The Covid-19 pandemic has led to a worldwide lockdown with unprecedented restrictions on economic activity. On March 24, Prime Minister Narendra Modi declared a nationwide lockdown to block the spread of the novel coronavirus. These restrictions, which came into force from March 25 were initially expected to be in place for three weeks. However, the lockdown in most states has been extended until May 3, with a possibility of further extensions. Other than a few essential services, all commercial, industrial, travel, religious, and cultural activity has been shut down.
The ensuing contraction of economic activity has led to intense discussions around the question of whether lockdowns are the appropriate policy response to the outbreak. How long of a lockdown can we afford? Is the stringency of the lockdown appropriate? Should the intensity of lockdown vary based on the exposure of regions to Covid-19 risk? These are all critical questions.
Unfortunately, this narrative seems to be premised on the notion that we need to save lives first; we can save the economy later. This is a false dichotomy. Saving the economy necessarily saves lives. Hence, policymakers need to assess the impact of the lockdown on public health and compare outcomes with similar ones for the outbreak. We, therefore, argue that the trade-off is between saving lives from Covid versus the lives lost due to contraction in economic activity. Evaluating the trade-off is far from easy. Nonetheless, we present a framework that will hopefully assist policymakers in thinking through these critical issues as they implement measures to contain the spread of the pandemic. Our exercise aims to translate the economic loss into a measure of lives to allow for an equivalent comparison of the outbreak and economic shutdown.
Saving the economy saves lives
A robust finding in epidemiological research is the inverse relationship between socioeconomic status and health status of individuals. Economic decline, especially in conjunction with unemployment and withdrawal from the labor force increases mortality and morbidity rates. There are other hard-to-quantify costs of a lockdown that affect human lives as well: Access to healthcare for non-Covid patients; rise in lifestyle-related illnesses; marital abuse; malnutrition among infants born during this period to the most vulnerable population; anxiety; and depression, among others. These costs may not immediately increase mortality but affect the longevity of individuals. Our framework can be expanded to include these costs as well.
We begin with an estimation of fatalities due to economic loss. Figure 1 below shows the relationship between real per capita GDP and mortality rates for the last two decades in India.
Between 1995 and 2019, India's per capita GDP quadrupled, while fatality rates fell by over 26%. We draw on this correlation to provide estimates of the direct impact on fatality as a result of economic loss. We start by examining the relationship between per capita GDP and mortality rate for Indian states between 1995 and 2019. We use these estimates to predict the baseline fatality rate in the absence of any Covid-related economic disruption. We next estimate the counterfactual fatality rates commensurate with a 5%-30% drop in GDP. The difference between the baseline and counterfactual fatality rates reflect the additional deaths due to loss in economic activity.
Figure 2A plots the estimated increase in fatality associated with a drop in GDP, assuming that the economy recovers within a year. Our estimates suggest that a 5% drop in GDP is associated with about 47,000-62,000 additional deaths, while a 30% drop is associated with 330,000-430,000 additional deaths.
Figure 2B plots the estimated deaths assuming that the economy would gradually recover over the next three years. Clearly, the number of lives lost is even greater if the effects are long-lasting.
However, these estimates do not provide the complete picture for at least two reasons. First, the cost of economic contraction is significantly higher for poorer regions such as Bihar and Orissa, implying that there are distributional consequences of stringent lockdowns with disproportionately larger effects on some fraction of the population.
The percentage increase in the fatality rate for a 30% drop in GDP is between 6-9% for states at the lower end of the income distribution in contrast to 1.2-5.5% for states at the higher end of the income distribution. Interestingly, as can be seen from Figures 3a and 3b below, some of these poorer states also have the lowest per capita Covid infection and fatality rate.
Second, the overall fatality masks significant heterogeneity in the effects across age groups. For instance, the median age of deceased due to COVID in Italy (UK) is 81 (80), while the life expectancy is 83 (81), suggesting a loss in human life of 2 (1) years for the deceased. Even in India, the fatality rate is greater for those above 60 years of age. In contrast, the increase in fatality rate due to economic contraction primarily affects the younger population. Table 1 below reports our estimates for the percentage increase in fatality rate for a 5% drop in per capita GDP. These estimates suggest that the economic coma is likely to cost most lives in the 0-29-year-olds, implying a greater loss in human-life years. The loss in human life due to the death of a 29-year-old translates into a decline in human-life years of about 40 years, given the life-expectancy in India is 69.2 years.
