The Size and Timing of Devaluations in Capital-Controlled Developing Countries
The Size and Timing of Devaluations in Capital-Controlled Developing Countries
复制标题
资本控制的发展中国家货币贬值的规模和时间
DOI：
10.3386/w4957
复制
发表时间：
1994
期刊：
影响因子：
--
通讯作者：
N. Marion
中科院分区：
文献类型：
--
作者：
Robert P. Flood;N. Marion
发展中国家通常将其汇率与美元等单一货币挂钩，即使它面临的通货膨胀率高于与之挂钩的国家。因此，它经历了实际汇率失调和一系列容易预期的货币贬值。虽然量化资本管制避免了围绕预期贬值的资本市场混乱事件，但这个国家仍面临着经典的贬值问题：何时应该贬值，以及贬值多少？在本文中，我们考虑一个政策制定者，他与名义汇率挂钩，并定期调整挂钩，以使一系列成本最小化。由于货币贬值的未来时间是未知的随机变量，控制问题变得困难。将实际汇率描述为受调节的布朗运动，可以明确地解决成本最小化问题。贬值的规模和时机是优化行为共同决定的结果。该框架有助于深入了解随机环境的变化如何影响货币贬值的规模和时机。这些见解为了解贬值事件的决定因素提供了指导，即使布朗运动不是与实际汇率相关的随机过程。利用17个拉丁美洲国家在1957-1990年期间的80次钉住汇率事件的横截面数据，我们发现了模型主要预测的实证支持。
A developing country often pegs its exchange rate to a single currency, such as the U.S. dollar, even though it faces a higher inflation rate than the country to which it is pegged. As a consequence, it experiences real exchange-rate misalignments and a series of easily-anticipated devaluations. While the chaotic capital market events surrounding anticipated devaluations are avoided through quantitative capital controls, the country is left with the classic devaluation problem: when should it devalue, and by how much? In this paper, we consider a policymaker who pegs the nominal exchange rate and adjusts the peg periodically so as to minimize a set of costs. The control problem is made difficult by the fact that the future times for devaluations are currently unknown stochastic variables. Characterizing the real exchange rate as regulated Brownian motion permits the cost minimization problem to be solved explicitly. The size and timing of devaluations are jointly determined outcomes of optimizing behavior. The framework yields insights into how changes in the stochastic environment affect both the size and timing of devaluation. These insights provide guidance about the determinants of devaluation episodes even when Brownian motion is not the relevant stochastic process for real exchange rates. Using cross-sectional data on 80 peg episodes from seventeen Latin American countries over the 1957-1990 period, we find empirical support for the model's main predictions.