Exchange-Rate Forecasting and Macroeconomic Analysis

Introduction

Exchange-rate forecasting is the process of estimating the future value of one currency relative to another. It is central to international financial management because exchange rates influence export competitiveness, imported inflation, foreign-currency debt, investment returns, multinational cash flows and hedging decisions. A treasury manager may need a forecast for next week, while a strategic planner may need assumptions for the next five years. No single forecasting method is consistently superior across all horizons and market conditions. Effective analysis therefore combines economic theory, market information, statistical evidence, scenario analysis and professional judgement.

An exchange rate is a price. In this chapter, the convention is that S is the domestic-currency price of one unit of foreign currency. Thus, an increase in S represents domestic-currency depreciation. For example, a movement from 1.10 to 1.15 euros per US dollar means that the dollar has appreciated against the euro, while the euro has depreciated against the dollar.

The purpose of forecasting is not to produce a certain number. It is to support decisions under uncertainty. A useful forecast identifies its assumptions, horizon, probability distribution, sources of risk and implications for action.

Learning Objectives

After studying this chapter, students should be able to:

1. Explain the economic meaning and limitations of exchange-rate forecasts.
2. Apply fundamental, technical, market-based and econometric forecasting methods.
3. Interpret purchasing power parity, interest-rate parity and balance-of-payments factors.
4. Construct scenarios and sensitivity analyses for international financial decisions.
5. Evaluate forecast accuracy, model uncertainty and behavioural influences.
6. Translate forecast information into treasury, investment and risk-management decisions.

1. The Macroeconomic Foundations of Exchange Rates

1.1 Monetary and real influences

Exchange rates are influenced by both nominal and real variables. Important macroeconomic factors include:

• relative inflation and price levels;
• interest-rate differentials;
• economic growth and productivity;
• fiscal and monetary policy;
• current-account and capital-account flows;
• commodity prices and terms of trade;
• political and sovereign risk;
• market liquidity and investor risk appetite.

The monetary approach emphasizes money supply, money demand, income and interest rates. If domestic money supply expands substantially relative to money demand, domestic prices may rise and the currency may weaken over time. The asset-market approach treats currencies as financial assets whose prices adjust to expected returns, risk and future economic conditions. The portfolio-balance approach adds that domestic and foreign assets are imperfect substitutes, so risk premia and the supply of government debt can affect exchange rates.

Short-run exchange-rate movements may be dominated by expectations, order flow, news and risk aversion. Long-run movements are more plausibly connected with inflation, productivity, trade structure and fiscal sustainability. A model that works at a monthly horizon may therefore be unsuitable for a five-year strategic forecast.

1.2 Purchasing power parity

Absolute purchasing power parity states that, after expressing prices in a common currency, comparable goods should have the same price. If P_d is the domestic price index and P_f is the foreign price index, the parity exchange rate is:

Here S is domestic currency per unit of foreign currency. The relationship assumes comparable goods, no transport costs, no taxes or trade barriers, and frictionless arbitrage. These assumptions rarely hold exactly, so absolute PPP is best viewed as a long-run benchmark.

Relative purchasing power parity focuses on changes in the exchange rate. Approximately:

where ∆S ÷ S is the proportional change in the domestic-currency price of foreign currency, and π_d and π_f are domestic and foreign inflation rates. If domestic inflation is expected to exceed foreign inflation by 3 percentage points, the domestic currency is expected to depreciate by approximately 3% under relative PPP.

PPP is useful for identifying currencies that appear misaligned over long periods. It is weak as a short-term timing tool because prices adjust slowly, non-traded goods are important, and productivity differences can produce persistent deviations.

1.3 Interest-rate parity

Covered interest parity links the spot rate, forward rate and interest rates when currency risk is hedged through a forward contract. With S quoted as domestic currency per foreign currency:

F is the forward exchange rate, i_d is the domestic interest rate and i_f is the foreign interest rate for the same maturity. The formula assumes comparable maturity, known interest rates, no transaction costs, and absence of capital controls. It is an arbitrage relationship rather than a guaranteed prediction of the future spot rate.

Uncovered interest parity removes the forward hedge and relates expected spot-rate change to the interest differential:

Equivalently, for modest rates:

Uncovered parity is a behavioural and expectations-based proposition. Empirical evidence shows that it often fails, especially over short horizons. High-interest-rate currencies may not depreciate as much as the parity condition predicts, producing the well-known carry-trade phenomenon. Risk premia, changing expectations and limits to arbitrage help explain this result.

