Evaluating the Effects of COVID-19 and Vaccination on Employment Behaviour: A Panel Data Analysis Acrossthe World

Abstract

COVID-19 is a fast-invading virus that quickly invaded the human body and made no human activity immune to its infections. The purpose of this study is to simulate the effects of COVID-19 on employment behaviour and vaccination’s weight in the recovery process. Based on quarterly panel data from 43 nations from 2018 to 2020, we built an adaptive employment model. The major findings demonstrate that COVID-19 has negative and large net and second effects, with parameters of −7049 and −15,768 employees each quarter for 100,000 infected people, respectively. While immunization has a positive net effect of 10,900 employees every quarter, it has a negative second effect of −29,817 employees. This last result may look strange, but it is rational and demonstrates that immunizations modify employees’ behaviour toward prevention measures, leading to actions such as resuming mobility, reopening, cancelling confinement, and so on, even though COVID-19 continues to spread. Demand, the labour force, the short-term multiplier, and immunization appear to have a positive and large impact on employment behaviour, while average labour productivity appears to have a negative impact.

1. Introduction

Since February 2020, the world has been facing an unexpected economic challenge due to COVID-19. Employment seems to have been severely, rapidly, and directly affected by this health shock, which started in China on 23 January 2020 before reaching Italy and then the United States in a few weeks. In the first quarter (Q1) of 2020, the pandemic reduced working hours across the world by 5.2%, and then the effect was amplified during the second quarter to reach 18.2%. Concerning the third and fourth quarters, the work hour losses dropped to 7.2% in the third and 4.6% in the fourth across the world [1].

The remarkable feature of COVID-19 is its severity. Compared to the last health crisis, COVID-19 qualified as a pandemic after only a few days due to the great number of infected people, which increased rapidly; in Q1−2020, Q2−2020, Q3−2020, and Q4−2020 across the world, the number of infections was about 875,800, 10.5 million, 34.012 million, and 83.559 million, respectively.At the end of the first quarter of 2021, infections numbered 128,896,320, with different levels across countries and continents [2].

COVID-19 may have a variety of effects on employment. First, as a net effect, workplace infections and deaths reduce job creation and cause labour market disruptions by impeding individual mobility. Consequently, workers hesitate to report to work, and employers hesitate to hire persons out of fear of an infection. Through a second effect, governments’ responses to COVID-19 infections result in the adoption of quarantine-related policies and procedures, which compel employees to stay at home instead of going to work, and businesses’ cancellations of worker recruitment plans rise as COVID-19 infections rise.

Nonetheless, humanity has not succumbed to the pandemic; vaccination has historically been the most effective measure against viruses (influenza, SARS, H1N1, etc.). With that in mind, pharmaceutical firms have invested in research to develop vaccines, namely Pfizer, AstraZeneca, Sputnik, etc. The vaccination campaign started in December 2020, by which time employment behaviour had recorded changes timidly indicating the commencement of a recovery period. Indeed, the stock of vaccines up until the end of December 2020 had been only about 9 million doses, and with rapid production, it reached 600.3 million vaccines at the end of the first quarter of 2021 [3]. Consequently, work-hour losses have been expected, from 2020 to 2021 (based on 48 work hours per week), to decrease from about 252.5 million to 88 million full-time equivalent employment losses [1].

Many researchers have attempted to predict the pandemic’s effects [4,5,6,7], the relative measures to tackle its spread [8,9,10,11,12,13,14], and, most importantly, the effect of vaccination [15]. Authors have attempted to suggest numerous scenarios of evolution trends [16,17,18], pandemic features [19,20], and the risk of the virus returning for a second wave. In this frame of COVID-19 investigations, the present research aims at simulating the net and second effects of COVID-19 infection and vaccination tracking on employment behaviour across the world. In this direction, the current article intends to simulate:

(a)At which level employment will decrease if one person is effectively infected by COVID-19, and what the second effect will be if we consider all of the facts that accompany the COVID-19 shock, such as lockdown, social distancing, people’s mobility hesitation, government subvention during the pandemic, etc.

(b)At which level employment will increase if one person is vaccinated, and what the second effect will be if we take into consideration that many facts accompany vaccination, such as changes in people’s behaviour, government policy restrictions, reopening, confinement cancellation, mobility resumption, etc.

