Exogenous shocks and citizens’ satisfaction with governmental policies: can empirical evidence from the 2008 financial crisis help us understand better the effects of the COVID-19 pandemic?

The idea for this study matured quickly. Since the beginning of March 2020, because of the coronavirus pandemic, the university where I work, shut down completely. The same happened with schools, restaurants, churches, bars, and gyms. All gatherings with more than 10 people were forbidden. In my case, all university personnel, whether with strictly academic duties, administrative or both, were forced to work from home using all the available web resources. Literally, from one day to the next many things that were taken for granted changed. Our lives were affected, directly and indirectly, in many ways.

How did all of this begin? At the end of 2019 news agencies around the world started reporting that a new flu-type virus called COVID-19 appeared to be infecting residents at the city of Wuhan, capital of the Hubei province in China at an ever alarming rate. Although the Chinese government attempted to downplay the news, the thousands of infected cases and the death toll soon made this impossible. New infections and casualties were reported daily. In order to contain the epidemic, at the end of January 2020, the Chinese government finally took swift action and imposed a lockdown on more than 50 million people living in that city, as well as in other provinces around the country.

By that time however, the epidemic had spread in many other countries as well. Soon from Italy came news of acute health conditions, of thousands of newly infected and deaths. Iran and Spain followed. Gradually other countries, thought to be immune, reported infections and deaths as well. This was the wake-up call for many governments to start informing their citizens of the perils of this epidemic and what needed to be done in terms of hygiene and social distancing. The United States was one of the last countries to act on this situation.

The World Health Organization (WHO) albeit hesitant initially, finally classified the phenomenon a pandemic and announced harsh warnings on what would happen if measures were not implemented by all. Because of the abnormally large amount of infected individuals, the biggest fear was that health services would not cope; there would not be enough specialized equipment in intensive care units to administer adequate treatment to those in most need.

Thus, most countries imposed restrictive rules on movements of individuals and ordered lockdowns of whole sectors in the economy. Sporting events, championship series and concerts were cancelled or postponed until further notice. Naturally, the immediate effects were felt in the tourism, transportation, restaurant and entertainment sectors, and where big individual financial commitments are made such as the real estate and the car industries. Economic activity came to an abrupt stop because everything closed within a few days. That in turn caused thousands of layoffs and an upsurge in unemployment claims. Governments and central banks hurried to support the collapsing system by promising subsidies, loans with better terms and occasionally free money to those immediately affected firms and individuals alike. Work from home suddenly became the norm. Schools and universities were closed. Despite initial difficulties, teaching was soon conducted via the web. To the apprehension of many parents, the responsibility for supervising their children’s school performance fell on them.

The financial markets which had already enjoyed one of the longest bull-run in history started feeling the effects of the uncertainty caused by the havoc in everyday life. At the beginning 2020, three of the best-known indexes, the Dow Jones Industrial Average, the SandP 500 and the Nasdaq recorded all-time highs. During the second week of March all collapsed with sharp declines of more than 20%.¹ Although all three have since rebounded, the extreme volatility has generated billions of dollars in losses to institutional as well as private investors worldwide. Oil prices fell sharply; at one point prices for the West Texas Intermediate were quoted with a minus sign.².

Governments that have managed to curtail the spread of the virus better seem to enjoy popularity from their constituents whereas other governments are blamed for not taking harsher measures early enough. People started perceiving themselves as “experts” with the R₀ indicator and the pandemic curve. They have familiarized themselves at what point then curve is estimated to reach its apex and start diminishing. The dilemma of social distancing and personal hygiene versus herd immunity became the forefront much debate. Many countries grappled with the questions of how long whole areas or even whole countries can remain under lockdown without generating irrevocable damage to the economy.

