Fletcher and Taylor (1996) and McBrady (2004) take
the analysis one step further. Rather than simply measuring
the magnitude of deviations from parity, they offer
more sophisticated analyses that directly incorporate
the effects of transaction costs. Fletcher and Taylor incorporate
direct measures of the bid–ask transaction costs
incurred by one-way arbitrageurs and find that deviations
from parity are on average smaller in magnitude.
On the other hand, they find that deviations from the
CIP in excess of bid–ask transaction costs are neither rare
nor short lived. Interestingly, they also find that the variance
of measured deviations from the CIP is a decreasing
function of the time spent outside of the ‘neutral
band’ established by transaction costs, suggesting that
some type of one-way covered interest arbitrage serves
to bring bond yields and currency swap rates closer to
parity in equilibrium.
McBrady (2004) performs a similar analysis on a
broader data set of corporate Eurobond yields, government
benchmark yields, and short-term currency rates.
Adopting a methodology frequently used to assess the
integration of traded goods markets, he fits a threshold
autoregression model of deviations from parity. The
model provides direct estimates of the width of noarbitrage
neutral bands and the half-lives of deviations
from parity that fall outside them. Overall, he finds that
both measures of capital mobility and market integration
suggest that longer term, lower credit quality bonds
are markedly less integrated than short-term currency
markets. Unlike Popper’s findings (and consistent with
the evidence in Figures 15.1 and 15.2), McBrady also
finds that government benchmark bond markets appear
the least integrated of all. On the other hand, McBrady
finds that relatively short-term high-grade corporate