For most of the pandemic, Hong Kong appeared to have accomplished something many Western countries could not. Through strict border controls, quarantine, contact tracing, isolation and aggressive restrictions whenever cases appeared, the city kept COVID infections and deaths remarkably low. While hospitals elsewhere were overwhelmed and governments moved repeatedly between lockdown and reopening, Hong Kong largely prevented the virus from circulating through its population.
Then Omicron arrived.
Over the last several weeks, Hong Kong has experienced an extraordinary reversal. Cases have surged, hospitals have struggled, isolation facilities have filled and deaths have risen sharply. Images of overcrowded hospitals have replaced the earlier picture of successful containment. The government had been preparing a compulsory citywide testing programme and discussing even tighter restrictions, but on Friday Chief Executive Carrie Lam announced that the mass-testing plan would be suspended and that some travel and social restrictions would begin easing in April.
At first, the obvious conclusion is that zero-COVID failed.
I think that is too simple.
For much of 2020 and 2021, the strategy achieved exactly what it was designed to achieve: extremely low transmission and very few deaths compared with many other wealthy societies. Calling the entire policy a failure because it eventually became unsustainable would be another version of judging every decision only through its final outcome.
The more interesting question is why a strategy that worked for so long became so difficult to adapt once circumstances changed.
This distinction matters because good policy is not necessarily a policy that remains correct forever. It is a policy that responds when the relationship between costs and benefits changes.
COVID in March 2022 is not the same policy problem governments faced in March 2020. Vaccines now exist. Large numbers of people possess immunity from vaccination or previous infection. Treatments have improved. Omicron is substantially more transmissible than the original virus. Public tolerance for restrictions has changed. Businesses and schools have absorbed two years of disruption.
A strategy designed around preventing almost every chain of transmission therefore operates in a completely different environment.
The policy objective should change when the constraints change.
That sounds obvious.
Governments are often surprisingly bad at doing it.
One reason is political reputation. If a government has spent two years telling citizens that a particular strategy is evidence of competent leadership, abandoning the strategy can look like admitting that critics were right.
Policy therefore becomes connected to identity.
A government stops asking only, “Does this still work?” and begins asking, “What does changing it say about us?”
That is dangerous.
Past success should be irrelevant to whether a policy remains optimal today except insofar as it provides evidence about how the policy works.
Economists would describe part of this as a sunk-cost problem. Resources already spent cannot be recovered, so rational decisions should depend primarily on future costs and benefits.
Politics rarely behaves that cleanly.
Governments invest political capital as well as money. Leaders make speeches defending policies. Bureaucracies are constructed around them. Officials develop expertise in administering them. Businesses adapt to the rules. Supporters begin identifying the policy with the government’s competence.
Changing course then imposes a reputational cost even when the original decision was reasonable.
The result is policy lock-in.
This happens far beyond pandemics.
Governments continue subsidies long after their original purpose disappears because industries organise around receiving them. Militaries continue weapons programmes because careers and budgets depend on them. Infrastructure projects survive changing economic assumptions because cancelling them would mean acknowledging that money was wasted. Foreign policies continue because reversing them appears weak.
Institutions develop constituencies.
Once a policy creates organisations and expectations around itself, those organisations begin helping the policy survive.
Hong Kong’s COVID system became extraordinarily elaborate. Quarantine hotels, testing systems, contact-tracing procedures, isolation facilities and border restrictions were built around the assumption that preventing widespread transmission remained both possible and desirable.
Omicron attacked the first assumption.
A variant capable of spreading extremely rapidly makes elimination much more expensive because every missed case can produce many more infections before authorities detect it. Measures that successfully contained earlier variants may therefore need to become dramatically more restrictive to achieve the same outcome.
At some point, the marginal cost of preventing the next infection becomes enormous.
That is where policy needs a different objective.
Instead of minimising infections, governments may need to minimise severe disease and death.
Vaccination becomes more important than border closure.
