Capital controls, the dual exchange rate, and devaluation
📄 Summarized from the full manuscript · Human-reviewed for faithfulness before publication
In brief
Does devaluation matter in the long run? The textbook monetary approach says no -- once private agents rebuild their money balances, nothing real has changed. Obstfeld shows that conclusion rests on an overlooked assumption: that central-bank reserves earn no interest. Once reserves are allowed to earn the market rate, a devaluation that produces a reserve-accumulating balance-of-payments surplus permanently raises national income -- and, under capital controls, the economy can have several long-run equilibria, though even a large devaluation can never flip a surplus into a deficit.
What this paper finds — and why it matters
This 1986 Journal of International Economics paper by Maurice Obstfeld re-examines the classic monetary-approach claim that devaluation under capital controls has no long-run real effects, by adding one realistic assumption the earlier literature had dropped: central-bank foreign reserves earn interest. The paper first lays out a frictionless benchmark – an optimizing representative-agent economy with full private capital mobility and interest-bearing reserves – and shows devaluation is fully neutral there, even in the short run, because instantaneous private portfolio adjustment exactly offsets the central bank’s reserve transactions and leaves the discounted value of resources available to the public unchanged. It then imposes capital controls: private residents may hold a fixed pool of foreign-currency “investment currency” bonds at a floating financial exchange rate, while current-account transactions clear at an officially pegged commercial rate, and the central bank intermediates between the two. In this dual-rate economy, an unanticipated devaluation of the official commercial rate is no longer neutral: it generates an immediate balance-of-payments surplus as residents rebuild real money balances, the counterpart of which is a permanent increase in the central bank’s interest-earning reserves – and because that extra reserve income is rebated to the public, long-run consumption and real balances end up permanently higher than before the devaluation. A second result is that the economy’s dynamics need not have a unique steady state: because the domestic real interest rate (tied to the floating financial rate) can diverge from the world rate available to the central bank, the consumption/real-balances system can admit several stationary equilibria, some saddle-path stable and some fully unstable. The paper shows, however, that this multiplicity does not overturn its central qualitative conclusion: starting from any stable equilibrium, even a large devaluation always improves the balance of payments and can never “leap-frog” an unstable steady state into producing a deficit instead. A by-product of the analysis is a closed-form pricing formula for the floating financial exchange rate as the discounted stream of future consumption-weighted real yields, consistent with contemporary consumption-based asset-pricing theory.
Summary of a classic paper, AI-assisted and human-reviewed. See the linked original for the authoritative claims and full conditions.
Questions & answers
Q1. What is the paper’s central methodological departure from the earlier monetary-approach literature on devaluation?
The paper modifies the standard intertemporal optimizing framework for devaluation analysis (as used, for example, by Calvo 1981) by assuming that central-bank foreign reserves earn interest at the market rate, rather than earning no interest as in the traditional monetary-approach literature (Introduction, pp. 1-2). The author shows this single change has two “important” consequences: devaluation becomes nonneutral in the long run even absent any other friction, and the economy’s steady state need not be unique.
Q2. Why is devaluation fully neutral in the paper’s frictionless capital-mobility benchmark (Section I)?
With unrestricted private capital mobility and interest-bearing central-bank reserves, an unanticipated devaluation causes an incipient excess demand for domestic real balances that is eliminated immediately as residents sell central-bank foreign bonds to rebuild their money holdings; this portfolio reshuffling moves foreign-currency claims from private to official ownership but leaves the economy-wide total of private bonds plus reserves, b₀ + f₀, unchanged (Sec. I, pp. 7-8, eq. 10). Because the shadow value of wealth λ̂₀ is a function only of this unchanged total (together with other predetermined variables), equilibrium consumption and real balances are unaffected by the devaluation, so it is “fully neutral” – agents can instantaneously restore real balances while keeping consumption constant because they still receive, via transfers, the same foreign interest income as before.
Q3. How does the paper set up the dual exchange-rate regime once capital controls are imposed?
Under capital controls (Sec. II), private residents can no longer freely buy or sell bonds abroad; instead they hold a fixed pool b̄ of “investment currency” – foreign bonds traded only domestically at an endogenously floating financial (dual) exchange rate q, which can differ from the officially pegged commercial rate used for current-account transactions E (Sec. II, pp. 9-10). The central bank continues to peg E by intermediating current-account foreign-exchange flows, accumulating or decumulating interest-bearing reserves f as the public’s desired real balances grow or shrink over time.
Q4. What is the closed-form solution for the dual (financial) exchange rate, and how is it interpreted?
