Does deglobalisation make economies more resilient to recessions?

Marco Garofalo

Are less open economies more resilient to downturns? There is general agreement on the benefits of openness, but its adverse link to volatility is ambiguous. On the one hand, globalisation makes countries less sensitive to domestic disturbances, yet it also makes them more exposed to foreign shocks. In this post, I use local projections (LP) to show that international business cycles since 1870 appear to support a positive effect of openness on the economic resilience of a country, and that we may thus expect the current international slowbalisation trend to worsen future recessions.  

The edges of the sword

Many dub globalisation as a ‘double-edged sword’. The ‘good’ edge arguments highlight that specialisation and better allocation of resources across countries should increase productivity and, by extension, output. Openness may also boost productivity through higher innovation and adoption of ideas, and economists have tried to estimate this positive relationship. The effects on trade and output that followed the closure of the Suez Canal between 1967 and 1975 illustrate what a sudden shock to openness might look like.

The ‘bad’ edge arguments are, however, anything but conclusive. Some find that globalisation increases volatility, while others argue the opposite; still others claim the openness-volatility relationship to be ambiguous. For example, increased specialisation due to openness may make economies less agile when hit by shocks, and more vulnerable to global disturbances. However, supply and demand diversification can reduce volatility, as open economies are less sensitive to domestic shocks: Caselli et al (2020) find this effect has played a larger role in recent decades for most countries. My findings below seem also to support the more positive view in this debate.

Empirical evidence from international historical data: a LP approach

One way of understanding whether deglobalisation is expected to make future recessions better or worse is by looking at the international historical data on previous downturns. To do this, I use the Jordà-Schularick-Taylor Macrohistory Database, which includes macroeconomic variables such as GDP for 18 advanced economies over the years from 1870 to 2017. Following Jordà et al (2013) I identify the peaks and troughs in the series for the logarithm of real GDP per capita. These delimit contractions (from peak to trough) and expansions (from trough to peak) at expected dates (Chart 1), eg the beginning of the global financial crisis (GFC) in 2007 and the downturns in the 1970s.

Chart 1: Turning points in UK Log GDP per capita (per cent)

Chart 2 shows the box-plot of the distribution over the considered countries and identified cycles of the cumulative percentage change in GDP growth at each year of the contraction starting after the peak. (Horizon 0 is the year of the peak.) These types of graphs aim at summarising a variable by drawing a box whose edges are the first and third quartile of its distribution; what you would normally expect to happen to GDP growth at that given stage of a contraction should fall within the bounds of the box. Notice how in Chart 2 there are many historical outliers falling very far from the box, ie certain countries have experienced extraordinarily deep contractions. For comparison, I have added in light blue the cumulative change in GDP growth the UK experienced during the GFC, and what we have seen so far since the Covid-19 outbreak in red. Both look very severe compared to the international historical experience (For another example of putting UK recessions in an historical context using hundreds of years of data, see Thomas et al (2010).)

Chart 2: Box-plots of international historical contractions for the four years after a peak

Armed with a panel of cross-country contractions over more than 100 years, we can employ some simple econometric techniques to estimate the effect of openness on the expected severity of downturns. (Disclaimer: should econometrics not really be your cup of tea, I would not be offended if you decided to skip the next couple of paragraphs and jumped to the description of the results in Chart 3.)  

To do this, I follow Jordà et al (2020), who estimated the typical trajectory in a recession conditioned on the behaviour of business versus household credit observed before a peak. I want, instead, to condition on the evolution of trade openness, and thus set up the following linear regression for local projections:

Where on the left-hand side I set the percentage change in real GDP (y) for every country i at 1 to 4 years (h) since the peak. On the right-hand side, beyond the error term (u), I include a constant (c), country fixed effects () to capture the influence of cycle-invariant characteristics at country-level, eg national language, and my main variable of interest, ie the change in the sum of exports and imports over GDP – a common measure of trade openness (OPEN). For this, I look at the difference over 10 years before the peak, as substantial changes in openness tend to occur over decades rather than from year to year. I also include a vector of control variables, namely a linear and quadratic time trend, the percentage change in population since the peak, current (at the peak) and two lags of annual growth real GDP, consumption, investment, CPI, population, exchange rate against the US$, exports and imports, as well as the first difference in the short-term interest rates on government debt. These controls, together with the predetermined status of the change in openness as occurring before the contraction, should help prevent my results from being driven by factors other than the conditioning on a certain change in openness. That said, my findings cannot be interpreted causally, meaning that what I estimate is just the typical path for a contraction after a change in openness occurs.

Before I move to my results, I wanted to point out that, by applying a simple normalisation of the country FEs to sum to zero and demeaning all the explanatory variables, we can interpret the estimated constant as the average cycle path.

Chart 3 shows the results of this estimation. The average contraction a country should expect given the international historical experience of the advanced economies considered here is of a somewhat sharp drop in the first year after the peak, followed by a quick and steady recovery. Interestingly though, an increase in openness is estimated to support growth and a decrease in openness (deglobalisation), to make downturns worse. Indeed, the red line in the chart suggests a worse contraction followed by a slower recovery for countries experiencing even just a moderate reduction in openness before a downturn (appropriately proxied by a one standard deviation change).

Chart 3: Expected average contraction in real GDP and associated worsening after a 1 SD deglobalisation ‘shock’

In the contention over the ‘double-edged sword’ of trade openness, my exercise seems to support that globalisation makes countries more resilient to adverse shocks. However, there are some caveats to bear in mind.

As already mentioned, these results cannot be interpreted causally, as the identification scheme described above may not be enough to do so – but they provide helpful evidence based on statistical regularities. Globalisation appears to be associated with less severe downturns and deglobalisation with worse.

Moreover, my framework does not tell us anything about the channels through which globalisation may be exerting a positive effect on a country’s economic resilience. To speak to this, I run the same exercise looking at the contraction in real consumption, investment, exports and imports (results not reported here). What I find is that deglobalisation seems to worsen downturns in consumption and investment, but its effect is insignificant for exports and imports. This could provide some evidence, although admittedly only tentative, that the support to growth may not come from the ability to tap on internationally diversified demand and supply, but perhaps from the economic strength gained through innovation and higher productivity following years of surging openness.        

Conclusion

In this post, I showed how international business cycles since 1870 seem to support that higher trade openness might be associated with milder contractions, suggesting that countries following the ongoing slowbalisation may find themselves in worse conditions when the next recession strikes. Strengthening international co-operation and other strategies of ‘safe trade openness’ may be a wiser alternative to global decoupling for countries trying to increase their economic resilience (D’Aguanno et al (2021)).  

Of course, we still need more investigation on such an important research and policy question, in order to move beyond simple empirical regularities and produce conclusive evidence of the ultimate effects of trade openness on an economy and its channels.

---

Marco Garofalo works in the Bank’s Structural Economics Division.

If you want to get in touch, please email us at bankunderground@bankofengland.co.uk or leave a comment below.

Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge — or support — prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.