Volume 21 - Issue 3

Research Article Biomedical Science and Research Biomedical Science and Research CC by Creative Commons, CC-BY

Did The Implementation of Government Mandates Reduce COVID-19-Related Mortality Rates?

*Corresponding author:Tilahun Emiru, Assistant Professor of Economics, Lake Forest College, USA.

Received: November 25, 2023; Published: January 30, 2024

DOI: 10.34297/AJBSR.2024.21.002835


The global response to the COVID-19 pandemic witnessed an array of public health interventions aimed at curtailing the virus’s spread and saving lives. Notable among these were mask mandates and vaccine requirements. This study utilizes quarterly statelevel panel data from the United States spanning 2020-2021 to estimate the effect of vaccine and mask mandates on COVID-19- related mortality rates.


The global response to the COVID-19 pandemic witnessed an array of public health interventions aimed at curtailing the virus’s spread and saving lives. Notable among these were mask mandates and vaccine requirements, which sparked debates regarding their efficacy and potential political motivations. As the discourse on these interventions intensified, researchers explored the nuanced effects, questioning whether decisions were driven by evidence driven research findings or political considerations. This study seeks to contribute to the ongoing dialogue by examining the impact of mask and vaccine mandates on COVID-19 mortality rates in the United States.

Previous research has explored the effectiveness of public health interventions in combating COVID-19. Meta-analyses by Chu, et al. (2020) and Liang, et al. (2020) provided insights into the positive effects of physical distancing, face masks, and eye protection [1,2]. Asadi, et al. (2020) investigated mask efficacy in reducing particle emission, while Howard, et al. (2021) also supported the reduction of respiratory particle transmission through mask-wearing [3,4]. Guerra & Guerra (2021) and Motallebi, et al. (2022) examined the impact of mask mandates at different levels, yielding varying results. Ahmad, et al. (2021) and other studies highlighted the overall consensus on the effectiveness of using masks. In terms of immunities and death rates, Gazit, et al. (2021) compared natural and vaccine-induced immunity, emphasizing the robustness of natural immunity [5-8]. Acton, et al. (2022) demonstrated the effectiveness of vaccine mandates at the college level in reducing COVID-19 cases and deaths [9]. The existing literature underscores the multifaceted nature of interventions and their impact on virus transmission and mortality.

This study utilizes quarterly data from all fifty states of the United States and the District of Columbia over 2020-2021. Incorporating vaccine and mask mandates, along with demographic, social, and clinical covariates in a multivariate regression analysis, the research assesses the association between vaccine and mask mandates and COVID-19-related death rates.

We find a negative and statistically significant relationship between vaccine mandates and mortality rates. Periods when vaccine mandates have been implemented are associated with a 2.2% decrease in COVID-19-related deaths. Contrarily, the association between mask mandates and COVID-19-related deaths is statistically significant but unexpectedly positive. This is potentially because mask mandates are implemented during periods of high COVID-19 cases giving rise to a positive correlation between mask mandates and death rates. Additionally, the study reveals a significant positive association between population sizes and COVID-19-related deaths, emphasizing the effect of population density on virus transmission. COVID-19-related deaths also exhibit an expected negative relationship with state-level Gross Domestic Product (GDP) – more affluent states are equipped with the resources to minimize the health effect of diseases.

The study has several limitations. As it is based on observational data, potential endogeneity issues may arise, stressing the need to interpret the results as correlations, not causations. Moreover, aggregating data at the quarterly level gets rid of important insights that are visible with higher frequency data.

The remainder of the study is structured as follows. Section 2 offers a review of the related literature. Section 3 discusses the data, descriptive statistics, and the econometric model. Section 4 presents and discusses the findings, while Section 5 concludes.

Data and Econometric Model

Data and Summary Statistics

The dataset is structured as a panel dataset with 408 observations for all fifty states and the District of Columbia, for all four quarters of 2020 and 2021. Each quarter is interpreted as a typical calendar year (Q1/Q2/Q3/Q4). Our data sources are presented in Table 1.

Biomedical Science &, Research

Table 1: Data Sources.