To measure and explain the effects of COVID-19 vs. vaccination on employment, we must first frame employment behaviour in a theoretical context. For instance, we will refer to the conjectural economic theory of employment, which is based on short-term analysis. According to Keynesian theory, national income and employment are determined by the interaction between the aggregate demand curve and the aggregate supply curve. Therefore, [21] adopted the point of view of Robinson’s short-period theory of employment, which involves the Keynesian demand deficiency hypothesis to show that employment responds to a change in effective demand.

Thus, the principle of effective demand stipulates that when effective demand increases, employment also increases, and viceversa, i.e., the principle suggests that full employment is not possible to achieve, and real investment must equal the gap between income and consumption. In other words, employment cannot expand unless investment expands.

The implication of the concept of effective demand, as noted by [22], is that it suggests what structural change is required to stabilize a laissez-faire economy and what macroeconomic policy (monetary and fiscal) should aim to achieve. Therefore, domestic policy aims to achieve full employment by proceeding through investment, which is the dynamo of employment. In the case of the United States, [23] stated that he was unsure whether the government could or would boost demand when the economy needed it. Moreover, the Federal Reserve Board actively manages interest rates, pushing them down when it believes employment is too low and raising them when it believes the economy is overheating.

As a result, the government must implement a suitable fiscal policy to increase capital’s marginal efficiency while also implementing monetary policy to lower interest rates. In the short term, monetary policy has three competing goals, or a trilemma, that are incompatible with the international financial iron rule, and an open economy can only accomplish two of them, as explained by [23]. These goals are to uphold a free monetary policy, a stable exchange rate, and/or full convertibility. These goals and the associated monetary policies heavily influence the interest rate.

In general, labour costs are an important determinant of employment behaviour in the short term. To explain this idea, it should be noted that labour costs and productivity widely appear in the fundamentals of employment theory in the short term. In this direction, microeconomics doesnot only provide a long-term analysis but also a short-term framework for employment fluctuations. The employment theory in the short term is based, on the one hand, on the hypothesis of profit maximization, whereby firms are supposed to maximize receipts minus costs (π= pQ-wL-cK; w: wage and c: cost of capital utilization). On the other hand, the hypothesis given is that a firm’s capital stock is constant in the short run. According to the above hypothesis, the derivation of the profit (π) considering labour is equal to zero, and it implies that marginal productivity (in value) minus wages is equal to zero. Finally, the procedure of calculation yields the result that the marginal productivity of labour is equal to real wages (MPL = W/P). In the short term, the equilibrium is not evocated, and we can find that W/P may be superior to MPL or inferior to it.

In his paper, [22] illustrates that the classical model of employment is very simplistic, and it generates some plausible conclusions about the relationship between productivity and wages. The author writes:

-If productivity per unit of labour input (or worker) increases while wages remain constant, this will increase labour demand.

-Given a fixed labour supply, increased labour demand will result in higher pay until a new profit-maximizing equilibrium is reached.

Thus, the level and structure of labour costs and productivity and the way they change over time could play a central role in wage negotiations or in implementing and assessing employment, wages, and other social policies [24]. The role and influence of labour costs isnot limited to this level, but it is larger in employment. In addition, labour costs have a significant impact on employment creation. As stated by [25], real minimum basic wages have a significant negative effect on private employment creation. Empirically, the cost effect is estimated at −1213.35 (t = 22.38), meaning that an increase in the real basic wage by a unit will reduce private employment by 1213 individuals. By industry, the results have been confirmed for mining, energy, building and civil engineering, manufacturing, and other services.According to [26], strong positive employment reduces an employer’s social security contribution—an embodied policy in a package of reform measures (compared to stand-alone measures)—and reduces an employer’s taxes or increases subsidies that must be financed. In addition, reduced non-wage labour costs (consumer spending) have a positive effect on employment.

Regarding labour productivity, the short-term framework for employment fluctuations supports the idea that increasing productivity will increase labour demand while wages remain constant [27]. This lasts, and given a fixed labour supply, it will result in higher pay until a new profit-maximizing equilibrium is reached.

