I rely on a historical natural experiment to provide, for the first time, a causal estimate of the effect of currency unions on international trade. Since the seminal paper by Rose (2000), a large literature has developed around currencies as a trade cost. However, self-selection and endogeneity bias implied by membership of a currency union are likely to be pervasive and might explain the large pro-trade effect of currency union found in the literature. I offer a quasi-experimental contribution by exploiting an exogenous variation in currency union membership, driven by an unexpected geopolitical shock – the 1861 Italian unification - and involving a French franc pan-European zone that existed throughout the 19th century. I employ original data and structural gravity equations to estimate an effect in the order of 35%, consistent with a large - if heterogenous - effect of common currencies on trade.
A key economic benefit commonly associated with currency unions is an increase in cross-border flows, and notably trade. The effect of the adoption of a common currency on international trade has been found to be very large in previous studies, with average estimates suggesting an effect as high as a doubling of bilateral trade.
However, it is unclear whether such estimates are useful in guiding policy. Indeed, several factors might be spuriously biasing the measured effect upward. Countries adopting a common currency might at the same time be more likely to share other characteristics positively associated with international trade. For example, they might enjoy cultural or historical links, or an ex-ante positive trend of economic or political integration. Historical episodes of switches out of common currencies also tend to coincide with events negatively affecting international trade, such as decolonisation or civil wars. It is therefore far from clear that the estimates found in the literature are shedding lights on a policy-relevant relationship going from currency unions to bilateral trade.
To estimate a causal link between a common currency and bilateral trade flows we would ideally want to run an experiment, randomly assigning currency union membership to a pool of suitably similar countries, and then observe the ex-post difference between treatment and control groups. Faced with the impossibility of such an approach, one must turn to a quasi-experimental setting where real world events mimic the characteristics of an experiment.
I focus on a historical setting where a sudden geopolitical shock, the Italian unification, led to random variations in currency union membership. In 1861, a small buffer state, the Kingdom of Piedmont-Sardinia, ended up unexpectedly annexing most of modern-day Italy, following a series of unintended diplomatic and military events. Piedmont was until then part of the French sphere of influence and as such a member of a French franc area that existed throughout the 19th century. The Italian unification therefore meant that, when Piedmont extended its institutions to the annexed territories, several Italian states exogenously found themselves in an accidental common currency relationship with France and its satellites.
The paper relies on original data, compiled by foreign embassies in Italy, that I collected from diplomatic archives and historical publications. These new data allow me to observe Italian bilateral trade before and after unification at the pre-unitary state level. I can therefore estimate the common currency effect on international trade relying on a gravity model, in line with the contemporary literature. My findings indicate that, based on the quasi-experimental environment I examine, we should expect an increase of bilateral trade in the region of 35% once a common currency is implemented.
The estimate is robust to different specifications of the gravity equation, sampling of the data and, more importantly, is not driven by any contemporaneous third factor. Looking at the staggered timing of the implementation of the Italian unification as well as to product-level tariffs data, I provide evidence that results are not driven by a change in tariff policy. I also show they cannot be explained by the adoption of a new unit of measurements, historical trends, or path dependency related to the Napoleonic occupation.
I find the effect to be heterogenous within each bilateral pair. In line with previous work, the estimated common currency effect is found to be larger for ex-ante lower trade shares. Additionally, while large in absolute terms, the magnitude of the effect I uncover is meaningfully smaller than the average one estimated in the literature. This might signal that an upward bias related to the endogeneity of currency unions might indeed be an issue in the existing literature. However, the paper largely confirms the early policy implications of the literature in a quasi-experimental setting, as we should expect a common currency to have a significant positive effect on bilateral trade.
Updated on: 12/31/2021 14:07