Abstract
This paper tries to quantify the impact of government policy intervention on the death due to COVID-19 in India at national, regional and sub-national levels. The data used for this study are collected from the Oxford COVID-19 Government Response Tracker (OxCGRT), a longitudinal database of daily government response from Jan 28th, 2020, when the first COVID case was diagnosed in India till December 31st, 2022. Here, stringency measures, which gauge the severity of interventions such as lock-downs and travel restrictions, indicative of government control; and containment measures, representing a spectrum of actions aimed at preventing or limiting virus transmission and the overall government support, providing a holistic assessment of the government’s efforts in mitigating the virus’s spread. Using the Panel Corrected Standard Error (PCSE) method, this paper finds out that the stringency and overall government support interventions by the government have been successful in reducing the death counts by 25% and 23% respectively however the containment intervention alone has failed to reduce the death at all levels. Exploring regional variations, event study plots reveal nuanced temporal dynamics. The daily and 24-day lagged dependent variables, representing overall government response and stringency measures, reveal a consistent impact post-intervention at the all-India level. Both current and lagged variables show a reduction in COVID-19 deaths, with a more pronounced effect emerging after a four-day lag. Event-study plots with a 24-day lagged dependent variable confirm the anticipated negative impact of overall government response on deaths. However, the pattern diverges for stringency and overall government interventions compared to daily death counts.
JEL Classifications I18, I38, C23, C54
1 Introduction
The emergence of the COVID-19 pandemic raised a multitude of questions that demanded contributions from various fields of science and policy-making. Scientific and health experts played a crucial role in developing vaccines and implementing infection control measures. However, the responsibility for vaccine distribution, the implementation of restrictions, and the preservation of livelihoods ultimately fell on one entity: the government. In India, operating within a quasi-federal structure, this paper endeavors to address the effectiveness of government policies in managing the COVID-19 situation. India undertook a mixture of pharmaceutical and non-pharmaceutical interventions. We try answering how effective the non-pharmaceutical interventions (NPIs) have been in controlling deaths due to COVID-19. Amongst others at times of such endemic, I try to quantify the importance of welfare state in controlling deaths due to COVID-19. The concept of a welfare state is embedded in the Indian constitution (Article 38(1)) (see (Bakshi & Kashyap, 1982)), obligating the government to safeguard lives and improve the well-being of its citizens during times of emergencies, encompassing strategic, medical, and climatic dimensions. The gravity of the COVID-19 pandemic’s spread and impact necessitated swift and effective government interventions focused on curtailing virus transmission, caring for the affected, and formulating responsive policies to mitigate the unforeseen consequences of stringent measures.
In a nation like India, poised as one of the world’s fastest-growing economiesa and on the cusp of surpassing China in terms of population, addressing the havoc wrought by the pandemic demanded systematic and well-coordinated actions from both state and central governments. The population prediction or India (see Figure-1a) and specific composition of its demographic groups (see Figure-1b) impacts its route for mitigating the COVID-19 challenge distinctively. As of January 27, 2023, India reported over 45 million COVID cases, with 0.1% of these cases resulting in fatalities (over 0.5 million deaths), making India second only to the USA in total deaths. However, India’s initial testing efforts paint a different picture. Despite ranking second in total tests conducted globally, India was only 12th in tests conducted per million people during the same time. Furthermore, India’s vast geographical and socio-economic diversity compelled the imposition of both nationwide and region-specific policies. In this context, this paper seeks to analyze the government’s response to COVID-19 in India and its various states, with a focus on their targeted policy responses and an economic evaluation.
The spread of COVID-19 was characterized by its high infectivity, especially in countries like India with dense populations (Shereen, Khan, Kazmi, Bashir, & Siddique, 2020). The virus’s higher infectiousness, coupled with increased mortality rates across its variants (Wang et al., 2020), had a profound impact on the economic, social, and psychological aspects of life. Displayed as a weekly moving average, the graph (Figure-3) vividly portrays the persistence of infection, the evolution of new case trends (waves), and their eventual decline over time. This graph also highlights the positive effect of collective efforts such as vaccination, social distancing, and lock-downs by the government on reducing peak infection spikes and shortening the duration of waves. Nevertheless, India’s testing efforts have been relatively modest in comparison to its vast population (Thiagarajan, 2021).
It is the nature of government and governance, especially in times of pandemic precarity, that necessitates large-scale government interventions. The commitment to control the spread of the pandemic, ensure sustenance for citizens, and prepare the economy for revival forms a significant part of government planning (Moser & Yared, 2022). In the absence of profitability, such planning can only be channeled through fiscal means. The considerable socio-economic impact of the pandemic necessitates urgent actions to address short-term stability and the longer-term revival of the nation (Kumar et al., 2020). The role of the government has expanded from national to individual aspects, as the psychological health along with medical for the of the citizens has become of critical importance. The emergence of workplace and household stress, especially due to lock-downs, necessitated free professional care on demand. While dealing with survival and preparing for revival, the government has played an increasingly pivotal role. Recognizing the importance of innovation, technological adaptation, and sustainable policies to better combat the pandemic, policies supporting and promoting innovative solutions have become inevitable (Pu, Qamruzzaman, Mehta, Naqvi, & Karim, 2021). Furthermore, in the context of rejuvenating businesses, the government’s role in bailing out and financially assisting private entities has been substantial (Abate, Christidis, & Purwanto, 2020). Institutions like the IMF have inadvertently termed the challenge of COVID a war analogous wherein, in which stronger health and economic policies play a pivotal role. Importantly, it is worth noting that the government’s role is irreplaceable on several dimensions.