To demonstrate this idea, we can start by describing the mechanism of the labour market under classical theory’s point of view in two scenarios. The first scenario involves the labour demand: firms attempt to maximize profits; they will employ labour until the MPL is equal to the given W/P. In this state, there is no involuntary unemployment, and full employment of labour prevails. The second scenario considers labour supply: households in the economy depend on their pattern of preference between income and leisure. In the short term, the population does not vary. The supply curve of labour slopes upward. The fact is that households or individual workers maximize their utility or satisfaction by choosing a combination of work and leisure. To sum up, according to classical theory, employment is determined by the equality of real wages and marginal productivity (MPL). Employers apply this principle when they decide to hire; when MPL is higher than the real wage, employers continue to hire workers until there is equality among them, and vice versa.

This thesis is criticized by the efficiency wage theory suggested by [28], under the argument that individual workers maximize their utility, taking account of wages, alternative incomes, and efforts, all of which contribute to determining their marginal product. Economists consider the conventional wage maximization theory in light of the conventional wage bargaining model of profit maximization and efficiency wages. It indicates a consensus that productivity influences wage-setting behaviour, and a firm’s policy enhances setting a wage to achieve a higher level of employment. In this direction of employment bargain analysis, [29] highlights that the Tunisian labour market is rigid. Their findings show that, in the absence of collective bargaining, the relationship between real wages and productivity cannot be used to determine employment changes.

Another dimension of employment fluctuation in the short term is that employment behaviour might receive many shocks that have different origins, such as economic [30,31], social [32], political [33], technological [34], and input cost shocks [35,36], as well as health shocks related to viral epidemics, e.g., the current crisis caused by COVID-19 [1,37,38,39,40]. As they take lessons from ancient employment shocks during previous recessionary periods [41], researchers are interested in examining the influence of COVID-19 on employment by benchmarking the periods before the pandemic vs. the era ofCOVID-19 [42], as they have learned from earlier recessionary periods [41]. In fact, [43] emphasizes the COVID-19 pandemic’s effects on flexible work relationships and careers to examine how they might be affected by it, its recurring consequences, and its impact on flexible employment relationships.

In [44], the authors reviewed employment-related determinants of health and health protection during the COVID-19 pandemic in Sweden, Belgium, Spain, Canada, the United States, and Chile. The results suggest that COVID-19 affects non-standard employment in terms of work hours, income, and benefits. In addition, the pandemic has influenced worker types, the duration of employment arrangements, and the ability to cover regular expenses during the pandemic. In [45], the authors analyse the effects of COVID-19 on airline employment by using a time series analysis that shows a job loss of between 7% and 13% for the airline workforce, while the recovery is estimated to take between 4 and 6 years.

In this direction, according to [46], “regular” recessions affect men’s employment more severely than women’s employment, while the COVID-19 shock has had a large impact on sectors with high female employment shares, namely those that employ working mothers. This finding can be confirmed by the research conducted by [47], which finds a negative effect of COVID_19 on employment of women, minorities, the less educated, and the young. Focusing on a high-risk group, [48] analysed the impact of COVID-19 on employment loss among people with severe mental disorders. The paper’s major findings show that during lockdown, employment in this group decreased significantly (13% vs. 9.2%; c2 = 126.228, p < 0.001).

In the same framework, [49] examines the decline in Asian American employment as a result of the imposed COVID-19 lockdown and finds that this group, particularly the less educated members, was more negatively affected than other racial groups after controlling for education, immigration, and other covariates. In [50], the author predicts that COVID-19 will put between 14% and 26% (1.9–3.4 million) of Australians out of work, and that unemployment will exceed 50% in hospitality, retail trade, education and training, and the arts; the author further predicts that the rate will be twice as high for lower-income workers, young people, and women, who are more vulnerable to the pandemic’s effects. Concerning the recovery, the authors note the recent recession is different from historical crises that have been deliberately engineered as a matter of public health, and substantial economic support is in place. Following the current global economic depression, the international community’s intervention-based rescue efforts have moved more quickly than in the past [23]. When a nation is in peril, a SWAT team responds right away. During the recent COVID-19 issue, international cooperation actually took centre stage. For example, vaccination is advised as the most significant tool that could prevent a serious crisis and is the subject of strong international cooperation. The partnership includes not just the financing and production of vaccines but also their international dissemination. The United States, in particular, places a high focus on population health and devotes significant resources to it. Some economists are not satisfied with this strategy, such as [51], who label policymakers as utterly irrational for deciding to prioritize health over wealth solely because they are subject to social pressure.