Integrating the first-order condition linking the marginal utility of real balances to the dual exchange rate’s expected capital gain yields q₀ = ∫₀^∞ [u_c(c_t,m_t)/u_c(c₀,m₀)] · r · exp(−∫₀^t ρ_s ds) dt (Sec. II, eq. 25, p. 13) – the dual exchange rate equals the present value of the constant physical yield r on investment-currency bonds, discounted both at consumers’ marginal rates of substitution between present and future consumption and at the path of market real interest rates – a formula the paper notes is “consistent with recent consumption-based models of asset pricing” (Grossman and Shiller 1981).
Q5. Why can the model have multiple, and even unstable, long-run equilibria?
A stationary equilibrium requires the Engel curve relating steady-state consumption to real balances, c̄ = y + rb̄ + ρ[m̄ − (D₀/P)], to intersect the locus along which desired real balances are constant, ū_m/ū_c = δ; because the Engel curve can cross this locus from above or below more than once, the system can have several stationary points* (Sec. III, pp. 13-14, Figure 1). The paper derives a precise saddle-path stability criterion (inequality 27): a stationary point is a saddle (locally stable) if the Engel curve cuts the ṁ=0 locus from below, and is unstable if it cuts it from above – a condition the author shows has the intuitive interpretation that no stable trajectory exists unless an increase in real balances raises desired consumption by more than the world interest rate times the increase, i.e., unless a current-account surplus is self-correcting rather than self-reinforcing.
Q6. What happens to the dual exchange rate along the adjustment path to a stationary equilibrium?
Along a convergent (saddle) path, a balance-of-payments deficit (consumption above its long-run level, marginal utility of consumption below its long-run level) is accompanied by a dual exchange rate above its long-run value that is falling — i.e., the financial currency is appreciating — while a surplus is accompanied by a rising (depreciating) dual rate (Sec. III, pp. 14-15), a result the paper proves rigorously in an appendix using the linearized dynamics around the steady state (Sec. V appendix, p. 20-21, eqs. A1-A3, using the strict concavity of the utility function to sign ω₁/ω₂ > 0).
Q7. What is the central result of the nonneutral-devaluation analysis (Section IV), both locally and globally?
Locally, near a stable steady state, an unanticipated devaluation shifts both the ṁ=0 and ċ=0 loci in a way that unambiguously raises long-run consumption and real balances (Sec. IV, pp. 16-17, Figure 5): the economy initially experiences a fall in real balances and a balance-of-payments surplus, but because the resulting reserve accumulation earns interest at rate ρ*, national income and therefore long-run consumption and money demand end up permanently higher than before the devaluation. The paper then shows (Sec. IV, pp. 18-19, Figure 6) that this conclusion survives globally even with multiple stationary states: a large devaluation always moves the economy to the nearest stable equilibrium in the direction associated with a balance-of-payments surplus, and can never “leap-frog” an intervening unstable steady state to produce a long-run deficit instead – because with both goods normal, an Engel curve cannot cross the ṁ’=0 locus from above between the initial point and a surplus outcome.
Q8. What other sources of devaluation nonneutrality does the paper distinguish itself from?
The paper explicitly isolates the reserve-interest channel from other conventional sources of nonneutrality it deliberately abstracts from: it notes that devaluation acts as an implicit tax reducing national income and consumption when reserves earn no interest (Frenkel and Rodriguez 1975); that nominal-bond effects on real wealth (Boyer 1977; Lapan and Enders 1978) and finite-horizon overlapping-generations redistribution can also make devaluation nonneutral (Sec. II, p. 9, footnote 9) – but the paper’s own nonneutrality result holds even in a representative-infinite-horizon setting with no nominal-bond wealth effects, precisely because of the interest-bearing-reserves assumption alone.
Key terms in this paper
Definitions below follow the paper's own usage.
- Dual (financial) exchange-rate pricing formula
- The paper's closed-form result (Sec. II, eq. 25) that the floating financial exchange rate applicable to the economy's fixed pool of investment-currency bonds equals the present value of the bonds' constant physical yield, discounted by the path of consumers' marginal rates of substitution and market real interest rates -- a devaluation-under-capital-controls analog of consumption-based asset-pricing formulas.
- Devaluation nonneutrality via interest-bearing reserves
- The paper's central departure from the prior monetary-approach literature (Secs. I-IV): because central-bank foreign reserves earn interest at the world rate, a devaluation that produces a balance-of-payments surplus permanently raises the central bank's interest income, which is rebated to the public and so permanently raises long-run national income, consumption, and real balances -- unlike in the traditional case of non-interest-bearing reserves, where devaluation under capital controls leaves the long-run real equilibrium unchanged.
- Multiple and unstable stationary equilibria (Engel-curve crossing)
- The paper's result (Sec. III, Figure 1 and inequality 27) that once the domestic real interest rate can diverge from the world rate under capital controls, the steady-state relationship between consumption and real balances (the Engel curve) can intersect the locus of constant desired real balances more than once, yielding several long-run equilibria; those where the Engel curve crosses from below are saddle-path stable, and those where it crosses from above are fully unstable.