1.4 Balance of payments and macroeconomic diagnosis

The balance of payments records transactions between residents and non-residents. A current-account deficit may create pressure for currency depreciation, but the outcome depends on how it is financed. A country with a current-account deficit financed by stable foreign direct investment may face less immediate pressure than one dependent on short-term portfolio inflows.

A macroeconomic forecast should examine:

Macroeconomic analysis should distinguish cause from correlation. A higher policy rate may strengthen a currency by increasing expected returns, but it may also signal severe inflation or financial stress. The same variable can therefore have different effects across countries and regimes.

2. Principal Forecasting Methods

2.1 Fundamental forecasting

Fundamental forecasting estimates an exchange rate from economic variables and theoretical relationships. A fundamental model may include inflation differentials, interest-rate differentials, income growth, fiscal conditions, commodity prices and external balances.

A simple conceptual specification is:

S_t is the exchange rate at time t; α is a constant; β coefficients measure estimated associations; π represents inflation; i represents interest rates; CA represents an external-balance measure; Risk represents a political or financial-risk indicator; and ε_t is the unexplained component. The exact signs and magnitudes must be estimated and tested rather than assumed universally.

Fundamental analysis is especially useful for medium- and long-term planning. Its weaknesses include uncertain coefficient estimates, structural breaks, data revisions, omitted variables and the difficulty of forecasting the explanatory variables themselves. A model can be economically sensible yet produce poor forecasts if policy regimes change.

2.2 Technical forecasting

Technical forecasting uses historical prices, returns, trends, momentum, support and resistance levels, moving averages and chart patterns. A moving average with window n is:

Smoothing can help identify trends, but it does not establish economic causation. A simple trading rule might generate a positive signal when a short-term moving average rises above a long-term moving average. Technical methods are often used at short horizons, especially when market participants react to order flow and positioning.

Their limitations include data mining, false signals, transaction costs and instability. A rule that performed well in historical data may have benefited from chance. Technical analysis should therefore be tested out of sample and evaluated after realistic spreads, financing costs and market impact.

2.3 Market-based forecasting

Market-based methods infer expectations from observable prices. The forward exchange rate is the most common benchmark. It is transparent and easy to obtain, but it is not necessarily an unbiased forecast of the future spot rate because it may include a risk premium.

Other market indicators include currency options, implied volatility, risk reversals, futures positions, cross-currency basis and sovereign bond spreads. An option-implied distribution can provide information about the market’s assessment of possible outcomes, although it reflects both expected probabilities and risk preferences.

For example, a high implied volatility indicates a wide range of priced uncertainty, not a definite direction. A risk reversal showing more expensive downside protection may indicate demand for protection against depreciation, but it may also reflect hedging pressure rather than a pure forecast.

2.4 Econometric forecasting

Econometric models formally estimate relationships in historical data. Common approaches include:

• autoregressive models, which use past exchange-rate observations;
• vector autoregressions, which model several variables jointly;
• error-correction models, which combine short-run dynamics with long-run relationships;
• panel models, which use observations across countries and time;
• volatility models, which estimate changing conditional variance;
• regime-switching models, which allow relationships to differ across states.

An autoregressive model of order 1 can be written as:

where c is a constant, ρ is the persistence parameter and ε_t is an error term with conditional mean zero. In practice, exchange-rate levels may be non-stationary, so researchers often model log changes:

where r_t is the continuously compounded exchange-rate return. Stationarity, residual behaviour, parameter stability and forecast performance must be examined before relying on results.

A model should be judged by out-of-sample performance, not merely by in-sample fit. A low error in historical data may indicate overfitting. Useful measures include mean absolute error:

and root mean squared error:

T is the number of forecast observations, S_t is the observed rate and Ŝ_t is the forecast. MAE treats errors linearly, while RMSE gives greater weight to large errors. Neither metric identifies the best model for every decision; a treasury department may care more about tail losses than average error.

3. Combining Forecasts and Managing Model Uncertainty

Forecast combination can be more reliable than selecting one model. If forecasts Ŝ_1, Ŝ_2 and Ŝ_3 are assigned weights w_1, w_2 and w_3, where weights sum to one, the combined forecast is:

Weights may reflect historical accuracy, economic rationale or equal weighting. Combination does not eliminate uncertainty, but it reduces dependence on one incorrect model.

