References
Alessandria, George, and Horag Choi, 2020 “The Dynamics of the U.S. Trade Balance and Real Exchange Rate: The J Curve and Trade Costs?” Journal of International Economics, 132 (3–4).
Allen, Treb, and Costas Arkolakis, 2016. “Elements of Advanced International Trade,” http://www.econ.yale.edu/-ka265/teach-ing/GradTrade/notes/ClassNotes.pdf.
Anderson, James E., 1979. “A Theoretical Foundation for the Gravity Equation,” American Economic Review, 69, 1, pp. 106–116.
Anderson, James E., and Eric van Wincoop, 2003. “Gravity with Gravitas: A Solution to the Border Puzzle,” American Economic Review, 93, 1, pp. 170–192.
Antràs, Pol, Teresa C. Fort, and Felix Tintelnot, 2017. “The Margins of Global Sourcing: Theory and Evidence from US Firms,” American Economic Review, 107 (9), pp.2514–2564.
Arkolakis, Costas, Arnaud Costinot, and Andrés Rodríguez-Clare, 2012. “New Trade Models, Same Old Gains?,” American Economic Review, 102, 1, pp. 94–130.
Armington, Paul S., 1969. “A Theory of Demand for Products Distinguished by Place of Production ,” IMF Staff Papers, 16, 1, pp. 159–178.
Baldwin, Richard, and Rebecca Freeman, 2021. “Risks and Global Supply Chains: What We Know and What We Need to Know,” Bank of England Staff Working Paper No. 942.
Bernard, Andrew B., and Andreas Moxnes, 2018. “Networks and Trade,” Annual Review of Economics, 2018, 10, pp. 65–85.
Bernard, Andrew B., Emmanuel Dhyne, Glenn Magerman, Kalina Man-ova, and Andreas Moxnes, 2021. “The Origins of Firm Heterogeneity: A Production Network Approach,” Journal of Political Economy, forthcoming.
Blanchard, Olivier, 1985. “Debt, Deficits, and Finite Horizons,” Journal of Political Economy, Vol. 93, No. 2. (Apr., 1985), pp. 223–247.
Bown, Chad P., 2019. “Trump’s Fall 2019 China Tariff Plan: Five Things You Need to Know,” PIIE Trade and Investment Policy Watch, 14 August.
Bown, Chad P., Euijin Jung, and Eva Zhang, 2019. “Trump Has Gotten China to Lower its Tariffs. Just Toward Everyone Else,” PIIE Trade and Investment Policy Watch, 12 June.
Caliendo, Lorenzo, and Fernando Parro, 2015. “Estimates of the Trade and Welfare Effects of NAFTA,” Review of Economic Studies, 82, 1, pp. 1–44.
Carrère, Céline, Monika Mrázová and J. Peter Neary, 2020. “Gravity without Apology: The Science of Elasticities, Distance and Trade,” Economic Journal, 130 (May), pp. 880–910.
Cerdeiro, Diego A., Eugster, Johannes, Mano, Rui C., Muir, Dirk, and Shanaka J. Peiris, 2021. “Sizing Up the Effects of Technological Decoupling,” IMF Working Paper WP/21/69.
Chaney, Thomas, 2008. “Distorted Gravity: The Intensive and Extensive Margins of International Trade,” American Economic Review, 98, 4, pp. 1707–1721.
Costinot, Arnaud, and Andrés Rodríguez-Clare, 2014. “Trade Theory with Numbers: Quantifying the Consequences of Globalization,” in Gita Gopinath, Elhanan Helpman and Kenneth S. Rogoff (eds.), Handbook of International Economics, 4, chapter 4, pp. 197–261.
Cuñat, Alejandro, and Robert Zymek, 2018. “International Value-Added Linkages in Development Accounting,” CESifo Working Paper No. 7196.
Davis, Donald R., and David E. Weinstein, 2002. “The Mystery of the Excess Trade (Balances),” American Economic Review, Papers and Proceedings, 92, 2, pp.170–174.
Dekle, Robert, Jonathan Eaton, and Samuel Kortum, 2007. “Unbalanced Trade,” American Economic Review, Papers and Proceedings, 97, 2, pp 351–355.
Dekle, Robert, Jonathan Eaton, and Samuel Kortum, 2008. “Global Rebalancing with Gravity: Measuring the Burden of Adjustment,” IMF Staff Papers, 55, 3, pp. 511–540.
Eaton, Jonathan, and Samuel Kortum, 2002. “Technology, Geography, and Trade,” Econometrica, 70, 5, pp. 1741–1779.
Eaton, Jonathan, Samuel Kortum, Brent Neiman, and John Romalis, 2016a. “Trade and the Global Recession,” American Economic Review, 106, 11, pp. 3401–3438.
Eaton, Samuel Kortum, and Brent Neiman, 2016b. “Obstfeld and Rogoff’s International Macro Puzzles: A Quantitative Assessment,” Journal of Economic Dynamics and Control, 72, pp. 5–23.
Eugster, Johannes, Florence Jaumotte, Margaux MacDonald, and Roberto Piazza, 2020. “Are Bilateral Trade Balances Irrelevant?” IMF Working Paper WP/20/210.
Fajgelbaum, Pablo, and Amit Khandelwal, 2021. “The Economic Impacts of the U.S.-China Trade War,” NBER Working Paper 29315.
Fally, Thibault, 2015. “Structural Gravity and Fixed Effects,” Journal of International Economics, 97, 1, pp. 76–85.
Feenstra, Robert C., Wen Hai, Wing T. Woo, and Shunli Yao, 1998. “The U.S.-China Bilateral Trade Balance,” NBER Working Paper No. 5910.
Feenstra, Robert C., Robert Inklaar, and Marcel P. Timmer, 2015. “The Next Generation of the Penn World Tables,” American Economic Review, 105, 10, pp. 3150–3182.
Felbermayr, Gabriel, and Yoto V. Yotov, 2021. “From Theory to Policy with Gravitas: A Solution to the Mystery of the Excess Trade Balances,” CESifo Working Paper 7825.
G7 Panel on Economic Resilience, 2021. “Global Economic Resilience: Building Forward Better,” ISBN 978-1-3999-0752-1.
Head, Keith, and Thierry Mayer, 2014. “Gravity Equations: Workhorse, Toolkit, and Cookbook,” in Gita Gopinath, Elhanan Helpman and Kenneth Rogoff (eds.), Handbook of International Economics, 2014, 4, pp. 131–195.
Hughes, Neil C., 2005. “A Trade War with China?” Foreign Affairs, 84, 4, pp. 94–106.
Janow, Merit E., 1994. “Trading with an Ally: Progress and Discontent in U.S.-Japan Trade Relations,” in Gerald L. Curtis. (ed.), The United States, Japan, and Asia: Challenges for U.S. Policy, pp. 53–95.
King, Mervyn, and David Low, 2014. “Measuring the ’World’ Real Interest Rate,” NBER Working Paper No. 19887.
Krugman, Paul R., 1980. “Scale Economies, Product Differentiation, and the Pattern of Trade,” American Economic Review, 70, 5, pp. 950–959.
Krugman, Paul R., 2017. “On The US-Germany Imbalance,” The New York Times, May 31st.
Matsuyama, Kiminori, 1987. “Current account dynamics in a finite horizon model,” Journal of International Economics, 23, 3–4, pp. 299–313.
Mayer, Thierry, Vincent Vicard, and Soledad Zignago, 2019. “The Cost of Non-Europe, Revisited,” Economic Policy, 98, pp. 145–199.
Melitz, Marc M., 2003. “The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry Productivity,” Econometrica, 71, 6, pp. 1695–1725.