The literature studying the impact of the degree of government response is scarce both in general and also in the case of India. In their study Mukherjee, Banerjee, Mitra, and Mukherjee (2022), the desired needs of the hour during the pandemic. They supported the government’s stringent measures in response to COVID-19, such as work-from-home and stringent lock-downs, while also suggesting the need for timely interventions. As attributed earlier, India presents a compelling landscape for the study of non-pharmaceutical interventions (NPIs) in the context of COVID-19 due to its diverse population and geographical variations. With a population exceeding 1.3 billion people, the country’s demographic mosaic offers a unique opportunity to examine the differential impact of interventions across various cultural and regional settings. High population density, particularly in urban areas (Nijman, 2012), contributes to the rapid transmission of infectious diseases, making it imperative to understand how NPIs operate in densely populated regions. Moreover, India exhibits significant variability in healthcare infrastructure (Chaturvedi et al., 2023), economic conditions, and socio-demographic factors, influencing the implementation and effectiveness of NPIs. The nation has implemented a range of interventions, including lock-downs and social distancing measures, providing a rich data-set (OXCGRT in this case) for analyzing policy implementation and compliance. Given the prevalence of infectious diseases and varying co-morbidities, examining the interplay between pre-existing health conditions and the outcomes of COVID-19 interventions in India contributes valuable insights to the global understanding of pandemic management. India’s global significance further underscores its importance as a lucrative study area for researchers exploring the impact of NPIs on COVID-19.
In the context of assessing the impact of NPIs in India I use the confirmed number of deaths as a dependent variable for my analysis. Similarly, I use the degree of government response in terms of stringency measures, containment measures, and overall government support to reduce the infection rate in India as the independent variables. This constitutes a strongly balanced panel but with gaps, that uses data from all the states (28) and union territories (8) of India. I use daily data on the mentioned dependent and independent variables over almost two years, spanning from 28th January 2021b until 31st December 2022.
1.1 COVID-19: The Indian COVID-19 Chronicle
The first of the novel coronavirus (COVID-19) was recorded in Wuhan, China, in December 2019. Swift transmission of the virus ensued, affecting a significant number of people within a month. India’s first reported case was reported in late January 2020 in the state of Kerala, with the individual having recently returned from China. Following this, there has been a notable upswing in COVID-19 cases across various Indian states. In recognition of the gravity of the situation, the government initiated a 21-day nationwide lockdown from March 25, 2020, to April 14, 2020. This stringent measure aimed to mitigate the spread of COVID-19, resulting in the closure of industries, academic institutions, markets, and public gatherings. Subsequent to the initial lockdown, three successive lockdown phases were implemented (April 15 to May 3, 2020; May 4 to May 17, 2020; May 18 to May 31, 2020). In an effort to revive the Indian economy, two unlock phases were subsequently introduced (June 1 to June 30, 2020, and July 1 to July 31, 2020). A detailed depiction of the government response has been pictorially presented in Figure-2.
On March 29, 2020, the Government of India established 11 empowered groups to address various aspects of COVID-19 management in the country. These groups were tasked with making informed decisions on a wide range of issues, including medical emergency planning, hospital availability, isolation and quarantine facilities, disease surveillance and testing, essential medical equipment availability, human resource and capacity building, supply chain and logistics management, coordination with the private sector, economic and welfare measures, information, communications, and public awareness, technology and data management, public grievances, and strategic issues related to lockdown. On September 10, 2020, these groups underwent restructuring to adapt to the changing needs and evolving scenario.
The Ministry of Health & Family Welfare (MoFHW) unveiled containment strategies to address both cluster and widespread outbreaks on March 2nd and April 4th, 2020, respectively. These plans underwent periodic updates. The containment strategies focus on disrupting the transmission chain through (i) delineating containment and buffer zones, (ii) implementing stringent perimeter control, (iii) conducting thorough house-to-house searches for cases and contacts, (iv) isolating and testing suspected cases and high-risk contacts, (v) quarantining high-risk contacts, (vi) intensifying risk communication to enhance community awareness regarding simple preventive measures and the importance of prompt treatment seeking, and (vii) reinforcing passive surveillance for Influenza Like Illness (ILI) and Severe Acute Respiratory Illness (SARI) in containment and buffer zones.
A system of health facilities consisting of three tiers has been established to effectively handle COVID-19 cases. This included: (i) COVID Care Centers equipped with isolation beds for mild or pre-symptomatic cases; (ii) Dedicated COVID Health Centers (DCHCs) providing oxygen-supported isolation beds for moderate cases; and (iii) Dedicated COVID Hospitals (DCHs) with ICU beds for severe cases. Additionally, tertiary care hospitals affiliated with organizations such as ESIC, Defence, Railways, paramilitary forces, and the Steel Ministry have been utilized for the management of cases. Further, instructions for the clinical management of COVID-19 were released, consistently revised, and widely disseminated. These guidelines encompassed various aspects, such as defining cases, implementing infection control measures, conducting laboratory diagnoses, initiating early supportive therapy, addressing severe cases, and managing complications. Furthermore, allowances for investigation therapies, including Remdesivir, Convalescent plasma, and Tocilizumab, were outlined for the treatment of severe cases under rigorous medical supervision. It is important to note that the indices of government response, used as major policy intervention variables, represent a combination of measures for quantification.
I use the sub-national (OxCGRT) database on the degree of government responses for all the indices (independent variable) and daily confirmed deaths (dependent variable) from the Worldometer database. The data on other control variables from Indiastat.com which is a repository. To ensure better representation, I avoid using indicator values as variables and use values of the composite indices of stringency, containment, economic support, and the total degree of government response as the independent variables. It benefits immensely from the fact that these indices values are the composite estimated values of their respective categorized components (under the heads of stringency, containment, the degree of government response, and economic support) and not mere comparable figures. As the project profile outlines, ”they record the number and strictness of government policies, and should not be interpreted as ‘scoring’ the appropriateness or effectiveness of a country’s response, and a higher position in an index does not necessarily mean that a country’s response is ‘better’ than others lower on the index”.c. This limitation of the values of the respective indices makes them suitable to be used as independent variables for the paper.
The observed national-level analysis reveals a notable negative correlation between stringency and overall government response measures (a reduction of 25% and 23%, respectively) with COVID-19-related deaths. Surprisingly, containment measures exhibit an unfortunate positive association. Despite variations in magnitudes, higher stringency levels consistently show a statistically significant negative impact on mortality rates, emphasizing the effectiveness of tailored interventions. Similar trends emerge in the assessment of government support measures. However, the consistently positive impact of containment measures raises concerns about their standalone efficacy. Event studies illustrate a post-intervention reduction in deaths for both stringency and overall government response, with a delayed impact, indicating a gestation period before tangible results manifest. Regional variations in the timing of impact suggest differing duration required for interventions to yield desired outcomes.
