Chapter 3

Forecasting Exchange Rates

Objectives
This chapter stresses the value of reliable forecasts, but suggests that reliable forecasts cant always be obtained. Because no technique has been singled out as superior, various techniques are mentioned. Whatever techniques the MNC chooses, it sho uld monitor performance ove r time. The specific objectives are to:
2

Objectives
explain how firms can benefit from forecasting exchange rates; describe the common techniques used for forecasting; explain how forecasting performance can be evaluated.

Why Firms Forecast Exchange Rates

MNCs need exchange rate forecasts for their:
hedging decisions, short-term financing decisions, short-term investment decisions, capital budgeting decisions (long-term investment decisions), long-term financing decisions, and earnings assessment.

Forecasting Techniques
The numerous methods available for forecasting exchange rates can be categorized into four general groups:
technical, fundamental, market-based,and mixed.

Technical Forecasting
Technical forecasting involves the use of historical data to predict future values. It includes statistical analysis and time series models. Speculators may find the models useful for predicting day-to-day movements. However, since they typically focus on the near future and rarely provide point/range estimates, they are of limited use to MNCs.
6

Fundamental Forecasting
Fundamental forecasting is based on the fundamental relationships between economic variables and exchange rates. A forecast may arise simply from a subjective assessment of the factors that affect exchange rates. A forecast may be based on quantitative measurements ( with the aid of regression models and sensitivity analysis) too.
7

Fundamental Forecasting
Example 1:
Assume that the corporate objective is to forecast the percentage change in the British pound with respect to U.S. dollar during the next quarter. For simplicity, assume that the firms forecast for the British pound is dependent on only two factors that affect the pounds value:

1. Inflation in the United States relative to inflation in the United Kingdom. 2. Income growth in the United States relative to income growth in the United Kingdom.
8

Fundamental Forecasting
*

The first step is to determine how these variables have affected the percentage change in the pounds value based on historical data.
The regression equation can be defined as BPt =b0 + b1INFt-1 + b2INCt-1 + t Where BP is the quarterly percentage change in the British pound value

INFt-1 is the previous quarterly percentage change in the inflation differential (U.S. inflation rate minus British inflation rate) INCt-1 is the previous quarterly percentage change in the income growth differential (U.S. Income growth minus British income growth) b0 is a constant b1 measures the sensitivity of BPt to changes in INFt-1 b2 measures the sensitivity of BPt to changes in INCt-1 t represents an error term

Fundamental Forecasting
* The second step is to generate the values of regression coefficients (b0, b1, and b2) by using the historical data of BP, INF and INC. To illustrate, assume the following values:
b0 = .002 b1 = .8 b2 =1.0
10

Fundamental Forecasting
*The third step: use the coefficients to forecast. (Assume the most recent quarterly percentage change in INFt-1 is 4 percent and INCt-1 is 2 percent.) The forecast for BPt is BPt =b0 + b1INFt-1 + b2INCt-1 = .002 + .8(4%) + 1(2%) = .2% + 3.2% + 2% = 5.4%
11

Fundamental Forecasting
* Conclusion: given the current figures for inflation rates and income growth, the pound should appreciate by 5.4 percent during the next quarter. * Note that this example is simplified to illustrate how fundamental analysis can be implemented for forecasting. A full-blown model might include more than two factors, but the application would still be similar.
12

Fundamental Forecasting
Use of sensitivity analysis for fundamental forecasting. Example 2: Phoenix Corp. develops a regression model to forecast the percentage change in the Mexican pesos value. It believes that the real interest rate differential and the inflation differential are the only two factors that affect exchange rate movements, as shown in 13 this regression model:

Fundamental Forecasting
et = a0 + a1INTt + a2INFt-1 + t
where
et = percentage change in the pesos exchange rate over period t INTt = real interest rate differential over period t INFt-1= inflation differential in the previous period a0, a1, a2 = regression coefficients t = error term

14

Fundamental Forecasting
* Assume that regression analysis has provided the following estimates for the regression coefficients:
a0 = .001 a1 = -.7 a2 = .6

* To forecast the pesos percentage change over the upcoming period, INTt and INFt-1 must be estimated. Assume that INFt-1 was 1 percent. However, INTt is not known at the beginning of the period and must therefore be forecasted. Assume that Phoenix Corp. has develop the following probability distribution for INTt:

15

Fundamental Forecasting
Probability Possible Outcome 20% -3% 50% -4% 30% -5% * A separate forecast of et can be developed from each possible outcome of INTt as follows: Forecast of INT Forecast of et Probability
-3% -4% -5% .1%+(-.7)(-3%)+.6(1%)=2.8% .1%+(-.7)(-4%)+.6(1%)=3.5% .1%+(-.7)(-5%)+.6(1%)=4.2% 20% 50% 30%