Note*: Definition of the study variables.

Definition Of the Study Variables

COVID-19-Related Data: COVID-19-related information was gathered from various sources. The vaccine mandate data was obtained from the Kaiser Family Foundation (KFF), which categorized each state based on the presence or absence of a vaccine mandate, along with the specific timeframe during which the mandate was active. This data was utilized to assess the effectiveness of vaccine mandates in comparison to states without such mandates. Additionally, the mask requirement data was extracted from Wikipedia, where each state was segmented by the start and end dates of the mask mandate. These dates were substantiated with references to external sources, including articles and government websites. Both the vaccine mandate and mask requirement variables are presented as binary variables, where (=1) signifies the presence of the mandate during that quarter, and (=0) indicates the absence of the mandate during that period.

Macroeconomic Data: Macroeconomic data was sourced from the U.S. Census Bureau, St. Louis FRED, and the Centers for Medicare & Medicaid Services (CMS). Quarterly GDP figures were extracted for each state from the St. Louis FRED database, providing insights into economic activity and to see if individuals were in close contact with one another. Total unemployment serves as another indicator of financial and economic health, with an uptick suggesting a decline in economic wellbeing during the specified period.

Medicare eligibility data, acquired from CMS, was utilized to establish a new metric termed ‘atrisk population.’ This metric was obtained by dividing the number of Medicare eligible individuals in a state by the total population of the state, offering a proportion that represents elderly individuals. This metric aids in assessing the impact of additional elderly individuals, who have a heightened mortality risk compared to under-sixty-five individuals, on COVID-19 deaths.

The population variable, sourced from the U.S. Census Bureau and aggregated to a quarter, contributes to the analysis of population density within each state. Additionally, it also played a role in the creation of the ‘at-risk population’ variable, helping gauge the proportion of elderly individuals relative to the total population in each state.

Mortality Data: Mortality data was gathered from the Centers for Disease Control and Prevention (CDC), presented on a monthly basis for each state spanning from 2020 to 2022. For the analysis, we aggregated the data for each state to a quarter by calculating the average over three-month intervals. The deaths were further categorized into those attributed to COVID-19 and those associated with all causes. Leveraging these variables, we computed the proportion of deaths specifically attributed to COVID-19 relative to the total number of deaths.

Descriptive Statistics

The average COVID-related death rate over the eight quarters in the sample for all fifty states and the District of Columbia is 693 people, whereas the average total death rate is 5,633 people. Both death rate measures exhibit significant standard deviations, indicating substantial cross-sectional and time variations across state-years. Notably, California in the first quarter of 2021 (with 10,036 deaths) and Utah in the first quarter of 2020 (0 deaths) had the highest and lowest COVID deaths respectively, while California in the first quarter of 2021 (35,003 deaths) and Alaska in the second quarter of 2020 (371 deaths) had the highest and lowest total deaths over the sample period. Approximately 10.6% of total deaths over the entire sample are attributed to COVID-19.

The average unemployment rate, standing at 6%, reflects the challenges the labor market faced in the recovery from the pandemic. Throughout the sample, 12% of state-years had a vaccine mandate, while approximately 43% had a mask mandate. The average state-quarter GDP for the sample period was $427,569 billion, with California boasting the largest economy in the last quarter of 2021 (over $3.5 trillion) and Vermont’s economy in the second quarter of 2020 ($30.6 billion) being the smallest. California also holds the highest population in first quarter of 2020 (about 39.5 million people), while Wyoming in the same quarter has the lowest population (577 thousand people).

An average of 1.2 million people were covered by Medicare, with California having the largest number benefiting from Medicare coverage, and the District of Columbia having the smallest. Our measure of ‘at-risk population,’ defined as the ratio of people eligible for Medicare coverage to the total state population, has an average value of 19.3%, meaning roughly 1 out of 5 individuals had an elevated mortality risk during the sample period. Maine tops the list in the last quarter of 2021, while Utah was at the bottom in the first quarter of 2020 (Table 2).