The subsequent sections are organized as follows: In §2, I delve into the government response metrics utilized for the analysis. §3 presents summary statistics and diagnostic tests, with detailed versions available in Appendix-A.1. §4 outlines the empirical model employed for data analysis, while §5 introduces the event-study specification. Findings are discussed in §6, with a focus on sub-national (state) levels in §7 and event-study results in §8. Robustness check results are detailed in §9, followed by critical questions and limitations in §10. The study concludes in §11.
2 Government Response Metrics
This study mainly uses The Oxford COVID-19 degree of government response Tracker (OxCGRT) database. The OxCGRT database (Hale et al., 2021) is an excellent repository by the Blavatnik School of Government, University of Oxford, and several other sources. This provides a systematically collected set of cross-national, longitudinal figures of the degree of government responses from 1st January 2020. The project tracks national and, for some countries, subnational governments’ policies and interventions across a standardized series of indicators. It also furthers a group of composite indices to measure the extent of these responses. OxCGRT’s design has mainly emphasized comparability, legibility, and transparency of the available informationd. The Economic Support Index (ESI)e most uniquely records the fiscal measures such as income support and debt relief. It is calculated using all ordinal economic policy indicatorsf. For further details on the variables in use in this paper, see Table-5 in the appendices.
2.1 Measures of Government response
I use three key measures of government response to the COVID-19 pandemic: Stringency Measures (STI), Containment Measures (CNI), and Overall Government Support (GRI). These measures were selected for several compelling reasons. Firstly, these measures offer a multifaceted view of government actions and policies in response to the pandemic, providing a comprehensive assessment of the overall government response. Stringency Measures gauge the severity of government interventions, encompassing measures such as lock-downs, travel restrictions, and the closure of non-essential businesses. These measures are crucial indicators of the extent to which the government enforces control and containment measures to mitigate the spread of the virus. Containment Measures encompass a spectrum of actions taken by the government to prevent or limit the virus’s transmission. This includes testing, contact tracing, quarantine protocols, and public health campaigns. By including containment measures, I aim to capture the proactive steps taken by the government to detect, control, and manage the pandemic.
Overall Government Support represents the broader policy and financial measures put in place by the government to aid individuals, businesses, and healthcare systems during the pandemic. This includes measures such as economic stimulus packages, investments in healthcare infrastructure, and social welfare programs. The inclusion of overall government support enables a comprehensive evaluation of the government’s holistic response in terms of providing resources and support to combat the pandemic’s impacts. By incorporating these measures into my analysis, I can delve into the multifaceted nature of government responses and assess their potential impact on COVID-19 infection rates and mortality.
3 Key Statistical Highlights on Response Metrics
In this section, based on Table-6, I present summary statistics for the key variables used in my study, providing insights into the characteristics and variability of the data set. With over 1.4 billion people, India conducted around 0.9 billion tests, translating to approximately 609 tests per 1000 people. This suggests that the actual infection figures could be different from the reported numbers. Due to this limitation I use daily deaths instead of infections as the dependent variable of this analysis. The variable “Death” represents the log of cumulative deaths per million population. With 32,071 observations, the mean death rate is approximately 9.984 per million, and the standard deviation is 2.966. The data range is substantial, spanning from −4.645 to 14.286, highlighting significant variation in the pandemic’s impact across regions. The “Stringency” variable captures the Stringency Index, reflecting government intervention strictness. The mean value is 47.885, with a standard deviation of 29.742. This index ranges from 0 (minimal restrictions) to 100 (comprehensive government measures).
The “Overall Government Response” variable is based on an index that encompasses various government actions. On average, this index is approximately 53.056, with a standard deviation of 23.192. Values range from 0 to 92.19, indicating diverse government responses to the pandemic. The “Containment” variable represents the Containment Health Index, measuring efforts to control the virus’s spread. The mean value is approximately 54.417, with a standard deviation of 23.265. This index also displays a broad range, from 0 to 96.43. Additionally, my data set includes variables for average temperature, average rainfall, population density, per capita income, the number of hospital beds, the count of government hospitals, and total migrants. These variables exhibit unique characteristics, which will be explored in more detail in my analysis. In summary, these summary statistics offer an initial understanding of the data set’s distribution and the variation in each variable. The wide range of values emphasizes the need for a comprehensive analysis to uncover potential relationships and correlations between these variables and the impact of the COVID-19 pandemic.
3.1 Diagnostics
I have conducted several robust diagnostic tests to ensure the validity of my analysis. These tests include checks for heteroskedasticity, auto-correlation, and panel unit-root tests. These diagnostic checks are essential to guaranteeing the reliability of my results and making accurate interpretations. It is worth noting that the number of variables considered for my study has significantly decreased after applying these robust diagnostic tests. This reduction underscores the importance of these tests in my analysis. The diagnostic test results are available in section-A.1 in the appendices.
4 A Model Demonstrating Impact of Government Responses
In line with (Liu, Dong, & Jiang, 2023), (Yadav & Rahman, 2017) and (Shah & Garg, 2023) I frame the following model which has been used for empirical calculation. The functional specifications of the same in case of confirmed deaths are used as the dependent variable is given by,
where respective abbreviated codes stand for the interpretations of independent variables as mentioned in the variable description table - 6. Here, “i” refers to the cross-sectional unit index, which represents the different sub-national regions or states in India. Each unique value of “i” corresponds to a specific region or state. Similarly, “t” refers to the time index, which represents different dates time periods, or waves of data collection. The dependent variable Death which is cumulative death for a million population has been transformed into natural log form (ldeath) to get the elasticity values and reduce the heteroscedasticity of the dependent variables.
where STI stands for Stringency, GRI stands for Overall government response, and CNI stands for Containment and Health Index. Moreover, I also control for several factors such as Θ stands for Weather Controls, Φ represents Demographic & Economic Controls, and Λ corresponds to Health Infrastructure Controls. I estimate equation-1 in three functional forms such as: without any effect, with region-fixed effect, and with state-fixed effect.