Obstfeld, Maurice and Kenneth Rogoff, 2000. “The Six Major Puzzles in International Macroeconomics: Is There a Common Cause?” in Ben Bernanke and Kenneth Rogoff (eds.), NBER Macroeconomics Annual 2000, 15, 1, pp. 339–390.
Ossa, Ralph, 2015. “Why Trade Matters After All,” Journal of International Economics, 97, 2, pp. 266–277.
Ravikumar, B., Santacreu, M., and Michael Sposi, 2019. “Capital Accumulation and the Dynamic Gains from Trade,” Journal of International Economics, 119, pp. 93–110.
Reyes-Heroles, Ricardo, 2016. “The Role of Trade Costs in the Surge of Trade Imbalances,” manuscript.
Santos Silva, João, and Silvana Tenreyro, 2006. “The Log of Gravity,” Review of Economics and Statistics, 88, 4, pp. 641–658.
Simonovska, Ina, and Michael E. Waugh, 2014. “The Elasticity of Trade: Estimates and Evidence,” Journal of International Economics, 92, 1, pp. 34–50.
Sposi, Michael, 2021. “Demographics and the Evolution of Global Imbalances,” Journal of Monetary Economics, in press.
Swanson, Ana, 2017. “Trump Lashes out at Germany’s Trade Practices — and He May Have a Point,” The Washington Post, May 30th.
Timmer, Marcel P., Erik Dietzenbacher, Bart Los, Robert Stehrer, and Gaaitzen J. de Vries, 2015. “An Illustrated User Guide to the World Input-Output Database: the Case of Global Automotive Production,” Review of International Economics, 23, pp. 575–605.
Tinbergen, Jan, 1962. “An Analysis of World Trade Flows,” in Jan Tin-bergen (ed.), Shaping the World Economy, New York, NY: Twentieth Century Fund.
Waugh, Michael E., 2010. “International Trade and Income Differences,” American Economic Review, 100, 5, pp. 2093–2124.
Zuleeg, Fabian, 2020. “The End of the Level Playing Field?” European Policy Centre, October 10th.
A Appendix
A.1 Approximating Bilateral Trade Imbalances
We can write the proportional bilateral imbalance between n and n’ as:
Note from (1) that
The first-order Taylor-series expansion of (39) centered at In (1 — NXn/Dn) = 0 for all n, lndsn = lnesn = In (Ds/D) for all s and n, and
and, hence,
A.2 Variance Decomposition for 1995–1999
A.2.1 Data
To compile the data for 1995–1999 decomposition of the variation in bilateral imbalances, we proceed as described in Sections 2.1 and 2.2 – with one exception: we use the 2013 release of WIOD (whose data tables start in 1995), instead of the 2016 release (whose data tables start in 2000). The data allow us to aggregate trade and spending values to the same 31 sectors as described in Section 2.2 (and shown in Table A2). However, in the 2013 release Croatia, Norway and Switzerland are not covered as individual countries but grouped with the “Rest of the World”. For this reason, the 1995–1999 data only cover 37 individual economies and the Rest of the World, which yields (38 x 37/2 =) 703 distinct bilateral trade imbalances.
Figure A1 correlates the bilateral imbalances available in both periods with one another, using only the 703 surpluses for 1995–1999. The figure indicates that there is a fairly high degree of persistence: the correlation of the 1995–99 surplus with the 2010–14 value of the same trade balance is .36. Moreover, more than two thirds of the bilateral balances which were in surplus in 1995–1999 were still in surplus in 2010–14.
Proportional bilateral imbalances, 2010–14 versus 1995–99
Citation: IMF Working Papers 2022, 090; 10.5089/9798400208843.001.A999
Proportional bilateral imbalance refers to (Mn'n — Mnn')/(Mn'nMnn') ‘1/2, where Mn'nt, where Mn'nt represents the total spending by economy n on goods and services from n' in period t. On the horizontal axis, all values are the average for the 2010–14 period. On the vertical axis, all values are the average for the 1995–1999 period. The 2010–14 data is based on WIOD (2016 release), the 1995–99 data on WIOD (2013 release). The chart covers 37 individual economies.Proportional bilateral imbalances, 2010–14 versus 1995–99
Citation: IMF Working Papers 2022, 090; 10.5089/9798400208843.001.A999
Proportional bilateral imbalance refers to (Mn'n — Mnn')/(Mn'nMnn') ‘1/2, where Mn'nt, where Mn'nt represents the total spending by economy n on goods and services from n' in period t. On the horizontal axis, all values are the average for the 2010–14 period. On the vertical axis, all values are the average for the 1995–1999 period. The 2010–14 data is based on WIOD (2016 release), the 1995–99 data on WIOD (2013 release). The chart covers 37 individual economies.Proportional bilateral imbalances, 2010–14 versus 1995–99
Citation: IMF Working Papers 2022, 090; 10.5089/9798400208843.001.A999
Proportional bilateral imbalance refers to (Mn'n — Mnn')/(Mn'nMnn') ‘1/2, where Mn'nt, where Mn'nt represents the total spending by economy n on goods and services from n' in period t. On the horizontal axis, all values are the average for the 2010–14 period. On the vertical axis, all values are the average for the 1995–1999 period. The 2010–14 data is based on WIOD (2016 release), the 1995–99 data on WIOD (2013 release). The chart covers 37 individual economies.A.2.2 Variance Decomposition
Figure A2 is the analogue for the 1995–1999 period of Figure 4 in the main text. The quantitative results of the variance decomposition are remarkably similar.
Variance decomposition for 1995–99
Citation: IMF Working Papers 2022, 090; 10.5089/9798400208843.001.A999
In each panel, the horizontal-axis variable is the first-order linear approximation of (Mn'n — Mnn')/(Mn'nMnn') ‘1/2 from equation (6), represents the total spending by economy n on output from W. The vertical-axis variable is one each of the four right-hand-side terms in expression (6). The red line represents the line of best fit, whose respective slope is also printed in red. All data is based on WIOD (2013 release), averaged for the years 1995–99. The data covers 37 individual economies and the Rest of the World.Variance decomposition for 1995–99
Citation: IMF Working Papers 2022, 090; 10.5089/9798400208843.001.A999
In each panel, the horizontal-axis variable is the first-order linear approximation of (Mn'n — Mnn')/(Mn'nMnn') ‘1/2 from equation (6), represents the total spending by economy n on output from W. The vertical-axis variable is one each of the four right-hand-side terms in expression (6). The red line represents the line of best fit, whose respective slope is also printed in red. All data is based on WIOD (2013 release), averaged for the years 1995–99. The data covers 37 individual economies and the Rest of the World.Variance decomposition for 1995–99
Citation: IMF Working Papers 2022, 090; 10.5089/9798400208843.001.A999
In each panel, the horizontal-axis variable is the first-order linear approximation of (Mn'n — Mnn')/(Mn'nMnn') ‘1/2 from equation (6), represents the total spending by economy n on output from W. The vertical-axis variable is one each of the four right-hand-side terms in expression (6). The red line represents the line of best fit, whose respective slope is also printed in red. All data is based on WIOD (2013 release), averaged for the years 1995–99. The data covers 37 individual economies and the Rest of the World.Variation in economies’ aggregate trade balances accounts for 3% of the variation in bilateral trade imbalances. Differences in production and spending patterns (“triangular trade”) account for 9% of the variation, and asymmetric trade wedges account for the remaining 88%.
