Faster, more flexible, more challenging: how COVID-19 has changed research at KOF

"Normally we produce our forecasts on a monthly and quarterly basis. Because of the pandemic – with its dynamic infection patterns and the drastic political interventions involved – we have now had to switch to a weekly frequency. The challenge here was that, in some cases, we could not simply resort to classic economic indicators, so we have had to become creative. In order to calculate private consumption, for example, we have included in our models the transaction volumes of cash withdrawals using debit cards, debit card payments in retail outlets, and credit card payments online and in retail outlets. We have mapped the political restrictions on freedom of movement by using a mobility indicator which includes the volume of private transport, the use of public transport, and flight movements. Back in early April of last year we developed a basic economic forecasting model tailored to the coronavirus pandemic, which we then refined and optimised over time. Today we are well practised in the preparation of economic forecasts in the age of coronavirus. For us economists the past year has been an extremely exciting time. Unfortunately, a lot of basic research has been left undone because we have concentrated mainly on forecasting. But the forecasting work itself has rarely been so challenging because the coronavirus crisis is not comparable with other crises like the financial crash or the Swiss franc shock. We have constantly had to work with different scenarios. We are basically used to that, as there are always political uncertainty factors such as Brexit, for example. But the uncertainty has probably never been as high historically as it was last year."

"I believe that we economists have learned more from the crisis year of 2020 than we have in a long time. The pandemic has forced us to research even faster and in more practical ways. Our forecasting models work relatively reliably in normal times. Based on decades of experience of industrial production, for example, we can draw fairly precise conclusions for the manufacturing sector.
At the beginning of the coronavirus crisis, however, we had to constantly readjust our forecasts because the news was changing almost every minute. Since classic indicators such as industrial production are often only reported with a month's delay, we had to switch to alternative data such as Google searches, credit card transactions and mobility data. While these do not always correlate perfectly with what we are really looking for, they are often a good approximation and, most importantly, immediately available. And speed was in greater demand than ever during the coronavirus year of 2020. The coronavirus crisis has produced many anomalies. We have had to rethink much of our economic experience. For example, it is not typical for consumer spending on things such as travel or restaurants – which is usually relatively stable – to suddenly drop to zero. And the hospitality sector is not normally so susceptible to economic crises. The coronavirus crisis will continue to keep us busy in 2021. For example, we will keep an eye on bankruptcies in Switzerland. Even though state aid has so far prevented a major wave of bankruptcies, we fear that these will increase in the spring."

"In response to the pandemic we were relatively quick to supplement our company surveys of more than 4,500 Swiss businesses by adding specific questions on coronavirus. Initially, when the extent of the crisis was not yet foreseeable, we were mainly interested in whether supply chains were still intact and whether there were any restrictions on staff deployment. Later we also asked about expectations around annual turnover and general uncertainty in order to make even more accurate economic forecasts. At first we were concerned that we would overtax these firms with our additional questions, because they already had enough other worries during the pandemic. But, despite the crisis, we had a high response rate to our questionnaires. It was particularly striking that the different sectors were affected to varying degrees. Normally an economic downturn is a broad-based macroeconomic phenomenon that drags down all sectors to a greater or lesser extent. This time there were clear winners, such as online retail, and clear losers, such as tourism. Some sectors, such as retail – unlike other industries – recovered very quickly from the first lockdown. Personally, as an economist who has been studying the business cycle for decades, I had actually believed that I had already witnessed the worst economic crises of my academic career after the financial and euro crises. But the coronavirus crisis has shown that things can get even worse."

"Government-imposed restrictions such as the closure of schools, businesses, restaurants, sports facilities and shops play an important role in any analysis of the coronavirus crisis. The so-called Oxford Stringency Index attempts to quantify the severity of such measures. Its values range from 0 (no restrictions) to 100 (total lockdown). However, this index is only available for the whole of Switzerland but not for the individual cantons. This is why we have developed the KOF Stringency Index – based on the Oxford Stringency Index – which makes comparisons between the individual cantons possible. We are constantly updating this index, which has proved to be a great challenge, especially in the run-up to Christmas, when new political restrictions were being imposed almost daily at the cantonal level. We are currently evaluating the effectiveness of the various cantonal coronavirus strategies in detail. Particularly for the fourth quarter of 2020, when the cantons were using differing measures to fight the pandemic, the KOF Stringency Index provides valuable insights that also feed into the work of the Swiss National COVID-19 Research Task Force."

