Every year since 2012, the International Monetary Fund (IMF) has released an External Sector Report (ESR) on the global patterns of trade balances and exchange rates.1 These reports rely on a comprehensive framework for determining desired levels of balances by country and understanding the policies and market forces that give rise to any undesired deviations. The latest ESR improves the statistical model and reaches sensible policy conclusions on trade imbalances in the largest economies. However, there remain some important areas for statistical improvement, and the ESR continues to understate the distortions created by very large surpluses in a few medium-sized economies.
The major advance of the first ESR was a statistical model that analyzed the trade balances of the most important economies in a common framework that included policy variables and other fundamental economic factors.2 Within this framework, desired balances are determined by economic fundamentals such as demographics and wealth plus policies set at desirable levels. Undesired gaps arise both because policies are sometimes not at desirable levels (for example, the current large US fiscal deficit) or because of mistakes and misperceptions in financial markets.
Statistical improvements in the 2018 ESR include better measurement of foreign exchange intervention, better controls for the feedback of trade balances on intervention that can bias the estimate of its effect, the replacement of an ad hoc control for measurement errors in trade balances with a sensible correction of the data, and more careful modeling of demographic effects on trade balances.
There is always room for improvement. For example, the latest revisions show an increase in the estimated effect of foreign exchange intervention on trade balances, but only for countries with restrictions on capital flows; thus, the model continues to assume that intervention has no effect when there are no capital flow restrictions. However, my own work shows that intervention has a smaller but still highly significant effect in economies such as Norway and Singapore that have no restrictions on capital flows. A related issue is that the model does not include sovereign wealth fund flows in the intervention data, despite the fact that they are also official purchases of foreign assets. What matters economically is what governments do, not the labels they apply to their actions.
In principle, restrictions on capital flows ought to reduce the impact of many underlying variables (such as fiscal policy or demographic trends) on trade imbalances because they make it harder to borrow or lend across borders to support those imbalances. My research shows that this effect can be important; this is an area IMF staff should pursue further.
The statistical model is the starting point for the ESR's policy analysis. The model's conclusions about desired and excessive trade imbalances are tempered by the informed judgment of the IMF staff. Country-specific circumstances that cannot be included in the statistical model lead to deviations from the model's conclusions that IMF staff incorporate in a way that preserves overall consistency across countries.
For the largest economies—the United States, China, the euro area, and Japan—staff adjustments to the model are small, and the desired current account balances, or norms, are sensible. The 2018 ESR current account balance norms (in percent of GDP) are 3 percent for Japan, 2 percent for the euro area, ‑1/4 percent for China, and ‑3/4 percent for the United States. China and the euro area are said to have trade balances that are "moderately stronger" than desired, by about 2 percentage points in each case. Japan's trade balance is said to be "broadly consistent," less than 1 percentage point above its norm. The US trade balance is "moderately weaker" than desired, by about 2 percentage points.
Many of the policy recommendations are sensible: China should take steps to increase consumption, including by strengthening the social safety net; the euro area should encourage investment by increasing the resiliency of the currency union and should adopt more growth-friendly fiscal policies; the United States should focus on fiscal consolidation while improving export capacity through education, training, and infrastructure.
Notably absent from these recommendations is any use of foreign exchange intervention to reduce imbalances. China is urged to use its foreign exchange reserves only "to smooth excessive volatility." Intervention policy is not even mentioned for the euro area and the United States, reflecting the false assumption that it would have no effect.
It is unfortunate that the 2018 ESR continues the pattern in previous editions of applying judgmental adjustments that partially ratify the large trade surpluses of some medium-sized economies: Singapore, Switzerland, and Thailand.3 Another economy with a large trade surplus, Norway, is not included in the policy analysis despite being in the statistical model. Better statistical modeling and a more limited and rigorous application of judgmental adjustments would lead to the conclusion that in all four of these economies, excessive official purchases of foreign assets were a major contributor to large trade surpluses.
1. The reports focus on the broadest measure of the trade balance, the current account balance. These terms are used synonymously here.
2. The ESR also has two statistical models of exchange rates. However, the ESR correctly places less emphasis on these models, noting that "exchange rates tend to be more volatile and difficult to explain."
3. The details differ across these economies. In Switzerland, the staff adjustment reflects a reasonable estimate of measurement error, but relaxing the assumption that foreign exchange intervention has no effect when capital is mobile and acknowledging that a very large share of net foreign assets reflect past intervention would lead to the conclusion that excessive Swiss intervention policy was a major factor behind the Swiss trade surplus. In Thailand, staff make an unreasonably large adjustment for "political uncertainty, terms of trade, [and a] temporary tourism boom." Singapore is not in the statistical model; the staff breakdown of the surplus into 13.4 percentage points desired and 5.5 percentage points excessive uses a high degree of judgment supported by the country's large net foreign assets, high income per capita, and rapid aging. It is worth noting that the vast majority of Singapore's net foreign assets reflect official purchases through foreign exchange reserves and two sovereign wealth funds and thus are the result of past official policies.