Young person visiting old person in lockdown
VoxEU Column

How public policies can curb pandemics: It’s important to take the age composition into account

In the ongoing debate on pandemic management, the role of age composition in shaping optimal policies has received little attention. This column develops an economic framework that specifically addresses the heterogeneity in age responses to pandemic risks, encompassing both individual voluntary behavioural changes and policy interventions, to show that targeted interventions based on demographic and disease-specific characteristics are often beneficial.

The ongoing debate on pandemic management has largely centred around the implementation of lockdown measures and their economic repercussions. One aspect that has received less attention is the role of age composition in shaping optimal pandemic policies. Infectious diseases are often associated with a substantial age gradient in their health consequences, and the Covid-19 pandemic was a stark reminder of this. This age component raises questions about the nature of voluntary protective behaviour across age demographics and its implications for policy design. More precisely, how should governments implement public policies aimed at curbing the effects of pandemics? And should these policies differ by age, and if so, how? The answer to this question depends on the characteristics of the disease at hand and the state of a particular economy.

The discourse surrounding pandemic response has evolved significantly in recent years, reflecting on contributions from various researchers who emphasise the complexity and diversity of effective strategies. While initial research contributions such as Acemoglu et al. (2021) advocated for strong lockdowns especially of the elderly based on a modelling framework where individuals do not themselves adjust their behaviour in response to the disease, in reality there was substantial heterogeneity in protective behaviour in response to Covid-19 (e.g. see Conti and Giustinelli 2023 for the UK). Papetti and Giagheddy (2021) and Turner et al. (2021) highlight the efficacy of age-specific containment measures and comprehensive public health policies in mitigating both epidemiological and economic consequences. We previously wrote about how alternative policies can affect different age groups in the context of Covid-19 (Brotherhood et al, 2020). In this column, we argue how optimal lockdowns that are sensitive to age and economic factors are crucial in dealing with pandemic challenges.

A new economic model of pandemics: Behavior, testing, and policies

In a recent study (Brotherhood et al. 2024), we develop an economic framework that specifically addresses the heterogeneity in age responses to pandemic risks, encompassing both individual voluntary behavioural changes and policy interventions. In the model, individuals weigh the benefits of social distancing against its costs, through lost income and reduced leisure, with options for remote work and safer leisure activities as partial substitutes. Due to initial symptom ambiguity, testing becomes crucial. Moreover, disease severity, economic needs, and mortality risks vary between younger and older populations. The government can limit outdoor activities to reduce social interactions differentially by age.

We initially focus on the Covid-19 pandemic, calibrating our model to the US experience including the actual lockdown measures imposed by the US government during the first year of the pandemic. This calibration tailors the model to reflect real-world behaviours and outcomes by aligning it with pre-existing time allocations across different age groups and key demographic information. The model replicates well the actual age-specific hospitalisation and death rates over time (see Figure 1a). Importantly, the time series of time spent at home in the model lines up well with the data (see Figure 1b).

Figure 1 Model versus data

Figure 1 Model versus data

Optimal lockdowns with behavioural responses by the old and the young

Behavioural changes – both voluntary and policy-induced – had a big impact on the dynamics of the Covid-19 pandemic, leading to an 80% reduction in deaths compared to a world in which individuals behaved as if no disease were present. Voluntary actions alone significantly reduced mortality rates: the model suggests that without any government intervention, voluntary protective behaviour, especially among the older population, would have reduced the death toll by 65% compared to a scenario with no behavioural adjustments.

We solve for the optimal lockdown and judge the strictness of policies by the additional hours of lockdown relative to a world without lockdowns. Unlike simple policies that reduce time outside uniformly for everyone, the model reveals that an age-targeted lockdown strategy, where younger individuals face more stringent restrictions compared to the old, results in a more effective reduction in mortality rates. This counterintuitive strategy arises from the differential impact of the virus across age groups and the economic and social activities characteristic of each group. In a ‘no lockdown’ world, the young tend to neglect taking strong precautions to limit the spread of the disease due to low personal risk, leaving the old to bear an undue burden (dashed lines in Figure 2a). The optimal lockdown thus prescribes less time outside for the young (solid blue line in Figure 2a). This more restrictive lockdown leads to a more contained pandemic. With a safer environment, the planner can allow the old to enjoy more time outside relative to a ‘no lockdown’ scenario (solid orange line in Figure 2a). The planner’s stringency is contingent on the duration required to sufficiently control the disease, a process spanning two winters. Figure 2b shows that deaths for both age groups are much lower under the optimal policy and they essentially disappear two years into the pandemic. Following this period, the combination of low disease prevalence and increased testing capacity allows the planner to ease restrictions significantly. Note that the planner allows systematically more time outside for both age groups throughout the first half of 2022. This recipe for the optimal lockdown highlights that integrating behaviour, testing, and policies in a unified framework is vital for assessing the optimal pandemic response.