Model uncertainty has several dimensions:

1. Parameter uncertainty: estimated coefficients are not known with certainty.
2. Variable uncertainty: future inflation, rates and growth are themselves uncertain.
3. Specification uncertainty: alternative models may imply different relationships.
4. Regime uncertainty: policy rules and market behaviour can change.
5. Measurement uncertainty: data may be revised or measured imperfectly.

Professional forecasts should communicate ranges, confidence intervals or probability-weighted scenarios where possible. A precise point estimate can create false confidence.

4. Scenario and Sensitivity Analysis

4.1 Scenario analysis

Scenario analysis evaluates coherent alternative futures rather than changing one variable mechanically. A useful set may include:

Scenarios should specify assumptions, time horizon, probability if appropriate, and decision triggers. They should not be confused with forecasts: a scenario is a conditional statement, not necessarily the most likely outcome.

4.2 Sensitivity analysis

Sensitivity analysis changes one or more assumptions to measure the effect on an outcome. Suppose a US importer must pay USD 2,000,000 in 90 days. Its home currency is the euro and the exchange rate is quoted as euros per dollar. If the rate is 0.92, the euro cost is:

If the dollar appreciates and the rate rises to 0.98:

The additional cost is EUR 120,000. The percentage increase in the euro cost is approximately:

This example assumes the invoice amount is fixed in dollars and ignores interest, transaction costs and hedging. It demonstrates why a forecast should be connected to exposure size, not considered in isolation.

Sensitivity tables can vary the exchange rate, sales volume, profit margin and hedge ratio. A treasury manager may find that operating margins are more sensitive to demand than to currency changes, or that a partial hedge provides a better balance between protection and participation in favourable movements.

5. Behavioural Effects and Forecast Interpretation

Foreign-exchange markets are not purely mechanical systems. Expectations, narratives and institutional incentives influence prices. Behavioural effects include:

• Anchoring: analysts remain attached to a previous exchange-rate level.
• Herding: investors follow widely observed trades rather than independent analysis.
• Overconfidence: forecasters underestimate uncertainty and assign excessive weight to their own model.
• Loss aversion: managers may hedge aggressively after losses but ignore comparable upside risks.
• Recency bias: recent currency movements are extrapolated too far into the future.
• Confirmation bias: evidence supporting an existing view receives disproportionate attention.

Central-bank communication can move exchange rates even before policy rates change because it alters expectations. Market narratives such as “safe haven,” “carry currency” or “commodity currency” can affect positioning and create self-reinforcing movements. These labels are useful shorthand but should not replace analysis.

A disciplined process separates data, interpretation and judgement. Analysts should record the forecast date, information set, assumptions, confidence level and reasons for revision. Forecast evaluation should consider whether the process was sound, not only whether the numerical outcome happened to be correct.

6. Applied Case: Hedging a Foreign-Currency Payable

Background

The following is an illustrative case. Meridian Components, a euro-based manufacturer, must pay a US supplier USD 5 million in six months. Management is preparing a budget and deciding whether to hedge the exposure.

Problem

The current spot rate is EUR 0.91 per USD. The six-month forward rate is EUR 0.93 per USD. The treasury team expects the euro to weaken because domestic inflation is high and the central bank may ease policy. Another team member argues that a recession could reduce US interest rates and strengthen the euro.

Analysis

The exposure is a payable in dollars. If the dollar appreciates, the euro cost rises. The forward contract fixes the cost at:

Suppose instead that the spot rate after six months is 0.88. An unhedged payment would cost EUR 4,400,000, so the forward hedge would be EUR 250,000 more expensive than paying at the future spot rate. If the future spot rate is 1.00, the unhedged payment would cost EUR 5,000,000, making the forward hedge save EUR 350,000.

The treasury team should not claim that the forward rate is the best forecast. It is a price for locking in a transaction. The relevant decision depends on budget certainty, risk tolerance, cash-flow flexibility, accounting treatment, credit exposure and the opportunity cost of favourable exchange-rate movements.

Approach and solution

Meridian could compare three policies:

1. No hedge: maximum exposure to exchange-rate movements.
2. Full forward hedge: maximum budget certainty but no benefit from euro appreciation.
3. Partial hedge: for example, hedge 60% immediately and review the remaining 40% as information develops.

A robust policy might hedge the amount that is contractually certain while leaving uncertain forecast purchases partially open. Management should establish a permitted hedge range, counterparty limits, documentation standards and escalation triggers.

