The Portfolio Balance Model of Exchange Rate Determination: Assets

The portfolio balance model explains the exchange rate as the price that clears the market for financial assets, not as the price that equalises goods prices or money supplies. Investors hold a portfolio of domestic money, domestic bonds and foreign bonds; the exchange rate adjusts so that they are content with the stock of each that exists.

Its distinguishing assumption is that domestic and foreign bonds are imperfect substitutes. Investors regard them as different assets — because of currency risk, political risk, default risk or simple home preference — and therefore require compensation for holding more of one than they would otherwise choose. That compensation is a risk premium, and its existence is what separates this family of models from every model that assumes uncovered interest parity holds.

The setup

A representative investor holds three assets: domestic money, which pays no interest; domestic bonds, paying the domestic rate; and foreign bonds, paying the foreign rate and denominated in foreign currency.

Total wealth is the sum of the three, with foreign bonds valued at the current exchange rate. The investor allocates wealth between them according to relative expected returns and risk, and equilibrium requires that the desired holdings equal the available stocks.

The exchange rate enters in two places, and this is what makes the model distinctive. It determines the domestic-currency value of foreign bonds already held, so a movement changes the composition of wealth without any transaction. And it adjusts to clear the market when desired and actual holdings diverge.

How the model works

An increase in the domestic money supply. Investors hold more money than they want and rebalance into bonds, domestic and foreign. Buying foreign bonds requires foreign currency, so the domestic currency depreciates. The result agrees with the monetary models, by a different route.

An open-market purchase of domestic bonds. The central bank buys domestic bonds for money. Investors hold fewer domestic bonds and more money, rebalance partly into foreign bonds, and the currency depreciates.

A sterilised intervention. The central bank sells foreign currency and simultaneously offsets the effect on the money supply by buying domestic bonds. The money supply is unchanged, so a monetary model predicts no effect on the exchange rate. The portfolio balance model predicts an effect, because the composition of the bond stock has changed: investors now hold more domestic bonds and fewer foreign ones, and the relative price must move to make them content.

This is the model's sharpest empirical claim and the reason it survives. Whether sterilised intervention works is a question on which central banks and monetary models disagree, and the portfolio balance channel is the standard theoretical account of how it could.

The current account as the dynamic

The model's long-run mechanism runs through the current account, and this is where it goes beyond a static asset-market story.

A current account surplus means the country is accumulating net foreign assets. Domestic residents hold a growing stock of foreign bonds, which they will not hold at unchanged relative returns, so the domestic currency must appreciate to restore balance — either by raising the domestic-currency value of the existing stock or by changing expected returns.

A deficit works in reverse: net foreign assets fall, foreign residents accumulate claims on the domestic economy, and the currency depreciates.

The adjustment is therefore gradual and stock-driven and not instantaneous. This is a genuine advantage over models in which the exchange rate jumps to a new equilibrium and stays there, because observed exchange rates do move persistently over years in ways that flow variables alone do not explain.

The risk premium, stated properly

The premium is the quantity the model turns on, and it is worth writing out because a great many answers refer to it without defining it.

Uncovered interest parity says the expected return on domestic and foreign bonds should be equal once expected currency movement is taken into account: the domestic interest rate equals the foreign rate plus the expected depreciation. Where the two differ, the gap is the risk premium.

In the portfolio balance model the premium is not a constant. It depends on the relative supplies of the two assets and on how much risk investors are willing to bear. Increase the outstanding stock of domestic bonds and investors must be paid more to hold them, so the premium rises; and because the premium enters the arbitrage condition, the exchange rate moves.

That dependence is the model's whole content in one sentence: relative asset supplies move the risk premium, and the risk premium moves the exchange rate. Everything else is bookkeeping around it.

Two consequences follow. A government financing a deficit by issuing domestic bonds is, on this account, exerting downward pressure on its currency independently of anything happening to the money supply. And a central bank that wants to influence the exchange rate without changing interest rates has exactly one channel available, which is the composition of the bond stock.

