the role of china in the recent trade slowdown, g. gaulier's slides, june 2016
TRANSCRIPT
The role of China in the trade slowdown
Guillaume Gaulier, Walter Steingress & Soledad Zignago Banque de France
• After the 2008 financial crisis, international trade of goods and global industrial production grew nearly at the same pace implying an implicit Global Trade Elasticity (GTE=growth of trade/growth of output) of 1
• This stands in sharp contrast to the pre-crisis period, when global trade increased twice as fast as industrial output
GTE>2 GTE=1
Source: Author’s computations using CPB World Trade Monitor
• Among the conditions to have a GTE of 1: balanced trade and symmetric shocks
• Not true in the short and medium run • Some countries or regions are big enough to
weight on the GTE for a long period of time
GTE instability: in search of a smoking gun
Does China hold the smoking gun? • Building on Gaulier, Santoni, Taglioni and
Zignago (2015) we argue that there was in China an asymmetric shock capable of moving the GTE – “With […] a large production base compared to
the world total, China generated a large export surplus that drove down the world price for goods in which it specialised and reinforced specialisation patterns based on Ricardian comparative advantages and the reallocation of global demand for those products towards Chinese exports”
Does China hold the smoking gun?
• We do not have a formal model from which we could estimate the contribution of China/Emerging countries integration into the global economy
• But we bring various empirical evidences that we think are consistent with China’s guilt!
Main results/evidences – A simple time-series model with a long run GTE of
1 fits the historical data and has the best predictive performance
– Emerging Asia contributes disproportionally to the (short or medium run) GTE
– Supply shocks in this region “cause” global trade – China’s contribution to global trade growth
through supply declined and became similar with its contribution through demand
– China’s supply shock triggered a decline of world prices of the goods it exports, which is now over
The long run GTE was always 1 • Before discussing China’s role, we show that based on
time series analysis one cannot reject the hypothesis of a long run GTE of 1
• Given large persistent deviations to the LR equilibrium, we generally observe elasticities different from 1
• We estimate an ECM using the CPB WTM dataset, varying the long run GTE:
𝑑𝑑𝑑𝑑𝑑𝑑 𝑇𝑇𝑇𝑇𝑇𝑇𝑑𝑑𝑇𝑇 𝑡𝑡 = 𝜌𝜌 𝑑𝑑𝑑𝑑 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑂𝑂𝑂𝑂𝑡𝑡𝑂𝑂𝑂𝑂𝑡𝑡 𝐺𝐺𝐺𝐺𝐸𝐸
𝑡𝑡−1+ 𝛽𝛽𝑑𝑑𝑑𝑑𝑑𝑑 𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂 𝑡𝑡−1 + 𝛼𝛼 𝑑𝑑𝑑𝑑𝑑𝑑 𝑇𝑇𝑇𝑇𝑇𝑇𝑑𝑑𝑇𝑇 𝑡𝑡−1 + 𝜀𝜀
– Fits and in-sample dynamic forecasts are as good with a long-run GTE set to 1 or to 2
– But speed of convergence to the long run (𝜌𝜌) is not stable for a GTE of 2
– And out-of sample dynamic forecast better with a GTE of 1
In-sample (1991-2007) forecasts are the same with GTE of 1 or 2
140
120
100
80
60
40
1992 1994 1996 1998 2000 2002 2004 2006
world trade volume CPBdynamic forecast elast=1dynamic forecast elast=2
Source: Author’s computations using CPB World Trade Monitor
Recursive estimates of the speed of convergence to the long run (𝜌𝜌) does