Outcome

If certainty of production cost is the primary objective, the full forward hedge is defensible even if the future spot rate later proves more favourable. If management has sufficient liquidity and can tolerate variation, partial hedging may preserve some upside. The quality of the decision should be evaluated against the stated objective, not solely against hindsight.

Learning points

The case illustrates the distinction between forecasting and hedging. A forecast expresses an expectation; a hedge manages the consequences of uncertainty. Fundamental analysis may inform the decision, but it does not remove risk. Scenario analysis makes the financial consequences visible and supports governance.

7. Governance, Technology and Contemporary Practice

Forecasting systems increasingly use automated data feeds, machine learning, natural-language processing and high-frequency market indicators. These tools can identify nonlinear patterns, process news and generate scenario probabilities. However, predictive performance can deteriorate when market regimes change. Machine-learning models may also reproduce biased training data, obscure decision logic or encourage excessive trading.

Best practice includes:

• maintaining a clear data dictionary and version-controlled data;
• separating training, validation and genuinely out-of-sample testing;
• monitoring model drift and forecast errors;
• documenting assumptions and human overrides;
• using explainability appropriate to the decision;
• applying access controls and cybersecurity safeguards;
• ensuring compliance with treasury mandates, accounting rules and market-conduct requirements.

A model should support, not replace, accountable judgement. Forecasts used in budgeting, valuation or risk reporting should be reproducible and subject to independent review. Sensitive information must be handled lawfully, and analysts must avoid presenting speculative forecasts as facts.

8. Best-Practice Forecasting Workflow

A practical workflow is:

1. Define the currency quotation, exposure, horizon and decision objective.
2. Identify the relevant macroeconomic and market variables.
3. Establish benchmark forecasts using PPP, forward rates and simple statistical models.
4. Develop a fundamental narrative and test its consistency with market prices.
5. Compare technical and econometric evidence without treating historical fit as proof.
6. Produce baseline, adverse, upside and stress scenarios.
7. Translate exchange-rate outcomes into cash flows, earnings and risk measures.
8. Document assumptions, confidence and model limitations.
9. Set monitoring indicators and revision triggers.
10. Review forecast errors and improve the process.

Chapter Summary

Exchange-rate forecasting supports international financial decisions but cannot eliminate uncertainty. Fundamental methods use macroeconomic relationships such as inflation, interest rates, growth and external balances. Purchasing power parity provides a long-run benchmark, while covered interest parity explains the relationship between spot, forward and interest rates. Technical methods use market history and are often applied at shorter horizons. Market-based forecasts use forward, futures and options information but may contain risk premia. Econometric models formalize historical relationships and require careful attention to stationarity, out-of-sample performance and structural change.

Scenario analysis evaluates coherent alternative futures, while sensitivity analysis measures the effect of changing assumptions on a financial outcome. Forecast combination can reduce dependence on one model, but model, parameter, data and regime uncertainty remain. Behavioural effects—including anchoring, herding and overconfidence—can influence both markets and managerial decisions. Effective practice connects forecasts to actual exposures, communicates ranges rather than false precision, documents assumptions and uses governance controls. Forecasting and hedging are distinct: a forecast expresses an expectation, whereas a hedge manages the financial consequences of uncertain exchange rates.

Key Terms

Exchange rate: The price of one currency expressed in another currency.

Spot rate: The current exchange rate for near-immediate settlement.

Forward rate: A contracted exchange rate for settlement on a future date.

Purchasing power parity: A theory linking exchange rates to relative price levels.

Relative PPP: The proposition that exchange-rate changes approximately reflect inflation differentials.

Covered interest parity: The no-arbitrage relationship among spot rates, forward rates and interest rates.

Uncovered interest parity: A proposition linking expected exchange-rate changes with interest-rate differentials.

Fundamental forecasting: Forecasting based on economic and financial determinants.

Technical forecasting: Forecasting based on historical prices, trends and market patterns.

Market-based forecasting: Inferring expectations from traded financial prices.

Econometric model: A statistical model used to estimate and forecast economic relationships.

Stationarity: A statistical property in which a process has stable distributional characteristics over time.

Model uncertainty: Uncertainty about variables, parameters, specification or economic regimes.

Scenario analysis: Evaluation of outcomes under coherent alternative assumptions.

Sensitivity analysis: Measurement of how an outcome changes when assumptions change.