While it is a solution against COVID-19 infections, at its first stage, the vaccine has been the subject of a great deal of controversy between those who reject it and those who accept it. On the side of reluctance, hesitation, and non-willingness to vaccinate, [52] studied the differences in COVID-19 vaccine concerns among Asian Americans and Pacific Islanders and found that, overall, 76% of the respondents reported having concerns about the vaccine, and 65% of those concerns were identified as regarding side effects. In trying to explain vaccine hesitancy in South Africa, [53] finds that many reasons, in particular those critical to personal health and the severity of the hesitancy, are correlated to age, race, education, politics, geographical location, and employment. On the other hand, a strong demand for and high acceptance of COVID-19 vaccination has been shown among the Chinese population [54] and American people [55], while concerns about vaccine safety may hinder the promotion of vaccine uptake.

In the same direction, a social valuation of COVID-19 vaccines is observed among Chile’s people (93.3%) who have expressed an individual willingness to pay for them [56]. The research conducted by [57] confirms this behaviour toward acceptance and willingness to pay (WTP) in Ecuador. The authors find that a large proportion of individuals (at least 97%) are willing to accept a COVID-19 vaccine. Furthermore, the WTP for the vaccine has been associated with income, employment status, the perceived probability of needing hospitalization if one contracts COVID-19, and region of residence.

Many researchers have attempted to assess the impact of vaccines. For instance, [58] evaluates the impact of a two-dose vaccination campaign on reducing COVID-19 outbreaks in the United States. The main findings suggest that vaccination could reduce the overall COVID-19 infection rate, specifically for people over the age of 65, by 54–62%, as well as reducing deaths and adverse outcomes. The authors of [59] reported the fact that any delay in a vaccination campaign can increase COVID-19 cases as well as hospitalizations and deaths, as observed in Chicago and New York City. The authors noted that the earlier the vaccination campaign begins, the greater the potential impact on COVID-19 infection reduction. These findings were confirmed in India by [60], where the authors uncover that COVID-19 vaccination programs for healthcare professionals (HCPs) have been critical in slowing the pandemic and suggest raising awareness about the importance of vaccines among HCPs in India. Regarding Africa, people have suffered a great deal from the delay in COVID-19 vaccination [61], mainly because of limited funds and concerns around vaccine safety, etc. In contrast, vaccination’s negative effects have been observed on people’s attitudes towards protective countermeasures [62]. The authors find that in China, vaccination reduces the frequency of hand washing by 1.75 times and physical distancing by 1.24 times, and that this has caused a resurgence of COVID-19.

Finally, COVID-19 vaccination gives small businesses a chance to recover from COVID-19 infections, but large businesses are more likely to close and less likely to reopen, and disadvantaged people are more likely to be laid off and less likely to return to work. In [63,64], researchers examined the influence of the COVID-19 problem on employment by categorizing three economic sectors based on European countries’ confinement decrees: Germany, Spain, Italy, and the United Kingdom; Poland; and Sweden. The findings reveal that the nations hardest struck by the pandemic (Spain, Italy, and the United Kingdom) were also the countries most likely to suffer the most negative effects of confinement on employment. In [65], researchers studied the employment loss in informal settlements before and during the COVID-19 pandemic in Chile, and the authors showed a decrease in employment of 30% and 40%, respectively. In addition, they concluded that employment loss was substantially higher in informal settlements and for the immigrant population.

2. Materials and Methods

Note that Keynesian theory [66] has suggested using the multiplier in the basic employment analysis, which had its principal origin in effective demand ED effects. When ED increases, employment level increases. For instance, we refer to the work by [25]. In establishing a relationship between real and desired labour (L) increments:

L_t − L_t−1 = η (L^*_t − L_t)

(1)

This equation shows that when a firm decides to raise L over the existing stock level at time t − 1, (L_t−1), to the stock level at time t, (L_t), it should hire new workers. After some mathematical developments of (1), we could write the employment (E_t) flow as:

E_t − E_t−1 = δ (E^*_t − E_t−1)

(2)

where E_t^* is the expected employment at time t. This shows that annual net employment increment equals its expected increment corrected by an adjustment cost (0 < δ < 1). Theoretically, [67] assumed the adjustment cost to be a function of the net employment change and demand fluctuation. The coefficient of partial adjustment could be defined as follows:

n = n₀ + [1/(E^*_t − E_t−1)] Φ(X_t)

