4.2. Recovery Outcome and CalibrationsAs shown in Table 2, this research selects 3 aspects to evaluate a city’s recovery, COVID recovery, economy recovery, and future protection capacity. All the indicators are as of 30 June 2021. If we choose a certain point in 2020 as the cut-off point of our data, due to the different start times of the pandemic and the weak test capacity of different cities, the pandemic data in the sample may include some data deviation, which cannot fully reflect current facts. The reason why we did not use duration period after local onset of pandemic or duration after first wave is that the onset time of the first pandemic is different in each city, and the duration period varies greatly. As a non-public health research, it is difficult for us to accurately define the exact time of the first pandemic in each city. In addition, we did not use the latest date because the virulence of the virus weakens as it mutates, increasing the rate of infection and decreasing the rate of severe illness in all regions. This research tends to focus on the community resilience and recovery during the period when the virus is still virulent, and the rate of severe illness is high.

COVID recovery: This research evaluates the city’s recovery from COVID-19 infection and death, respectively. Our assessment of the recovery situation is to calculate the proportion of the number of infections/deaths in the city on 30 June 2021, to the city’s historical peak data. The smaller the ratio, the better the recovery of the city. Since many countries do not provide city-level pandemic data, we corrected this ratio through the population of the country and the city.

Economy recovery: This research evaluates economy recovery through GDP and unemployment. GDP is the most commonly used indicator to measure the development status of a region. Unemployment is also a measure of how well a place has recovered from a pandemic. Regions that provide enough jobs tend to recover well.

Future protection capacity: At present, coronavirus is still spreading wantonly, and the newly mutated strains are highly contagious, making the duration of the epidemic full of uncertainty. Immunization by vaccination is one of the best ways to avoid large-scale transmission again. Cities with higher vaccination rates can more easily return to normal life and restore economic development.

Raw data collected are calibrated to a number between 0 and 1 according to the requirement of QCA analysis. All calibration follows the formula as shown below. World maximum and minimum are used to calibrate data so that countries not included in the list could also make use of the QCA results obtained from this research.

Calibrated data=(X−Wmin)(Wmax−Wmin)

(1)

In addition, we must ensure that these four indicators have isotonicity. The closer the calibrated number is to 1, the better the local resilience. Therefore, for some indicators, we subtracted the initial calculated value by 1.

Finally, we take the average of these 3 calibrated indicators as the policy outcome index, as shown in Table 3. 4.3. Contributing Factors and CalibrationsAs we selected 16 cases as a sample as the fsQCA methods apply to 2n principle, 4 factors can be defined to analyze 16 cases (24 = 16). Given the types of political regimes, economy, external communication, the severity of COVID-19, and the ability of medical care of the cities, we concluded that the most critical four factors of recovery from COVID-19, which are made up of two to four indicators as shown in Table 4.Robustness: Each city possessed some extent of robustness, and those with specific characteristics are more stable and less likely to be disturbed by the pandemic. People in areas with high life expectancy are in better health, are more immune to the pandemic, and less likely to get infected. High age dependency ratio indicates that the elderly and children, who are vulnerable to infection, account for a large proportion in the city, which deserves government departments’ attention. Government and community policies are easier to implement if the local population has primary education and can read [84]. In diverse communities, there is a big gap in the behavior pattern and way of thinking of residents. Especially in the context of COVID-19, diverse communities are not conducive to the implementation of a unified epidemic prevention policy [85]. Political regime represents the governing ability of a city government, and an efficient government can control the pandemic in a timely manner [86].

Preparedness: If a region has experience dealing with a pandemic and has some material reserves, we think that region can recover quickly from the pandemic. Stockpile of PPEs is used for evaluating the reserves of protective equipment. The medical capacity of a city is an important factor in the government’s consideration of whether to implement lockdown or other policy. If the city has adequate medical resources, it can relax controls appropriately to protect the economy.

Resource and social capital: During the pandemic, the subjective initiative of communities was fully exerted. With limited government resources, many communities organized themselves to deal with the pandemic. It also reflects the resilience of the city. During the raging period of the virus, the production of agricultural products was negatively affected, and the import link was hindered. How the city supplies basic living materials is one of the important evaluation factors to measure the city’s resilience.

Government response: There are three dimensions to measure government response, namely closures and containment, economic measures, and health measures. These three policies can reflect whether the government has taken a response policy after the outbreak and whether the policy is comprehensive.

