Divergence from de jure exchange rate regime: a stochastic process of learning

Abstract

This article provides a model framework to characterize the process of stochastic learning during the period of divergence from a de jure currency regime. The model outcome shows that divergence from a de jure regime is a process of slow learning in small steps, which indicates that the learning completes over time. Therefore, divergence does not have a long-term effect on the distribution of exchange rate regime. Empirical illustration indicates that countries work toward the development of their financial sector during the period of divergence.

Acknowledgements

The author thanks Kenichi Ohno, T. Oyama and Yoichi Okita for their helpful comments on the earlier version of the paper.

Notes

¹A fixed regime consists of hard pegs such as formal dollarization, currency union and currency board arrangement; an intermediate regime consists of all soft pegs such as conventional fixed pegs (basket pegs with published or secret weights, single currency peg), crawling pegs, crawling bands and tightly managed floats; and a floating regime consists of managed floats without a predetermined range and free floats.

²To better understand the Equation 2(2), consider a linear transformation x′ = (1 – θ) x + θλ (θ ≥ 0, λ ≤ 1). Rewrite as x – x′ = θ (λ – x), where the change in x is a proportion θ of the distance λ, that is, θ denotes the rate of change in x. If conditioning is ineffective, then θ = 0 and λ is immaterial. If O ₁ C ₁ occurs, then x′ – x ≥ 0. Thus, if θ > 0, λ = 1. Similarly λ = 0 for O ₀ C ₁. Thus, it follows that ΔX _n  = θ(1 – X _n ), if R _in O _1n C _1n . Similarly, ΔX _n  = –θ X _n , if R _in O _0n C _1n .