(3)

Φ(X_t): determinants of the coefficient of adjustment of employment. Thus:

D E = (n₀ + [1/(E^*_t − E_t−1)] Φ(X_t) ) . (E^*_t − E_t−1)

(4)

The developed form follows the specification of adaptive behaviour from [68], and the final form of the estimated model could be written as:

E_it = α n₀ ν (GDP_i(t−1) − (1 − δ) GDP_i(t−2)) + (n₁ AP_it + n₂ LabF_it + n₃ Tcases_it + n₄ TVacc_it) + (2 − n₀ − ν (1 + g)) E_i(t−1) − (1 − n₀) (1 − ν(1 + g)) E_i(t−2)

(5)

Data is collected by quarter (Supplementary Materials). Quarterly total employment and labour force data were collected from the database of [1]. Concerning quarterly real GDP, presented in 2010 constant USD, data was collected from the database of [24]. Average productivity is calculated by dividing real GDP by employment for each quarter.

Data concerning total cases of COVID-19 and total vaccinations were collected from the global database of [3]. Variables used in the model are defined in the following

Table 1. Variables and Parameters.

Variables	Parameters	Definition
Eit	−	Employment at date t for country i.
−( . )GDPit	Short-term multiplier	The variation of GDPi(t−1) with consideration of depreciation GDPi(t−2) has a positive effect equal to αn0 ν.
LabFit	Effect of labour force changes	Labour force at date t has an effect: n2.
TCasesit	Net effect of infected people	The total number of people infected (stock) by COVID-19 at date t has an effect: n3.
Dumm1	Second effect of COVID-19	The effect of the government’s anti-COVID-19 measures on employment is defined as the net effect increased by the effect of social distancing, lockdowns, worker alternance measures, etc. A dummy variable called Dumm1 has a value of 1 if measures are present, and 0 otherwise.
TVaccit:	Net effect of vaccinated people	The total number of people vaccinated (stock) at date t has an effect: n4
Dumm2	Second effect of vaccination	The effect of restriction of all measures connected to vaccination on employment: it includes people regaining their mobility, breaking out of quarantine and lockdown, overlooking social distancing, restricting work alternation, etc. Dumm2 expresses the second effect, where a value of 1 denotes complete disregard for laws, lack of fear, etc. A result of 0 denotes the continuation of COVID-19 prevention measures
Ei(t-1)	Autoregressive adjustment coefficient of lagged−1 variable	Employment at date t-1 for country i has an effect $\left(2-n_0-v(1+g)\right)$:
$\frac{\alpha n_0\gamma }{1-\{(2-n_0-\gamma (1+g))-(1-n_0(1-\gamma (1+g)))\}}$		Long-term multiplier represents the short-term multiplier adjusted by sum of autoregressive coefficient of lagged variables 1 and 2.

.

The model is conducted using panel data from 43 nations across 12 quarters from 2018 to 2020 using the SPSS statistical software. The included countries are: Austria, Belgium, Brazil, Bulgaria, Canada, Chile, Costa Rica, Croatia, Cyprus, Czechia, Denmark, Estonia, Finland, Greece, France, Georgia, Hungary, Iceland, Ireland, Italy, the Republic of Korea, Latvia, Lithuania, Luxembourg, Malta, Mexico, the Netherlands, Norway, Peru, Poland, Portugal, Romania, Serbia, Slovakia, Slovenia, South Africa, Sweden, Switzerland, Thailand, Turkey, the United Kingdom, the United States, and Vietnam.

3. Results

The panel data estimation consists of three parts. The first, model estimation used in panel data analysis, is pooled regression, the second is fixed effects, and the third is random effects.

We should emphasize that before running the model, a criterion related to data stationarity must be met. We use the unit root test to do this, which establishes a stationary state for all structural variables such as employment, weighted variation of real GDP, labour force, and productivity (See Supplementary Materials: Tables S1–S3).