A.3 Dynamic Model
A.3.1 Agents’ Optimality
The utility maximisation problem of an agent born in t’ can be written as
subject to
where Int (f) is the agent’s investment in t; Bnt (t1) denotes bond holdings; Knt (ξ’) denotes capital holdings;
and the optimal portfolio requires
A.3.2 Steady-State Optimal Savings
We can analytically characterise the steady-state consumption and savings decisions of an agent born in period t' as a function of their period-t asset and human wealth:
Define
where ant ≡ Ant/Hnt. There is a stationary distribution of assets in steady state as long as
A.3.3 Steady-State Net Exports
In steady state,
This in turn implies
From the definition of GDP,
This, together with (52) and (54), gives us the steady-state trade balance-to-GDP ratio.
A.4 Exact-Hat Algebra
A.4.1 Key Outcomes and “Own Spending” Shares
In the spirit of Arkolakis et al. (2012), we can re-write a number of key conditions in terms of “own spending” shares. Specifically, from (20)-(28),
where vsn’n = Msn'n/ Σn' Msn'n = (τsn’npsn’)-θs/ Σn’ (τsn’npsn’)-θs is the economy-n’ trade share in economy-n expenditure in sector s.
A.4.2 Changes in Trade Costs and Productivity
For any steady-state outcome xn, define
Then:
where
Equations (60)-(62) describe the exact-hat algebra for our model conditional on given changes in trade balances and per-worker capital stocks,
A.4.3 Financial Autarky
We only consider the transition from our baseline assumption of perfectly integrated international asset markets (no barriers to international asset trade) to complete financial autarky (prohibitive barriers to international asset trade). The latter requires all net holdings of the international bond to be zero in equilibrium: Bnt = 0 for all n and t. Since economies differ in their production technologies and intertemporal preferences, each economy must have its “own” interest rate Rnt (instead of Rt) for this to be an equilibrium outcome.
Assuming an economy-specific interest rate Rnt, we can proceed as in Section A.3 to show that in steady state,
Financial autarky requires Bn = 0, which implies Ant = ηnPnKnt. Equating (68) and (69) yields a quadratic equation in permissible values of Rn.47 This quadratic equation has only one positive root, which corresponds to the steady-state interest rate:
It is straightforward to show that Bnt = 0 implies NXnt = 0 for all n and t.
The exact-hat algebra required to compute outcomes in the new financial-autarky steady state is now summarised by the following system of equations:
where Rn is given in equation (70) and, from the reasoning above,
A.5 Additional Details on Counterfactuals
A.5.1 Bilateral Exposure and Global Trade-Wedge Symmetry
Beyond real GDP and consumption effects, one way in which counterfactual global trade-wedge may impact the global economy is by altering the relative dependence of economies on different trade partners. We refer to this as economies’ bilateral “exposures”. Specifically, we define the “exposure” of economy n to n’ as the percent change in economy-n steady-state real GDP in response to a permanent 1 percent increase in the aggregate productivity of economy
We compute the matrix of bilateral exposures, as defined above, for all economies in our data using the exact-hat algebra in equations (60)-(66). Figure A3 gives an overview of the results in matrix form. The matrix shows the row economy’s GDP response to a 1 percent aggregate-productivity increase in the column economy. Diagonal elements showing economies’ exposures to themselves are omitted, and the off-diagonal elements are colour-coded: darker shades of green indicate greater positive exposures (economy-n real GDP rises in economy-n’ aggregate productivity); darker shares of red indicate greater negative exposures (economy-n real GDP declines in economy-n’ aggregate productivity).
Bilateral exposures
Citation: IMF Working Papers 2022, 090; 10.5089/9798400208843.001.A999
Each cell shows the percentage change in the row economy’s steady-state real GDP per capita in response to a 1 percent increase in the aggregate productivity of the column economy. See Section A.5.1 for details. Calibration on data from PWT (edition 9.0) and WIOD (2016 release), average for the years 2010–14.Bilateral exposures
Citation: IMF Working Papers 2022, 090; 10.5089/9798400208843.001.A999
Each cell shows the percentage change in the row economy’s steady-state real GDP per capita in response to a 1 percent increase in the aggregate productivity of the column economy. See Section A.5.1 for details. Calibration on data from PWT (edition 9.0) and WIOD (2016 release), average for the years 2010–14.Bilateral exposures
Citation: IMF Working Papers 2022, 090; 10.5089/9798400208843.001.A999
Each cell shows the percentage change in the row economy’s steady-state real GDP per capita in response to a 1 percent increase in the aggregate productivity of the column economy. See Section A.5.1 for details. Calibration on data from PWT (edition 9.0) and WIOD (2016 release), average for the years 2010–14.As would be expected, most bilateral exposures are small in absolute value. However, productivity changes in the larger economies — notably, the U.S., China, and Germany — have economically significant effects on the real incomes of all economies. China in particular stands out, with the median country gaining .12 percent of real income for from a 1 percent increase in Chinese aggregate productivity. This reflects China’s centrality in global value chains that has been widely noted elsewhere.48
By our measures, most economies’ trade wedges in importing from China are lower than China’s wedges in importing from them. As a result, for most economies global trade-wedge symmetry implies a trade liberalisation vis-à-vis trade partners other than China. We now explore to what extent this changes the patterns of bilateral exposure. We do so by using the exact-hat algebra in equations (60)-(66) to compute bilateral exposures for all economies after global trade wedges have been made symmetric. Figure A4 presents the changes in bilateral exposures relative to what is shown in Figure A3. Again, matrix elements are colour-coded: darker shades of green indicate greater positive changes; darker shades of red indicate greater negative changes. As expected the exposure to China declines for the large majority of economies, with a .01 percentage point decline for the median economy. By contrast, global trade-wedge symmetry increases almost all economies exposure to the Rest of the World.
Changes in bilateral exposures due to global trade-wedge symmetry
Citation: IMF Working Papers 2022, 090; 10.5089/9798400208843.001.A999
Each cell shows the percentage change in the row economy’s “exposure” to the column economy. See Section A.5.1 for details. Calibration on data from PWT (edition 9.0) and WIOD (2016 release), average for the years 2010–14.Changes in bilateral exposures due to global trade-wedge symmetry
Citation: IMF Working Papers 2022, 090; 10.5089/9798400208843.001.A999
Each cell shows the percentage change in the row economy’s “exposure” to the column economy. See Section A.5.1 for details. Calibration on data from PWT (edition 9.0) and WIOD (2016 release), average for the years 2010–14.Changes in bilateral exposures due to global trade-wedge symmetry
Citation: IMF Working Papers 2022, 090; 10.5089/9798400208843.001.A999
Each cell shows the percentage change in the row economy’s “exposure” to the column economy. See Section A.5.1 for details. Calibration on data from PWT (edition 9.0) and WIOD (2016 release), average for the years 2010–14.While these changes are quantitatively small, they point to an intriguing possibility. In recent months, G7 policy makers have expressed a concern that the global economy’s resilience to economic shocks may have been undermined by an over-reliance on China’s manufacturing capacity in (some) global value chains.49 Our results suggest that, in addition to China’s size and comparative advantages, the current prominence of China in cross-border production networks may also be owed to a particular configuration of trade-wedge asymmetries.
A.5.2 Financial Autarky
Figure A5 gives a graphical overview of the macroeconomic impact of financial autarky across economies. The real-GDP and real-consumption changes primarily reflect a dramatic relocation of capital. Economies with net negative international bond holdings under full financial integration (towards the left-hand side of Figure A5) see their capital stocks and real income levels shrink in financial autarky. Meanwhile, economies with net positive bond holdings under financial integration (towards the right-hand side of Figure A5) see their capital stocks and real incomes grow. However, both groups experience a decline in their real consumption levels. This is because the former lose the benefit of higher wages supported by externally financed capital investments, while the latter lose the benefit of higher foreign investment returns.
