Q2. What is the main argument for the CU?
As abenchmark, the authors follow Rose and van Wincoop (2001), who used Eq. (1) to estimate the CUeffect on trade, and assume that τij is a log-linear function of observablesτij = M∏m=1(zmij ) γm × expCUij −γ, (2)where (CU) is the common Currency Union dummy variable and zmij is a set of observablearguments, m = 1, . . . ,M , which affect bilateral trade.
Q3. Where did Rose find the CU effect?
†University of Paris Sud and CES, University of Paris 1, France (jdesousa@univ-paris1.fr).of the gravity equation: year by year, from 1948 to 2009, in its multiplicative form by thepoisson pseudo-maximum likelihood (PPML) estimator with importer and exporter fixedeffects.
Q4. What is the dummy for Eq. (4)?
3To get theoretically consistent parameter estimates and account for the multilateral resis-tance terms (Pi and Pj), The authorrun Eq. (3) and (4) year by year with directional country fixedeffects from 1948 to 2009.
Q5. What is the effect of currency unions on trade?
In 1948, holding all other factors fixed, two countries that share a currency trade eight timesas much as they would with different currencies, while sixty years later, in 2009, currencyunions are found to have no positive effect on trade.
Q6. What are the dummy variables in zij?
Bilateral distance and the construction of the dummy variables contained in zij come fromthe CEPII distance database,2 except the Free Trade Agreement dummy and the CurrencyUnion dummy.
Q7. What is the effect of the currency union on trade?
The CU effect is found to be economically and statistically large untilthe seventies, then negative and finally insignificant at the beginning of the 21st century.
Q8. What is the effect of PPML on trade between two countries?
In 1948, the PPML CU effect implies that, other things being equal, trade be-tween two countries that share a currency is 8 times larger than trade between two countriesusing different currencies [(exp(2.11)) ≈ 8].