not stabilize when the GTE is 2
Source: Author’s computations using CPB World Trade Monitor
GTE=1
GTE=2
Break in 2008-2009
The long run GTE was always 1
• We continue the horse race with out-of-sample dynamic forecasts for the crisis and post-crisis period
• Models with GTE of 1, 2 or zero (no long run equilibrium) are tested [estimate over 1991-2007]
• To start addressing the question of country shocks in the short run part, we take country or regional Industrial Production (IP) instead of global IP (US, Japan, EA, Emerging Asia, CEEC, Latin America, Africa and Middle East)
𝑑𝑑𝑑𝑑𝑑𝑑 𝑇𝑇𝑇𝑇𝑇𝑇𝑑𝑑𝑇𝑇 𝑡𝑡 = 𝜌𝜌 𝑑𝑑𝑑𝑑 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇
𝑂𝑂𝑂𝑂𝑡𝑡𝑂𝑂𝑂𝑂𝑡𝑡 𝐺𝐺𝐺𝐺𝐸𝐸𝑡𝑡−1
+ ∑ 𝜷𝜷𝒓𝒓𝑑𝑑𝑑𝑑𝑑𝑑 𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑇𝑇 𝑡𝑡−1𝑇𝑇 +…
Whereas short run elasticity (𝛽𝛽𝑇𝑇 ) for EA is in line with the region’s weight in global trade, the elasticity for Emerging Asia is much higher than its share in trade
Source: Author’s computations using CPB
World Trade Monitor
Out-of-sample dynamic forecasts (growth rate of actual and predicted global trade)
Model with GTE=1 tracks well Global Trade whereas models with GTE=2 or GTE=0 over-estimate trade by 2 to 3% pts per year during the recent period
Source: Author’s computations using CPB World Trade Monitor
ZOOM
Does China/Emerging supply “causes” global trade, or does the causality goes the
other way? IP X M IP X M IP X M
IP 5.4 4.5 7.5 4.4 8.1 8.5 5.2 9.2 4.6 X 0.7 4.8 3.2 2.3 11.6 3.3 4.3 4.5 1.5 M 2.7 1.7 4.6 2.5 9.6 2.8 2.7 4.8 1.4 IP 2.8 2.8 2.4 5.0 9.8 3.2 5.3 3.9 5.7 X 1.6 4.1 5.4 2.1 6.8 4.6 3.9 4.7 4.4 M 1.3 2.6 0.9 1.3 4.0 4.5 5.8 3.1 4.6 IP 1.4 3.0 2.1 3.5 2.8 11.0 10.0 13.8 8.7 X 2.1 2.5 2.3 2.9 2.4 13.1 5.7 2.9 2.4 M 1.5 1.1 1.6 3.2 1.8 8.4 3.3 1.0 4.8
2.3 3.2 4.2 4.9 5.1 14.9 8.7 5.1 7.6
Emerging Asia
US
EA
Global Trade
Emerging Asia US EA Global Trade
Granger-causality analysis This Table gives the Fisher test statistic for each pair of variables. The null hypothesis is that “variable in row does not cause variable in column”. A high Fisher means that non causality can be rejected. Significant statistics at the 1% level are in bold.
Source: Author’s computations using CPB World Trade Monitor
Emerging Asia IP is the only variable that “causes” other variables, including Global Trade
But is not (significantly) caused by any other variable
IP X M IP X M IP X MIP 5.4 4.5 7.5 4.4 8.1 8.5 5.2 9.2 4.6 X 0.7 4.8 3.2 2.3 11.6 3.3 4.3 4.5 1.5 M 2.7 1.7 4.6 2.5 9.6 2.8 2.7 4.8 1.4 IP 2.8 2.8 2.4 5.0 9.8 3.2 5.3 3.9 5.7 X 1.6 4.1 5.4 2.1 6.8 4.6 3.9 4.7 4.4 M 1.3 2.6 0.9 1.3 4.0 4.5 5.8 3.1 4.6 IP 1.4 3.0 2.1 3.5 2.8 11.0 10.0 13.8 8.7 X 2.1 2.5 2.3 2.9 2.4 13.1 5.7 2.9 2.4 M 1.5 1.1 1.6 3.2 1.8 8.4 3.3 1.0 4.8
2.3 3.2 4.2 4.9 5.1 14.9 8.7 5.1 7.6
Emerging Asia
US
EA
Global Trade
Emerging Asia US EA Global Trade
Notice that US imports caused by all variables, but does not cause Emerging Asia variables, including exports
Supply (adjusted exports) and Demand (adjusted imports) contributions to global trade growth
Decomposition using Measuring Export Competitiveness (mec.worldbank.org)
China provided much less stimulus to global trade through demand/import than through supply/export.