16

Fundamental Forecasting
Use of PPP for fundamental forecasting. Known relationships like the PPP can be used for forecasting Example 3: The U.S. inflation rate is expected to be 1 percent over the next year, while the Australian inflation rate is expected to be 6 percent. According to PPP, the Australian dollars exchange rate should change as follows: ef = (1 + IU.S. ) / (1 + If) -1 = 1.01 / 1.06 1 ~ -4.7%
17

Fundamental Forecasting
This forecast of the percentage change in the Australian dollar can be applied to its existing spot rate to forecast the future spot rate at the end of the year. If the existing spot rate of the Australian dollar is $.20, the expected spot rate at the end of one year will be : E(St+1) = St (1 + ef ) = $.20 [ 1 + (-.047)] = $.1906
18

Fundamental Forecasting
Note that in reality, the inflation rates of two countries over an upcoming period are uncertain and would have to be forecasted when using PPP to forecast future exchange rate. This complicates the use of PPP to forecast future exchange rates. Even if the inflation rates were known with certainty, problems may arise for the following reasons:

the timing of the impact of inflation on trade behavior is not known for sure, prices may be measured inaccurately, trade barriers may disrupt the trade patterns that should emerge, and other influential factors may exist.

19

*Comparison of IRP, PPP, and IFE Theories

The IRP theory focuses on why the forward rate differs from the spot rate and on the degree of difference that should exist. The PPP theory and IFE theory focuses on how a currencys spot rate will change over time. PPP theory suggests that the spot rate will change in accordance with inflation differential; IFE theory suggests that it will change in accordance with interest rate differential.

20

Fundamental Forecasting
In general, fundamental forecasting is limited by :
the uncertain timing of the impact of the factors, the need for forecasts of factors with instantaneous impact, the possibility that other relevant factors may be omitted from the model, and changes in the sensitivity of currency movements to each factor over time.

21

Market-Based Forecasting
Market-based forecasting involves developing forecasts from market indicators. Usually, either the spot rate or the forward rate is used, since speculation should push the rates to the level that reflect the market expectation of the future exchange rate.
22

Market-Based Forecasting
Long-term exchange rate forecasts can be derived from long-term forward rates. Since long-term forward contracts have low trading volumes and are not widely quoted, the interest rates on risk-free instruments can be used to determine what the forward rates should be according to IRP for long-term forecasting.
23

Market-Based Forecasting
* Example: The U.S. five-year interest rate is currently 10 percent annualized, while the British five-year interest rate is 13 percent. The five-year compounded return on investments in each of these countries is computed as follows: Country Five-year Compounded Return U.S. (1.10) 5 1 = 61% 5 U.K. (1.13) 1 = 84%
24

Market-Based Forecasting
Thus, the appropriate five-year forward rate premium (or discount ) of the British pound would be ) / (1 + iU.K.) 1 = 1.61 / 1.84 1 = -.125, or 12.5% The results suggests that the five-year forward rate of the pound should contain a 12.5 percent discount.
U.S. 25

p = (1+ i

Mixed Forecasting
Mixed forecasting refers to the use of a combination of forecasting techniques. The actual forecast is a weighted average of the various forecasts developed.

26

Evaluation of Forecast Performance

An MNC that forecasts exchange rates should monitor its performance over time to determine whether its forecasting procedure is satisfactory. The MNC may also want to compare the various forecasting methods.

27

Evaluation of Forecast Performance

One measure of forecast performance is the absolute forecast error as a percentage of the realized value:
| forecasted value realized value | realized value

Over time, MNCs are likely to have more confidence in their forecasts when they know the mean error for their past forecasts.
28

Evaluation of Forecast Performance

The ability to forecast currency values may vary with the currency of concern. In particular, the value of a less volatile currency is likely to be forecasted more accurately.

29

Forecast Bias
If the forecast errors are consistently positive or negative over time, then there is a bias in the forecasting procedure.

30

Forecast Bias
The following regression model can be used to test for forecast bias:
realized = a0 + a1 forecast + m Where a0 = intercept

a1 = regression coefficient m = error term

31

Forecast Bias
*If the predictor is unbiased, the intercept should equal zero, and the regression coefficient should equal 1.0. *If a0 = 0 and a1 is significantly less than 1.0, this implies the predictor is systematically overestimating the spot rate, for example, if a0 = 0 and a1 = .90, the realized value is estimated to be 90 percent of the forecasted value. Vice versa. *If a predictor is found to be biased, the estimated a0 and a1 values can be used to correct the systematic 32 error.

Graphic Evaluation of Forecast Performance

Forecast performance can be examined with the use of a graph that compares forecasted values with the realized values for various time periods.