In this context, I estimate the same equation for all three different waves of COVID-19 in India, for different waves or time periods of data collection. The first wave spans from August 15, 2020, to January 17, 2021, covering 156 days. The second wave extends from March 13, 2021, to June 19, 2021, encompassing 99 days, and the third wave ranges from January 1, 2022, to March 31, 2022, comprising 90 days. Individually, the first wave accounts for 14.5% of the observations, the second wave for 9.26%, and the third wave for 8.42%. These dummy variables capture period effects in the model. Consequently, the primary model includes more observations than the sum of these COVID-19 waves. Similarly, I estimate the influence of non-pharmaceutical interventions across regions and sub-national (state) levels as well.
4.1 Rationale behind using PCSE Method
Here I employ a Panel Corrected Standard Error (PCSE) analysis, as suggested by Bailey and Katz (2011). The PCSE has been widely employed in the literature, as evidenced by studies such as Rahman and Lamsal (2021), Cameron and Trivedi (2010), and Nguyen and Nagase (2019). This approach is particularly suitable for my analysis as it effectively addresses several issues that may arise in standard fixed or random effect models, such as auto-correlation, heteroscedasticity, and cross-sectional dependence. I use PCSE to take care of the cross-sectional dependence issue arising out of the dataset, as observed in section-A.1.4.To specify the corrected standard error using variables in a Panel Corrected Standard Error (PCSE) analysis, the formula for calculating the corrected standard error in the model would be as follows:
where: se(βi) represents the corrected standard error for the coefficient βi, T is the number of time periods in the panel data, N is the number of entities or regions in the panel data, σ^2 is the estimated residual variance, ρ^ is the estimated first-order autocorrelation coefficient, typically obtained through a diagnostic test such as the Wooldridge test, V (βi) is the estimated variance-covariance matrix of the coefficient βi obtained from the regression. To ensure the robustness of my estimates, I adopted the standard error technique developed by (Driscoll & Kraay, 1998). This technique provides consistent and reliable estimates of standard errors, further enhancing the validity of my findings and allowing for appropriate inference. The Driscoll and Kraay standard error for coefficient βi is given by:
where the variables and terms have been previously explained, and: Kst represents the element at the s-th row and t-th column of the kernel matrix K, diag(·) extracts the diagonal elements of a matrix.
5 Event-Study Regression Model for Causality Linkages
I use event-study regression models to examine how non-pharmaceutical measures of stringency, and overall government response impact the death numbers due to COVID-19 in India at both national and regional levels. While the usual causality analysis can’t be done here which (has been discussed in Section-10). The main aim behind using event study plots is two fold. Firstly, it is to see if the impact can be interpreted in causal terms and secondly to know the delay in time take for the intervention to have some desirable impact on the outcome variable death. Moreover, we only use the event plots for stringency and overall government response and exclude the containment measure as it demonstrates counter-intuitive impact on the deaths. It adds a visual narrative of the intervention’s impact over time. As PCSE estimates fall short of claiming causality, and the fact some interventions might have delayed impacts might not be interpreted from the PCSE coefficients alone.
The model is represented as follows:
In this model, αi represents the unit-fixed effect, γt is the time-fixed effect, and Q′ ψ is the vector of controls with conformable coefficients ψ. The potentially unobserved confounds are captured in Cit, and ɛit represents unobserved disturbances. The variable of interest is denoted 
, where βm reflects policy dynamic effects.
I use event study plots to represent the findings from the event study regression. Notably, Yit corresponds to the dependent variable (which is the log of cumulative deaths per million population). In this context, i denotes the region, and t denotes time. It is crucial to highlight that as I extend my results to national and regional levels, i takes the value of the region as the entire country, i.e., India, and respective regions accordingly.
Following (Gupta et al., 2021), I set the focal length of the event time window to run from 20 days before and after the event of NPI intervention. However, for better picturing, I reduced the same to 15 without losing any significant pattern of the impact of the interventions. However, in some regions (like North-eastern (see Figure-13)), I use 30-days window as we see a more delayed impact of the intervention.
6 Key Findings on The Impact of Government Interventions
The impact of non-pharmaceutical interventions from the government’s side on COVID-19 needs to be perceived through the lens of policy appropriateness. Unequivocally, policy appropriateness invariably involves the nature, meaning, and timing of the policy intervention. In this section, I discuss the impact of government intervention on COVID-19 deaths at the national level, followed by sub-national levels.
6.1 Impact on Deaths at National Level
At the national level, the impact of COVID-19 has been observed across various dimensions, spanning from socio-economic (Kumar et al., 2020) to pharmaceutical aspects (Chatterjee & Dev, 2023). However, this section is dedicated to quantitatively measuring the degree to which policy decisions at the national level and precautions implemented by the Indian government, as discussed theoretically by Mukherjee et al. (2022). The estimates of PCSE for the overall impact of government intervention on deaths due to COVID-19 are presented in table - 7g. The three key government intervention variables considered are stringency, containment measures, and overall government support. The results are presented for three different models: the pooled model, a model with the inclusion of regional fixed effects (Region FE), and a model with state-level fixed effects (State FE).
In the pooled model (Column 1), the coefficient for stringency (−0.250) is highly statistically significant. This suggests that an increase in stringency is associated with a 25% decrease in the cumulative number of deaths due to COVID-19. A similar reduction in the cumulative number of deaths is associated with the impact of overall government response i.e. 23%. However, the coefficient for containment measures (0.494) is also statistically significant, indicating that stronger containment measures are associated with a close to 50% increase in cumulative deaths. The constant term (5.687) represents the expected log number of deaths when all independent variables are zero.
The introduction of regional fixed effects (Column 2 of table-7) aims to control for regional differences. The coefficients for stringency and containment remain statistically significant, with the same direction of impact. However, the magnitudes of the coefficients change slightly, implying that regional variations play a role in these associations. Moving to the model with state-level fixed effects (Column 3 of table-7), the results become more nuanced. Stringency and containment measures remain significant, but the magnitudes of their impacts change once more, highlighting the importance of state-specific factors in shaping the relationship between interventions and cumulative deaths. Interestingly, the overall government support variable in this model becomes statistically significant and negative (−0.107), indicating that increased overall government support is associated with a decrease in cumulative deaths, controlling for state-level fixed effects. The impact of non-pharmaceutical interventions across different waves are presented in Figure-6.