Impact of financial autarky on real GDP and consumption
Citation: IMF Working Papers 2022, 090; 10.5089/9798400208843.001.A999
Percent change in real per-capita GDP and consumption relative to data in the financial-autarky steady state, as described in Section 3.4 and Appendix A.5.2. Calibration on data from PWT (edition 9.0) and WIOD (2016 release), average for the years 2010–14.Impact of financial autarky on real GDP and consumption
Citation: IMF Working Papers 2022, 090; 10.5089/9798400208843.001.A999
Percent change in real per-capita GDP and consumption relative to data in the financial-autarky steady state, as described in Section 3.4 and Appendix A.5.2. Calibration on data from PWT (edition 9.0) and WIOD (2016 release), average for the years 2010–14.Impact of financial autarky on real GDP and consumption
Citation: IMF Working Papers 2022, 090; 10.5089/9798400208843.001.A999
Percent change in real per-capita GDP and consumption relative to data in the financial-autarky steady state, as described in Section 3.4 and Appendix A.5.2. Calibration on data from PWT (edition 9.0) and WIOD (2016 release), average for the years 2010–14.The disappearance of macro trade surpluses and deficits also prompts changes in real incomes via the “transfer effect”: expenditure shifts towards the output of former trade-surplus economies, which causes a terms-of-trade in their favour, raising their real incomes, and lowering the real incomes of former trade-deficit economies. However, as found in Dekle et al. (2007, 2008), these effects are quantitatively small, and they are dwarfed for most economies by the impact of financial autarky on their capital stocks.
A.5.3 Sources and Concordances for U.S.-China Trade War Tariffs
We obtain data on tariff changes and import values at the 10-digit level of HS for the U.S. from Bown (2019). For China, we take data on tariff changes and import values at the 8-digit level of HS from Bown et al. (2019).50 Using a concordance from HS to ISIC Rev. 4, we aggregate the tariff changes at the (roughly) 2-digit level of ISIC used in the WIOD (2016 release). We then aggregate further to obtain tariff changes for the coarser set of sectors used throughout this paper (see Section 3). The resulting changes in trade wedges, upon which we base our counterfactual, are shown in Table A4.
A.6 Eaton and Kortum (2002)
This section presents a version of the Eaton-Kortum (2002) model that delivers the same steady-state relationships as our benchmark Armington model. We maintain most of the assumptions made in Section 3, but replace the Armington side of the model, equations (17) and (18), with the assumption that the non-tradable sector-s input is assembled from tradable varieties according to the CES production function
where χs ≥ 0. xsnt represents the use of variety i in the production of the sector-s input by economy n. Varieties are produced with technology
where αn,µsn ∈ (0,1). Ksnt(i), Hsnt(i), and Jsnt(i) respectively represent the capital, efficiency units of labour, and economy-n final good used in the production of variety i. Productivity shifter zsn(i) is the realisation of a random variable drawn independently for each i from a place-specific Fréchet probability distribution:
where
Goods markets continue to be perfectly competitive, and international trade is subject to the same iceberg transport costs: Ksn'n > 1 units of the economy-n’, sector-s variety must be shipped for one unit to arrive in economy n. Production factors can move freely between activities within economies, but cannot move across borders.
Under these assumptions, the steady-state relationships in section 3.1.4 must be adjusted as follows:
respectively replace equations (20) and (21), where Ξs ≡ {Γ [{βs + 1 — χs) /βs]}1/(1−χs), Γ [·] is the gamma function, and psn is still as defined in equation (21); and
replaces equation (24).
Re-defining κsn’n ≡ τsn'n, βs = θs and
A.7 Dollar-Value versus Proportional Bilateral Imbalances
In their pioneering analysis of bilateral trade balances, Davis and Weinstein (2002) investigate how much of the variation across country pairs in the US-dollar value of bilateral trade balances can be explained using a gravity equation under the assumption of symmetric trade barriers. They conclude that a large portion remains unexplained — and term this the “mystery of the excess trade balances”. However, their gravity equation does not control for multilateral resistance either through appropriate fixed effects or a theory-consistent non-linear regression model.
Felbermayr and Yotov (2021) revisit the estimation of Davis and Weinstein (2002) in a recent paper, updating it to control for multilateral resistance. As they find that the resulting gravity-predicted trade flows can be used to explain variation in the dollar value of bilateral trade balances well, they argue that this solves the “mystery”.
By contrast, our analysis focuses on variation across trade-partner pairs in proportional bilateral balances (bilateral trade balances relative to the geometric average of bilateral trade flows). Our approach is fully structural, and takes account of multilateral resistance. We find that, with respect to proportional bilateral trade balances, the “mystery” remains: a large part of their variation cannot be explained unless we allow for black-box asymmetries in trade wedges. As we argue in Section 2.1.2, we consider this the most appropriate test of the ability of structural gravity to explain trade imbalances: an analysis of the variation in the unnormalised dollar value of bilateral trade balances conflates the (well-understood) ability of structural gravity to explain variation in average trade flows across trade-partner pairs with the (less well-studied) inability of gravity to account for variation in the proportional gap between bilateral flows.
To give a sense of the effect of conflating the two, we use PPML to estimate a gravity regression of the form
where Ωn' is an economy-n’-exporter dummy; Πn is an economy-n-importer dummy; δn'n = δnn' is a pair dummy; and εn'n ≠ εnn' is a mean-zero error.51 As the left-hand-side variable, we use the 1995–1999 average value of bilateral trade flows from WIOD (2013 release) for 37 individual economies and the “Rest of the World” (= 1406 pairs). We use this data to facilitate comparison with Davis and Weinstein (2002), who use data for 1995. Based on our estimates, we then construct
i.e. the gravity trade value exempting any trade-wedge asymmetries (the magnitude of which is captured by In
Figure A6 plots
Accounting for economy-pair variation in the simple differences of bilateral flows
Citation: IMF Working Papers 2022, 090; 10.5089/9798400208843.001.A999
“Unnormalised difference in bilateral trade flows” refers to Mn'n — Mnn', where Mn'n is the value (in million US$) of imports by economy n from n'. “Data imbalances” are the unnormalised differences observed in the data. “Explained...” are the unnormalised differences predicted on the basis of equations (83) and (84) in Appendix A7. All data is based on WIOD (2013 release), averaged for the years 1995–99. The data covers 37 individual economies and the Rest of the World.Accounting for economy-pair variation in the simple differences of bilateral flows
