The Advantage of Tying One's Hands: EMS Discipline and Central Bank Credibility

Within the EMS, countries with above-average inflation lose competitiveness relative to Germany: between 1978 and 1986 Denmark and Italy saw real appreciations of 9 and 11 percent respectively (Sec. 1, footnote 1, source EEC). This loss has two components – excess inflation produces one-for-one nominal appreciation between realignments, and realignments themselves are typically “insufficient to make up for the real appreciation experienced since the previous realignment,” so high-inflation countries accumulate a long-run trend of real appreciation rather than simply oscillating around PPP. The puzzle, as the authors frame it quoting Stanley Fischer (“The EMS is currently an arrangement for France and Italy to purchase a commitment to low inflation by accepting German monetary policy”) and a 1985 Economist piece on the possibility of sterling joining, is why an inflation-prone country would ever accept this arrangement – the paper’s stated aim is to determine not just whether EMS is an effective disciplinary device (obviously, yes) but whether it is actually welfare-improving from the standpoint of the disciplined country’s own monetary authority.

The authors model a monetary authority with an instantaneous objective function V = integral of e^{-rho t}[h q_t + c(pi_t - pi_t^e) - (a/2) pi_t^2] dt, h, a, c > 0 (eq. 1), where pi_t is domestic inflation, q_t is the (log) real exchange rate (entering positively because the authority values export-sector profitability), and the middle term captures the authority’s temptation to engineer inflation surprises (pi_t - pi_t^e) to raise output via a standard wage-contracting mechanism (footnote 6 sketches a simple Nash wage-setting game in the spirit of Fischer 1986). Under the EMS, q_t evolves as the level set at the last realignment minus cumulated inflation since then (eq. 2), with realignments occurring every T periods and resetting q to a pre-assigned q_0. Solving for the time-consistent (rational-expectations, pi^e = pi) inflation path yields eq. (4): a monotonically rising, “saw-tooth” inflation profile between realignments, concentrated toward the end of each interval precisely so the resulting loss of competitiveness is carried for as short a time as possible.

The paper shows the EMS time-consistent inflation rate pi(t) is strictly below the flexible-rate inflation rate pi-tilde = c/a (eq. 5) for all t < T, converging to it only at realignment dates – so average EMS inflation is unambiguously lower, confirming the EMS “disciplines” inflation** (Sec. 2). But the authors stress this is “quite obvious” and not the interesting question; what matters normatively is the welfare difference Delta = V* - V-tilde* between the two regimes (eq. 6, derived in the appendix), which depends on T: Delta = 0 as T -> 0 (EMS collapses to flexible rates) and rises to a finite positive asymptote as T -> infinity (EMS approaches permanently fixed rates), with Delta shown to be uniformly increasing in T (eq. A.4-A.5). The intuition: by attaching an extra, publicly known penalty to inflation, the EMS lowers the authority’s incentive to inflate, and since that disincentive is common knowledge, the public’s rational inflation expectations fall correspondingly, partially curing the inefficiency that stems from the public’s mistrust of policy-makers’ incentives.

The baseline setup (q_0 = 0 at every realignment, i.e., the real exchange rate always returns exactly to PPP) implies the country’s real exchange rate fluctuates below PPP between realignments – shifting domestic demand toward foreign goods and steadily worsening the trade balance – so that, absent capital mobility, foreign exchange reserves (or foreign debt) would be exhausted and the country forced out of the system (Sec. 2, end). The authors identify two fixes: (i) redesign realignments so the real exchange rate fluctuates around PPP rather than below it (Section 3), which restores sustainability; or (ii) accept that membership can only be temporary and ask whether even a finite spell of EMS membership is still beneficial (Section 4).

Section 3 derives the sustainability condition q_0 = (1/T) * integral_0^T [integral_0^t pi_s ds] dt (eq. 7): at each realignment the real exchange rate must jump above PPP by exactly the average loss of competitiveness the country will incur before the next realignment. For this “competitiveness bonus” to still deliver a credibility gain, it is crucial that the policy-maker treat q_0 as outside its own control – otherwise (footnote 14) the authority could exploit the sustainability relationship to set q_0 so high that the real exchange rate term drops out of its objective function entirely, and all disciplinary/credibility effects vanish. Given exogenous q_0 at its sustainability-implied value, the welfare gain from permanent membership is A’ = (h/rho) q_0* + A (eq. 6’), strictly larger than the gain A computed under q_0 = 0 because of the added competitiveness bonus. The EMS regime is unambiguously superior to flexible rates whenever c > h/rho (the incentive to generate inflation surprises exceeds the discounted marginal value of real depreciation); it is unambiguously inferior for all T whenever c < h/(2rho); and for the intermediate range h/(2rho) < c < h/rho, the EMS is superior only if T is sufficiently large (Sec. 3, summarized in the paper’s Fig. 1 and footnote 16).

