GALSTYAN AND LANE
Since μ0 > 0, the level and composition of spending matters for the real exchange rate.
In particular, a country that is a long-run creditor (B > 0) experiences real apprecia-
tion, since the interest income on the creditor position enables an increase in the steady-
state level of consumption.11 In the traded sector, this translates into a long-run trade
deficit (TB = -rB), while the increase in demand for the nontraded good generates real
appreciation due to the presence of the fixed factor in the nontraded sector.12 More-
over, an increase in government consumption that is concentrated on nontraded goods
([GN -GT] > 0) also generates real appreciation by shifting the composition of aggregate
consumption towards the nontraded sector.13
Finally, the effect of an increase in the public capital stock on the real exchange rate is
given by the coefficient μ1, which has an ambiguous sign. If an increase in public capital
has a symmetric impact on productivity in both sectors (αZ = βZ) and both sectors have
similar capital shares (αK = βK), the real exchange rate is unaffected by the level of
the public capital stock. If αZ = βZ but the nontraded sector is less capital intensive
(αK > βK), then an increase in public capital generates real appreciation, by the same
logic as a symmetric improvement in the sector-specific productivity terms AT and AN .
However, even if αK > βK, it is possible to construct scenarios in which an increase
in the public capital stock generates real depreciation if productivity in the nontraded
sector is more sensitive to the level of public capital than is productivity in the traded
sector (αZ < βZ). Accordingly, the sign of the relation between public investment and
the real exchange rate is ultimately an empirical matter.
3. Data and Empirical Specification
3.1. Data
The data covers years 1980 to 2004. The relative price of non-traded goods is constructed
using the ratio of the deflators of services sectors to manufacturing sectors from the EU
KLEMS database (Timmer et al 2008).14 The relative productivity differential variable is
11See also Lane and Milesi-Ferretti (2002, 2004) and Galstyan (2007).
12As noted earlier, if the share of the fixed factor in the non-traded sector were equal to zero, then μ0 = 0,
and the demand side is redundant for the real exchange rate.
13A balanced increase in government consumption across the two sectors has no effect on the relative
price of non-traded goods. Since the aggregate labour supply is inelastic, the level of production in each
sector is unchanged by a shift in the mix between public and private consumption that leaves aggregate
consumption in each sector unchanged (if GT and GN are simultaneously increased by one unit each, CT
and CN each decline by one unit).
14The services sector is defined as the aggregate of ‘Wholesale and Retail Trade’, ‘Hotels and Restaurants’,
‘Transport and Storage and Communication’, ‘Finance, Insurance, Real Estate and Business Services’ and