3.1. Pooled (Regression) Model

There are two different versions of the employment model. To assess the respective net impacts of COVID-19 and vaccines, Model 1 includes variables for total COVID-19 cases (Tcases) and total vaccinations (Tvacc). In order to evaluate the second effects of COVID-19 and vaccination, Model 2 uses the variables Dumm1 and Dumm2, respectively, rather than Tcases and Tvacc. Given that the estimable form of the model is autoregressive-moving-average, the GMM method must be used to run the model (which represents the version of OLS applied to simple regression). The obtained results show the model’s fit and its significance for the specified variables, as well as an explanation of job behaviour (see

Table 2. Goodness of Fit Test.

Model 1			Model 2
Likelihood Ratio Chi-Square	df	Sig.	Likelihood Ratio Chi-Square	df	Sig.
2718.635	7	0.000	2710.231	7	0.000

). Either Model Form1 or Model Form2 appears to fit across both identifications; factors accurately describe employment behaviour; or the likelihood ratio chi-square is significant. (See Supplementary Materials: Tables S4–S7. Descriptive Statistics and Correlation.)

The regression form is separated into two scenarios: the first involves total cases (number) of people infected by COVID-19and total number of people who have been vaccinated (number). The second scenario includes two dummy variables; the first is represented by Dumm1 generating the second effect of COVID-19 on employment, and it includes COVID-19′s impact on employment, which includes total cases of COVID-19, social distancing, quarantines, and government grant and subsidy policies.The second involved measures to fight COVID-19, which include vaccination, employers, employees, and people’s reactions to resuming mobility, among other things.

Table 3. Regression Estimation Using GMM.

Parameters	Model Form1: T-Cases and T-Vacc					Model Form2: Dumm1 and Dumm2
	Coeff.	Std. Error	Sig.	95% Wald CI		Coeff.	Std. Error	Sig.	95% Wald CI
				Lower	Upper				Lower	Upper
(Intercept)	39.950	108.732	0.713	−173.161	253.061	163.523	118.0876	0.166	−67.925	394.970
T_Casest	0	0.0001	0.000	−1	0
T_Vacct	4	0.0010	0.000	2	0
Dumm1						−331.552	124.2950	0.008	−575.166	−87.938
Dumm2						406.646	250.8786	0.105	−85.067	898.359
VGDPt−1	0.005	0.0017	0.004	0.002	0.008	0.005	0.0015	0.001	0.002	0.008
LF15+t	0.444	0.0321	0.000	0.382	0.507	0.399	0.0294	0.000	0.342	0.457
APt	−0.002	0.0046	0.718	−0.011	0.007	−0.003	0.0046	0.571	−0.012	0.006
Et−1	0.343	0.0621	0.000	0.221	0.465	0.463	0.519	0.000	0.361	0.565
Et−2	0.170	0.0525	0.001	0.067	0.273	0.095	0.0476	0.045	0.002	0.189

summarizes the findings (see Supplementary Materials Tables S8–S17).

The regression estimation results (Table 3) demonstrate that only the intercept and average productivity (AP_t) are not statistically significant in the first scenario (Model Form 1), with probabilities of 0.713 and 0.718, respectively. The probabilities for the second scenario (Model Form 2) are 0.166, 0.571, and 0.103, respectively, with the addition of a non-significant effect of Dumm2. The net effects of COVID-19 and vaccinations confirm their expectations, and they are statistically significant. Thus, total cases (TCases_t) of COVID-19 have a low negative effect (between −0.001 and 0) in the Model Form1 results. This indicates that for every 1000 infected individuals, there will only be a maximum loss of 1 job. In contrast, total vaccination (TVacc_t) has a negligible positive effect (0.004), which suggests that for every 1000 persons who receive vaccinations, there are an additional 4 jobs created. In terms of dummy variables (Dumm1 and Dumm2), Dumm1 is negative, and Dumm2 is positive. The two effects confirm the projected values of 331,552 and 406,646; however, Dumm2 is not significant at 10%. Due to its high Akaike test value, this regression (pooled) estimation does not seem to be the better fit. We conclude this by estimating the model’s fixed and random effects.

3.2. Choosing between Fixed and Random Effect

Regarding the estimation of fixed and random effects (see Supplementary Materials) of the models (Forms 1 and 2), it is shown that the fixed effect model’s Akaike information criterion (AIC) is lower than the random effect model’s AIC, as shown in

Table 4. Information Criteria (*).