6.2 Impact on Deaths at Regional Levels
In this section I shall discuss the effects of comparable non-pharmaceutical interventions (NPIs) at regional levels. my analysis, which relies on regional classifications, is designed to highlight the efficacy of these interventions within administrative regions. This approach is essential for deriving important policy insights. I have summarized my findings in the Table 8 and Figure-5. In all regions (Southern, Western, Eastern, Central, Northern, and North-Eastern), increased stringency measures have a statistically significant negative impact on cumulative deaths. For example, in the Southern Region, a one-unit increase in stringency is associated with a substantial decrease of approximately 28.4% in cumulative deaths. Similarly, in the Western Region, a one-unit increase in stringency leads to a significant reduction of about 25.0% in cumulative deaths. This negative relationship is consistent across all regions, with varying magnitudes of impact. The overall government support variable also shows a significant negative relationship with cumulative deaths in all regions. For instance, in the Eastern Region, a one-unit increase in overall government support results in a noteworthy decrease of approximately 9.69% in cumulative deaths. In the Western Region, this decrease is about 18.0%. These findings suggest that greater government support is associated with a reduction in COVID-19-related deaths in all regions, but the magnitude of this impact varies.
Containment measures, on the other hand, exhibit a significant positive impact on cumulative deaths across all regions. In the Central Region, a one-unit increase in containment measures is associated with a considerable increase of approximately 29.4% in cumulative deaths. The Northern Region demonstrates a similar pattern, with a one-unit increase leading to an increase of approximately 44.6% in cumulative deaths. This positive relationship between containment measures and cumulative deaths holds consistently across all regions, with varying degrees of impact. In the upcoming section, I shift the focus to specific states within each region, enabling a more detailed examination of the impact of non-pharmaceutical interventions at a localized level.
7 Impact on Deaths at State Levels
7.1 Southern States
In my examination of the Southern region states of India (presented in the Table-9), the state-level analysis provides a granular view of the impact of non-pharmaceutical interventions on the cumulative number of COVID-19 deaths. The dependent variable remains the log of cumulative deaths, allowing us to gauge the percentage changes with precision. Stringency measures exhibit varied impacts across the states in the Southern region. Notably, in Andhra Pradesh (Column 2), a one-unit increase in stringency is associated with a substantial decrease of approximately 24.9% in cumulative deaths, whereas Tamil Nadu (Column 6) experiences an even more pronounced reduction of around 32.5%. Conversely, Telangana (Column 7) demonstrates a different pattern, where the reduction in cumulative deaths is more substantial, at approximately 55.3%. These regional variations underscore the importance of tailoring interventions to the unique dynamics of each state.
Examining overall government support, I find that its impact is also state-specific. In Kerala (Column 4), a one-unit increase in overall government support results in a significant reduction of approximately 24.7% in cumulative deaths. In contrast, Telangana (Column 7) experiences a less pronounced reduction, at approximately 12.5%. These findings emphasize that the effectiveness of government support measures can differ widely from state to state. Containment measures exhibit a significant positive impact on cumulative deaths across states. For instance, in Tamil Nadu (Column 6), a one-unit increase in containment measures is associated with a notable increase of approximately 54.2% in cumulative deaths, while in Telangana (Column 7), this increase is even more pronounced, at approximately 83.1%. These results highlight the critical role of containment measures but also underscore the need for nuanced approaches in their implementation.
7.2 Western and Eastern Region States
In this state-level analysis, I focus on the Western and Eastern regions of India (presented in the Table-10), delving into the impact of non-pharmaceutical interventions on the cumulative number of COVID-19 deaths. Across these states, the stringency of interventions displays significant and region-specific effects on cumulative deaths. In Goa (Column 1), a one-unit increase in stringency is associated with a reduction of approximately 9.73% in cumulative deaths. Gujarat (Column 2) exhibits a more substantial reduction of around 24.9%, while Maharashtra (Column 3) follows suit with a decrease of about 14.7%. In contrast, Bihar (Column 4) and Jharkhand (Column 5) witness a milder impact, with decreases of approximately 12.6% and 6.80%, respectively. Odisha (Column 6) experiences a substantial decrease of approximately 32.5%, and West Bengal (Column 7) shows a striking reduction of about 55.3%. These diverse impacts underscore the importance of tailoring interventions to the unique dynamics of each state.
The influence of overall government support on cumulative deaths also varies across states. In Goa, a one-unit increase results in a reduction of around 11.7% (Column 1), while Gujarat sees a more pronounced reduction of about 15.4% (Column 2). Maharashtra (Column 3) experiences a decrease of approximately 16.5%, and Bihar (Column 4) faces a substantial reduction of about 24.7%. In Jharkhand (Column 5), the decrease is approximately 9.33%, whereas Odisha (Column 6) records a decrease of about 18.3%. West Bengal (Column 7) demonstrates a reduction of approximately 12.5%. These findings highlight the variability in the impact of government support measures across states. Containment measures exhibit varying effects on cumulative deaths within the Western and Eastern regions. While Goa (Column 1) experiences a significant increase of approximately 19.2% for a one-unit rise in containment measures, Gujarat (Column 2) records a more substantial increase of about 36.0%. Maharashtra (Column 3) observes an increase of around 28.0%, and Bihar (Column 4) sees a notable increase of approximately 30.2%. In contrast, Jharkhand (Column 5) experiences a slight decrease, but this effect is not statistically significant. Odisha (Column 6) showcases a considerable increase of about 54.2%, and West Bengal (Column 7) demonstrates a substantial rise of around 83.1%. These diverse impacts underline the necessity of considering containment strategies tailored to the unique circumstances in each state.
