from that sector is larger. Although these measures are imperfect, they are the most
sensible and commonly used measures available.
To reduce potential endogeneity problems, the five measures of sectoral influence
are lagged one year. The influence of the manufacturing export sector and the
agriculture export sector are manufacturing exports/GDP and agriculture raw material
exports/GDP, respectively. The models also include the variable exports, measured as
exports/GDP, to determine if other sectors’ exports also affect currency valuation.
Imports/GDP is used to measure the influence of import-competing sectors.
The
political influence of finance is measured by liquid liabilities/GDP, using the Financial
Following common practice, veto points is the log of the number of
checks and balances from the Database of Political Institutions.
Given the problems
with declaratory measures of exchange rate regimes, a behavioral dummy measure of
Several control variables are used in the analysis. I control for real per capita
income (divided by 1000), the terms of trade and external debt as a share of GDP.
Indebtedness to foreigners may encourage appreciation to reduce the debt burden.
Including the terms of trade helps control for changes in economic fundamentals. I also
control for inequality, using the Deiniger and Squire gini index, because several
economists have suggested that it contributes to overvaluation.
I control for US interest
rates, using data from the Federal Reserve’s website, because this international factor
72
See Broz and Frieden 2001; Frieden 2002.
73
These variables are taken from the World Bank’s World Development Indicators.
74
Accessed from the World Bank’s website. Broz 2002 uses the same measure for financial sector
influence.
75
Keefer and Stasavage 2003.
76
Levy-Yeyati and Sturznegger 2005.
77
Data were obtained from Easterly 2001a.
78
Easterly 2001b; Sachs 1989.
29
Convention