Protecting vulnerable people becomes more important than identifying every mild case.
Hospital capacity becomes a more useful measure than raw infection numbers.
This is not “giving up”.
It is changing the optimisation problem.
But Hong Kong entered this transition with a major vulnerability: vaccination among elderly residents remained far too low.
That is one of the most important failures exposed by the current wave.
A society can suppress transmission so successfully that the urgency of vaccination becomes less visible. If almost nobody around you has COVID, the perceived benefit of vaccination may seem smaller, particularly to older people worried about side effects.
Paradoxically, successful containment can therefore reduce the behavioural incentive to prepare for containment eventually failing.
This is a version of the prevention paradox.
The policy succeeds, so the threat becomes invisible.
Because the threat becomes invisible, people underinvest in protection against it.
Then circumstances change.
The success has created a new vulnerability.
Hong Kong’s experience therefore should not be reduced to restrictions versus vaccines. The strongest pandemic strategies probably required both at different stages: restrictions to reduce transmission before vaccines were available, then aggressive vaccination so restrictions could eventually become less necessary.
A temporary policy needs an exit strategy.
Without one, success can become dependence.
This is where Hong Kong’s relationship with mainland China complicates the policy further. Beijing continues pursuing a dynamic zero-COVID strategy and has treated suppression of the virus as evidence that China’s political system can protect life more effectively than Western governments that tolerated much higher death tolls.
Hong Kong therefore does not make pandemic policy in a purely technocratic vacuum.
Its choices carry political meaning within China’s wider narrative about governance.
Moving away from zero-COVID can consequently look like more than a public-health adjustment. It can appear to challenge a policy strongly associated with the central government’s approach.
That creates another source of lock-in.
When policy becomes ideological, evidence has to fight identity.
This is not uniquely Chinese. Western governments did it too. Masks, vaccines, lockdowns and school closures became attached to political identities. Once a policy becomes a marker of which political group someone belongs to, changing one’s mind becomes socially expensive.
People stop asking whether new evidence changes the optimal policy and start asking whether accepting the evidence helps the other side.
That is terrible decision-making.
A competent government should be capable of saying: the policy was correct under the previous conditions; those conditions have changed; therefore the policy is changing too.
There is no contradiction in that statement.
In fact, refusing to change would be less rational.
This is why I think the political obsession with “U-turns” can be harmful. Opposition parties and journalists understandably scrutinise governments when they reverse policy because reversals can reveal incompetence or poor planning.
But not every reversal is failure.
Sometimes refusing to reverse is the failure.
Imagine a central bank raises interest rates because inflation is high. Inflation subsequently collapses and unemployment rises. Cutting rates does not prove the original increase was mistaken. It proves policy responds to information.
Pandemic policy should work similarly.
The relevant standard is not consistency.
It is responsiveness.
Consistency is valuable only when the underlying world remains sufficiently consistent.
Omicron is not consistent with the virus Hong Kong designed its original containment system around.
There is also an economic cost that has become increasingly difficult to ignore. Hong Kong is one of the world’s major financial and commercial centres. Its prosperity depends heavily on flows of people, capital and business across borders. Extended quarantine requirements and travel restrictions therefore impose particularly large costs.
A remote island with little international business might tolerate border closure relatively easily.
A global financial hub pays a different price.
Again, context determines policy.
The same restriction can produce very different cost-benefit calculations in different economies.
Hong Kong’s business community has increasingly warned that prolonged isolation risks damaging the city’s competitiveness. International companies can relocate employees. Executives can choose Singapore or another regional hub. Talent can leave.
These effects can become persistent because economic networks depend on expectations. Once a company moves regional operations elsewhere, removing the restriction that caused the move does not guarantee the company returns.
Temporary policy can therefore create permanent economic effects.
This is another reason governments need to think dynamically rather than only about immediate health outcomes.
A restriction may reduce infections this month while changing investment decisions for years.
That does not automatically make the restriction wrong.