7.3 Northern Region States
The state-level analysis in the Northern region (presented in the Table-11) provides a detailed look at the impact of non-pharmaceutical interventions on the cumulative number of COVID-19 deaths. In this region, I consider several states, each demonstrating unique characteristics in response to these interventions. When examining the effect of stringency measures, it’s evident that the impact varies among states. In Chandigarh (Column 1), a one-unit increase in stringency is associated with a significant reduction of approximately 11.7% in cumulative deaths, emphasizing the effectiveness of such measures in this state. Delhi (Column 2) displays a similar trend, with a notable decrease of around 19.0% in cumulative deaths. In contrast, Himachal Pradesh (Column 4) shows a smaller effect, with a decrease of approximately 3.0%. The variations across states highlight the importance of tailoring interventions to specific regional contexts. The analysis of overall government support reveals further diversity. While Haryana (Column 3) displays a significant reduction of approximately 14.6% in cumulative deaths for a one-unit increase in support, Himachal Pradesh (Column 4) experiences a more pronounced decrease of around 44.4%, underlining the crucial role of government assistance in mitigating the impact of the pandemic in certain states. Containment measures also exhibit regional differences. For instance, Jharkhand (Column 5) demonstrates a noteworthy increase of approximately 30.7% in cumulative deaths for a one-unit increase in containment. On the other hand, Rajasthan (Column 7) experiences a less pronounced impact, with a 19.4% increase. These variations underscore the need for state-specific strategies to effectively manage the pandemic.
7.4 North-Eastern Region States
Table-12 presents a comprehensive analysis of non-pharmaceutical interventions’ impact on the cumulative number of COVID-19 deaths in the various states of the North-Eastern region of India. The dependent variable in these models is the log of cumulative deaths, which allows for a more precise assessment of percentage changes. Examining the impact of stringency measures in each state, I find significant variation. In Assam (Column 1), a one-unit increase in stringency results in a substantial decrease of approximately 35.4% in cumulative deaths, underscoring the effectiveness of stringent measures in this state. Arunachal Pradesh (Column 2) also sees a significant reduction of approximately 6.45% in cumulative deaths due to increased strin-gency. However, Mizoram (Column 5) exhibits a contrasting pattern with a positive coefficient, suggesting that stringency measures may not have had a significant impact. Understanding these variations is vital for tailoring intervention strategies to the unique circumstances in each state. State-level analysis reveals that the impact of overall government support varies. In Assam (Column 1), an increase in government support is associated with a significant increase of approximately 8.47% in cumulative deaths, suggesting a need for a more targeted approach. Arunachal Pradesh (Column 2), on the other hand, experiences a decrease of approximately 12.0% in cumulative deaths in response to increased government support. These findings emphasize the need to adapt government support policies to the specific conditions of each state. Containment measures exhibit a significant positive impact on cumulative deaths in several states. In Assam (Column 1), a one-unit increase in containment measures leads to a substantial increase of approximately 29.5% in cumulative deaths. Arunachal Pradesh (Column 2) also observes a similar pattern with an increase of approximately 16.3% in cumulative deaths. Notably, Mizoram (Column 5) experiences a decrease in cumulative deaths, indicating that containment measures may have been effective in this state.
7.5 Central Region States
In the Central region of India, I explore (presented in the Table-13) the impact of non-pharmaceutical interventions on the cumulative number of COVID-19 deaths across several states. The dependent variable remains the log of cumulative deaths, allowing us to assess the percentage changes in a nuanced manner. The stringency of interventions exhibits variable impacts in the Central region. In Chhattisgarh (Column 1), a one-unit increase in stringency is associated with a statistically significant decrease of approximately 8.02% in cumulative deaths. Meanwhile, in Madhya Pradesh (Column 2), the effect is even more pronounced, with a decrease of approximately 22.7%. Conversely, in Uttarakhand (Column 3) and Uttar Pradesh (Column 4), the reduction in cumulative deaths is around 27.6% and 24.0%, respectively. These variations underscore the importance of tailoring intervention strategies to the specific context of each state within the Central region.
Across the states in the Central region, overall government support also exhibits variable impacts. In Chhattisgarh, it results in a statistically significant reduction of approximately 15.1% in cumulative deaths. Madhya Pradesh shows a similar trend, with a decrease of approximately 10.7% (Column 2). However, in Uttarakhand (Column 3), the impact is even more substantial, with a reduction of approximately 22.5%. Notably, in Uttar Pradesh (Column 4), the impact of overall government support appears to be negligible, with a coefficient of −0.0538, indicating a limited effect. Containment measures demonstrate a consistent and statistically significant positive impact on cumulative deaths across all states in the Central region. In Chhattisgarh, a one-unit increase is associated with an increase of approximately 20.3% in cumulative deaths (Column 1). Madhya Pradesh exhibits a similar pattern, with a notable increase of around 36.2% (Column 2). Uttarakhand shows the most pronounced effect, with containment measures leading to an increase of approximately 49.8% in cumulative deaths (Column 3). In Uttar Pradesh, the impact is also substantial, with a 30.4% increase in cumulative deaths (Column 4).
8 Results from Event Studies
In Figure 7, I present the daily and 24-day lagged dependent variables as columns across measures of the overall government response and stringency on the rows. At the all-India level, both the lagged and current dependent variables exhibit a similar pattern of impact post-intervention. It is evident that all the interventions lead to a reduction in deaths due to COVID-19 post-intervention, although the impact becomes more apparent after a lag of four days. The event-study plots with a 24-day lagged dependent variable (death) confirm the expected negative impact of the overall government response on deaths due to COVID-19 in India. However, for both stringency and overall government interventions, the pattern (similar across both interventions) depicts a story different from that of the daily death counts. It suggests that post-intervention, the impact gestation period is over a week (7-8 days), and before effectively reducing death, the immediate increase in deaths post-interventions seeks further explanations.
I further investigate the impact of interventions of stringency and overall government response using event study plots across different regions of the nation. The impact varies across regions and duration (in days) interventions require to have a desirable impact. Given the fact that desirable impact is a reduction in deaths, the central region has a 2-day gap in showing the impact of stringency restrictions on reducing daily deaths and deaths with a 24-day lag, with the same number of days taken for the intervention to work, see Figure 8. A one-week delayed impact of the overall government response has also been observed in daily and 24-day lagged death numbers. As shown in Figure 9, in the eastern region states, the impact of stringency interventions in reducing death numbers due to COVID-19 is quicker (2-3 days) compared to the overall government response (6-7 days).