"I joined KOF in the middle of the coronavirus crisis in April 2020. Unfortunately, because I was working from home, I was not able to say a proper goodbye to my former employers – the Swiss Federation of Trade Unions – nor to meet all of my new colleagues at KOF in person. Professionally, however, I didn't need a lengthy period of training as I was able to continue some of my old projects at KOF. For example, I had experimented with Google search queries even before the pandemic. The question was what economic trends could be derived from this data in real time. I joined forces with economist friends and new colleagues from KOF to participate in the Coronavirus Hackathon #VersusVirus and launch the ‘trendEcon’ project. We subsequently managed to secure funding to refine this methodology. trendEcon is basically a tool that is used to gauge economic activity in Switzerland based on Google search queries. While it is not a substitute for analysis of classic economic indicators such as the unemployment rate, it is a supplementary method that can highlight trends on a daily basis – and it was precisely this timeliness that was in demand when the coronavirus crisis emerged. One of the challenges in developing trendEcon is that you have to adopt the mindset of people and their search behaviour on Google, which goes far beyond simply measuring economic variables. Coronavirus has also had a tangible impact on my own field of research, namely economic inequality. The coronavirus crisis has significantly increased public and media interest in social inequality. Moreover, some previously invisible groups, such as the Sans Papiers, have suddenly become visible. They do not appear in any statistics because they do not have legal residency status. But because they are often precariously employed and are not entitled to any short-term financial support or unemployment benefits, the Sans Papiers have been one of the groups most seriously affected by the coronavirus crisis, which has reinforced existing social inequalities in many areas."

"The year 2020 will probably go down in the annals of economic and labour market research. Never before have we been forced to find answers to new labour market policy questions within such a short period of time. Instead of using monthly unemployment figures, for example, we have switched to daily updated data on the number of jobseekers, which is a good leading indicator of unemployment. Fortunately, we had already started collecting this data for another project two years ago. This gave us a head start in terms of experience and enabled us to distinguish seasonal fluctuations in jobseeker data from the underlying economic trends. Another positive factor is that we have not yet seen a major wave of layoffs in Switzerland, which is mainly due to the use of short-time working. Whereas this form of work was still controversial among academics before the coronavirus crisis, there is now widespread agreement that this instrument effectively prevents redundancies and is suitable for mitigating the effects of economic crises. This is demonstrated by a comparison with the United States, where unemployment rose extremely sharply as a result of COVID-19. At the beginning of the coronavirus pandemic there were still instinctive hopes among liberals that the situation would be rectified by self-correcting market forces. The crisis was seen as helpful in some circles because it would force structurally weak firms out of the market in a form of creative destruction. But this liberal paradigm does not apply in the current crisis because, in order to contain the virus, firms were prevented from operating. Consequently, even highly productive and profitable firms were facing bankruptcy through no fault of their own. In order to safeguard these healthy market structures it was therefore right to act on a large scale and to respond to the COVID-related closures by making loans, short-time working and economic stimulus packages available, otherwise the crisis would only have got worse."

"The wheels of academia often turn quite slowly. For example, it can sometimes take years for a research paper in the field of economics to be published. We economists have been forced to do our research faster and in an even more applied way during the coronavirus crisis. This has been and still is especially true of my work as part of the COVID-19 Research Task Force, where policymakers have to make decisions quickly and rely on scientific expertise to do so. Since we have been and, in part, still are in a situation of uncertainty, research has been in great demand in policy advice and in the media. Accordingly, scientists have had to work under considerable pressure. For example, a vaccine has been developed in record time. In economics, too, we have had to swiftly increase our pace. We have only been able to draw on our previous experience from the financial crisis to a limited extent, because any comparison of the two crises fails on several levels. Their causes are completely different and the coronavirus pandemic has hit the general public much harder – and not only economically. However, the current crisis also holds the potential for a quicker return to normal. Once we have the pandemic under control in terms of healthcare, the economy will be able to recover faster than it did in 2008/2009. Any academic assessment of the coronavirus crisis, on the other hand, will continue for some time. Just as the Great Depression of the 1930s is still being researched today, the pandemic will keep us academics busy for years and decades to come."

"At KOF we usually collect data for research purposes ourselves through business surveys or draw on data sets from the Swiss National Bank, the Federal Statistical Office or the State Secretariat for Economic Affairs. The fast-moving nature of the coronavirus crisis has meant that we have had to switch from these monthly or quarterly data to new types of data sets from completely different sources. For example, we have used data from a parking app and flight data from Zurich Airport to measure Swiss citizens’ changing mobility behaviour. In addition, we teamed up with the market research company Intervista to analyse tracking data from over 2,500 Swiss nationals who voluntarily participated in the study. The KOF Mobility Indicator was then created from these mobility data. We then made all of these data sets, which were originally only intended for internal research, available to the general public in an open-source project on our so-called High Frequency Dashboard, where we have compiled 18 data sets on a website, some of which are updated several times a day. The ideas of almost 50 researchers were incorporated into the selection of these data, which are entered automatically and can be read out by the users electronically. We computer scientists call this machine-to-machine communication. Even though the platform still has the potential for technical improvements, it was important for us to develop this interim solution quickly. In computer science it's all about agility. You don't have a year to develop the perfect car like in Formula 1."