By Brian Peters

The Chinese current-account surplus has averaged approximately 2 percent of GDP since 2015, declining from a 2007 peak near 10 percent yet remaining far above what its income level and demographic structure would predict in a global panel. We estimate this gap as the residual against the income-conditional demographic baseline fitted on 140 countries with China held out. The residual averages +5 to +12 percentage points of GDP across four sub-periods of 2000-2024, peaks at +14.3 pp in 2007, and decays at a 16.6-year half-life (bootstrap 95% CI 9.5 to 58.7 years). Country-cluster bootstrap and within-year joint-demographic permutation tests reject the null of zero residual in every period (p < 0.002). The residual is concentrated on the saving side. China’s private-saving residual exceeds prediction by +23 to +28 percentage points/GDP across all four sub-periods; the government-saving residual flipped from +3.5 in the 2000s to-4.0 in the 2020s; the gross-investment residual is +12 to +20 pp. Adding household consumption share and gross investment share to the baseline absorbs 70 percent of the residual under a contemporaneous specification and 30 to 50 percent under five-year-lagged regressors that mitigate simultaneous-determination bias. The aging-conditional channel survives lagging: the Z₁ × household-consumption interaction is-0.14 (p < 0.001) contemporaneously and-0.045 (p = 0.021) at five-year lags. A peer-group placebo on Thailand, Vietnam, Malaysia, Korea, Indonesia, India, Singapore, and Japan shows the absorption pattern is shared by Thailand and India but not by the other peers; the residual structure is not generic to developmental-state Asia. Sex-ratio and pension/financial-repression mechanisms produce null Z₁ interactions in crosscountry variation. Projecting the residual decay alongside the United Nations medium-variant demographic projection implies China’s current account flips to deficit around 2030 and reaches-2.8 pp of GDP by 2040 (95% CI-4.0 to-0.6), with the deepening demographic baseline rather than the residual driving the reversal.

Source: SSRN