Figure 2 The optimal lockdown for the US Covid-19 pandemic

Figure 2 The optimal lockdown for the US Covid-19 pandemic

Looking beyond Covid

The framework can also be used to look at other diseases beyond Covid-19. We start with the Spanish flu of the 1910s and recalibrate our model to replicate features of the Spanish flu pandemic. Optimal policy would entail milder reductions in social interactions for the Spanish flu, even though the overall death rate of this disease was higher. This result is based on a combination of different factors that were different 100 years prior, most importantly a younger population and a virus that was deadlier for the young. These results for the Spanish flu highlight the importance of both the characteristics of the economy and the characteristics of the disease itself.

Since disease characteristics are important when comparing Covid and Spanish flu, and since a possible next epidemic might yet carry other characteristics, we explore the implications of different disease characteristics in shaping the optimal lockdown policy systematically. To this end, we vary the infectiousness and deadliness and study the consequences for the design of optimal lockdowns. Several lessons emerge. Stringent lockdowns are warranted when the basic reproduction number (R0) is high, but less so when only the case fatality rate (CFR) is elevated. Figure 3 illustrates these points by plotting the increase in hours at home by the young under different disease scenarios. Take Covid-19, the middle bar in the figure. The dark portion shows the increase in time at home that the young engage in without any lockdowns. As noted previously, the optimal lockdown (light blue portion of the bar) prescribes more time at home for the young. The left two bars plot the same statistics for a disease with lower deadliness. The first bar has low infectiousness and the second bar, high infectiousness. With a less infectious disease, changes in hours are smaller. For the high-infection disease, prevalence is higher, and agents respond by spending more time at home. However, a higher infection rate means the externality is more pronounced. Hence, the planner prescribes a lot more time at home (light blue portion of the second bar). Similar results materialise for the more deadly disease. The difference is that this higher deadliness leads to more voluntary precautions (dark blue portions) and consequently less need for policy intervention. In other words, the externality is larger when deadliness is lower, counterintuitively requiring more policy intervention.

The age gradient is a crucial factor: if the case fatality rate is high among the young, a sizable and active group, fewer additional restrictions are necessary due to increased voluntary precautions. Two other important lessons are that optimal policy may not completely prevent all deaths, and that welfare benefits can be quite unevenly distributed across age groups. Economic conditions also have a big impact on the optimal lockdown. In scenarios where the older population is smaller, life expectancy is lower, or remote working is easy, a less restrictive policy is optimal. These features appear together in many developing countries, partially rationalising the significantly milder policy reactions there.

Figure 3 Increase in time at home for different diseases, young, hours per week

Figure 3 Increase in time at home for different diseases, young, hours per week

The importance of testing

A novel feature of our approach is the uncertainty of individuals regarding their health status, making testing an effective tool. We thus explore various dimensions of testing. While testing by itself does not eradicate Covid-19, it substantially lessens the impact of the virus. The structure of the optimal lockdown policy is also affected by the testing strategy, allowing for less stringent measures, diminishing economic losses, and enabling faster relaxation of restrictions. Testing boosts overall welfare, yet the value derived from the most effective lockdown strategy decreases as tests become more available. When tests are expensive and in short supply, allocating them primarily to younger individuals is advantageous.


In sum, our study highlights the importance of integrating age-specific responses, testing availability, and economic factors into pandemic policy design. Through the lens of our economic model, we discuss the balance between public health and economic activity. It emerges that targeted interventions based on demographic and disease-specific characteristics are often beneficial. The insights from our analysis, spanning from Covid to Spanish flu and other synthetic diseases, offer guidance for crafting more effective responses to future pandemics. It turns out that a one-size-fits-all approach falls short. Instead, policies must be tailored, leveraging testing and behavioural insights, to safeguard both lives and livelihoods in the face of evolving health threats.


Acemoglu, D, V Chernozhukov, I Werning, and M D Whinston (2021), “Optimal Targeted Lockdowns in a Multigroup SIR Model”, American Economic Review: Insights 3(4): 487-502.

Brotherhood, L, P Kircher, C Santos and M Tertilt (2024), “Optimal Age-based Policies for Pandemics: An Economic Analysis of Covid-19 and Beyond”, CEPR Discussion Paper 18759.

Brotherhood, L, P Kircher, C Santos and M Tertilt (2020), “The importance of testing and age-specific policies during the COVID-19 pandemic”,, 12 June.

Conti, G and P Giustinelli (2023), “One size does not fit all! Policy-relevant heterogeneity in health behaviour: Lessons from Covid-19 in the UK”,, 19 October.

Papetti, A and M Giagheddy (2021), “The macroeconomics of age-specific containment measures for COVID-19”,,

Turner, D, Y Guillemette, F Murtin and B Égert (2021), “Epidemiological and economic consequences of government responses to the COVID-19 pandemic”,, 2 January.