How much liquidity is needed to be injected into the economy so that people and enterprises survive the first shock? For how long can this support continue? Should central banks reduce the reserve requirement ratios for commercial banks, should they reduce discount rates, should they buy bonds held by commercial banks or should they even print new money³? Must countries with exceedingly high debt to GDP ratios nonetheless attempt to borrow from the financial markets to revive their economies?⁴

The COVID-19 pandemic is still an ongoing event that no one knows how it will actually develop in the future. The second wave observed during the autumn of 2020 struck with greater intensity than many had estimated. Some acknowledge that a third wave might be on its way in the spring of 2021. Regardless of the virus’s potency, however, its repercussions on billions of people will be felt for years to come. The sudden and extreme measures imposed by governments to fight the pandemic, combined with the fear of contracting the lethal virus itself, have generated considerable anxiety and mental illness amongst the general population (Fofana et al. 2020). Vaccines to immunize against the virus have been produced and have recently received licensing. Nonetheless, even if these vaccinations are implemented in sufficient numbers for the general population, it will take a considerable amount of time for socioeconomic conditions to return to pre-virus period levels. It is thus interesting from a sociological perspective to investigate how individuals reacted in the short- and long-term, during another calamity of the recent past.

The financial crisis of 2008 was not health-related. It was the subprime problem in the US that burst, bankrupted Lehman Brothers and created a tsunami of financial collapses initially and fiscal problems in many countries later (Ervasti et al., 2019, p. 1210). The two crises are indeed not completely comparable. A significant difference is the mindset of the people affected by both. In 2008 many were suddenly faced with a financial collapse and the uncertainty of the potential unemployment ahead. With the COVID-19 pandemic the negative financial repercussions are only one aspect. The restrictions on movement due to lockdowns and, as discussed above, the fear of contracting the lethal virus have created even harsher living conditions for millions. And yet, the two crises are similar at a more macro level. They were both sudden crises, were generated by exogenous events and they effected directly or indirectly all countries around the globe.⁵

In examining the 2008 crisis we have the opportunity to investigate how different countries coped with the sudden exogenous shock, both short and long term. In particular, we can examine the opinions of citizens about their governments and how they evaluate the policies on the economy, education and health services. This might provide a gauge in the future for comparing it to the current COVID-19 crisis. The European Social Survey (ESS) has the suitable data for this. Respondents are asked how satisfied they are with their government, the economy, education and the health services. All these four indicators provide a reasonable assessment of how citizens evaluate their government. Although the majority of the literature is focusing on the level of trust citizens have in political and impartial institutions, examining the satisfaction level with respect to specific policies is a more direct and practical approach in measuring the success or failure of government activities concretely.

My main goal is to measure dynamically the short-as well as the long-term effects of the crisis. For this I use data from 14 countries and responses collected in all 9 rounds of the ESS.⁶ Of the 38 countries that at some point in time were surveyed, only those 14 participated in every round (2002-2018).

Policy impacts are best measured by conducting so called social experiments. The focus is on the population of interest or a sample of its units, which is randomly divided into 2 groups: one that is exposed to the policy (the treatment) of interest and one that is not. The latter group is often used to depict the counterfactual situation—what would have happened to the policy indicator of interest had the experimental group not been exposed to the treatment. The random assignment to the two groups ensures that whatever different characteristics exist between them cancel out and the differences that remain on the indicator is the impact of the policy intervention.

In standard notation,

The effect of the intervention on the units of interest is always based on a relative comparison. It is the average difference of the impact on the units given (“|”) they were treated, less the impact on the units given they were not treated, or  As discussed, through randomization, we attempt to make the two groups as identical as possible, because we cannot observe simultaneously the same units under 2 different treatment regimes. For a non-experimental ex-post evaluation- as it is the case here - the best case scenario is having 2 populations, with measurements on the indicator of interest; one that is exposed to the treatment and the other that is not exposed. Preferably the observations cover the periods before and after the intervention, for some time. However since the classification is not done at random, it becomes problematic because we do not have two groups of data that differ by just that exposure to the intervention, on average. This, in turn, implies that their measured differences in the indicator might not be solely due to this intervention (Holland 1986; Imbens and Wooldridge 2009).