No – Section 3.1 adds a linear seignorage term b*pi_t to the objective function (eq. 1’), which shifts the optimal inflation path upward by a constant b/a at every date (eq. 4’), but once the sustainability-implied q_0’’ is recomputed to account for the higher average inflation, the seignorage terms exactly cancel in the welfare-gain expression: Delta’’ = Delta’ (the welfare gain computed with q_0 = b = 0), eq. (6’’), footnote 18 derivation. The intuition given is that the seignorage loss from lower EMS-regime inflation is exactly offset by a correspondingly larger competitiveness bonus the country must be granted at each realignment to remain sustainable – so the paper’s central welfare comparison is robust to whether or not the authority also values seignorage revenue.

Assuming q_0 = 0 and that the regime can only be sustained for T_K periods before the country is forced to abandon it and revert to flexible rates, the welfare gain from temporary membership is Delta(K) = (1 - e^{-rho T_K}) * Delta (eq. 7 restated), a strictly positive monotone transformation of the permanent-membership gain Delta – so even an EMS spell the country knows it cannot sustain indefinitely can still be beneficial, provided the country can afford to remain for at least one full interval of length T (Sec. 4). The authors also show that explicitly costing the foreign-exchange reserves burned during temporary membership (by raising h, the shadow cost of real appreciation) only strengthens the case for EMS membership, since Delta is increasing in h.

Introducing a parameter lambda such that only a fraction (1 - lambda) of the cumulated real appreciation is reversed at each realignment (so a proportion lambda is “lost forever” and the real exchange rate trends persistently rather than merely fluctuating), the optimal inflation path (eq. 4’’) now rises over the life of the regime as the date the system must be abandoned approaches, and the welfare gain relative to flexible rates, computed over the sustainable horizon, is strictly higher than in the lambda = 0 case for all lambda > 0 (eq. 7’). The authors interpret this as the persistent real-appreciation penalty unambiguously raising the (shadow) cost of inflation and so unambiguously improving welfare – in contrast to the sustainable permanent-membership case, where a larger penalty (h) can either help or hurt welfare depending on whether c exceeds h/rho.

Adding uncovered interest parity and a policy-maker who dislikes volatility in the nominal interest rate on deposits of maturity tau (objective eq. 9), the welfare gain of perfect capital mobility relative to an EMS with capital controls is shown to be strictly positive for all T (eq. 12) and strictly decreasing in tau – the shorter the maturity the policy-maker cares about, the larger both the equilibrium disincentive to inflate and the resulting welfare gain. The intuition: because interest rates on short-maturity deposits must jump sharply at realignment dates to compensate holders for the anticipated capital loss, a policy-maker who dislikes that volatility becomes even more cautious about generating inflation than in the capital-controls case, strengthening the credibility channel. The paper notes (footnote 23, citing the authors’ own 1985 work) that capital controls in France and Italy over 1980-84 sharply reduced the variability of domestic short rates (from a standard deviation of 5.2 to 0.5 in France; 4.8 to 1.8 in Italy), motivating why this extension matters empirically.

The paper states its result “parallels that in Rogoff 1985,” who shows the non-cooperative inflation rate can be lowered “through a system of rewards and punishments which alters the incentives of the central bank,” e.g., by placing direct weight on a nominal target such as the price level, nominal GNP, or money supply (Sec. 1). The paper’s distinct contribution is to identify the real exchange rate – specifically, the mechanism by which EMS membership constrains the nominal exchange rate to zero growth except at realignment dates – as the concrete, institutionally available “nominal variable” that can play this disciplining role for an individual country, and to work out exactly which design features (sustainability via an exogenous competitiveness bonus; sufficiently high net incentive to inflate; low-frequency realignment) are necessary for that discipline to be not merely effective but welfare-improving from the disciplined country’s own standpoint.