	Regression Effect	Fixed Effect Model	Random Effect
2 Restricted Log Likelihood	−3581.316	5481.580	9504.084
Akaike’s Information Criterion (AIC)	7180.632	5591.580	9630.084
Hurvich and Tsai’s Criterion (AICC)	7181.067	5608.691	9652.993
Bozdogan’s Criterion (CAIC)	7226.080	5868.268	9947.017
Schwarz’s Bayesian Criterion (BIC)	7217.080	5813.268	9884.017

* The information criteria are displayed in smaller-is-better form.

.

As a result, the study used fixed effect results to determine employment behaviour and to mimic COVID-19 and immunization effects across the world using Model 1 and Model 2.

3.3. Fixed Effect Model Results

3.3.1. Simulation of Net effects of COVID-19 vs. Vaccination

The results issued from fixed effect Model 1 are presented in

Table 5. Results of Fixed Effect Model Form1 (Estimation using GMM).

Parameters	Estimate	Std. Error	df	t	Sig.	95% CI
						Lower Bound	Upper Bound
Intercept	314.3142	84.7513	4.274	3.709	0.018	84.8300	543.7985
VGDPt−1	0.0006	0.0001	48.327	4.639	0.000	0.0004	0.0009
LFt	0.9395	0.0036	21.918	261.490	0.000	0.9321	0.9470
APt	−0.01300	0.0037	6.966	−3.514	0.010	−0.0218	−0.0042
Tcasest (100 thousands)	−70.493	5.6552 × 10−6	37.391	−12.465	0.000	−81.947	−59.038
TVacct (100 thousands)	109	2.1472 × 10−5	39.621	5.071	0.000	65.484 × 10−5	21.52
Et−1	0.0454	0.003902	65.616	11.628	0.000	0.0376	0.0532
Et−2	0.0026	0.0016	9.483	1.699	0.122	−0.0008	0.0061

. In terms of net demand, it has a favourable and considerable impact on the generation of jobs at 1%. According to economic theory, the coefficient with t = 4.639 (probability = 0.000) is 0.000636. The impact known as the short-term multiplier shows that a one-unit (USD 1000) change in lagged GDP between the last two quarters causes an increase in employment at quarter t.

The influence of average productivity (which replaces labour cost (LC) in the model) on the other hand, verifies the theory. Indeed, workers’ increased average productivity (AP) improves an employer’s ability to minimize employment by allowing numerous workers to be replaced. This could be seen as similar to theway that an increase in LC reduces employment numbers. With t = −3.514 and probability = 0.01, the coefficient is negative, equalling−0.013, and it is statistically significant. This suggests that if AP increases by one unit (USD 1000 per person), total employment will fall by 13 individuals.

The employment theory is supported by the fact that the labour force effect (LF) has a positive and significant coefficient of 1%. With t = 261.49, the coefficient value is 0.939, and the probability is 0.000. When LF rises by one unit (1000 people), employment rises by 940 people.

On the other hand, employment lag variables 1 and 2 (Et−1, Et−2) could provide information on both the adaptive and long-term multiplier coefficients. The Et−1 adaptation coefficients are positive (0.0454) and significant (t = 11.628 at 1%). It means that employment at date t−1 rises by one unit, resulting in a 0.0454 increase in employment at date t. This behaviour could be explained by the fact that, in the near term, employment in the previous quarter could boost present employment due to adaptive plans regarding COVID-19, which resulted in a large number of lost working hours. Et−2 has a positive adaptation coefficient of 0.0026, but it isn’t statistically significant.

The long-term multiplier is equal to 0.0133, which is a short-term multiplier modified by the adaptive coefficient, but it is not significant. These findings could be explained by the fact that GDP has a positive long-term impact on employment, with a USD 1 million rise increasing employment by 13,247. The fact that net demand has a favourable long-term impact on employment supports the theory.

The net effect of COVID-19 represents the number of people infected by COVID-19, which causes employment loss (number of people infected and in stock who donot reach their workplaces). It indicates how many job opportunities will be lost if the overall number of instances rises by one more infected person. The data demonstrates that increasing the number of total instances in the stock by one unit reduces employment by −7.049 × 10⁻⁵. This suggests that infected supply 100,000 individuals will result in the loss of 7049 jobs.