Econometric Model and Method of Estimation

The econometric model adopted in this study is a multiple linear regression model designed to explain COVID-19-related

deaths. Given that we are utilizing state-level quarterly panel data from 2020 and 2021, the formulation of the econometric model is as follows:


Where the subscript 𝑖  denotes states and 𝑑  denotes quarters spanning from the first quarter of 2020 to the last quarter of 2021. The term 𝛼𝑖  captures time-invariant fixed effects for each state, while 𝑒𝑖𝑑  represents the error term.

The dependent variable in our model (π‘Œπ‘–π‘‘) is the ratio of COVID- 19-related deaths to total deaths in each state every quarter. The explanatory variables (the π‘₯𝑖𝑑’s) incorporated into our econometric model aim to encompass macroeconomic, demographic, and health factors for each state. These variables include the natural logarithm of the state’s quarterly GDP, the unemployment rate, and the natural logarithm of the state’s population. Additionally, our regression equation includes a measure of the at-risk population.

Biomedical Science &, Research

Table 2: Summary Statistics.

Note*: Source: Own computations, data from Census Bureau, St. Louis FRED, and the Centers for Medicare & Medicaid Services (CMS), Centers for Disease Control and Prevention (CDC).

Therefore, the regression equation we estimate is given by

π·π‘’π‘Žπ‘‘β„Žπ‘…π‘Žπ‘‘π‘’π‘ π‘–π‘‘= 𝛽0 + 𝛽1π‘‰π‘Žπ‘π‘π‘–π‘›π‘’π‘€π‘Žπ‘›π‘‘π‘Žπ‘‘π‘’π‘ π‘–π‘‘+ 𝛽2π‘€π‘Žπ‘ π‘˜π‘€π‘Žπ‘›π‘‘π‘Žπ‘‘π‘’π‘ π‘–π‘‘

+ 𝛽3π‘ˆπ‘›π‘’π‘šπ‘π‘™π‘œπ‘¦π‘šπ‘’π‘›π‘‘π‘…π‘Žπ‘‘π‘’π‘–π‘‘+ 𝛽4log (𝐺𝐷𝑃)𝑖𝑑+ 𝛽5log (π‘ƒπ‘œπ‘π‘’π‘™π‘Žπ‘‘π‘–π‘œπ‘›) 𝑖𝑑+ 𝛽6π΄π‘‘π‘…π‘–π‘ π‘˜π‘ƒπ‘œπ‘π‘’π‘™π‘Žπ‘‘π‘–π‘œπ‘›π‘–π‘‘+ 𝛼𝑖+ 𝑒𝑖𝑑 

𝛼𝑖 and 𝑒𝑖𝑑 are the fixed effects and the error term, respectively, as before.

We estimate our regression using both pooled Ordinary Least Squares (OLS) and fixed-effects panel data techniques. Pooled OLS treats the 408 observations as distinct entities, disregarding the panel structure of the data. Given that each state and the District of Columbia in our sample have 8 observations (four quarters for 2020 and 2021 each), the pooled OLS estimate results in biased coefficient estimates due to state-specific variables that are related to the geographic location of the state, the state’s laws, and so on. etc. Therefore, we present it for comparison purposes only.

Our primary inferences are drawn from the fixed-effects regression. This technique acknowledges the panel structure of the data and eliminates individual fixed effects through data transformations. Consequently, state-specific factors do not introduce bias to the fixed-effects results.

Results and Discussion

We present the regression results in Table 3 below. As mentioned previously, our primary interpretations and inferences will rely on the fixed effects regression results. However, it is noteworthy that several variables that are estimated to be insignificant with the pooled OLS technique attain significance when accounting for the panel structure of the data. Additionally, the R-squared from the fixed effects model is 28.5%, markedly higher than that of the pooled OLS. This suggests that the covariates included in the regression equation collectively account for 28.5% of the variation in COVID-19 related deaths as a ratio of total deaths with the fixed effects regression (Table 3).

Biomedical Science &, Research

Table 3: Regression Results.