The event-study plot in Figure 10 of the Southern region tells a different story with daily death terms having a 4-day delay period to show the desirable impact. While stringency seems to have a similar delay period (5 days) when considering the 24-day lagged death variable, overall government response displays a 7-day delay in revealing the impact on deaths. Figure 11 shows a longer delay in death impact when considering the 24-day lagged death as the dependent variable. It demonstrates a 9-10 day delay in demonstrating a significant negative impact on death numbers reduction in this case, as opposed to daily death counts, which takes a 3-day delay. This holds true for both the interventions of stringency and overall government response.
In the northern region states, the display paints an appealing picture in terms of delays in the reduction of daily deaths, as the term is 3 days, shorter compared to most other regions. Yet, the 24-day daily lagged death shows interesting inputs. In overall government interventions cases, the lag shown in Figure 12 presents a 7-day lag in the reduction of deaths. However, in the stringency implementation segment, though the post-intervention impact has only reduced the COVID-19 related deaths, the coefficient has become negative only after a 12-day delay. The north-eastern region, with an 18-day delay (highest across regions), has been perceived from Figure 13 for the intervention to have a desirable negative impact on COVID-related deaths through stringency intervention. Unlike stringency, the overall impact of the government responses, though seen in the reduction in the increasing number of deaths, a significant decrease in deaths post-intervention can only be observed after 21 days. However, with the 24-day lagged death being the dependent variable, both interventions demonstrate a desirable impact post 16-17 days of the intervention.
9 Robustness Checks
The robustness check tables present results from a regression analysis investigating the impact of COVID-19 containment measures, represented by stringency, on death outcomes during different waves in India. The tables are organized to show the effects withh and without lags for each wave, incorporating region and state fixed effects. In the first table-14, the pooled model and the first wave are examined, while the second table-15 focuses on the second and third waves. The coefficients indicate the estimated effects, and negative/positive values suggest negative/positive associations, respectively. The tables present the fact that estimates are consistent in terms statistical significance and direction of association among all the interventions in the model.
10 Critical Questions and Limitations
In this section, we delve into potential queries arising from the estimates and claims presented in this paper. Some of these questions may serve as avenues for future research, while others highlight limitations inherent in the current study.
Problem of Causality
The fundamental constraint constraining the scope of this work is its inability to assert causal implications instead of mere statistical associations. Although the assertions presented in this paper are robust and merit attention, establishing causality empirically remains elusive. Employing causal analysis tools such as Different-in-Difference and SIRD modelingi for SIRD model applications, and Chiesa, Antony, Wismar, and Rechel (2021) for a review, is challenging. This challenge arises because a natural control group is unattainable due to simultaneous interventions implemented across all states in India. Furthermore, the synthetic control approachj is impractical due to significant diversity in socio-economic and geographical terms within India. Constructing a statistically viable donor group becomes challenging without making very strong assumptions, rendering the synthetic control approach less feasible. Consequently, the primary limitation of this study is its ability only to establish association rather than meaningful causality.
Regional Disparities in Intervention Impact
The analysis prompts an exploration of the varied impact of interventions across different states. Regionally, as illustrated in Table-8, the Southern states exhibit notable effectiveness in terms of both stringency and overall government support. This trend is further emphasized in the state-level analysis presented in Table-9, where the interventions demonstrate significant impact. The reasons for such disparities are multifaceted. For instance, Kerala’s success is often attributed to the acclaimed “Kerala Model” of management and a robust healthcare system (Chathukulam & Tharamangalam, 2021). Comparable claims regarding performance differentials among states in the context of COVID-19 interventions extend beyond the scope of this paper and warrant further research.
Temporal Dynamics in Intervention Impact
The temporal dimensions of non-pharmaceutical interventions are pivotal factors influencing their efficacy during the COVID-19 pandemic. While the duration (Coccia, 2021) and timing (Silverio et al., 2020) of these interventions are crucial, the debate surrounding the time taken (delay) from intervention implementation to a noticeable reduction in deaths persists. Our findings reveal 3-4 day intervals in daily death reduction following both stringency and overall government response interventions, whereas a one-week gap in 24-day lagged death numbers is observed to have an impact (Figure-7). Contrastingly, early-stage studies on COVID-19 indicate a 15-day delay (see Meo et al. (2020)). A discernible trend from the event study graphs is the consistent, longer gap in 24-day lagged death numbers compared to daily death numbers.
Conclusion
In examining the impact of interventions at national, sub-national and regional levels, this analysis has highlighted distinct differences in effectiveness. This exploration of temporal dynamics in event studies sheds light on the critical factors influencing the efficacy of non-pharmaceutical interventions. The debate over the delay in observing impacts following intervention implementation remains complex, with the findings suggesting varying time frames for daily death reduction and lagged death numbers. The nuances revealed in our analyses underscore the multifaceted nature of intervention impact dynamics during the course of the pandemic. In conclusion, this study contributes valuable insights into the nuanced effectiveness of interventions, emphasizing regional variations and temporal intricacies. The observed disparities underscore the need for tailored strategies and continuous research to refine our understanding and response to the ongoing challenges posed by the COVID-19 pandemic.
Data Availability
All data produced are available online at OXCGRT Database
https://www.bsg.ox.ac.uk/research/covid-19-government-response-tracker
A Tables
A.1 Diagnostic Checks
A.1.1 Heteroskedasticity and auto-correlation tests
It was needed on the part of the data to be tested against heteroskedasticity and autocorrelation hypotheses. The results of the same have been replicated in the table - 1 below. In the first case, where Confirmed cases are the dependent variable, values of the Modified Wald test for the group-wise heteroscedasticity test statistick, and the Wooldridge test for autocorrelation (Papke & Wooldridge, 2008) in panel data are 2448.58 and 95.822, which are significant at 1% level of significance. Similarly, in the second case, where Confirmed deaths are the dependent variable, values of the Modified Wald test for the group-wise heteroscedasticity test statistic, and the Wooldridge test for autocorrelation (Papke & Wooldridge, 2008) in panel data are 23754.89 and 4.801, which are significant at 1% and 5% level of significances. Thus, the results in both models confirmed the presence of autocorrelation and heteroscedasticity.