An even weaker scenario is where the intervention covers all the available population. As I show later, this compels us to approximate the counterfactual population using a very weak assumption on its behavior; that it will continue to behave the same post-treatment, as during pre-treatment. This in fact is the main methodological constraint with the current data and the research questions. On the other hand, measurements on the indicators of interest are present, both pre-and post-intervention. Hence, we can perhaps identify trends in growth, positive or negative, before and after 2008, the seminal year during which the crisis erupted.

In brief, we may think of the financial crisis as the policy treatment. By definition the crisis is an exogenous event that produces a global shock, thus covers all the population of the countries examined. The objective is to establish whether the crisis has had an impact on the levels of satisfaction of the people surveyed. In my analysis I use an Ordinary Least Squares (OLS) fixed effects⁷ regression specification called Interrupted Time Series.

In this study I examined the individual level of satisfaction in 14 European countries, for a 16-year period, from 2002 to 2018. The data analyzed was gathered from the European Social Survey. The indicators of satisfaction were linked to the performance of government overall and in relation to three specific policy areas; the economy, health services and education. During the period under scrutiny the exogenous shock generated by the global financial crisis of 2008 compelled many governments to implement different fiscal, monetary and reform policies to combat the severe recession that followed. With the available time series data, I measured how the crisis was associated with these indicators of satisfaction, both short-and long-term.

The motivation for the analysis is twofold. First it is due to the COVID-19 global crisis which resembles somewhat the events of 2008, but on a much wider and severer scale. Governments are now under immense pressure to solve several problems, such as economic, health and education-related without delays. The losses in human life have easily surpassed those in many battle fields of the twentieth century and that, in a much shorter time span. Businesses- especially small and medium sized (SMEs) - have closed or are in the brink of closing, while unemployment has risen sharply. Schools and universities have had to lock their facilities while distance teaching has been enforced on a wide scale.

Governments and public sector officials are constantly being evaluated on how they are handling the unprecedented calamity generated by this exogenous shock. Thus, examining not only the overall performance of governments during the 2008 crisis but also how they fared particularly in these three vital policy areas - the economy, health services and education- may provide us with a baseline to match and compare with the respective levels generated by the current COVID-19 crisis.

Second, at a more theoretical level, policies influence people’s wellbeing and behavior. Economic policies in particular play a vital role, especially in the light of exogenous events (Gahner-Larsen et al. 2019, p. 234). Governments need feedback mechanisms (e.g. through satisfaction surveys) from citizens on their merits and faults. This in turn influences future policy planning and implementation. Hence and irrespective of the COVID-19 or the 2008 crisis, examining how governments perform is of academic interest in itself.

On aggregate, the results indicate that during the period immediately after the crisis, from 2008 up until 2010, individual satisfaction with governments in general and the economy in particular, dropped considerably; for health and education related policies satisfaction grew. After the crisis, the slopes of all indicators came out statistically significant with a positive sign. Comparing the pre-and post-crisis slopes, satisfaction with the government in general and with the economy in particular grew faster after the crisis. The opposite was the case for health services and education policies. Their slopes grew as well, but at a pace slower than before the crisis. Applying the same analysis for each country separately, the results were mixed. There was no indication of any country with consistently low or high coefficient values for any of the four satisfaction indicators. In the four pairs of Figures for each type of satisfaction (Figs. 1 and 5, 2 and 6, 3 and 7, 4 and 8) we see interesting behavior patterns within each country. In the majority of cases (20 + 23 = 43 out of 56) if the short-term effect (STE) is negative (N), the long-term effect (LTE) is positive (P) and vice-versa. Table 7 depicts these effects for each country and each type of measured satisfaction. Overall, short- and long-term effects of the 2008 crisis seem to have balanced out. The switch from short-term negative to long-term positive satisfaction could be interpreted as reflecting the successful efforts of governments to correct the immediate dissatisfaction felt by many due to the 2008 crisis. In contrast, the long-term reduction in satisfaction compared to the short-term positivity is not as easily explained. It could be that some individuals seeing the problems other countries faced applauded their own government’s short-term performance in handling the crisis. Long-term however, they became more critical, both overall and regarding the three individual policies under scrutiny.