Citation: IMF Working Papers 2022, 090; 10.5089/9798400208843.001.A999
“Unnormalised difference in bilateral trade flows” refers to Mn'n — Mnn', where Mn'n is the value (in million US$) of imports by economy n from n'. “Data imbalances” are the unnormalised differences observed in the data. “Explained...” are the unnormalised differences predicted on the basis of equations (83) and (84) in Appendix A7. All data is based on WIOD (2013 release), averaged for the years 1995–99. The data covers 37 individual economies and the Rest of the World.Accounting for economy-pair variation in the simple differences of bilateral flows
Citation: IMF Working Papers 2022, 090; 10.5089/9798400208843.001.A999
“Unnormalised difference in bilateral trade flows” refers to Mn'n — Mnn', where Mn'n is the value (in million US$) of imports by economy n from n'. “Data imbalances” are the unnormalised differences observed in the data. “Explained...” are the unnormalised differences predicted on the basis of equations (83) and (84) in Appendix A7. All data is based on WIOD (2013 release), averaged for the years 1995–99. The data covers 37 individual economies and the Rest of the World.By contrast, Figure A7 plots
Accounting for economy-pair variation in proportional bilateral imbalances
Citation: IMF Working Papers 2022, 090; 10.5089/9798400208843.001.A999
“Proportional bilateral imbalances” refers toAccounting for economy-pair variation in proportional bilateral imbalances
Citation: IMF Working Papers 2022, 090; 10.5089/9798400208843.001.A999
“Proportional bilateral imbalances” refers toAccounting for economy-pair variation in proportional bilateral imbalances
Citation: IMF Working Papers 2022, 090; 10.5089/9798400208843.001.A999
“Proportional bilateral imbalances” refers toA.8 Appendix Tables
Sample of economies
Sample of economies
| WIOD (2016) | Final data | ||
|---|---|---|---|
| Economy | Code | Economy | Code |
| Australia | AUS | Australia | AUS |
| Austria | AUT | Austria | AUT |
| Belgium | BEL | Belgium | BEL |
| Brazil | BRA | Brazil | BRA |
| Bulgaria | BGR | Bulgaria | BGR |
| Canada | CAN | Canada | CAN |
| China | CHN | China | CHN |
| Croatia | HRV | Croatia | HRV |
| Cyprus | CYP | Rest of the World | RoW |
| Czech Republic | CZE | Czech Republic | CZE |
| Denmark | DNK | Denmark | DNK |
| Estonia | EST | Estonia | EST |
| Finland | FIN | Finland | FIN |
| France | FRA | France | FRA |
| Germany | DEU | Germany | DEU |
| Greece | GRC | Greece | GRC |
| Hungary | HUN | Hungary | HUN |
| India | IND | India | IND |
| Indonesia | IDN | Indonesia | IDN |
| Ireland | IRL | Ireland | IRL |
| Italy | ITA | Italy | ITA |
| Japan | JPN | Japan | JPN |
| Korea | KOR | Korea | KOR |
| Latvia | LVA | Latvia | LVA |
| Lithuania | LTU | Lithuania | LTU |
| Luxembourg | LUX | Rest of the World | RoW |
| Malta | MLT | Rest of the World | RoW |
| Mexico | MEX | Mexico | MEX |
| Netherlands | NLD | Netherlands | NLD |
| Norway | NOR | Norway | NOR |
| Poland | POL | Poland | POL |
| Portugal | PRT | Portugal | PRT |
| Rest of the World | RoW | Rest of the World | RoW |
| Romania | ROU | Romania | ROU |
| Russia | RUS | Russia | RUS |
| Slovakia | SVK | Slovakia | SVK |
| Slovenia | SVN | Slovenia | SVN |
| Spain | ESP | Spain | ESP |
| Sweden | SWE | Sweden | SWE |
| Switzerland | CHE | Switzerland | CHE |
| Taiwan, Prov. of China | TWN | Taiwan, Prov. of China | TWN |
| Turkey | TUR | Turkey | TUR |
| U.K. | GBR | U.K. | GBR |
| U.S. | USA | U.S. | USA |
Sample of economies
| WIOD (2016) | Final data | ||
|---|---|---|---|
| Economy | Code | Economy | Code |
| Australia | AUS | Australia | AUS |
| Austria | AUT | Austria | AUT |
| Belgium | BEL | Belgium | BEL |
| Brazil | BRA | Brazil | BRA |
| Bulgaria | BGR | Bulgaria | BGR |
| Canada | CAN | Canada | CAN |
| China | CHN | China | CHN |
| Croatia | HRV | Croatia | HRV |
| Cyprus | CYP | Rest of the World | RoW |
| Czech Republic | CZE | Czech Republic | CZE |
| Denmark | DNK | Denmark | DNK |
| Estonia | EST | Estonia | EST |
| Finland | FIN | Finland | FIN |
| France | FRA | France | FRA |
| Germany | DEU | Germany | DEU |
| Greece | GRC | Greece | GRC |
| Hungary | HUN | Hungary | HUN |
| India | IND | India | IND |
| Indonesia | IDN | Indonesia | IDN |
| Ireland | IRL | Ireland | IRL |
| Italy | ITA | Italy | ITA |
| Japan | JPN | Japan | JPN |
| Korea | KOR | Korea | KOR |
| Latvia | LVA | Latvia | LVA |
| Lithuania | LTU | Lithuania | LTU |
| Luxembourg | LUX | Rest of the World | RoW |
| Malta | MLT | Rest of the World | RoW |
| Mexico | MEX | Mexico | MEX |
| Netherlands | NLD | Netherlands | NLD |
| Norway | NOR | Norway | NOR |
| Poland | POL | Poland | POL |
| Portugal | PRT | Portugal | PRT |
| Rest of the World | RoW | Rest of the World | RoW |
| Romania | ROU | Romania | ROU |
| Russia | RUS | Russia | RUS |
| Slovakia | SVK | Slovakia | SVK |
| Slovenia | SVN | Slovenia | SVN |
| Spain | ESP | Spain | ESP |
| Sweden | SWE | Sweden | SWE |
| Switzerland | CHE | Switzerland | CHE |
| Taiwan, Prov. of China | TWN | Taiwan, Prov. of China | TWN |
| Turkey | TUR | Turkey | TUR |
| U.K. | GBR | U.K. | GBR |
| U.S. | USA | U.S. | USA |
Sector sample
Sector sample
| WIOD (2016) | Final data | ||
|---|---|---|---|
| Sector | ISIC 2-dg. | New code | |
| Crop and animal production... | 1 | 1 | |
| Forestry and logging | 2 | 1 | |
| Fishing and aquaculture | 3 | 1 | |
| Mining and quarrying | 5–9 | 2 | |
| Manufacture of food products,... | 10–12 | 3 | |
| Manufacture of textiles,... | 13–15 | 4 | |
| Manufacture of wood and... | 16 | 5 | |
| Manufacture of paper and... | 17 | 6 | |
| Printing and reproduction... | 18 | 6 | |
| Manufacture of coke and... | 19 | 7 | |
| Manufacture of chemicals... | 20 | 8 | |
| Manufacture of basic pharma... | 21 | 8 | |
| Manufacture of rubber and... | 22 | 9 | |
| Manufacture of other non-metal... | 23 | 10 | |
| Manufacture of basic metals | 24 | 11 | |
| Manufacture of fabricated metal... | 25 | 11 | |
| Manufacture of computer,... | 26 | 13 | |
| Manufacture of electrical equip... | 27 | 13 | |
| Manufacture of machinery and... | 28 | 12 | |
| Manufacture of motor vehicles,... | 29 | 14 | |
| Manufacture of other transport... | 30 | 14 | |
| Manufacture of furniture; other... | 31–32 | 15 | |
| Repair and installation... | 33 | 15 | |
| Electricity, gas, steam and... | 35 | 16 | |
| Water collection, treatment... | 36 | 16 | |
| Sewerage; waste collection,... | 37–39 | 16 | |
| Construction | 41–43 | 17 | |
| Wholesale and retail... | 45 | 18 | |
| Wholesale trade, except... | 46 | 18 | |
| Retail trade, except of... | 47 | 19 | |
| Land transport and... | 49 | 21 | |
| Water transport | 50 | 22 | |
| Air transport | 51 | 23 | |
| Warehousing and support... | 52 | 24 | |
| Postal and courier activities | 53 | 25 | |
| Accommodation and food... | 55–56 | 20 | |
| Publishing activities | 58 | 6 | |
| Motion picture, video and... | 59–60 | 6 | |
| Telecommunications | 61 | 25 | |