Note that even after adjusting for the above caveats, these estimates are likely to be a lower bound on the estimated loss of human lives. This is because we have not accounted for the non-fatality related loss in life years due to the reasons mentioned earlier. Nonetheless, these estimates provide a starting point to examine the trade-offs associated with lockdowns.
Our intention is not to argue in favour of or against the lockdown, but rather to put forth a framework for evaluating this decision. Ultimately, policy decisions on a lockdown and its tenure must necessarily consider the tradeoff between lives lost due to economic contraction versus lives lost to the pandemic. In brief, the optimal decision should hinge on the net savings in lives or human-life years.
Lives lost to COVID-19
While we do not focus in this article on the estimation of lives lost to the pandemic, it is important to note that such an estimation requires a nuanced approach. First, economists are increasingly pointing out that epidemiological modelers have not adequately adapted their estimates of the incidence and fatality of the disease to account for two factors. That tests used to detect cases do not capture people who were infected and recovered. Second, testing rates were very low for a long period of time in most countries and typically reserved for the very ill. As a result, confirmed cases assumed by epidemiological models may be an order of magnitude lower than the true number of infections, resulting in significant overestimation of fatalities from the outbreak.
Further, it is important to consider the abnormal or excess mortality beyond the norm for the country-season. Many of the deaths flagged as Covid related may not have been caused by Covid but rather, already prevalent conditions in the deceased. Alternatively, some deaths caused by Covid may not have been reported as such if the person died without being tested. Thus, what we should focus on is excess mortality, which would capture additional deaths above what is the norm for the season in any country.
It is also important to consider heterogeneity across countries in contagion rates, age distribution, the health ecosystem, and income levels in estimating fatality rates. For instance, the percent change in excess mortality in the UK for the period Jan-March 2020 relative to the equivalent period in 2019 is 6.6%, and relative to the average of past 5 years (2014-2019) is 1.4%. Figure 4A plots the total deaths in UK between Jan and March for the years 2019 and 2020 (Data at high frequency wasn't readily available for other countries). Figure 4B plots the excess mortality for these two time periods across different age groups.
India registered approximately 10 million deaths in 2019. The assumption of a similar increase in baseline mortality as in the UK yields a 6.6% (1.4%) increase in Covid related deaths for the country or about 670,000 (142,000) additional deaths. However, such extrapolation of estimation of fatalities in the UK to the Indian context is flawed. The median age of the population is 40.6 years in the UK in contrast to 28.7 years in India. Since death rates due to Covid are primarily amongst the older population, the related fatality rate in India should be lower than in the UK. Indeed, with 496 deaths so far, the case fatality rate in India is 3% in contrast to 13% in the UK. As noted earlier, lack of enough testing is unlikely to affect this ratio adversely as more testing should increase the denominator (the number of confirmed cases) more than the numerator (total covid deaths). The positivity rate, which captures the ratio of positive cases to the total number of tests is also significantly lower for India at 4% versus 30% for the UK. Therefore, in working with the UK's excess death rate, we are likely to significantly overestimate the excess mortality rate for India.
These estimations do not make Covid a non-issue for India or the world. Reports from the U.S. and Europe demonstrate evidence of struggles with overwhelmed health systems. These are especially salient to countries like India that have relatively poorer access to health care. The number of hospital beds per 1000 in India is 0.5 compared to 2.5 in the UK. But, given the enormous consequences of decisions around responding to Covid, including the loss in lives from economic losses, obtaining and using clear data to guide decisions is critical.
Our analysis suggests that we can't have a "one size fits all" model. The intensity of the lockdown must vary across countries, and across regions within countries as well. The economic activity needs to restart, albeit cautiously. The government does exhibit this intent with districts now categorised into green, yellow, and red zones with progressively more stringent restrictions across these zones. Through a careful data-driven policy approach, we hope that we can save both lives and the economy (oops! lives).
Views are personal.
Deepa Mani is a faculty member in the Information Systems area and Executive Director of the Srini Raju Center for IT and the Networked Economy at the Indian School of Business.
Shashwat Alok is a faculty member in the Finance area at the Indian School of Business.