The net result of vaccination is a gain in employment as a result of a one-person increase in the stock of vaccinated individuals. The results show that employment increases by 0.000109 for every unit increase in the number of immunized individuals. According to this, creating 10,900 additional instances of employment would require immunizing 100,000 people. In conclusion, the recovery is reliant on the balance between COVID-19infection and vaccination speed, with the frequency of vaccination accelerating the spread of COVID-19 infections.

3.3.2. Simulation of Second Effects of COVID-19 vs. Vaccination

Two dummy variables, Dumm1 and Dumm2, are used to replicate the second effects of COVID-19 and immunization. The COVID-19pandemic’s second effect (Model 2) is reflected by Dumm1, which includes measures of social distancing, quarantine procedures and duration, and people’s mobility. The findings (

Table 6. Results of Fixed Effect Model Form 2 (Estimation using GMM).

Parameters	Value	Std. Error	Df	t	Sig.	95% CI
						Lower Bound	Upper Bound
Intercept	−16.4788	93.3925	50.350	−0.176	0.861	−204.0308	171.0732
Dumm1	−15.7684	7.6780	42.945	−2.054	0.046	−31.2532	−0.2836
Dumm2	−29.8170	7.6645	14.513	−3.890	0.002	−46.2014	−13.4325
VGDPt−1	0.0006	9.6659 × 10−5	20.742	6.591	0.000	0.0004	0.0008
LF15+t	0.9472	0.0035	62.514	267.201	0.000	0.9401	0.9543
APt	−0.0099	0.0027	15.999	−3.718	0.002	−0.0155	−0.0042
Et−1	0.0552	0.0029	69.336	18.967	0.000	0.0494	0.0610
Et−2	−0.0073	0.0018	20.448	−4.114	0.001	−0.0110	−0.0036

) show that the negative effect is substantial, at 5%. Furthermore, the second effect is two times as powerful as the net effect of COVID-19:−15,768 personnel. This means that COVID-19 is accompanied by other information that could have a detrimental impact on employment.

The second effect of vaccination is represented by the variable Dumm2, which includes vaccination; people’s mobility behaviour (optimism, social distancing, mobility resumption, and so on) has an unanticipated negative influence on employment of −29,817 employees.

4. Discussion

The outcomes of this investigation back up the theory that COVID-19′s health shock has a significant negative net effect. As a result, the obtained data supports the conclusions of other studies. The benchmarking conducted for the European Union’s labour market [42] before and during COVID-19 noted the same negative effect. For example, in Sweden, Belgium, Canada, the United States, and Chile, COVID-19 has been shown to have a negative impact on non-standard work hours, and [44,45] identified a job loss in the aviation industry along a similar line. In [51], the authors came to the same conclusion about Australian workers quitting their jobs.

As a result of steps taken by the government and businesses to prevent employees from returning to their places of employment, the net effect of COVID-19, as we call it, will be accompanied by a second negative effect on employment. Lockdowns, shorter working hours, a reduction in the number of workers, work alternating for distance, and so on were among the measures used. The conclusions of [50], which looked at the decline in Asian American employment as a result of the mandated lockdowns, are supported by the current study’s assessment of COVID-19′s negative second effect.

According to our findings, vaccination is an important part of resuming employment. The vaccine does, in fact, have a considerable positive net effect. The fact that immunization has a bigger net effect than COVID-19 shows that the net job loss caused by COVID-19 could be counterbalanced by the vaccine’s heft. Vaccination, in reality, helps and accelerates the recovery of people’s health, allowing them to return to work. The study must come to a conclusion, and the debate over whether or not to vaccinate children among Asian Americans and Pacific Islanders [52] must be addressed. The research supports the first viewpoint, which is shared by Chinese people [54], Americans [55], and the vast majority of Chileans [56].

Vaccination, despite its beneficial advantages, has a negative side effect, according to the current study. This could be explained by the fact that people’s views toward preventative countermeasures change following vaccination [61]. In China, for example, immunization restricts handwashing and physical isolation, leading to the re-emergence of COVID-19. As a result, immunization must be complemented by additional measures such as increased investment, subventions, and public awareness initiatives that promote rapid job recovery.

The usual variables that affect employment have significant effects. Indeed, the employment lag 2 adjustment coefficient, labour-force effect, average production influence, and short-term multiplier all have expected and substantial effects. Additionally, employment lag 1 has a favourable impact.