| Computer programming,... | 62–63 | 28 | |
| Financial service activities,... | 64 | 26 | |
| Insurance, reinsurance and... | 65 | 26 | |
| Activities auxiliary to financial... | 66 | 26 | |
| Real estate activities | 68 | 27 | |
| Legal and accounting activities;... | 69–70 | 28 | |
| Architectural and engineering... | 71 | 28 | |
| Scientific research and... | 72 | 28 | |
| Advertising and market research | 73 | 28 | |
| Other professional, scientific... | 74–75 | 28 | |
| Administrative and support... | 77–82 | 28 | |
| Public administration and... | 84 | 31 | |
| Education | 85 | 29 | |
| Human health and social work... | 86–88 | 30 | |
| Other service activities | 90–96 | 31 | |
| Activities of households... | 97–98 | 28 | |
| Activities of extraterritorial... | 99 | 31 | |
Sector sample
| WIOD (2016) | Final data | ||
|---|---|---|---|
| Sector | ISIC 2-dg. | New code | |
| Crop and animal production... | 1 | 1 | |
| Forestry and logging | 2 | 1 | |
| Fishing and aquaculture | 3 | 1 | |
| Mining and quarrying | 5–9 | 2 | |
| Manufacture of food products,... | 10–12 | 3 | |
| Manufacture of textiles,... | 13–15 | 4 | |
| Manufacture of wood and... | 16 | 5 | |
| Manufacture of paper and... | 17 | 6 | |
| Printing and reproduction... | 18 | 6 | |
| Manufacture of coke and... | 19 | 7 | |
| Manufacture of chemicals... | 20 | 8 | |
| Manufacture of basic pharma... | 21 | 8 | |
| Manufacture of rubber and... | 22 | 9 | |
| Manufacture of other non-metal... | 23 | 10 | |
| Manufacture of basic metals | 24 | 11 | |
| Manufacture of fabricated metal... | 25 | 11 | |
| Manufacture of computer,... | 26 | 13 | |
| Manufacture of electrical equip... | 27 | 13 | |
| Manufacture of machinery and... | 28 | 12 | |
| Manufacture of motor vehicles,... | 29 | 14 | |
| Manufacture of other transport... | 30 | 14 | |
| Manufacture of furniture; other... | 31–32 | 15 | |
| Repair and installation... | 33 | 15 | |
| Electricity, gas, steam and... | 35 | 16 | |
| Water collection, treatment... | 36 | 16 | |
| Sewerage; waste collection,... | 37–39 | 16 | |
| Construction | 41–43 | 17 | |
| Wholesale and retail... | 45 | 18 | |
| Wholesale trade, except... | 46 | 18 | |
| Retail trade, except of... | 47 | 19 | |
| Land transport and... | 49 | 21 | |
| Water transport | 50 | 22 | |
| Air transport | 51 | 23 | |
| Warehousing and support... | 52 | 24 | |
| Postal and courier activities | 53 | 25 | |
| Accommodation and food... | 55–56 | 20 | |
| Publishing activities | 58 | 6 | |
| Motion picture, video and... | 59–60 | 6 | |
| Telecommunications | 61 | 25 | |
| Computer programming,... | 62–63 | 28 | |
| Financial service activities,... | 64 | 26 | |
| Insurance, reinsurance and... | 65 | 26 | |
| Activities auxiliary to financial... | 66 | 26 | |
| Real estate activities | 68 | 27 | |
| Legal and accounting activities;... | 69–70 | 28 | |
| Architectural and engineering... | 71 | 28 | |
| Scientific research and... | 72 | 28 | |
| Advertising and market research | 73 | 28 | |
| Other professional, scientific... | 74–75 | 28 | |
| Administrative and support... | 77–82 | 28 | |
| Public administration and... | 84 | 31 | |
| Education | 85 | 29 | |
| Human health and social work... | 86–88 | 30 | |
| Other service activities | 90–96 | 31 | |
| Activities of households... | 97–98 | 28 | |
| Activities of extraterritorial... | 99 | 31 | |
Sector sample and trade elasticities
Sector sample and trade elasticities
| Final data | |||
|---|---|---|---|
| New code | Sector | Trade elasticity | |
| 1 | Agriculture, hunting, forestry and fishing | 8.11 | |
| 2 | Mining and quarrying | 15.72 | |
| 3 | Food, beverages and tobacco | 2.55 | |
| 4 | Textiles and textile products; leather, leather apparel and footwear | 5.56 | |
| 5 | Wood and products of wood and cork | 10.83 | |
| 6 | Pulp, paper; paper, printing and publishing | 9.07 | |
| 7 | Coke, refined petroleum and nuclear fuel | 51.08 | |
| 8 | Chemicals and chemical products | 4.75 | |
| 9 | Rubber and plastics | 1.66 | |
| 10 | Other non-metallic, mineral products | 2.76 | |
| 11 | Basic metals and fabricated metal | 7.99 | |
| 12 | Electrical and optical equipment | 10.60 | |
| 13 | Machinery, nec | 1.52 | |
| 14 | Transport equipment | 0.37 | |
| 15 | Manufacturing, nec; recycling | 5 | |
| 16 | Electricity, gas and water supply | 5 | |
| 17 | Construction | 5 | |
| 18 | Wholesale trade, commission trade, including motor vehicles and motorcycles | 5 | |
| 19 | Retail trade, except of motor vehicles and motorcycles | 5 | |
| 20 | Hotels and restaurants | 5 | |
| 21 | Inland transport | 5 | |
| 22 | Water transport | 5 | |
| 23 | Air transport | 5 | |
| 24 | Other supporting and auxiliary transport activities; activities of travel agencies | 5 | |
| 25 | Post and telecommunications | 5 | |
| 26 | Financial intermediation | 5 | |
| 27 | Real estate activities | 5 | |
| 28 | Other business activities | 5 | |
| 29 | Education | 5 | |
| 30 | Health and social work | 5 | |
| 31 | Public admin, defence, social security and other public services | 5 | |
Sector sample and trade elasticities
| Final data | |||
|---|---|---|---|
| New code | Sector | Trade elasticity | |
| 1 | Agriculture, hunting, forestry and fishing | 8.11 | |
| 2 | Mining and quarrying | 15.72 | |
| 3 | Food, beverages and tobacco | 2.55 | |
| 4 | Textiles and textile products; leather, leather apparel and footwear | 5.56 | |
| 5 | Wood and products of wood and cork | 10.83 | |
| 6 | Pulp, paper; paper, printing and publishing | 9.07 | |
| 7 | Coke, refined petroleum and nuclear fuel | 51.08 | |
| 8 | Chemicals and chemical products | 4.75 | |
| 9 | Rubber and plastics | 1.66 | |
| 10 | Other non-metallic, mineral products | 2.76 | |
| 11 | Basic metals and fabricated metal | 7.99 | |
| 12 | Electrical and optical equipment | 10.60 | |
| 13 | Machinery, nec | 1.52 | |
| 14 | Transport equipment | 0.37 | |
| 15 | Manufacturing, nec; recycling | 5 | |
| 16 | Electricity, gas and water supply | 5 | |
| 17 | Construction | 5 | |
| 18 | Wholesale trade, commission trade, including motor vehicles and motorcycles | 5 | |
| 19 | Retail trade, except of motor vehicles and motorcycles | 5 | |
| 20 | Hotels and restaurants | 5 | |
| 21 | Inland transport | 5 | |
| 22 | Water transport | 5 | |
| 23 | Air transport | 5 | |
| 24 | Other supporting and auxiliary transport activities; activities of travel agencies | 5 | |
| 25 | Post and telecommunications | 5 | |
| 26 | Financial intermediation | 5 | |
| 27 | Real estate activities | 5 | |
| 28 | Other business activities | 5 | |
| 29 | Education | 5 | |
| 30 | Health and social work | 5 | |
| 31 | Public admin, defence, social security and other public services | 5 | |
Trade-cost changes as a result of the USA-CHN trade war
Trade-cost changes as a result of the USA-CHN trade war
| Final data | |||
|---|---|---|---|
| New code | Sector | ||
| 1 | Agriculture, hunting, forestry and fishing | 1.16 | 1.25 |
| 2 | Mining and quarrying | 1.06 | 1.10 |
| 3 | Food, beverages and tobacco | 1.19 | 1.19 |
| 4 | Textiles and textile products; leather, leather apparel and footwear | 1.05 | 1.14 |