There are some caveats in the study. They stem mainly from the limitation of the data utilized and the implemented analytical method. First, external validity is not achieved. We cannot generalize the findings because of the low number of countries in the data. Nonetheless, by choosing only those countries that participated in all the survey rounds, I created a longer time series cross-sectional data set, and thus was able to run the ITS regression models. Furthermore, the ITS design requires at least 8 observational instances for the two periods, before and after the exogenous intervention. Since in my dataset I had only 9 in total (4 + 5), I bootstrapped all the models to ensure that the confidence intervals of the variable coefficients were of the correct size. Consequently this made the statistical significance of the coefficients more robust. Lastly, there is the school of thought which argues that, regardless of survey results, people respond to their evaluation of governments based not on facts, but on pre-existing beliefs (Hvidman 2019, p. 265). I did not test this assertion. This may indeed be a topic for future research wherein one combines evaluative responses on specific government policies with general perceptions on the functions and ethics of public sector officials and politicians.

At the beginning of August 2020 there were almost 20 million people in 213 countries that were infected with the virus; approximately 725 thousand had perished. In less than four months later, by the end of November 2020 the respective number of infections globally had increased threefold to 60 million, whereas the number of deaths had almost doubled, to a shocking 1.4 million. In the 14 countries I examine in this study, there were more than 9.3 million people infected and 225 thousand dead. The lockdown is going to shrink global economic growth dramatically, with the GDP in many countries estimated to drop by an average of 6% to 8% during 2020 (IMF, 2020, p.v). In its latest report, the European Commission (2020) forecasts that the European economy in particular will contract by an average of 8.75% in 2020, before recovering at an annual growth rate of 6% in 2021.¹⁷

In Table 8 the mortality rate per million inhabitants ranges between 58 in Norway to 1373 in Belgium. The same variability is observed in the GDP growth. During the summer of 2020, it ranged between -4.6% in Poland and -10.9% in Spain. In its forecasts for 2021, the European Commission predicts that all the economies of the 14 countries will recover, at different paces. The net GDP loss during the next 2 years will be-3.8% for Spain and Portugal, but only-0.3% for Poland and-0.5% for Switzerland.

Based on these indicators we can safely assume that that peoples’ satisfaction levels will vary from country to country, both during and after the coronavirus pandemic. In countries such as Finland and Norway, where the death toll per million is relatively low, satisfaction with health services could end up being much higher compared for example to Belgium, Spain, the UK or France. Further, satisfaction with the economy will most probably be inversely correlated to the magnitude of the net GDP loss during 2020 and 2021. This is because reduction in GDP usually means revenue losses, bankruptcies, higher unemployment and fewer investments. In addition, both the COVID-19 number of victims and the GDP figures may also affect satisfaction with the government, in general. Finally, when it comes to the satisfaction with education, it is difficult to evaluate in advance how it will be affected by the COVID-19 pandemic. It  will perhaps depend on the operational instructions given by each country’s ministry of education, on the technological infrastructure already in place for distance learning in each country, on whether employers encourage work from home or other arrangements.

The European Social Survey announced that during the fall of 2020 it will begin conducting its 10^th round of surveys with at least 26 countries participating. Although we cannot have a complete time series data for which we could examine past trends in their behavior, a simple pre-and post-crisis data set with responses from the 2018 and the 2020 surveys will be available from most of the participating countries. This will provide us with sufficient information to measure the short term effects of the pandemic on the topics investigated in this study so that a comparison of results can be made.¹⁸