Abstract
Economic nationalism is on the rise. What are the costs of cutting back international economic integration and rising policy uncertainty? We use the unexpected outcome of the Brexit vote in June 2016 as a natural macroeconomic experiment to study the costs of economic disintegration and their causes. As a methodological innovation, we propose a novel combination of synthetic control methods to identify the output loss caused by the Brexit vote, conjoined with an expectations-augmented vector autoregression to understand its drivers. Using the synthetic control, we first show that forward looking households and businesses have lowered spending in response to the Brexit vote, causing an output loss of close to 2 percent. Decomposition of the VAR estimates shows that shocks to economic policy uncertainty and growth expectations in the quarter following the Brexit vote explain almost the entire output loss. While higher uncertainty accounts for much of the initial output drop, the economic costs increase over time because of a downward revision of output growth expectations. Overall, both uncertainty and growth expectations account for about one half of the total economic costs of the Brexit vote. Linking quasi-experimental identification to structural time series variance decomposition, our study exposes not only the aggregate costs of the UK’s expected economic disintegration from Europe but also specifies the channels through which the Brexit vote affected the macroeconomy.
Introduction
The specter of economic nationalism is haunting the global economy. Supporters of the rule-bound liberal world economic order that was constructed after World War II are on the defensive. For economists, the recent rise of protectionism represents a particular challenge. From its beginnings, the benefits of an international division of labor have been a central tenet of the discipline. Foreshadowing a large literature, Adam Smith diagnosed, in disarmingly simple words, that foregoing the gains from trade would harm the wealth of nations.
It seems therefore all but possible that the recent rise of economic nationalism could take a toll on future economic prosperity. And to the extent that market participants act in a forward-looking manner, expectations of economic disintegration and de-globalization could already affect investment and consumption today. In addition, as trade agreements are torn apart, old alliances nullified, and protectionist measures contemplated, policy uncertainty has increased substantially. Increased uncertainty, too, could negatively impact the global economy.
Can we measure its economic costs? In this paper we make an attempt to do so as we exploit a unique natural experiment: the decision of the UK to leave the European Union. Two aspects are key for interpreting the vote for Brexit as a natural experiment. First, the outcome of the referendum on June 23, 2016 came as a major surprise. Both voter polls and betting markets indicated a “Remain” victory by a considerable margin. Second, the voting behavior was largely unrelated to UK’s recent macroeconomic performance. Rather, according to many observers, the case for Brexit was predominantly based on the political imperative to “take back control.”
The Brexit experiment allows us to measure the costs of economic nationalism because the UK’s departure from the Single European Market entails significant economic disintegration. The disintegration shock will extend beyond trade in goods and services. The British labor market will become less open to foreign workers, and capital markets will be affected through disintegration from the common European market for financial services. However, while the direction of the change is clear, the exact extent of disintegration remains uncertain, not least because the details of Brexit are still negotiated. Hence, the Brexit experiment nests both a disintegration shock and a policy uncertainty shock: it is a showcase of economic nationalism.
Methodologically, our paper breaks new ground by combining two different approaches in empirical macroeconomics. Crucially, this combination allows us to both identify the overall costs of the Brexit vote to the UK’s macroeconomy, and also to understand the channels driving these causal effects. First, we use the synthetic control method that was recently added to the toolbox of empirical macroeconomics by Abadie and Gardeazabal (2003) and Abadie et al. (2010, 2015). Under fairly mild assumptions we are able to identify the causal effect of the Brexit vote on UK’s macroeconomic performance since the referendum. But while the synthetic control approach exposes causal effects at the aggregate level, the underlying channels remain in the dark. In a second step, we therefore map the results of the synthetic control method into a structural time-series framework. Using an expectations-augmented vector autoregression, similar to e.g. Perotti (2014), allows us to quantify the contribution of different channels to the aggregate causal impact of the Brexit vote estimated by the synthetic control method.
Specifically, the synthetic control method makes it possible to measure the causal impact of the Brexit vote on the UK economy by estimating its synthetic “doppelganger”. It does so by letting an algorithm determine which combination of other economies matches the growth trend of the UK economy before the Brexit vote with the highest possible accuracy. Which economies get picked and what weight they are assigned is entirely data-driven. The better the algorithm constructs a doppelganger for the UK economy as a weighted combination of other economies before the referendum, the more precise our results will be. We use the largest available country dataset to obtain the best match possible.
Comparing the evolution of this synthetic doppelganger to actual data for the UK economy directly quantifies the aggregate costs of the Brexit referendum. Identification hinges on the very notion that the Brexit vote is a natural experiment: because the vote was unanticipated and unrelated to macroeconomic performance, the doppelganger continues to evolve in the way the UK economy would have in absence of the referendum. The difference in output between the UK economy and its doppelganger after the referendum is the causal effect of the experiment. Importantly, our approach does not depend on having the right economic model for the British, the European, or the global economy, nor do we need to assume a particular Brexit deal emerging from future negotiations.
We find that the economic costs of the Brexit vote are already visible and quite large: there is a sizable output gap between the doppelganger and actual output in the UK. By year-end 2017, the “doppelganger gap” amounts to approximately 1.8 percent, and the cumulative loss of GDP is close to 25 billion pounds. Following Abadie et al. (2015), we also conduct a number of time- and country-placebo tests, reassuring us of the causal effect of the Brexit vote.
Mueller_Brexit
Originally posted by CESifo, to find the original post, click here.