| 5 | Wood and products of wood and cork | 1.20 | 1.19 |
| 6 | Pulp, paper; paper, printing and publishing | 1.20 | 1.16 |
| 7 | Coke, refined petroleum and nuclear fuel | 1.18 | 1.25 |
| 8 | Chemicals and chemical products | 1.14 | 1.11 |
| 9 | Rubber and plastics | 1.13 | 1.08 |
| 10 | Other non-metallic, mineral products | 1.17 | 1.12 |
| 11 | Basic metals and fabricated metal | 1.18 | 1.19 |
| 12 | Electrical and optical equipment | 1.18 | 1.10 |
| 13 | Machinery, nec | 1.11 | 1.08 |
| 14 | Transport equipment | 1.23 | 1.00 |
| 15 | Manufacturing, nec; recycling | 1.10 | 1.06 |
| 16 | Electricity, gas and water supply | 1.23 | 1.06 |
Trade-cost changes as a result of the USA-CHN trade war
| Final data | |||
|---|---|---|---|
| New code | Sector | ||
| 1 | Agriculture, hunting, forestry and fishing | 1.16 | 1.25 |
| 2 | Mining and quarrying | 1.06 | 1.10 |
| 3 | Food, beverages and tobacco | 1.19 | 1.19 |
| 4 | Textiles and textile products; leather, leather apparel and footwear | 1.05 | 1.14 |
| 5 | Wood and products of wood and cork | 1.20 | 1.19 |
| 6 | Pulp, paper; paper, printing and publishing | 1.20 | 1.16 |
| 7 | Coke, refined petroleum and nuclear fuel | 1.18 | 1.25 |
| 8 | Chemicals and chemical products | 1.14 | 1.11 |
| 9 | Rubber and plastics | 1.13 | 1.08 |
| 10 | Other non-metallic, mineral products | 1.17 | 1.12 |
| 11 | Basic metals and fabricated metal | 1.18 | 1.19 |
| 12 | Electrical and optical equipment | 1.18 | 1.10 |
| 13 | Machinery, nec | 1.11 | 1.08 |
| 14 | Transport equipment | 1.23 | 1.00 |
| 15 | Manufacturing, nec; recycling | 1.10 | 1.06 |
| 16 | Electricity, gas and water supply | 1.23 | 1.06 |
Cuñat (alejandro.cunat@univie.ac.at) is University of Vienna and CESifo. We are grateful to Pol Antràs, Tommaso Aquilante, Kirill Borusyak, Thomas Chaney, Giancarlo Corsetti, Arnaud Costinot, Harald Fadinger, Thibault Fally, Niko Hobdari, Pete Klenow, Nan Li, David Miles, Stephen Millard and Esteban Rossi-Hansberg as well as seminar participants at Edinburgh, the Bank of England, Georgetown Qatar, NIESR, Danmarks Nationalbank, the IMF, Graz, SED 2018, Vienna, Durham, the GEA Christmas Meeting 2018, Humboldt University, WIEN 2019, Strathclyde, IFS, Manchester, HSE Moscow, WIIW, Mainz and the Virtual ITM Seminar for helpful comments and suggestions. Zymek gratefully acknowledges the hospitality of the Bank of England while this paper was conceived, and financial support from the UK Economic and Social Research Council (ESRC) under award ES/L009633/1.
Cuñat (alejandro.cunat@univie.ac.at) is University of Vienna and CESifo. We are grateful to Pol Antràs, Tommaso Aquilante, Kirill Borusyak, Thomas Chaney, Giancarlo Corsetti, Arnaud Costinot, Harald Fadinger, Thibault Fally, Niko Hobdari, Pete Klenow, Nan Li, David Miles, Stephen Millard and Esteban Rossi-Hansberg as well as seminar participants at Edinburgh, the Bank of England, Georgetown Qatar, NIESR, Danmarks Nationalbank, the IMF, Graz, SED 2018, Vienna, Durham, the GEA Christmas Meeting 2018, Humboldt University, WIEN 2019, Strathclyde, IFS, Manchester, HSE Moscow, WIIW, Mainz and the Virtual ITM Seminar for helpful comments and suggestions. Zymek gratefully acknowledges the hospitality of the Bank of England while this paper was conceived, and financial support from the UK Economic and Social Research Council (ESRC) under award ES/L009633/1.
The figure is based on data from the 2016 release of the World Input-Output Database (WIOD), the latest currently available. Section 2.2.1 discusses this data source in more detail.
Since the data represented in Figure 1 covers the total value of all U.S. exports and imports divided across 40 economies/regions, the average bilateral trade balance equals the overall U.S. trade balance divided by 40.
In the last three decades, the U.S. trade deficit vis-à-vis Japan (Janow, 1994), China (Feenstra et al., 1998; Hughes, 2005) and, most recently, Germany (Swanson, 2017; Krug-man, 2017) has been in the spotlight as a possible symptom of “unfair” trade practices on the part of these countries against American producers.
Head and Mayer (2014) survey the use of structural gravity models in empirical studies of international trade. Costinot and Rodríguez-Clare (2014) provide an overview of the common analytical properties of “gravity class” quantitative trade models.
A popular trick in quantitative trade modelling, going back to Dornbusch et al. (1977), is to introduce macro trade imbalances as exogenous international transfers into an otherwise static model. We set up a dynamic model in which macro trade imbalances are an endogenous steady-state outcome that is independent of initial conditions. This allows us to derive a dynamic block of exact-hat equations which is modular to the exact-hat algebra often employed to perform counterfactuals in static trade models.
Two recent papers re-visit the question what explains observed bilateral imbalances. The first, by Felbermayr and Yotov (2021), re-estimates the gravity model of Davis and Weinstein (2002) and suggest that the inclusion of theory-consistent multilateral resistance terms goes some way in resolving the “mystery” as originally conceived. We discuss the differences between our approach and the analyses of Davis and Weinstein (2002) and Felbermayr and Yotov (2021) in more detail in Appendix A7. The second, by Eugster et al. (2020), focuses on changes in bilateral imbalances over time. Consistently with our finding that bilateral trade-wedge asymmetries are fairly persistent, they find that these changes are primarily driven by macro factors (such as the macro trade balance).
Carrère et al. (2020) summarise some of the main facts established and remaining open questions in this literature, including in relation to bilateral trade balances.
Reyes-Heroles (2016) studies the contribution of trade globalisation to the emergence of current account imbalances. Eaton et al. (2016b) and Ravikumar et al. (2019) perform trade policy experiments in dynamic models that permit trade imbalances. Sposi (2021) shows that bilateral trade barriers influence how a shock that causes a trade and current-account imbalance in one country is reflected in the trade balances and current accounts of its trade partners.
Since we restrict our analysis to a comparison of steady states, it primarily speaks to the long-run drivers of trade balances. In a recent paper, Alessandria and Choi (2021) employ a similar decomposition to the one we develop below to investigate the drivers of short-run changes in the U.S. trade balance in the period 1980–2015. They find that a significant role for asymmetric movements in trade barriers in explaining these changes.
In principle, these sectors could represent very narrowly defined goods or services. In our data application, they will correspond to broad sectors at the 2-digit level of ISIC, e.g. “Transport equipment”. At that level of aggregation, the output of a given sector is likely to comprise both intermediate and final goods, and expenditure flows will represent a combination of value-chain and final trade.
