By the end of 2023, did Hong Kong lose fewer lives than Sweden?
Hong Kong kept COVID out for two years. Sweden became the famous example of not locking down. The fair score is excess deaths: deaths above what the years before the pandemic predicted, from any cause. Count everything from January 2020 to December 2023.
Your guess first
The answer
Read this before the charts
- Excess is measured against a projection. "Expected deaths" is a trend fitted to 2015–19 (the World Mortality Dataset method used by Our World in Data). Another baseline gives other numbers; panel 2 lets you switch.
- Outcomes, not causes. Four other places that kept COVID out early (New Zealand, Australia, Singapore, Taiwan) all ended below Sweden. So "zero-COVID" alone does not explain Hong Kong. Half its loss came in one Omicron wave, and local factors such as how many older people were vaccinated by then are not measured here. The page also does not weigh what each policy cost in other ways.
- Hong Kong had excess deaths before COVID arrived. By end-2021 it was deaths over the projection, against recorded COVID deaths. That is either an indirect effect or the baseline running low. If it was the baseline, and that drift ran on through 2023, Hong Kong would end near per million: only just above Sweden.
- No age adjustment, for any place. Hong Kong people live long, so its expected death rate is lower than Sweden's. Measured against its own expected deaths (panel 2, "% above expected"), Hong Kong's gap over Sweden gets wider, not narrower.
- Series differ. Hong Kong, Macao, Singapore, Taiwan and Japan report monthly; the others weekly. South Korea's series stops on 1 Oct 2023, so it has no end-2023 value (NEED DATA) and is not ranked.
Four years in one chart
Cumulative excess deaths per million since January 2020. Tap or hover for values.
Source: Our World in Data excess mortality (World Mortality Dataset, Human Mortality Database), projected 2015–19 baseline, all ages. Data through 31 Dec 2023, fetched .
Change the rules
Pick a baseline, a start date and a cutoff. Does Hong Kong's place survive?
Per million views show approximate excess deaths underneath (≈, using the end-2023 population). "Plain 2015–19 average" ignores population growth and ageing, so it inflates places whose deaths were already rising (Japan, Hong Kong, the US). "% above expected" divides excess by expected deaths on the projected trend, a rough way to allow for each place's own death rate. Places with no data within 31 days of the cutoff show NEED DATA.
Where Hong Kong's deaths came from
Bars: Hong Kong's monthly excess deaths. Dots: COVID deaths the Centre for Health Protection recorded that month (by report date). The CHP count is "death cases" to 29 Jan 2023, then "deaths related to COVID-19"; neither is a cause-of-death finding.
Sources: Our World in Data excess mortality (Hong Kong, monthly). CHP "Latest situation of reported cases of COVID-19" file: cumulative deaths to 29 Jan 2023, then daily COVID-related deaths to ; after that the file stops. Recorded deaths are by report date, so the March 2022 backlog may spill across months.
How this was made
One public CSV from Our World in Data (excess mortality), joined with the Hong Kong CHP daily situation file. For each place, monthly or weekly excess = deaths − expected deaths, summed. Per million uses the population implied by OWID's own per-million column. Ranking needs data within 31 days of the cutoff. The pipeline recomputes OWID's cumulative column and checks it matches. An independent check script recomputes the headline numbers. Built unattended on 10 Oct 2026 by the Night Build routine. Education only; not medical advice.