation of Equations (5.1) through (5.8)
The balance of payments condition, equation (5.7), can be rewritten as follows:
Similarly, the forward-market equilibrium condition, equation (5.8), can be rewritten as follows:
Solving the balance of payments condition, equation (5.7′), for the exchange rate e, yields the following:
where Ω = β + γg + γSDB + γAT.
Similarly, solving the forward-market equilibrium condition, equation (5.8′), for f, yields the following:
Substituting equation (5.10) into (5.9) to eliminate forward rate f yields a solution for exchange rate e in terms of exogenous variables:
Conversely, substituting equation (5.9) into (5.10) results in an expression for f in terms of the exogenous variables:
References
Baba, Naohiko, and Ilhyock Shim. 2011. “Dislocations in the Won-Dollar Swap Markets During the Crisis of 2007–09.” BIS Working Paper 344, Bank for International Settlements, Basel.
Becker, Chris, and Michael Sinclair. 2004. “Profitability of Reserve Bank Foreign Exchange Operations: Twenty Years after the Float.” Research Discussion Paper 2004–06, Reserve Bank of Australia, Sydney.
Garcia, Marcio, and Tony Volpon. 2014. “DNDFS: A More Efficient Way to Intervene in FX Markets?” Working Paper 501, Stanford University Center for International Development, Stanford, CA.
Ghosh, Atish R., Jonathan D. Ostry, and Marcos Chamon. 2016. “Two Targets, Two Instruments: Monetary and Exchange Rate Policies in Emerging Market Economies.” Journal of International Monetary and Finance 60 (C): 172–96.
Nedeljkovic, Milan, and Christian Saborowski. 2017. “The Relative Effectiveness of Spot and Derivatives Based Intervention: The Case of Brazil.” Working Paper 17/11, International Monetary Fund, Washington, DC.
This is not precisely true in the case of nondeliverable forwards, for which the gains or losses at the maturity of a forward contract are settled in a given currency, and there is no exchange of currencies. For this study, nondeliverable forwards are treated as standard forward contracts.
To avoid changes in domestic monetary conditions, the Reserve Bank of Australia would combine a spot US dollar sale with a currency swap, resulting in zero net change in its US dollar holdings and a reduction in its forward US dollar position.
Sales of a US dollar swap by the central bank would entail selling US dollars in the spot market and buying them back in the forward market.
Divergence from CIP determines what is known to market participants as the “basis spread.” The terms are used interchangeably in the text.
The long-term equilibrium exchange rate ē could change in response to a permanent change in fiscal policy, for example, or to a permanent external shock. However, these scenarios are not considered in the present study.
This conjecture is supported empirically in the case of Brazil by Nedeljkovic and Saborowski (2017).
The divergence from CIP must widen to increase arbitrage inflows and decrease synthetic US dollar borrowing, thereby compensating for the negative effect of a stronger exchange rate on trade and other capital inflows (g).
It is assumed that changes in the central bank’s forward position are analogous to changes in the spot reserves position, in that a change in the stock of reserves or forwards is assumed to be permanent. This implies that forward contracts are rolled over on expiration, absent further intervention.
This corresponds to results obtained by Garcia and Volpon (2014), who note the differential impact of spot and forward intervention, and the effect of forward intervention in inducing capital inflows.