All of the data are calibrated by Formula 1 and standardized to 0–1, and according to our hypothesis, the closer the index was to 1, the more likely the city was to recover from COVID-19. The averaged indexes of all factors are shown in Table 5. 4.4. Research OutcomesThe first step of QCA analysis is to determine the extent to which a combination of factors leads to the outcome; this is known as consistency in fsQCA. According to the QCA method, if the consistency score is 1, it shows that there is a perfect subset relationship between the antecedent condition and outcome, the antecedent condition being considered to be the necessary condition leading to the outcome. However, social science data often do not fully realize the perfect subset relationship, so the standard of 0.95 can be set, which means the variable constitutes a necessary condition for the outcome if the consistency score is greater than 0.95. If the consistency is close to 0.95, the antecedent condition can be considered as an important necessary condition [80].From Table 6, we can find out that no variable is consistent with 1, so there is no absolutely necessary condition. None of the variables can be considered a necessary condition of recovery, as the consistency of each variable is lower than the critical value of 0.95. It is noteworthy that “X-resources and social capital” have the consistency of 0.94 (very close to 0.95), which indicates that the resource of a city may have a greater influence on recovery, which needs further research and verification. Overall, it is necessary to perform configuration analysis on these condition variables and consider the conjectural synergistic effect of multiple conditions.According to the principle of QCA, the number of configurations formed by multiple antecedent conditions is logarithmic to the number of selected conditions; that is, for a fuzzy set with k antecedent conditions, 2k configurations can be constructed, and each configuration corresponds to a row in the truth table. This research selects four antecedent conditions, and there will be 16 configurations. Set the consistency threshold to 0.8 (if the result is greater than 0.8, it is 1; if the value is less than 0.8, the result does not exist). In QCA research, 0.8 and 0.75 are the most commonly used thresholds. When the sample size is small, the consistency threshold should be higher, while when the sample size is large, the consistency threshold can be lower. In this research, only 16 samples are selected; thus, we choose 0.8 as the threshold [80]. Meanwhile, the threshold of case frequency is set to 1 (the case result below this value is considered as a logical remainder). According to the above settings, the simplified truth table contains 16 configurations, of which 10 configurations exist in policy outcomes and 6 configurations do not exist as shown in Table 7. There is no contradictory configuration (the configuration with the same condition but the opposite result is called a contradictory configuration). From the truth table, it can be shown that the combination of causes leading to the recovery of cities is diverse, which proves that there is a complex causal relationship between the antecedents and results of recovery from COVID-19 [79].According to the results of the truth table, fsQCA 3.0 is further used for Boolean minimization. “Standard analyses” identify five distinct patterns comprising different combinations of attributes that determine the resilience of a society and, hence, its ability to recover from the COVID-19 pandemic. These five patterns illustrated in Table 8 together explain the main reasons of good recovery. The first pattern is characterized by a high level of robustness and low level of preparedness. The second pattern features a high level of resources and social capital and low level of preparedness The third pattern refers to cities that possess high levels of robustness and resources and social capital. The fourth pattern features high levels of robustness and government response. The final pattern comprises cities with high levels of resources and social capital, as well as government response. The consistency scores of the five configurations are 0.833491, 0.83596, 0.861196, 0.869426, and 0.83779 respectively.

More specifically, the data reveal that robustness and resources and social capital are essential to disaster recovery in the context of COVID-19. Pattern 1 shows that a city with high robustness could recovery well, even though it may lack sufficient preparedness. In other words, robustness represents resilience to some extent, which makes a city recovery faster and better from a disaster. Compared with other factors, a city’s robustness is an intrinsic characteristic and, hence, not easy to change in the short term.

Another important factor of recovery is resource and social capital, as a lot of social resources need to be consumed at all stages of the pandemic. During a concentrated outbreak of cases, cities need to concentrate medical resources to treat patients and provide residents with sufficient supplies to prevent more infections. After the number of infected people has dropped to a stable level, whether the government has sufficient funds for individuals and enterprises to achieve economic recovery is also a big test. In the future, in order to prevent social order from being overwhelmed by the pandemic again, the governments need to invest funds to support scientific research institutions and pharmaceutical manufacturers to promote vaccines and vaccination. It is shown that cities in pattern 3 and pattern 5 are well-developed in the world or at least in their country.

In addition, we found that low preparedness combines with other contributing factors lead to good recovery in both pattern 1 and 2, while high preparedness does not appear in any of the patterns. Specifically, most case cities in pattern 1 and pattern 2 are from developing countries with a relatively low level of preparedness. The case cities in pattern 3.4.5 are mostly from developed countries, where social capital and other contributing factors have a greater impact on recovery than preparedness. This conclusion suggests that low preparedness is a relatively common phenomenon for cities in developing countries and requires government attention. However, for cities in developed countries, a higher level of preparedness does not contribute resilience as much to other factors such as resources and social capital and government response. Future efforts to improve resilience should focus on raising the level of these conditions

The solution consistency is 0.934025, which means that 93% of cases that meet these five patterns could explain the recovery situation of cities. The solution coverage is 0.786694, which means that the five patterns can explain 86% of the cases. The solution consistency and solution coverage both are higher than the critical value, indicating that the empirical analysis is effective. Perhaps most importantly, our research shows that disaster recovery outcomes as they pertain to COVID-19 do not depend on any single factor. While robustness is critical in influencing COVID-19 disaster recovery outcomes, these often operate in tandem with other resilience attributes or drivers, such as social capital, preparedness, or government response.