Risk premium: Compensation investors require for bearing risk.

Volatility: A measure of the variability of an asset’s returns or price.

Hedging: Using a financial position to reduce exposure to an adverse price movement.

Behavioural bias: A systematic pattern of judgement that can affect decisions or market prices.

Out-of-sample test: Evaluation using observations not used to estimate a model.

Review Questions

5 Short Answer Questions

1. Why is purchasing power parity generally more useful as a long-run benchmark than as a short-run forecast?
2. Distinguish between a forward rate and an expected future spot rate.
3. State two limitations of technical forecasting.
4. What is the purpose of out-of-sample forecast evaluation?
5. Distinguish scenario analysis from sensitivity analysis.

5 Long Answer Questions

1. Explain how inflation, interest rates, growth and the balance of payments can influence an exchange rate.
2. Compare fundamental, technical, market-based and econometric forecasting methods.
3. Explain covered and uncovered interest parity, including their assumptions and limitations.
4. Discuss how behavioural biases can affect professional exchange-rate forecasting.
5. Explain why a firm may combine several forecasts rather than rely on one model.

3 Analytical / Application Questions

1. A domestic importer must pay USD 3 million. Compare its domestic-currency cost at exchange rates of 0.90, 0.95 and 1.02 domestic currency units per dollar. Explain how the results could influence its hedge ratio.
2. A country has higher inflation, a widening current-account deficit and falling foreign reserves, but its policy interest rate is increased sharply. Develop a baseline and stress scenario for its currency and explain why the interest-rate increase may not guarantee appreciation.
3. Design a forecasting process for a multinational planning a three-year investment in a foreign country. Specify the models, data, scenarios, governance controls and performance measures you would use.

Multiple Choice Questions with Answers

Question 1

Under relative purchasing power parity, if domestic inflation is expected to exceed foreign inflation by 4 percentage points, the domestic currency is expected to:

A. Appreciate by approximately 4%

B. Depreciate by approximately 4%

C. Remain fixed by definition

D. Change only if the forward rate changes

Correct Answer: B. Depreciate by approximately 4%

Question 2

With the quotation S as domestic currency per unit of foreign currency, an increase in S represents:

A. Domestic-currency appreciation

B. Foreign-currency depreciation

C. Domestic-currency depreciation

D. No change in either currency

Correct Answer: C. Domestic-currency depreciation

Question 3

Which relationship is primarily an arbitrage condition involving spot rates, forward rates and interest rates?

A. Covered interest parity

B. Absolute purchasing power parity

C. The moving-average rule

D. The balance-of-payments identity alone

Correct Answer: A. Covered interest parity

Question 4

A forward exchange rate should not automatically be interpreted as:

A. A contractual price for future settlement

B. A market-observable exchange-rate quote

C. An unbiased prediction of the future spot rate

D. A possible hedging instrument

Correct Answer: C. An unbiased prediction of the future spot rate

Question 5

Which measure gives relatively greater weight to large forecast errors?

A. Mean absolute error

B. Root mean squared error

C. Median error

D. Forecast horizon

Correct Answer: B. Root mean squared error

Question 6

A company owes USD 2 million and the exchange rate is quoted as EUR per USD. If the rate rises from 0.90 to 0.95, the euro cost of the payable will:

A. Decrease by EUR 100,000

B. Increase by EUR 100,000

C. Remain EUR 2 million

D. Decrease by EUR 50,000

Correct Answer: B. Increase by EUR 100,000

Question 7

Which statement best describes sensitivity analysis?

A. It evaluates only the most likely future

B. It changes assumptions to measure effects on an outcome

C. It guarantees the forecast will be accurate

D. It eliminates model uncertainty

Correct Answer: B. It changes assumptions to measure effects on an outcome

Question 8

Which is an example of model uncertainty?

A. The invoice amount is contractually fixed

B. The analyst is unsure which variables and specification to use

C. The settlement date is known

D. The currency quotation is clearly defined

Correct Answer: B. The analyst is unsure which variables and specification to use

Question 9

A technical forecast based on a short-term moving average is most directly using:

A. Historical price information

B. Current-account accounting rules

C. Absolute PPP only

D. A legal exchange-control requirement

Correct Answer: A. Historical price information

Question 10

Which behavioural bias involves following the actions of other market participants?

A. Anchoring

B. Herding

C. Loss recognition

D. Stationarity

Correct Answer: B. Herding