It means the cost belongs in the calculation.
There is also a question of what Hong Kong’s current crisis tells us about comparisons between political systems.
Throughout the pandemic, governments have used COVID outcomes as evidence for broader ideological claims. China pointed towards low deaths as evidence of state capacity. Western critics pointed towards coercive restrictions as evidence of authoritarian excess. Later, Western vaccination campaigns and reopening were presented as evidence of liberal societies’ adaptability.
Most of these comparisons select whichever period makes the preferred political system look best.
Pandemics are poor ideological scoreboards.
Different political systems possess different advantages and weaknesses.
Centralised governments may coordinate rapidly and impose difficult restrictions. Democracies may receive more open criticism and feedback. Wealthy states possess more fiscal capacity. Societies with greater trust may achieve compliance with less coercion.
But none of these relationships is deterministic.
A democracy can manage a pandemic badly.
An authoritarian state can manage one effectively.
The relevant question is which institutions produce accurate information, adapt policy when evidence changes and maintain enough legitimacy that citizens cooperate.
Adaptation may be the most important of those.
A system that implements the perfect policy for yesterday can still fail tomorrow.
This is where Hong Kong’s decision on Friday may be more important than whether every element of its new approach is correct.
The government has acknowledged, at least implicitly, that the previous trajectory cannot continue unchanged. Compulsory mass testing has been suspended rather than pursued simply because officials previously announced it. Travel restrictions will begin easing. The emphasis is gradually shifting.
That willingness to adjust should be encouraged rather than treated automatically as evidence that everything before it was wrong.
But adaptation also requires accountability.
If elderly vaccination was inadequate, why?
Did authorities communicate poorly?
Were citizens given insufficient incentives?
Did the success of suppression create complacency?
Were resources directed too heavily towards testing and isolation rather than vaccination?
Those questions matter because policy learning requires identifying which assumptions failed.
Otherwise governments merely move from one strategy to another without understanding why.
There is also a larger lesson about uncertainty.
In early 2020, nobody knew whether vaccines would arrive quickly, how the virus would mutate or how long immunity would last. Governments had to choose strategies with incomplete information.
Looking backwards from 2022, it is easy to say which restrictions were unnecessary or which preparations should have been made earlier.
That is hindsight.
The fairer test is whether governments updated their policies as new information became available.
A decision can be reasonable when made and unreasonable when maintained.
That distinction is crucial.
Political accountability should therefore ask two questions rather than one.
Was the original policy justified by the evidence available at the time?
Did the government change it quickly enough when the evidence changed?
Hong Kong may receive different answers to those questions.
Its early suppression strategy saved lives.
Its preparation for the eventual arrival of a highly transmissible variant, particularly through vaccination of vulnerable elderly residents, appears much harder to defend.
Both can be true.
This is more useful than forcing the entire pandemic response into “success” or “failure”.
Policies have stages.
Institutions learn—or fail to learn.
Circumstances change.
The political system has to change with them.
Perhaps this is one of the most important differences between ideology and policy. Ideology seeks consistency because it provides a stable framework for interpreting the world. Policy should be much more willing to be inconsistent because the world itself changes.
Governments should have principles that remain relatively stable: protect life, preserve liberty, distribute burdens fairly, maintain economic opportunity.
The instruments used to pursue those principles should change constantly when evidence demands it.
Zero-COVID is an instrument.
It should never become the objective itself.
The objective is reducing the total harm caused by the pandemic.
For two years, preventing transmission was an unusually effective way for Hong Kong to accomplish that.
Omicron has changed the relationship between the instrument and the objective.
The government now has to change with it.
There is nothing inherently embarrassing about that.
The embarrassing outcome would be continuing a policy after its costs exceed its benefits simply because admitting change feels politically harder than imposing another restriction.
Governments often describe consistency as strength.
Sometimes the stronger institution is the one capable of admitting that the policy which worked yesterday no longer works today.