As we illustrate in our formal model in Section 3.1, τsn'n might reflect a combination of trade barriers arising from physical and policy barriers to the delivery of sector-s output from country n’ to n, and biases in sector-s spending by country n in relation to other economies’ outputs (e.g. home bias). Therefore, τsn'n can be thought of more generally as an ad-valorem equivalent trade wedge that captures the country-pair-specific forces shaping sector-level bilateral expenditure patterns.
See Anderson (2011), Costinot and Rodríguez-Clare (2014) and Head and Mayer (2014) for surveys of this literature.
The compatibility of structural gravity with key assumptions of the Melitz (2003) model also implies a potential point of connection with the emerging literature on buyer-seller interactions in cross-border production networks. That literature has been using Melitz-style assumptions in combination with micro data to model cross border trade and production from the bottom up. See, for example, Antràs et al. (2017), Bernard and Moxnes (2018), and Bernard et al. (2021).
See Davis and Weinstein (2002), p. 171: equation (5).
Note that this follows because Psn = Osn for all s and n in this case.
These patterns are not specific to the U.S. example. In Appendix A.7, we discuss more formally how an exclusive focus on unnormalised dollar-value bilateral net exports has led some studies to conclude that “country effects” can account for most of the variation in bilateral imbalances.
Note that a key requirement is that the estimation is performed on a full matrix of bilateral expenditures, including economies’ expenditures on their own outputs in sector s, {Msnn}s,n.
For simplicity, we will refer to the 41 “economies” in our data from now on, instead of the more accurate “40 economies and one region”. None of the stylised facts presented throughout the rest of the paper are sensitive to dropping the Rest of the World, and reporting statistics for only the 40 genuine economies (or the corresponding 780 trade-partner pairs) instead.
For our sample of economies, and our chosen level of sectoral aggregation, zero-valued flows are very uncommon. Out of a total of (31 x 41 x 41 =) 52,111 sector-country-pair flows, less than 2% are zero-valued.
To see this, note that the slope of a univariate linear regression of xi on y is Cov (xi, y) /Var (y); and that for y ≡ Σi xi, we can write Var (y) = Σi Cov (Σi, y).
A possible concern is that small-value bilateral trade flows may be more prone to measurement error, and that our measure of proportional imbalances gives the imbalances calculated from such flows an outsized weight in the variance decomposition. To address this concern, we repeat the variance decomposition for only the proportional imbalances associated with the top 50% of country pairs by the geometric-average value of their bilateral trade flows. The results we obtain are almost identical: macro trade balances account for 3% of the variation; differences in production and spending patterns account for 12%; and asymmetric trade wedges for 85%.
In a similar spirit, Matsuyama (1987) uses an open-economy version of the Blanchard model in order to analyse the current-account dynamics of a small open economy whose rate of time preference differs from the world interest rate.
Increasing-returns models such as Krugman (1980) and Chaney (2008) would also deliver the same structural gravity equation. However, the presence of market-size effects makes an analysis of model dynamics more intricate. The extent of the isomorphisms between all these models is discussed in detail in Costinot and Rodríguez-Clare (2014).
However, note that we allow for exogenous human capital growth below which could be re-interpreted as reflecting combined population and productivity growth.
After this uncertainty is resolved, the wealth of a surviving cohort-t’ member in period t is Ant (t’) / (1 — ξ).
Ravikumar et al. (2019) develop a framework for the analysis of the macroeconomic impact of trade-cost changes in a world of financially integrated economies in steady state as well as along the transition path.
In our calibration below, γ > R turns out to be the relevant case. If γ < R, the interest payments an impatient country makes in steady state outstrip its new international liabilities. In this case, the country’s steady-state GDP exceeds expenditure, causing a trade surplus. Conversely, a patient country will run a trade deficit in steady state.
Note that the labour share in PWT is computed as the share of labour income in GDP, which corresponds to 1 — αn in our model.
Note from equations (22) and (28) that ηn > 1 implies αnyn/kn > R—1+δ, so despite the assumption of fully integrated international asset markets, our model is consistent with the observed differences in the marginal product of capital across economies. One interpretation of a “low investment efficiency” (ηn > 1) is that it captures frictions in the flow of capital to certain economies in a black-box fashion.
Note that we only need to choose a value of the parameter θ for expositional reasons, so as to be able to interpret the numbers in Table 5 and Figure 9 in terms of ad-valorem trade costs.
China has a lower aggregate import than export wedge for only one fifth of its trade partners.
In line with these findings, if we return to our global symmetry counterfactual from Section 3.3.1, but only apply (31) in the top five sectors from Table 6, about half of the variation in proportional bilateral imbalances disappears. This is equivalent to roughly two thirds of the effect of full global trade-wedge symmetry. The distribution of macroeconomic impacts across economies is qualitatively similar to that described in Section 3.3.3, but the magnitude of impacts is commensurately smaller.
Three manufacturing sectors have fewer than 10,660 triplets as a result of zero-valued trade flows.
Note that Croatia only joined the E.U. in 2013, and that we group Cyprus, Luxembourg and Malta with the “rest of the world” as discussed in Section 2.2.1.
For example, Zuleeg (2020) notes that “the E.U.’s Single Market (SM) has been built around the concept of a level playing field, going further than the rules which exist to govern global interactions. There is an extensive body of law that ensures that European companies face the same conditions no matter which member state’s markets they enter, with EU institutions executing supranational implementation, arbitration and enforcement.”
We exclude the “Rest of the World” from these regressions because its definition differs between the two datasets. See Appendix A.2 for a description of the properties of the 1995–99 data. After excluding the “Rest of the World”, we have 40 individual economies in our 2010–14 data, but only 38 of these are also in our 1995–99 data. This leaves (38 x 37/2 =) 666 unique pairs on which to perform the regression in (35). In line with (36) we define unique pairs such that their 1995–99 trade-wedge gap is positive, and we can thus assess by means of (35) if E.U. membership shrinks this gap.
See Mayer et al. (2019) for a discussion of the literature on the trade-promoting effects of E.U. membership as well as updated estimates.
See Bown (2019) and Bown et al. (2019).
Note that we continue to abstract from tariff revenue, as we are primarily interested in the impact of the trade war on trade imbalances, not welfare. Other studies have found similar-sized economic impacts from the U.S.-China trade war (see Fajgelbaum and Khan-delwal, 2021 for a survey) or alternative scenarios for U.S.-China decoupling (see Cerdeiro et al., 2021).
Unsurprisingly, the changes in U.S. imbalances as a result of the trade war have little effect on the global distribution of proportional bilateral imbalances: the correlation across all our pairs of economies between the empirical bilateral imbalances across and their post-trade-war counterfactual counterparts is .98.
Recently, regional trade agreements in Asia (CPTPP and RCEP), Latin America (Pacific Alliance) and Africa (AfCFTA) have sought to create conditions for deeper trade integration among their member countries.
Note that Rn ∈ [γ(1 + ρn), γ (1 + ρn) / (1 — ξ)] is required for
See, for example, Baldwin and Freeman (2021) for a recent discussion and evidence on China’s centrality in global value chains.
See, for example, G7 Panel on Economic Resilience (2021).
We would like to thank Chad Bown for making this data available.
This is a simplified version of the structural gravity model estimated by Felbermayr and Yotov (2021). The authors use PPML to estimate a non-linear model under the inclusion of theory-consistent mass variables and multilateral resistance terms. As Fally (2015) shows, this is equivalent to estimating a PPML gravity equation with a full set of importer and exporter fixed effects.