Where the model is most useful

Three applications survive the empirical weakness, and they are the ones worth naming in an answer.

Explaining persistent misalignment. A country running current account deficits for a decade accumulates external liabilities, and the model predicts a currency under sustained downward pressure that has nothing to do with the interest differential at any moment. This matches observed behaviour better than models that treat the rate as a jump variable.

Explaining why intervention sometimes works. Central banks intervene, and monetary models say they cannot affect anything without changing the money supply. Practitioners generally believe intervention has some effect, and this is the only mainstream model that supplies a mechanism.

Explaining safe-haven behaviour. When risk appetite falls, investors want less of the riskier asset at any given return, which is a shift in the demand side of exactly the relationship the model describes. Currency movements during episodes of stress are readable in these terms even where the model's quantitative predictions fail.

What it assumes about investors

Two assumptions about behaviour sit underneath the asset-market machinery, and both are worth stating because both are contestable.

The first is that investors optimise over a portfolio with a defined risk preference, so that a change in relative supplies produces a predictable change in required return. This is the standard finance assumption and it carries the usual objection: actual currency positions are held by a mixture of hedgers, exporters, reserve managers and speculators with different objectives, and only some of them are balancing a portfolio in the sense the model requires.

The second is that expectations of future exchange rates are formed consistently. The premium is defined as the gap between the interest differential and expected depreciation, so any test of the model is jointly a test of how expectations are measured. Survey-based expectations and model-consistent expectations give different answers, and neither is obviously right.

An answer that notes the joint-hypothesis problem — that the model and the expectations assumption are tested together and cannot be separated — is making the point that most of this literature turns on.

Against the monetary models

Three differences matter and are the usual examination question.

Substitutability. The monetary approach assumes domestic and foreign bonds are perfect substitutes, so uncovered interest parity holds and no risk premium exists. The portfolio balance model assumes imperfect substitutability, which admits a risk premium and makes the relative supply of assets matter.

What the exchange rate clears. In the monetary approach it clears the money market, so relative money supplies and demands determine it. Here it clears the whole asset market, so relative bond supplies matter too.

The role of wealth and the current account. The monetary models have no place for net foreign asset positions. The portfolio balance model makes them central, which gives it a dynamic story the others lack.

Dornbusch's overshooting model sits between them: it keeps perfect substitutability but adds sticky goods prices, producing a different explanation for the same observed volatility. An answer that treats the three as a family of responses to the same puzzle, rather than as unrelated theories, reads considerably better.

The evidence, and why it is thin

The model is theoretically well specified and empirically weak, and stating that plainly is more credible than defending it.

Risk premia are hard to measure. The premium is unobservable and must be inferred from the difference between the forward rate and the subsequent spot rate, which conflates the premium with expectational error. Tests therefore rest on an auxiliary assumption about expectations that is itself contested.

Asset supply data are poor. The model requires the stocks of domestic and foreign bonds held by each sector, and the data are incomplete, infrequent and inconsistently defined across countries.

The estimated effects are small. Where studies do find a portfolio balance effect from sterilised intervention, the magnitudes are generally modest and not always durable, which is consistent with the model being correct in mechanism and minor in scale.

The Meese and Rogoff problem applies here too. No structural exchange-rate model, this one included, reliably outperforms a random walk at short horizons. That finding is now decades old and has survived repeated attempts to overturn it.

It is also worth saying that the model's assumptions have become harder to test rather than less plausible. Capital mobility has risen enormously since the 1970s, which should reduce home preference and push bonds towards closer substitutability; at the same time the stock of government debt outstanding has grown, which magnifies exactly the supply effect the model describes. The two changes work in opposite directions, so the net prediction is ambiguous, and the data cannot settle it.

The reasonable conclusion for an assignment is that the model provides the clearest available explanation of how sterilised intervention could work and of why net foreign asset positions should matter, while sharing the whole literature's inability to forecast.