Note*: Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

The estimated coefficient for vaccine mandates is indeed negative and statistically significant at the 10% level. Holding all else constant, this coefficient implies that states and periods with implemented vaccine mandates are associated with a significant 2.2% decrease in COVID-19related deaths relative to total deaths, compared to states and periods without such mandates.

While the negative sign aligns with our expectations, significance only at the 10% level and seemingly modest magnitude warrant further elaboration. The impact of vaccine mandates extends beyond individual immunization; it contributes significantly to herd immunity, amplifying the overall effectiveness of the vaccine in curbing the spread of viruses. Thus, the observed modest magnitude and lower significance may be attributed in part to the constraints imposed by our small sample.

Furthermore, vaccination rates tend to increase over time, while our dummy variable remains fixed at 1 from the first quarter of its implementation. This static nature may not fully capture the evolving impact of rising vaccination rates. Incorporating the actual number of vaccinated individuals into the regression analysis could yield more robust and meaningful results, potentially offering a better understanding of the role played by vaccine mandates in mitigating COVID-19 cases and deaths.

The association between mask mandates and COVID-19-related deaths is statistically highly significant. However, the unexpected positive sign of the coefficient contradicts intuition. Intuitively, one would expect that the implementation of mask mandates, holding all else constant, should result in fewer deaths as it helps contain the spread of the virus. The coefficient obtained, however, is positive, suggesting that periods with mask mandates are associated with 4.2% higher COVID-19-related deaths compared to periods without mask mandates.

While the counterintuitive sign might raise eyebrows, a compelling explanation supports this finding. Mask mandates were likely implemented during periods of high COVID-19 cases. The positive association between mask mandates and COVID-19-related deaths captures this nuanced relationship – mask mandates are potentially a response to high COVID-19-related deaths, giving rise to a positive association between mask mandates and COVID-19 deaths. This aligns with the key limitation we highlighted in the introduction of our study – the results we report signify correlation rather than causation.

We also found an intuitive result related to our measure of the at-risk population. The coefficient for the at-risk population is 0.181, signifying that a 1% increase in the proportion of at-risk population is associated with an 18.1% increase in COVID- 19-related mortality as a proportion of total deaths, keeping all other factors constant. This finding aligns with our expectations, considering that older populations, constituting the at-risk segment, were significantly more affected than their younger counterparts during the pandemic.

The additional findings we found not only align with intuition but also confirm our initial expectations. Firstly, we observed that unemployment rates are not significantly linked to death rates. This could be attributed to the dual impact of unemployment on COVID-19 cases, where the potential negative effect of unemployment on the standard of living and health conditions may be offset by the positive health aspects of reduced social contacts due to people staying home. Secondly, in terms of Gross Domestic Product (GDP), our results indicate that higher GDP, which translates into the existence of more resources to combat the virus’s spread and provide adequate medical care to individuals upon contracting the virus, is significantly associated with lower COVID-19-related deaths. Finally, when considering the effect of population size, a higher population size, all else being equal, correlates with increased population density. This higher density makes social distancing more challenging and contributes to high spread of the virus.


This study examines the link between public health interventions and COVID-19-related mortality rates in the United States. Utilizing state-level quarterly data spanning 2020-2021, we conduct a fixed effects regression, examining the impact of vaccine and mask mandates alongside demographic, health, and economic variables on death rates. Our analysis reveals an expected decrease in death rates associated with vaccine mandates, while unexpectedly, mask mandates exhibit a positive correlation with mortality rates. This counterintuitive finding may stem from the fact that these policy interventions are a response to high death rates, establishing a positive correlation between mask mandates and high death rates. Additionally, we find a positive relationship between population size and COVID-19-related deaths and a negative correlation between the level of GDP and COVID-19-related death rates. As an area for future research, a study that utilizes higher frequency data, along with the actual number of vaccinations and a proxy for actual usage of facial coverings, can offer additional insights into our findings.


The authors would like to thank participants of the research seminar at Lake Forest College.

Conflicts of Interest



Sign up for Newsletter

Sign up for our newsletter to receive the latest updates. We respect your privacy and will never share your email address with anyone else.