33

Graphic Evaluation of Forecast Performance

$.28 $.26 $.24 $.22 $.20 $.18 $.16 $.14 $.12 $.10

.
Region of downward bias (underestimating)

Realized Value (in U.S. dollars)

Perfect forecast line

Region of upward bias (overestimating)

. . .

$.10 $.12 $.14 $.16 $.18 $.20 $.22 $.24 $.26 $.28 $.30

Predicted Value (in U.S. dollars)

34

Graphic Evaluation of Forecast Performance

If the points appear to be scattered evenly on both sides of the perfect forecast line, then the forecasts are said to be unbiased. Note that a more thorough assessment can be conducted by separating the entire period into subperiods.

35

Comparison of Forecasting Techniques

The different forecasting techniques can be evaluated
graphically - by comparing the distances from the perfect forecast line, or statistically - by computing the mean of the absolute forecast errors, and then using a t-test or a nonparametric test to determine whether there is a significant difference in the accuracy of the forecasting techniques.
36

Exchange Rate Volatility

MNCs also forecast exchange rate volatility. This enables them to specify a range (confidence interval) and develop best-case and worst-case scenarios along with their point estimate forecasts. Popular methods for forecasting volatility include: the use of recent exchange rate volatility,
37

Exchange Rate Volatility

the use of a historical time series of volatilities

(there may be a pattern in how the exchange rate volatility changes over time), and the derivation of the exchange rates implied standard deviation from the currency option pricing model.

38

Questions and Applications

*1.You are hired as a consultant to assess a firms
ability to forecast. The firm has developed a point forecast for two different currencies, presented in the table below. The firm asks you to determine which currency was forecasted with greater accuracy.
Period Yen Forecast Actual Yen Value Pound Forecast Actual Pound Value 1 $.0050 $.0051 $1.50 $1.51 2 .0048 .0052 1.53 1.50 3 .0053 .0052 1.55 1.58 4 .0055 .0056 1.49 1.52

39

Questions and Applications

*2. Assume that the four-year annualized interest rate in the United States is 9 percent and the four-year annualized Interest rate in Singapore is 6 percent. Assume interest rate parity holds for a four-year horizon. Assume that the spot rate of the Singapore dollar is $.60. If the forward rate is used to forecast exchange rates, what will be the forecast for the Singapore dollars spot rate in four years? What percentage appreciation or depreciation does this forecast imply over the four-year period?
40

Questions and Applications

3. If the euro appreciates substantially against the dollar during a specific period, would market-based forecasts have overestimated or underestimated the realized values over this period? Explain.

41

Questions and Applications

4. Royce Co. Is a U.S. firm with future receivables one year from now in Canadian dollars and British pounds. Its pound receivables are known with certainty, and its estimated Canadian dollar receivables are subject to a 2 percent error in either direction. The dollar values of both types of receivables are similar. There is no chance of default by the customers involved. Royces treasurer says that the estimate of dollar cash flows to be generated from the British pound receivables subject to greater uncertainty than that of Canadian dollar receivables. Explain the rationale for the treasurers statement.

42

Questions and Applications

5. Cooper, Inc., a U.S.-based MNC, periodically obtains euros to purchase German products. It assesses U.S. and German trade patterns and inflation rates to develop a fundamental forecast for the euro. How could Cooper possibly improve its method of fundamental forecasting as applied to the euro?
43

Questions and Applications

* 6. New York Co. has agreed to pay 10 million Australian dollars (A$) in two years for equipment that it is importing from Australia. The spot rate of the Australian dollar is $.60. The annualized U.S. interest rate is 4%, regardless of the debt maturity. The annualized Australian dollar interest rate is 12% regardless of the debt maturity. New York plans to hedge its exposure with a forward contract that it will arrange today. Assume that interest rate parity exists. Determine the amount of U.S. dollars that New York Co. will need in 2 years to make its payment.
44

Questions and Applications

* 7. Purdue Co. (based in the U.S.) exports cable wire to Australian manufacturers. It invoices its product in U.S. dollars, and will not change its price over the next year. There is intense competition between Purdue and the local cable wire producers that are based there. Purdues competitors invoice their products in Australian dollars and will not change their prices over the next year. The annualized risk-free interest rate is presently 8% in the U.S., versus 3% in Australia. Today the spot rate of the Australian dollar is $.55.
45

Questions and Applications

Purdue Co. uses this spot rate as a forecast of future exchange rate of the Australian dollar. Purdue expects that revenue from its cable wire exports to Australia will be about $2 million over the next year. If Purdue decides to use the international Fisher effect rather than the spot rate to forecast the exchange rate of the Australian dollar over the next year, will its expected revenue from its exports be higher, lower, or unaffected? Explain.
46