Exchange Rate Uncertainty and Trade Growth - A Comparison of Linear and Nonlinear (Forecasting) Models

the selection of variables in the next subsection. A more detailed description of the com-
putation of weights, variable transformations, and data sources is given in the Appendix.
Thereafter, we will discuss the approximation of FX uncertainty and provide estimation
results on this issue.

2.2.1 Trade figures, economic activity and the real exchange rate

We concentrate on a set of 15 industrialized countries, k = 1, . . . 15, namely Austria (AT),
Belgium (BE), Canada (CA), Finland (FI), France (FR), Germany (GE), Greece (GC),
Italy (IT), Ireland (IR), Japan (JP), the Netherlands (NL), Norway (NO), Sweden (SW),
the United Kingdom (UK), and the USA (US). The set of countries includes the G7, most pre
2004 members of the European Union and Norway. Each country reports sectoral exports
and imports over ten sectors, j = 0, . . . 9, on a multilateral basis. We employ one-digit SITC
categories which as described in Table I.

insert Table I about here

Given presample values to implement the VAR model, our sample of seasonally unad-
justed monthly observations covers for most samples the period October 1981 to December
1998, thus providing 207 observations. The sample period ends immediately before the ad-
vent of the Euro reducing FX rate risk to zero for a major part of bilateral relationships
covered by the cross section. With respect to GR sectoral import data is available only until
December 1997. Owing to data limitations we exclude BE from the set of countries when
analyzing import dynamics.

We compute country specific quantities of sectoral trade, foreign economic activity, of the
real effective FX rate, and of a volatility measure taking into account the multilateral nature
of trade flows. To augment the covered fraction of trade of the cross section members the
latter quantities are based a second, larger set of nineteen economies containing the former
cross section and, in addition, Switzerland (CH), Spain (SP), Mexico (MX), and Portugal
(PR). We exclude the latter countries from the investigated cross section since CH and SP do
not report trade flows on a sectoral basis and PR and MX report respective figures starting
in 1984.

insert Table II about here

For the considered cross section Table II shows the average relative size of export (import)
sectors over the period 1993:01 to 1998:01 in terms of total exports (imports). For most
countries the majority of trade flows settles in sectors 6, 7, and 8. However, heavy oil
exports of a volume exceeding 50% shift the main weight of the sectoral distribution of
NO’s exports to sector 3. The last column of Table II gives the percentage coverage the
set of partner countries contributes to country k’s total exports (imports). On average, the

<<   2   3   4   5   6   7   8   >>

More intriguing information