In the recent period, the stimulus was balanced
Supply-side contribution has shrunk
Demand-side contribution was more resilient
Before the global trade collapse: China supply-push effect was strong, but export growth in value was hampered by
slower world prices for the mix of goods exported
World market share gain >10% per year once we control for specialization
Declining relative prices of the good China is specialized in
More recently (2011q3-2015q2) the product mix became favorable: terms of trade increased
Push effect still positive, but much weaker
Favorable product mix (in terms of prices, not volume)
• Our interpretation is that the pre-crisis terms of trade loss was self inflicted, China was big enough to weight on the prices of the goods it exports
• This was necessary for the goods produced in China to find sufficient demand
• Local competitors were ousted by Chinese export prices • It contributed to a decline in the relative price of
tradables, causing an increase in the trade to output ratio at the world level
• It ceased as China’s growth model became more balanced: Higher relative costs, more diversified supply, use of more local inputs, etc.
• Less asymmetric shocks : industrial production growth rate more similar across countries/regions
Link with global current account imbalances?
• Notice that excess supply in China implies a current account surplus
• It was possible because the exchange rate was not allowed to appreciate
• This contributed to growing global imbalances during the “hyperglobalization” period
• It may not be a pure coincidence that trade slew down at the same time as global imbalances narrowed
• Recently: Capital outflow from China + higher rates in the US=> reemergence of global imbalances + GTE>1?
A zoom on the recent period BACK
Decomposition of market share growth
• Measuring Export Competitiveness (BdF, WB, ITC) • From bilateral (200×200 countries) disaggregated (5000
products) data, using fixed effects model, we extract exporter, importer and product contributions to market-share growth
• Exporter effects are associated to supply, importer effects to demand
• For each country we can compute the contribution to its export growth of his own specific performance (push effect or adjusted growth) and of the dynamism of foreign demand he faces: how is the demand in his partner countries and for the products in which he is specialized?
BACK
Growth of industrial production Due to emerging Asia: asymmetric shocks before the
global crisis, more symmetric shocks recently
Source: Author’s computations using CPB World Trade Monitor
BACK
• As stated in the task force report, the GTE should be one, except if:
• Demand is non homothetic: when income changes, relative demand of tradable changes. – For instance, higher demand for non tradable services
when income increases, or higher demand for electronic goods…
• The trade wedge changes: the relative price of international trade changes. This can be due: – Biased technological progress (containerization, ICT,
renewables) – Liberalization/protectionism – Asymmetric shocks: (big) shocks in the export sector
of (big) countries
other aspects we don’t address
• Cyclical component of the trade slowdown: – Role of the composition of demand and the
weakness in business investment among advanced economies (Boz, Bussière & Marsilli 2014)
– Role of low demand in the EU/EA, the most trade integrated region in the world (Ollivaud & Schwellnus 2015)
• Commodity-price decline, contraction of global value chains, etc
Prospects • In the short run, very high GTE are possible
(for instance due to a strong recovery in the EA, with countries trading with each others)
• But given limited scope for further trade liberalization (new treaties could favor FDI more than trade) and unless another big country or region integrates into world trade with a very export-oriented growth model, we shouldn't expect to see GTE persistently above 1
Short run/Long run
• The global elasticity was high when global imbalances were building up in the 90’s or the 00’s
• But it was low in the early 90’ and 00’s global recessions
• And now