A.1.2 Panel Cointegration test
To see their relationship and interdependence, I conduct the Kao test (Kao, 1999) for cointegration, the results of which are presented in the table-2 which is celebrated in this area of work.
The p-values of the different t-test statistics are very highly (1%) significant. This indicates that the null hypothesis has to be rejected to ensure a cointegrating relation in the panel.
A.1.3 Panel Unit-root test
The panel unit root tests are very crucial to be conducted beforehand as the data set does involve an element of time. These estimates presented in the table-3 are based on the Fisher type unit root test (Choi, 2001) and Im-Pesaran-Shin unit root test, as other tests need a panel to be strongly balanced (Banerjee, 1999). Here the dependent variables were found to be stationary at level. However, the presented table also has their log-transformed first difference term which has been to capture the change in confirmed cases and deaths.
A.1.4 Cross-sectional dependence test
To find out the cross-sectional dependence across the panel, the (Pesaran, 2021) test has been employed, results of which are presented in the table-4. The estimated statistics of the same have been presented in the table below. Under the null hypothesis that there is cross-sectional independence, the results indicate a rejection of the null hypothesis with a 1% level of significance. This dependence across cross-sectional units of the panel indicates the unobserved common factor impact in the model.
A.2 OxCGRT indices and variables of the study
A.3 Variable Description
A.4 Table of impact on Deaths at National Levels
A.5 Table of impact on Deaths at Regional Levels
A.6 Table for impact on Deaths at State Levels
A.7 Robustness Check Tables
B Figures
B.1 India Population and Demographic Composition Figures
These figures sourced in the country profile of World Population Prospects: India demonstrate the total population of the country is on a rising trend and simultaneously it has the highest number of people in 25-64 years age group. Though the figure also depicts a declining trend in aforementioned metrics of population at present the existing demographics are vital for out analysis of non-pharmaceutical interventions in COVID-19 related deaths.
B.2 A Framework of Government Response to COVID-19 in India
This figure depicts the different work-action measures taken as a government response since the first COVID case in India in January 2020. It shows, along with a phased implemented (and withdrawn) lockdown, there were containment strategies in place. For better functioning, 11 empowered groups were formed and further restructured to implement pharmaceutical and non-pharmaceutical guidelines.
B.3 Weekly moving average of cases and deaths due to COVID-19 in India
The Figure presents the weekly moving average of cases and deaths due to COVID-19 in India. The confirmed cases numbers look more volatile than the number of deaths which can be attributed to reporting discrepancy and other related factors.
B.4 Kernel-density plots of confirmed cases and confirmed deaths of COVID-19 in India
B.5 Coefficients of interventions at regional levels
Coefficient plot of PCSE regression coefficients of NPIs against COVID-19 in India
B.6 Coefficient plot of PCSE regression coefficients of different NPIs against COVID-19 in India
Graphical representation illustrating the regression coefficients of Non-Pharmaceutical Interventions (NPIs) in the context of the COVID-19 data in India across the three different waves and the pooled model.
B.7 Event study plot(s): National Level
National level event study plots of stringency and overall government response interventions on COVID-19 in India.
B.7.1 Event study plot: Central region states
Event study plots of stringency and overall government response interventions on COVID-19 in the Central region states of India
B.7.2 Event study plot: Eastern region states
Event study plots of stringency and overall government response interventions on COVID-19 in the Eastern region states of India
B.7.3 Event study plot: Southern region states
Event study plots of stringency and overall government response interventions on COVID-19 in the Southern Region States in India
B.7.4 Event study plot: Western region states
Event study plots of stringency and overall government response interventions on COVID-19 in Western Region States in India
B.7.5 Event study plot: Northern region states
Event study plots of stringency and overall government response interventions on COVID-19 in the Northern Regions of India
B.7.6 Event study plot: North-Eastern region states
Event study plots of stringency and overall government response interventions on COVID-19 in the North-Eastern Regions of India
Footnotes
↵† I express my sincere gratitude to the anonymous referees whose invaluable feedback significantly enhanced the quality of earlier versions of this manuscript. Special appreciation goes to Niyaz Mohammed, Shiv Nadar University, India, Alok Aditya, CESP, ISEC, and Niranjan Padhan, CESP, ISEC for their insightful contributions. I extend my thanks to Lipsa Mishra, B.J.B. Junior College, Bhubaneswar, Odisha, and Madhubrota Chatterjee, PRC, ISEC, whose thoughtful insights greatly influenced the development of earlier drafts of this paper. I also appreciate the thoughtful comments from participants in the conference on “Indian Economy in the Aftermath of COVID-19 “ held at B.J.B. College, Bhubaneswar.”
↵a With a projected real GDP growth rate of 6.3%, compared to China (5%), USA (2.1%), U.K. (0.5%). For more details, see World Economic Outlook (October 2023).
↵b The first COVID-19 case was confirmed in India; (See (Andrews et al., 2020)) for more).
↵c For further information, see the webpage of OxCGRT
↵d A detailed display of the indicators and indices has been presented in Appendix-5.
↵e As I use sub-national level data here, I will not be using values of ESI as they are available only at the national level.
↵f For further details on the methodology of these indices which are calculated as a function of the individual indicators, see Methodology for calculating indices
↵g Furthermore, it’s worth noting that the models are adjusted for a range of crucial factors, including weather conditions, demographics, economic variables, and health infrastructure quality. The detailed table coefficients are available as supplementary material to the paper.
↵h I took 24-days lag of the pooled model, 19-days lag of the first-wave, 21-days lag of the second wave and 22-days lag of the third wave model.
↵i Refer to Calafiore, Novara, and Possieri (2020), te Vrugt, Bickmann, and Wittkowski (2021)
↵j As utilized in Alfano, Ercolano, and Cicatiello (2021), Mader and Rüttenauer (2022), Yang (2021), and Group (2023) for an extensive review of methods in quantifying the impact of COVID-19.
↵k XTTEST3: Stata module to compute Modified Wald statistic for group-wise heteroskedasticity, Christopher Baum (2001), Boston College Department of Economics
References
